diff --color-moved-ws: optimize allow-indentation-change
[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];
7a4252c4 834 int al = cur->es->len, bl = match->es->len, cl = l->len;
ca1f4ae4
SB
835 const char *a = cur->es->line,
836 *b = match->es->line,
837 *c = l->line;
7a4252c4 838 const char *orig_a = a;
ca1f4ae4
SB
839 int wslen;
840
841 /*
7a4252c4
PW
842 * We need to check if 'cur' is equal to 'match'. As those
843 * are from the same (+/-) side, we do not need to adjust for
844 * indent changes. However these were found using fuzzy
845 * matching so we do have to check if they are equal. Here we
846 * just check the lengths. We delay calling memcmp() to check
847 * the contents until later as if the length comparison for a
848 * and c fails we can avoid the call all together.
ca1f4ae4 849 */
7a4252c4 850 if (al != bl)
ca1f4ae4
SB
851 return 1;
852
74d156f4 853 if (!pmb->wsd.string)
ca1f4ae4 854 /*
74d156f4
PW
855 * The white space delta is not active? This can happen
856 * when we exit early in this function.
ca1f4ae4
SB
857 */
858 return 1;
859
860 /*
74d156f4 861 * The indent changes of the block are known and stored in
ca1f4ae4
SB
862 * pmb->wsd; however we need to check if the indent changes of the
863 * current line are still the same as before.
864 *
865 * To do so we need to compare 'l' to 'cur', adjusting the
866 * one of them for the white spaces, depending which was longer.
867 */
868
74d156f4
PW
869 wslen = strlen(pmb->wsd.string);
870 if (pmb->wsd.current_longer) {
ca1f4ae4
SB
871 c += wslen;
872 cl -= wslen;
873 } else {
874 a += wslen;
875 al -= wslen;
876 }
877
7a4252c4 878 if (al != cl || memcmp(orig_a, b, bl) || memcmp(a, c, al))
ca1f4ae4
SB
879 return 1;
880
881 return 0;
882}
2e2d5ac1 883
ee1df66f
SB
884static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
885 const void *entry,
886 const void *entry_or_key,
2e2d5ac1
SB
887 const void *keydata)
888{
ee1df66f
SB
889 const struct diff_options *diffopt = hashmap_cmp_fn_data;
890 const struct moved_entry *a = entry;
891 const struct moved_entry *b = entry_or_key;
b3095712
SB
892 unsigned flags = diffopt->color_moved_ws_handling
893 & XDF_WHITESPACE_FLAGS;
ee1df66f 894
ca1f4ae4
SB
895 if (diffopt->color_moved_ws_handling &
896 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
897 /*
898 * As there is not specific white space config given,
899 * we'd need to check for a new block, so ignore all
900 * white space. The setup of the white space
901 * configuration for the next block is done else where
902 */
903 flags |= XDF_IGNORE_WHITESPACE;
904
01be97c2
SB
905 return !xdiff_compare_lines(a->es->line, a->es->len,
906 b->es->line, b->es->len,
b3095712 907 flags);
2e2d5ac1
SB
908}
909
910static struct moved_entry *prepare_entry(struct diff_options *o,
911 int line_no)
912{
913 struct moved_entry *ret = xmalloc(sizeof(*ret));
914 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
b3095712 915 unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
2e2d5ac1 916
b3095712 917 ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
2e2d5ac1
SB
918 ret->es = l;
919 ret->next_line = NULL;
920
921 return ret;
922}
923
924static void add_lines_to_move_detection(struct diff_options *o,
925 struct hashmap *add_lines,
926 struct hashmap *del_lines)
927{
928 struct moved_entry *prev_line = NULL;
929
930 int n;
931 for (n = 0; n < o->emitted_symbols->nr; n++) {
932 struct hashmap *hm;
933 struct moved_entry *key;
934
935 switch (o->emitted_symbols->buf[n].s) {
936 case DIFF_SYMBOL_PLUS:
937 hm = add_lines;
938 break;
939 case DIFF_SYMBOL_MINUS:
940 hm = del_lines;
941 break;
942 default:
943 prev_line = NULL;
944 continue;
945 }
946
947 key = prepare_entry(o, n);
948 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
949 prev_line->next_line = key;
950
951 hashmap_add(hm, key);
952 prev_line = key;
953 }
954}
955
e2fe6abc
SB
956static void pmb_advance_or_null(struct diff_options *o,
957 struct moved_entry *match,
958 struct hashmap *hm,
74d156f4 959 struct moved_block *pmb,
e2fe6abc
SB
960 int pmb_nr)
961{
962 int i;
963 for (i = 0; i < pmb_nr; i++) {
74d156f4 964 struct moved_entry *prev = pmb[i].match;
e2fe6abc
SB
965 struct moved_entry *cur = (prev && prev->next_line) ?
966 prev->next_line : NULL;
967 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
74d156f4 968 pmb[i].match = cur;
e2fe6abc 969 } else {
74d156f4 970 pmb[i].match = NULL;
e2fe6abc
SB
971 }
972 }
973}
974
ca1f4ae4
SB
975static void pmb_advance_or_null_multi_match(struct diff_options *o,
976 struct moved_entry *match,
977 struct hashmap *hm,
74d156f4 978 struct moved_block *pmb,
ca1f4ae4
SB
979 int pmb_nr, int n)
980{
981 int i;
982 char *got_match = xcalloc(1, pmb_nr);
983
984 for (; match; match = hashmap_get_next(hm, match)) {
985 for (i = 0; i < pmb_nr; i++) {
74d156f4 986 struct moved_entry *prev = pmb[i].match;
ca1f4ae4
SB
987 struct moved_entry *cur = (prev && prev->next_line) ?
988 prev->next_line : NULL;
989 if (!cur)
990 continue;
74d156f4 991 if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
ca1f4ae4
SB
992 got_match[i] |= 1;
993 }
994 }
995
996 for (i = 0; i < pmb_nr; i++) {
997 if (got_match[i]) {
74d156f4
PW
998 /* Advance to the next line */
999 pmb[i].match = pmb[i].match->next_line;
fab01ec5 1000 } else {
74d156f4 1001 moved_block_clear(&pmb[i]);
fab01ec5 1002 }
ca1f4ae4 1003 }
9c1a6c2b
PW
1004
1005 free(got_match);
ca1f4ae4
SB
1006}
1007
74d156f4 1008static int shrink_potential_moved_blocks(struct moved_block *pmb,
2e2d5ac1
SB
1009 int pmb_nr)
1010{
1011 int lp, rp;
1012
1013 /* Shrink the set of potential block to the remaining running */
1014 for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
74d156f4 1015 while (lp < pmb_nr && pmb[lp].match)
2e2d5ac1
SB
1016 lp++;
1017 /* lp points at the first NULL now */
1018
74d156f4 1019 while (rp > -1 && !pmb[rp].match)
2e2d5ac1
SB
1020 rp--;
1021 /* rp points at the last non-NULL */
1022
1023 if (lp < pmb_nr && rp > -1 && lp < rp) {
1024 pmb[lp] = pmb[rp];
74d156f4
PW
1025 pmb[rp].match = NULL;
1026 pmb[rp].wsd.string = NULL;
2e2d5ac1
SB
1027 rp--;
1028 lp++;
1029 }
1030 }
1031
1032 /* Remember the number of running sets */
1033 return rp + 1;
1034}
1035
09153277
JT
1036/*
1037 * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1038 *
f0b8fb6e
JT
1039 * Otherwise, if the last block has fewer alphanumeric characters than
1040 * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
09153277
JT
1041 * that block.
1042 *
1043 * The last block consists of the (n - block_length)'th line up to but not
1044 * including the nth line.
f0b8fb6e 1045 *
b0a2ba47
PW
1046 * Returns 0 if the last block is empty or is unset by this function, non zero
1047 * otherwise.
1048 *
f0b8fb6e
JT
1049 * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1050 * Think of a way to unify them.
09153277 1051 */
b0a2ba47 1052static int adjust_last_block(struct diff_options *o, int n, int block_length)
09153277 1053{
f0b8fb6e
JT
1054 int i, alnum_count = 0;
1055 if (o->color_moved == COLOR_MOVED_PLAIN)
b0a2ba47 1056 return block_length;
f0b8fb6e
JT
1057 for (i = 1; i < block_length + 1; i++) {
1058 const char *c = o->emitted_symbols->buf[n - i].line;
1059 for (; *c; c++) {
1060 if (!isalnum(*c))
1061 continue;
1062 alnum_count++;
1063 if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
b0a2ba47 1064 return 1;
f0b8fb6e
JT
1065 }
1066 }
09153277
JT
1067 for (i = 1; i < block_length + 1; i++)
1068 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
b0a2ba47 1069 return 0;
09153277
JT
1070}
1071
2e2d5ac1
SB
1072/* Find blocks of moved code, delegate actual coloring decision to helper */
1073static void mark_color_as_moved(struct diff_options *o,
1074 struct hashmap *add_lines,
1075 struct hashmap *del_lines)
1076{
74d156f4 1077 struct moved_block *pmb = NULL; /* potentially moved blocks */
2e2d5ac1 1078 int pmb_nr = 0, pmb_alloc = 0;
b0a2ba47 1079 int n, flipped_block = 0, block_length = 0;
2e2d5ac1
SB
1080
1081
1082 for (n = 0; n < o->emitted_symbols->nr; n++) {
1083 struct hashmap *hm = NULL;
1084 struct moved_entry *key;
1085 struct moved_entry *match = NULL;
1086 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
b0a2ba47 1087 enum diff_symbol last_symbol = 0;
2e2d5ac1
SB
1088
1089 switch (l->s) {
1090 case DIFF_SYMBOL_PLUS:
1091 hm = del_lines;
1092 key = prepare_entry(o, n);
3783aad4 1093 match = hashmap_get(hm, key, NULL);
2e2d5ac1
SB
1094 free(key);
1095 break;
1096 case DIFF_SYMBOL_MINUS:
1097 hm = add_lines;
1098 key = prepare_entry(o, n);
3783aad4 1099 match = hashmap_get(hm, key, NULL);
2e2d5ac1
SB
1100 free(key);
1101 break;
1102 default:
b0a2ba47 1103 flipped_block = 0;
2e2d5ac1
SB
1104 }
1105
1106 if (!match) {
74d156f4
PW
1107 int i;
1108
09153277 1109 adjust_last_block(o, n, block_length);
74d156f4
PW
1110 for(i = 0; i < pmb_nr; i++)
1111 moved_block_clear(&pmb[i]);
2e2d5ac1
SB
1112 pmb_nr = 0;
1113 block_length = 0;
b0a2ba47
PW
1114 flipped_block = 0;
1115 last_symbol = l->s;
2e2d5ac1
SB
1116 continue;
1117 }
1118
2034b473 1119 if (o->color_moved == COLOR_MOVED_PLAIN) {
b0a2ba47 1120 last_symbol = l->s;
2034b473 1121 l->flags |= DIFF_SYMBOL_MOVED_LINE;
176841f0 1122 continue;
2034b473 1123 }
176841f0 1124
ca1f4ae4
SB
1125 if (o->color_moved_ws_handling &
1126 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1127 pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1128 else
1129 pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
2e2d5ac1
SB
1130
1131 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1132
1133 if (pmb_nr == 0) {
1134 /*
1135 * The current line is the start of a new block.
1136 * Setup the set of potential blocks.
1137 */
1138 for (; match; match = hashmap_get_next(hm, match)) {
1139 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
ca1f4ae4
SB
1140 if (o->color_moved_ws_handling &
1141 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
74d156f4
PW
1142 if (compute_ws_delta(l, match->es,
1143 &pmb[pmb_nr].wsd))
1144 pmb[pmb_nr++].match = match;
ca1f4ae4 1145 } else {
74d156f4
PW
1146 pmb[pmb_nr].wsd.string = NULL;
1147 pmb[pmb_nr++].match = match;
ca1f4ae4 1148 }
2e2d5ac1
SB
1149 }
1150
b0a2ba47
PW
1151 if (adjust_last_block(o, n, block_length) &&
1152 pmb_nr && last_symbol != l->s)
1153 flipped_block = (flipped_block + 1) % 2;
1154 else
1155 flipped_block = 0;
f0b8fb6e 1156
f0b8fb6e 1157 block_length = 0;
2e2d5ac1
SB
1158 }
1159
2034b473
PW
1160 if (pmb_nr) {
1161 block_length++;
2034b473
PW
1162 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1163 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1164 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1165 }
b0a2ba47 1166 last_symbol = l->s;
2e2d5ac1 1167 }
09153277 1168 adjust_last_block(o, n, block_length);
2e2d5ac1 1169
74d156f4
PW
1170 for(n = 0; n < pmb_nr; n++)
1171 moved_block_clear(&pmb[n]);
2e2d5ac1
SB
1172 free(pmb);
1173}
e6e045f8 1174
86b452e2
SB
1175#define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1176 (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1177static void dim_moved_lines(struct diff_options *o)
1178{
1179 int n;
1180 for (n = 0; n < o->emitted_symbols->nr; n++) {
1181 struct emitted_diff_symbol *prev = (n != 0) ?
1182 &o->emitted_symbols->buf[n - 1] : NULL;
1183 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1184 struct emitted_diff_symbol *next =
1185 (n < o->emitted_symbols->nr - 1) ?
1186 &o->emitted_symbols->buf[n + 1] : NULL;
1187
1188 /* Not a plus or minus line? */
1189 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1190 continue;
1191
1192 /* Not a moved line? */
1193 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1194 continue;
1195
1196 /*
1197 * If prev or next are not a plus or minus line,
1198 * pretend they don't exist
1199 */
1200 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1201 prev->s != DIFF_SYMBOL_MINUS)
1202 prev = NULL;
1203 if (next && next->s != DIFF_SYMBOL_PLUS &&
1204 next->s != DIFF_SYMBOL_MINUS)
1205 next = NULL;
1206
1207 /* Inside a block? */
1208 if ((prev &&
1209 (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1210 (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1211 (next &&
1212 (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1213 (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1214 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1215 continue;
1216 }
1217
1218 /* Check if we are at an interesting bound: */
1219 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1220 (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1221 (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1222 continue;
1223 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1224 (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1225 (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1226 continue;
1227
1228 /*
1229 * The boundary to prev and next are not interesting,
1230 * so this line is not interesting as a whole
1231 */
1232 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1233 }
1234}
1235
091f8e28 1236static void emit_line_ws_markup(struct diff_options *o,
9d1e16bf
SB
1237 const char *set_sign, const char *set,
1238 const char *reset,
bc9feb05 1239 int sign_index, const char *line, int len,
091f8e28 1240 unsigned ws_rule, int blank_at_eof)
6957eb9a 1241{
b8767f79 1242 const char *ws = NULL;
bc9feb05 1243 int sign = o->output_indicators[sign_index];
6957eb9a 1244
091f8e28
SB
1245 if (o->ws_error_highlight & ws_rule) {
1246 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
b8767f79
JH
1247 if (!*ws)
1248 ws = NULL;
1249 }
1250
f7c3b4e2 1251 if (!ws && !set_sign)
017ac45e 1252 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
f7c3b4e2 1253 else if (!ws) {
29ef759d 1254 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
f7c3b4e2 1255 } else if (blank_at_eof)
6957eb9a 1256 /* Blank line at EOF - paint '+' as well */
017ac45e 1257 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
6957eb9a
JH
1258 else {
1259 /* Emit just the prefix, then the rest. */
017ac45e 1260 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
f7c3b4e2 1261 sign, "", 0);
091f8e28
SB
1262 ws_check_emit(line, len, ws_rule,
1263 o->file, set, reset, ws);
6957eb9a
JH
1264 }
1265}
1266
e6e045f8
SB
1267static void emit_diff_symbol_from_struct(struct diff_options *o,
1268 struct emitted_diff_symbol *eds)
36a4cefd 1269{
b9cbfde6 1270 static const char *nneof = " No newline at end of file\n";
f7c3b4e2 1271 const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
0911c475 1272 struct strbuf sb = STRBUF_INIT;
e6e045f8
SB
1273
1274 enum diff_symbol s = eds->s;
1275 const char *line = eds->line;
1276 int len = eds->len;
1277 unsigned flags = eds->flags;
1278
36a4cefd 1279 switch (s) {
b9cbfde6
SB
1280 case DIFF_SYMBOL_NO_LF_EOF:
1281 context = diff_get_color_opt(o, DIFF_CONTEXT);
1282 reset = diff_get_color_opt(o, DIFF_RESET);
1283 putc('\n', o->file);
017ac45e 1284 emit_line_0(o, context, NULL, 0, reset, '\\',
b9cbfde6
SB
1285 nneof, strlen(nneof));
1286 break;
f3597138
SB
1287 case DIFF_SYMBOL_SUBMODULE_HEADER:
1288 case DIFF_SYMBOL_SUBMODULE_ERROR:
1289 case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
0911c475 1290 case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
146fdb0d 1291 case DIFF_SYMBOL_SUMMARY:
0911c475 1292 case DIFF_SYMBOL_STATS_LINE:
4eed0ebd 1293 case DIFF_SYMBOL_BINARY_DIFF_BODY:
68abc6f1
SB
1294 case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1295 emit_line(o, "", "", line, len);
1296 break;
f2bb1218 1297 case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
c64b420b
SB
1298 case DIFF_SYMBOL_CONTEXT_MARKER:
1299 context = diff_get_color_opt(o, DIFF_CONTEXT);
1300 reset = diff_get_color_opt(o, DIFF_RESET);
1301 emit_line(o, context, reset, line, len);
1302 break;
36a4cefd
SB
1303 case DIFF_SYMBOL_SEPARATOR:
1304 fprintf(o->file, "%s%c",
1305 diff_line_prefix(o),
1306 o->line_termination);
1307 break;
091f8e28
SB
1308 case DIFF_SYMBOL_CONTEXT:
1309 set = diff_get_color_opt(o, DIFF_CONTEXT);
1310 reset = diff_get_color_opt(o, DIFF_RESET);
f7c3b4e2
JS
1311 set_sign = NULL;
1312 if (o->flags.dual_color_diffed_diffs) {
1313 char c = !len ? 0 : line[0];
1314
1315 if (c == '+')
1316 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1317 else if (c == '@')
1318 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1319 else if (c == '-')
1320 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1321 }
7648b79e 1322 emit_line_ws_markup(o, set_sign, set, reset,
bc9feb05 1323 OUTPUT_INDICATOR_CONTEXT, line, len,
091f8e28
SB
1324 flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1325 break;
1326 case DIFF_SYMBOL_PLUS:
86b452e2
SB
1327 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1328 DIFF_SYMBOL_MOVED_LINE_ALT |
1329 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1330 case DIFF_SYMBOL_MOVED_LINE |
1331 DIFF_SYMBOL_MOVED_LINE_ALT |
1332 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1333 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1334 break;
1335 case DIFF_SYMBOL_MOVED_LINE |
1336 DIFF_SYMBOL_MOVED_LINE_ALT:
2e2d5ac1 1337 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
86b452e2
SB
1338 break;
1339 case DIFF_SYMBOL_MOVED_LINE |
1340 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1341 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1342 break;
1343 case DIFF_SYMBOL_MOVED_LINE:
2e2d5ac1 1344 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
86b452e2
SB
1345 break;
1346 default:
2e2d5ac1 1347 set = diff_get_color_opt(o, DIFF_FILE_NEW);
86b452e2 1348 }
091f8e28 1349 reset = diff_get_color_opt(o, DIFF_RESET);
f7c3b4e2
JS
1350 if (!o->flags.dual_color_diffed_diffs)
1351 set_sign = NULL;
1352 else {
1353 char c = !len ? 0 : line[0];
1354
1355 set_sign = set;
1356 if (c == '-')
a7be92ac 1357 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
f7c3b4e2
JS
1358 else if (c == '@')
1359 set = diff_get_color_opt(o, DIFF_FRAGINFO);
a7be92ac
JS
1360 else if (c == '+')
1361 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1362 else
1363 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
0b91faa0 1364 flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
f7c3b4e2 1365 }
7648b79e 1366 emit_line_ws_markup(o, set_sign, set, reset,
bc9feb05 1367 OUTPUT_INDICATOR_NEW, line, len,
091f8e28
SB
1368 flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1369 flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1370 break;
1371 case DIFF_SYMBOL_MINUS:
86b452e2
SB
1372 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1373 DIFF_SYMBOL_MOVED_LINE_ALT |
1374 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1375 case DIFF_SYMBOL_MOVED_LINE |
1376 DIFF_SYMBOL_MOVED_LINE_ALT |
1377 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1378 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1379 break;
1380 case DIFF_SYMBOL_MOVED_LINE |
1381 DIFF_SYMBOL_MOVED_LINE_ALT:
2e2d5ac1 1382 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
86b452e2
SB
1383 break;
1384 case DIFF_SYMBOL_MOVED_LINE |
1385 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1386 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1387 break;
1388 case DIFF_SYMBOL_MOVED_LINE:
2e2d5ac1 1389 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
86b452e2
SB
1390 break;
1391 default:
2e2d5ac1 1392 set = diff_get_color_opt(o, DIFF_FILE_OLD);
86b452e2 1393 }
091f8e28 1394 reset = diff_get_color_opt(o, DIFF_RESET);
f7c3b4e2
JS
1395 if (!o->flags.dual_color_diffed_diffs)
1396 set_sign = NULL;
1397 else {
1398 char c = !len ? 0 : line[0];
1399
1400 set_sign = set;
1401 if (c == '+')
a7be92ac 1402 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
f7c3b4e2
JS
1403 else if (c == '@')
1404 set = diff_get_color_opt(o, DIFF_FRAGINFO);
a7be92ac
JS
1405 else if (c == '-')
1406 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1407 else
1408 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
f7c3b4e2 1409 }
7648b79e 1410 emit_line_ws_markup(o, set_sign, set, reset,
bc9feb05 1411 OUTPUT_INDICATOR_OLD, line, len,
091f8e28
SB
1412 flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1413 break;
ff958679
SB
1414 case DIFF_SYMBOL_WORDS_PORCELAIN:
1415 context = diff_get_color_opt(o, DIFF_CONTEXT);
1416 reset = diff_get_color_opt(o, DIFF_RESET);
1417 emit_line(o, context, reset, line, len);
1418 fputs("~\n", o->file);
1419 break;
1420 case DIFF_SYMBOL_WORDS:
1421 context = diff_get_color_opt(o, DIFF_CONTEXT);
1422 reset = diff_get_color_opt(o, DIFF_RESET);
1423 /*
1424 * Skip the prefix character, if any. With
1425 * diff_suppress_blank_empty, there may be
1426 * none.
1427 */
1428 if (line[0] != '\n') {
1429 line++;
1430 len--;
1431 }
1432 emit_line(o, context, reset, line, len);
1433 break;
3ee8b7bf
SB
1434 case DIFF_SYMBOL_FILEPAIR_PLUS:
1435 meta = diff_get_color_opt(o, DIFF_METAINFO);
1436 reset = diff_get_color_opt(o, DIFF_RESET);
1437 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1438 line, reset,
1439 strchr(line, ' ') ? "\t" : "");
1440 break;
1441 case DIFF_SYMBOL_FILEPAIR_MINUS:
1442 meta = diff_get_color_opt(o, DIFF_METAINFO);
1443 reset = diff_get_color_opt(o, DIFF_RESET);
1444 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1445 line, reset,
1446 strchr(line, ' ') ? "\t" : "");
1447 break;
4acaaa7a 1448 case DIFF_SYMBOL_BINARY_FILES:
a29b0a13
SB
1449 case DIFF_SYMBOL_HEADER:
1450 fprintf(o->file, "%s", line);
1451 break;
4eed0ebd
SB
1452 case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1453 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1454 break;
1455 case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1456 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1457 break;
1458 case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1459 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1460 break;
1461 case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1462 fputs(diff_line_prefix(o), o->file);
1463 fputc('\n', o->file);
1464 break;
5af6ea95
SB
1465 case DIFF_SYMBOL_REWRITE_DIFF:
1466 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1467 reset = diff_get_color_opt(o, DIFF_RESET);
1468 emit_line(o, fraginfo, reset, line, len);
1469 break;
f3597138
SB
1470 case DIFF_SYMBOL_SUBMODULE_ADD:
1471 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1472 reset = diff_get_color_opt(o, DIFF_RESET);
1473 emit_line(o, set, reset, line, len);
1474 break;
1475 case DIFF_SYMBOL_SUBMODULE_DEL:
1476 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1477 reset = diff_get_color_opt(o, DIFF_RESET);
1478 emit_line(o, set, reset, line, len);
1479 break;
1480 case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1481 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1482 diff_line_prefix(o), line);
1483 break;
1484 case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1485 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1486 diff_line_prefix(o), line);
1487 break;
0911c475
SB
1488 case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1489 emit_line(o, "", "", " 0 files changed\n",
1490 strlen(" 0 files changed\n"));
1491 break;
1492 case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1493 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1494 break;
bd033291
SB
1495 case DIFF_SYMBOL_WORD_DIFF:
1496 fprintf(o->file, "%.*s", len, line);
1497 break;
30b7e1e7
SB
1498 case DIFF_SYMBOL_STAT_SEP:
1499 fputs(o->stat_sep, o->file);
1500 break;
36a4cefd 1501 default:
033abf97 1502 BUG("unknown diff symbol");
36a4cefd 1503 }
0911c475 1504 strbuf_release(&sb);
36a4cefd
SB
1505}
1506
e6e045f8
SB
1507static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1508 const char *line, int len, unsigned flags)
1509{
1510 struct emitted_diff_symbol e = {line, len, flags, s};
1511
1512 if (o->emitted_symbols)
1513 append_emitted_diff_symbol(o, &e);
1514 else
1515 emit_diff_symbol_from_struct(o, &e);
1516}
1517
f3597138
SB
1518void diff_emit_submodule_del(struct diff_options *o, const char *line)
1519{
1520 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1521}
1522
1523void diff_emit_submodule_add(struct diff_options *o, const char *line)
1524{
1525 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1526}
1527
1528void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1529{
1530 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1531 path, strlen(path), 0);
1532}
1533
1534void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1535{
1536 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1537 path, strlen(path), 0);
1538}
1539
1540void diff_emit_submodule_header(struct diff_options *o, const char *header)
1541{
1542 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1543 header, strlen(header), 0);
1544}
1545
1546void diff_emit_submodule_error(struct diff_options *o, const char *err)
1547{
1548 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1549}
1550
1551void diff_emit_submodule_pipethrough(struct diff_options *o,
1552 const char *line, int len)
1553{
1554 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1555}
1556
6957eb9a
JH
1557static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1558{
1559 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1560 ecbdata->blank_at_eof_in_preimage &&
1561 ecbdata->blank_at_eof_in_postimage &&
1562 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1563 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1564 return 0;
018cff70 1565 return ws_blank_line(line, len, ecbdata->ws_rule);
6957eb9a
JH
1566}
1567
b8767f79 1568static void emit_add_line(const char *reset,
0e383e18
JH
1569 struct emit_callback *ecbdata,
1570 const char *line, int len)
1571{
091f8e28
SB
1572 unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1573 if (new_blank_line_at_eof(ecbdata, line, len))
1574 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1575
1576 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
b8767f79 1577}
0e383e18 1578
b8767f79
JH
1579static void emit_del_line(const char *reset,
1580 struct emit_callback *ecbdata,
1581 const char *line, int len)
1582{
091f8e28
SB
1583 unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1584 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
0e383e18
JH
1585}
1586
1587static void emit_context_line(const char *reset,
1588 struct emit_callback *ecbdata,
1589 const char *line, int len)
1590{
091f8e28
SB
1591 unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1592 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
0e383e18
JH
1593}
1594
89cb73a1
BW
1595static void emit_hunk_header(struct emit_callback *ecbdata,
1596 const char *line, int len)
1597{
8dbf3eb6 1598 const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
89cb73a1
BW
1599 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1600 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1601 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
f7c3b4e2 1602 const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
89cb73a1
BW
1603 static const char atat[2] = { '@', '@' };
1604 const char *cp, *ep;
2efcc977
BY
1605 struct strbuf msgbuf = STRBUF_INIT;
1606 int org_len = len;
1607 int i = 1;
89cb73a1
BW
1608
1609 /*
1610 * As a hunk header must begin with "@@ -<old>, +<new> @@",
1611 * it always is at least 10 bytes long.
1612 */
1613 if (len < 10 ||
1614 memcmp(line, atat, 2) ||
1615 !(ep = memmem(line + 2, len - 2, atat, 2))) {
c64b420b 1616 emit_diff_symbol(ecbdata->opt,
091f8e28 1617 DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
89cb73a1
BW
1618 return;
1619 }
1620 ep += 2; /* skip over @@ */
1621
1622 /* The hunk header in fraginfo color */
f7c3b4e2
JS
1623 if (ecbdata->opt->flags.dual_color_diffed_diffs)
1624 strbuf_addstr(&msgbuf, reverse);
cedc61a9 1625 strbuf_addstr(&msgbuf, frag);
2efcc977 1626 strbuf_add(&msgbuf, line, ep - line);
cedc61a9 1627 strbuf_addstr(&msgbuf, reset);
2efcc977
BY
1628
1629 /*
1630 * trailing "\r\n"
1631 */
1632 for ( ; i < 3; i++)
1633 if (line[len - i] == '\r' || line[len - i] == '\n')
1634 len--;
89cb73a1
BW
1635
1636 /* blank before the func header */
1637 for (cp = ep; ep - line < len; ep++)
1638 if (*ep != ' ' && *ep != '\t')
1639 break;
2efcc977 1640 if (ep != cp) {
8dbf3eb6 1641 strbuf_addstr(&msgbuf, context);
2efcc977 1642 strbuf_add(&msgbuf, cp, ep - cp);
cedc61a9 1643 strbuf_addstr(&msgbuf, reset);
2efcc977
BY
1644 }
1645
1646 if (ep < line + len) {
cedc61a9 1647 strbuf_addstr(&msgbuf, func);
2efcc977 1648 strbuf_add(&msgbuf, ep, line + len - ep);
cedc61a9 1649 strbuf_addstr(&msgbuf, reset);
2efcc977 1650 }
89cb73a1 1651
2efcc977 1652 strbuf_add(&msgbuf, line + len, org_len - len);
dfb7728f 1653 strbuf_complete_line(&msgbuf);
68abc6f1 1654 emit_diff_symbol(ecbdata->opt,
091f8e28 1655 DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
2efcc977 1656 strbuf_release(&msgbuf);
89cb73a1
BW
1657}
1658
479b0ae8
JK
1659static struct diff_tempfile *claim_diff_tempfile(void) {
1660 int i;
1661 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1662 if (!diff_temp[i].name)
1663 return diff_temp + i;
033abf97 1664 BUG("diff is failing to clean up its tempfiles");
479b0ae8
JK
1665}
1666
479b0ae8
JK
1667static void remove_tempfile(void)
1668{
1669 int i;
a8344abe 1670 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
076aa2cb 1671 if (is_tempfile_active(diff_temp[i].tempfile))
284098f1 1672 delete_tempfile(&diff_temp[i].tempfile);
a8344abe
NR
1673 diff_temp[i].name = NULL;
1674 }
479b0ae8
JK
1675}
1676
5af6ea95 1677static void add_line_count(struct strbuf *out, int count)
6973dcae
JH
1678{
1679 switch (count) {
1680 case 0:
5af6ea95 1681 strbuf_addstr(out, "0,0");
6973dcae
JH
1682 break;
1683 case 1:
5af6ea95 1684 strbuf_addstr(out, "1");
6973dcae
JH
1685 break;
1686 default:
5af6ea95 1687 strbuf_addf(out, "1,%d", count);
6973dcae
JH
1688 break;
1689 }
1690}
1691
7f7ee2ff
JH
1692static void emit_rewrite_lines(struct emit_callback *ecb,
1693 int prefix, const char *data, int size)
6973dcae 1694{
7f7ee2ff 1695 const char *endp = NULL;
7f7ee2ff
JH
1696 const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1697
1698 while (0 < size) {
1699 int len;
1700
1701 endp = memchr(data, '\n', size);
1702 len = endp ? (endp - data + 1) : size;
1703 if (prefix != '+') {
1704 ecb->lno_in_preimage++;
0e383e18 1705 emit_del_line(reset, ecb, data, len);
7f7ee2ff
JH
1706 } else {
1707 ecb->lno_in_postimage++;
1708 emit_add_line(reset, ecb, data, len);
13e36ec5 1709 }
7f7ee2ff
JH
1710 size -= len;
1711 data += len;
1712 }
b9cbfde6 1713 if (!endp)
091f8e28 1714 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
6973dcae
JH
1715}
1716
1717static void emit_rewrite_diff(const char *name_a,
1718 const char *name_b,
1719 struct diff_filespec *one,
13e36ec5 1720 struct diff_filespec *two,
d9bae1a1
JK
1721 struct userdiff_driver *textconv_one,
1722 struct userdiff_driver *textconv_two,
eab9a40b 1723 struct diff_options *o)
6973dcae
JH
1724{
1725 int lc_a, lc_b;
d5625091 1726 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
a5a818ee 1727 const char *a_prefix, *b_prefix;
840383b2 1728 char *data_one, *data_two;
3aa1f7ca 1729 size_t size_one, size_two;
7f7ee2ff 1730 struct emit_callback ecbdata;
5af6ea95 1731 struct strbuf out = STRBUF_INIT;
a5a818ee 1732
0d1e0e78 1733 if (diff_mnemonic_prefix && o->flags.reverse_diff) {
a5a818ee
JH
1734 a_prefix = o->b_prefix;
1735 b_prefix = o->a_prefix;
1736 } else {
1737 a_prefix = o->a_prefix;
1738 b_prefix = o->b_prefix;
1739 }
1a9eb3b9 1740
8a13becc
JH
1741 name_a += (*name_a == '/');
1742 name_b += (*name_b == '/');
1a9eb3b9 1743
d5625091
JH
1744 strbuf_reset(&a_name);
1745 strbuf_reset(&b_name);
a5a818ee
JH
1746 quote_two_c_style(&a_name, a_prefix, name_a, 0);
1747 quote_two_c_style(&b_name, b_prefix, name_b, 0);
d5625091 1748
6afaf807
NTND
1749 size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1750 size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
3aa1f7ca 1751
d91ba8fa 1752 memset(&ecbdata, 0, sizeof(ecbdata));
daa0c3d9 1753 ecbdata.color_diff = want_color(o->use_color);
26d024ec 1754 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
a3c158d4 1755 ecbdata.opt = o;
d91ba8fa
JH
1756 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1757 mmfile_t mf1, mf2;
1758 mf1.ptr = (char *)data_one;
1759 mf2.ptr = (char *)data_two;
1760 mf1.size = size_one;
1761 mf2.size = size_two;
1762 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1763 }
1764 ecbdata.lno_in_preimage = 1;
1765 ecbdata.lno_in_postimage = 1;
1766
3aa1f7ca
JK
1767 lc_a = count_lines(data_one, size_one);
1768 lc_b = count_lines(data_two, size_two);
3ee8b7bf
SB
1769
1770 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1771 a_name.buf, a_name.len, 0);
1772 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1773 b_name.buf, b_name.len, 0);
1774
5af6ea95 1775 strbuf_addstr(&out, "@@ -");
467ddc14 1776 if (!o->irreversible_delete)
5af6ea95 1777 add_line_count(&out, lc_a);
467ddc14 1778 else
5af6ea95
SB
1779 strbuf_addstr(&out, "?,?");
1780 strbuf_addstr(&out, " +");
1781 add_line_count(&out, lc_b);
1782 strbuf_addstr(&out, " @@\n");
1783 emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1784 strbuf_release(&out);
1785
467ddc14 1786 if (lc_a && !o->irreversible_delete)
d91ba8fa 1787 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
6973dcae 1788 if (lc_b)
d91ba8fa 1789 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
b76c056b 1790 if (textconv_one)
aed6ca52 1791 free((char *)data_one);
b76c056b 1792 if (textconv_two)
aed6ca52 1793 free((char *)data_two);
6973dcae
JH
1794}
1795
f59a59e2
JS
1796struct diff_words_buffer {
1797 mmfile_t text;
071bcaab 1798 unsigned long alloc;
2e5d2003
JS
1799 struct diff_words_orig {
1800 const char *begin, *end;
1801 } *orig;
1802 int orig_nr, orig_alloc;
f59a59e2
JS
1803};
1804
1805static void diff_words_append(char *line, unsigned long len,
1806 struct diff_words_buffer *buffer)
1807{
23c1575f 1808 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
f59a59e2
JS
1809 line++;
1810 len--;
1811 memcpy(buffer->text.ptr + buffer->text.size, line, len);
1812 buffer->text.size += len;
2b6a5417 1813 buffer->text.ptr[buffer->text.size] = '\0';
f59a59e2
JS
1814}
1815
9cba13ca 1816struct diff_words_style_elem {
882749a0
TR
1817 const char *prefix;
1818 const char *suffix;
1819 const char *color; /* NULL; filled in by the setup code if
1820 * color is enabled */
1821};
1822
9cba13ca 1823struct diff_words_style {
882749a0 1824 enum diff_words_type type;
63a01c3f 1825 struct diff_words_style_elem new_word, old_word, ctx;
882749a0
TR
1826 const char *newline;
1827};
1828
c2e86add 1829static struct diff_words_style diff_words_styles[] = {
882749a0
TR
1830 { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1831 { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1832 { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1833};
1834
f59a59e2 1835struct diff_words_data {
f59a59e2 1836 struct diff_words_buffer minus, plus;
2e5d2003 1837 const char *current_plus;
4297c0ae
BY
1838 int last_minus;
1839 struct diff_options *opt;
2b6a5417 1840 regex_t *word_regex;
882749a0
TR
1841 enum diff_words_type type;
1842 struct diff_words_style *style;
f59a59e2
JS
1843};
1844
bd033291 1845static int fn_out_diff_words_write_helper(struct diff_options *o,
882749a0
TR
1846 struct diff_words_style_elem *st_el,
1847 const char *newline,
bd033291 1848 size_t count, const char *buf)
882749a0 1849{
4297c0ae 1850 int print = 0;
bd033291 1851 struct strbuf sb = STRBUF_INIT;
4297c0ae 1852
882749a0
TR
1853 while (count) {
1854 char *p = memchr(buf, '\n', count);
4297c0ae 1855 if (print)
bd033291
SB
1856 strbuf_addstr(&sb, diff_line_prefix(o));
1857
882749a0 1858 if (p != buf) {
bd033291
SB
1859 const char *reset = st_el->color && *st_el->color ?
1860 GIT_COLOR_RESET : NULL;
1861 if (st_el->color && *st_el->color)
1862 strbuf_addstr(&sb, st_el->color);
1863 strbuf_addstr(&sb, st_el->prefix);
1864 strbuf_add(&sb, buf, p ? p - buf : count);
1865 strbuf_addstr(&sb, st_el->suffix);
1866 if (reset)
1867 strbuf_addstr(&sb, reset);
882749a0
TR
1868 }
1869 if (!p)
bd033291
SB
1870 goto out;
1871
1872 strbuf_addstr(&sb, newline);
882749a0
TR
1873 count -= p + 1 - buf;
1874 buf = p + 1;
4297c0ae 1875 print = 1;
bd033291
SB
1876 if (count) {
1877 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1878 sb.buf, sb.len, 0);
1879 strbuf_reset(&sb);
1880 }
882749a0 1881 }
bd033291
SB
1882
1883out:
1884 if (sb.len)
1885 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1886 sb.buf, sb.len, 0);
1887 strbuf_release(&sb);
882749a0
TR
1888 return 0;
1889}
1890
4297c0ae
BY
1891/*
1892 * '--color-words' algorithm can be described as:
1893 *
5621760f 1894 * 1. collect the minus/plus lines of a diff hunk, divided into
4297c0ae
BY
1895 * minus-lines and plus-lines;
1896 *
1897 * 2. break both minus-lines and plus-lines into words and
1898 * place them into two mmfile_t with one word for each line;
1899 *
1900 * 3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1901 *
1902 * And for the common parts of the both file, we output the plus side text.
1903 * diff_words->current_plus is used to trace the current position of the plus file
1904 * which printed. diff_words->last_minus is used to trace the last minus word
1905 * printed.
1906 *
1907 * For '--graph' to work with '--color-words', we need to output the graph prefix
1908 * on each line of color words output. Generally, there are two conditions on
1909 * which we should output the prefix.
1910 *
1911 * 1. diff_words->last_minus == 0 &&
1912 * diff_words->current_plus == diff_words->plus.text.ptr
1913 *
1914 * that is: the plus text must start as a new line, and if there is no minus
1915 * word printed, a graph prefix must be printed.
1916 *
1917 * 2. diff_words->current_plus > diff_words->plus.text.ptr &&
1918 * *(diff_words->current_plus - 1) == '\n'
1919 *
1920 * that is: a graph prefix must be printed following a '\n'
1921 */
1922static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1923{
1924 if ((diff_words->last_minus == 0 &&
1925 diff_words->current_plus == diff_words->plus.text.ptr) ||
1926 (diff_words->current_plus > diff_words->plus.text.ptr &&
1927 *(diff_words->current_plus - 1) == '\n')) {
1928 return 1;
1929 } else {
1930 return 0;
1931 }
1932}
1933
7c61e25f
JK
1934static void fn_out_diff_words_aux(void *priv,
1935 long minus_first, long minus_len,
1936 long plus_first, long plus_len,
1937 const char *func, long funclen)
f59a59e2 1938{
f59a59e2 1939 struct diff_words_data *diff_words = priv;
882749a0 1940 struct diff_words_style *style = diff_words->style;
2e5d2003 1941 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
4297c0ae 1942 struct diff_options *opt = diff_words->opt;
30997bb8 1943 const char *line_prefix;
f59a59e2 1944
4297c0ae 1945 assert(opt);
30997bb8 1946 line_prefix = diff_line_prefix(opt);
4297c0ae 1947
2e5d2003
JS
1948 /* POSIX requires that first be decremented by one if len == 0... */
1949 if (minus_len) {
1950 minus_begin = diff_words->minus.orig[minus_first].begin;
1951 minus_end =
1952 diff_words->minus.orig[minus_first + minus_len - 1].end;
1953 } else
1954 minus_begin = minus_end =
1955 diff_words->minus.orig[minus_first].end;
1956
1957 if (plus_len) {
1958 plus_begin = diff_words->plus.orig[plus_first].begin;
1959 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
1960 } else
1961 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
1962
4297c0ae
BY
1963 if (color_words_output_graph_prefix(diff_words)) {
1964 fputs(line_prefix, diff_words->opt->file);
1965 }
1966 if (diff_words->current_plus != plus_begin) {
bd033291 1967 fn_out_diff_words_write_helper(diff_words->opt,
882749a0
TR
1968 &style->ctx, style->newline,
1969 plus_begin - diff_words->current_plus,
bd033291 1970 diff_words->current_plus);
4297c0ae
BY
1971 }
1972 if (minus_begin != minus_end) {
bd033291 1973 fn_out_diff_words_write_helper(diff_words->opt,
63a01c3f 1974 &style->old_word, style->newline,
bd033291 1975 minus_end - minus_begin, minus_begin);
4297c0ae
BY
1976 }
1977 if (plus_begin != plus_end) {
bd033291 1978 fn_out_diff_words_write_helper(diff_words->opt,
63a01c3f 1979 &style->new_word, style->newline,
bd033291 1980 plus_end - plus_begin, plus_begin);
4297c0ae 1981 }
2e5d2003
JS
1982
1983 diff_words->current_plus = plus_end;
4297c0ae 1984 diff_words->last_minus = minus_first;
f59a59e2
JS
1985}
1986
2b6a5417
JS
1987/* This function starts looking at *begin, and returns 0 iff a word was found. */
1988static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
1989 int *begin, int *end)
1990{
1991 if (word_regex && *begin < buffer->size) {
1992 regmatch_t match[1];
b7d36ffc
JS
1993 if (!regexec_buf(word_regex, buffer->ptr + *begin,
1994 buffer->size - *begin, 1, match, 0)) {
2b6a5417
JS
1995 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
1996 '\n', match[0].rm_eo - match[0].rm_so);
1997 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
1998 *begin += match[0].rm_so;
1999 return *begin >= *end;
2000 }
2001 return -1;
f59a59e2
JS
2002 }
2003
2b6a5417
JS
2004 /* find the next word */
2005 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2006 (*begin)++;
2007 if (*begin >= buffer->size)
2008 return -1;
f59a59e2 2009
2b6a5417
JS
2010 /* find the end of the word */
2011 *end = *begin + 1;
2012 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2013 (*end)++;
2014
2015 return 0;
f59a59e2
JS
2016}
2017
23c1575f 2018/*
2e5d2003
JS
2019 * This function splits the words in buffer->text, stores the list with
2020 * newline separator into out, and saves the offsets of the original words
2021 * in buffer->orig.
23c1575f 2022 */
2b6a5417
JS
2023static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2024 regex_t *word_regex)
f59a59e2 2025{
2e5d2003 2026 int i, j;
2b6a5417 2027 long alloc = 0;
f59a59e2 2028
2e5d2003 2029 out->size = 0;
2b6a5417 2030 out->ptr = NULL;
f59a59e2 2031
2e5d2003
JS
2032 /* fake an empty "0th" word */
2033 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2034 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2035 buffer->orig_nr = 1;
2036
2037 for (i = 0; i < buffer->text.size; i++) {
2b6a5417
JS
2038 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2039 return;
2e5d2003
JS
2040
2041 /* store original boundaries */
2042 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2043 buffer->orig_alloc);
2044 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2045 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2046 buffer->orig_nr++;
2047
2048 /* store one word */
2b6a5417 2049 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2e5d2003
JS
2050 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2051 out->ptr[out->size + j - i] = '\n';
2052 out->size += j - i + 1;
2053
2054 i = j - 1;
f59a59e2
JS
2055 }
2056}
2057
2058/* this executes the word diff on the accumulated buffers */
2059static void diff_words_show(struct diff_words_data *diff_words)
2060{
2061 xpparam_t xpp;
2062 xdemitconf_t xecfg;
f59a59e2 2063 mmfile_t minus, plus;
882749a0 2064 struct diff_words_style *style = diff_words->style;
f59a59e2 2065
4297c0ae 2066 struct diff_options *opt = diff_words->opt;
30997bb8 2067 const char *line_prefix;
4297c0ae
BY
2068
2069 assert(opt);
30997bb8 2070 line_prefix = diff_line_prefix(opt);
4297c0ae 2071
2e5d2003
JS
2072 /* special case: only removal */
2073 if (!diff_words->plus.text.size) {
bd033291
SB
2074 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2075 line_prefix, strlen(line_prefix), 0);
2076 fn_out_diff_words_write_helper(diff_words->opt,
63a01c3f 2077 &style->old_word, style->newline,
4297c0ae 2078 diff_words->minus.text.size,
bd033291 2079 diff_words->minus.text.ptr);
2e5d2003
JS
2080 diff_words->minus.text.size = 0;
2081 return;
2082 }
2083
2084 diff_words->current_plus = diff_words->plus.text.ptr;
4297c0ae 2085 diff_words->last_minus = 0;
f59a59e2 2086
9ccd0a88 2087 memset(&xpp, 0, sizeof(xpp));
30b25010 2088 memset(&xecfg, 0, sizeof(xecfg));
2b6a5417
JS
2089 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2090 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
582aa00b 2091 xpp.flags = 0;
2b6a5417 2092 /* as only the hunk header will be parsed, we need a 0-context */
2e5d2003 2093 xecfg.ctxlen = 0;
7c61e25f 2094 if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
9346d6d1 2095 diff_words, &xpp, &xecfg))
3efb9880 2096 die("unable to generate word diff");
f59a59e2
JS
2097 free(minus.ptr);
2098 free(plus.ptr);
2e5d2003 2099 if (diff_words->current_plus != diff_words->plus.text.ptr +
4297c0ae
BY
2100 diff_words->plus.text.size) {
2101 if (color_words_output_graph_prefix(diff_words))
bd033291
SB
2102 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2103 line_prefix, strlen(line_prefix), 0);
2104 fn_out_diff_words_write_helper(diff_words->opt,
882749a0 2105 &style->ctx, style->newline,
2e5d2003 2106 diff_words->plus.text.ptr + diff_words->plus.text.size
bd033291 2107 - diff_words->current_plus, diff_words->current_plus);
4297c0ae 2108 }
f59a59e2 2109 diff_words->minus.text.size = diff_words->plus.text.size = 0;
f59a59e2
JS
2110}
2111
76fd2828
JH
2112/* In "color-words" mode, show word-diff of words accumulated in the buffer */
2113static void diff_words_flush(struct emit_callback *ecbdata)
2114{
e6e045f8
SB
2115 struct diff_options *wo = ecbdata->diff_words->opt;
2116
76fd2828
JH
2117 if (ecbdata->diff_words->minus.text.size ||
2118 ecbdata->diff_words->plus.text.size)
2119 diff_words_show(ecbdata->diff_words);
e6e045f8
SB
2120
2121 if (wo->emitted_symbols) {
2122 struct diff_options *o = ecbdata->opt;
2123 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2124 int i;
2125
2126 /*
2127 * NEEDSWORK:
2128 * Instead of appending each, concat all words to a line?
2129 */
2130 for (i = 0; i < wol->nr; i++)
2131 append_emitted_diff_symbol(o, &wol->buf[i]);
2132
2133 for (i = 0; i < wol->nr; i++)
2134 free((void *)wol->buf[i].line);
2135
2136 wol->nr = 0;
2137 }
76fd2828
JH
2138}
2139
acd00ea0
NTND
2140static void diff_filespec_load_driver(struct diff_filespec *one,
2141 struct index_state *istate)
77d1a520
TR
2142{
2143 /* Use already-loaded driver */
2144 if (one->driver)
2145 return;
2146
2147 if (S_ISREG(one->mode))
acd00ea0 2148 one->driver = userdiff_find_by_path(istate, one->path);
77d1a520
TR
2149
2150 /* Fallback to default settings */
2151 if (!one->driver)
2152 one->driver = userdiff_find_by_name("default");
2153}
2154
acd00ea0
NTND
2155static const char *userdiff_word_regex(struct diff_filespec *one,
2156 struct index_state *istate)
77d1a520 2157{
acd00ea0 2158 diff_filespec_load_driver(one, istate);
77d1a520
TR
2159 return one->driver->word_regex;
2160}
2161
2162static void init_diff_words_data(struct emit_callback *ecbdata,
6440d341 2163 struct diff_options *orig_opts,
77d1a520
TR
2164 struct diff_filespec *one,
2165 struct diff_filespec *two)
2166{
2167 int i;
6440d341
TR
2168 struct diff_options *o = xmalloc(sizeof(struct diff_options));
2169 memcpy(o, orig_opts, sizeof(struct diff_options));
77d1a520
TR
2170
2171 ecbdata->diff_words =
2172 xcalloc(1, sizeof(struct diff_words_data));
2173 ecbdata->diff_words->type = o->word_diff;
2174 ecbdata->diff_words->opt = o;
e6e045f8
SB
2175
2176 if (orig_opts->emitted_symbols)
2177 o->emitted_symbols =
2178 xcalloc(1, sizeof(struct emitted_diff_symbols));
2179
77d1a520 2180 if (!o->word_regex)
acd00ea0 2181 o->word_regex = userdiff_word_regex(one, o->repo->index);
77d1a520 2182 if (!o->word_regex)
acd00ea0 2183 o->word_regex = userdiff_word_regex(two, o->repo->index);
77d1a520
TR
2184 if (!o->word_regex)
2185 o->word_regex = diff_word_regex_cfg;
2186 if (o->word_regex) {
2187 ecbdata->diff_words->word_regex = (regex_t *)
2188 xmalloc(sizeof(regex_t));
2189 if (regcomp(ecbdata->diff_words->word_regex,
2190 o->word_regex,
2191 REG_EXTENDED | REG_NEWLINE))
1a07e59c
NTND
2192 die("invalid regular expression: %s",
2193 o->word_regex);
77d1a520
TR
2194 }
2195 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2196 if (o->word_diff == diff_words_styles[i].type) {
2197 ecbdata->diff_words->style =
2198 &diff_words_styles[i];
2199 break;
2200 }
2201 }
2202 if (want_color(o->use_color)) {
2203 struct diff_words_style *st = ecbdata->diff_words->style;
63a01c3f
BW
2204 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2205 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
8dbf3eb6 2206 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
77d1a520
TR
2207 }
2208}
2209
f59a59e2
JS
2210static void free_diff_words_data(struct emit_callback *ecbdata)
2211{
2212 if (ecbdata->diff_words) {
76fd2828 2213 diff_words_flush(ecbdata);
e6e045f8 2214 free (ecbdata->diff_words->opt->emitted_symbols);
6440d341 2215 free (ecbdata->diff_words->opt);
8e0f7003 2216 free (ecbdata->diff_words->minus.text.ptr);
2e5d2003 2217 free (ecbdata->diff_words->minus.orig);
8e0f7003 2218 free (ecbdata->diff_words->plus.text.ptr);
2e5d2003 2219 free (ecbdata->diff_words->plus.orig);
ef5644ea
BC
2220 if (ecbdata->diff_words->word_regex) {
2221 regfree(ecbdata->diff_words->word_regex);
2222 free(ecbdata->diff_words->word_regex);
2223 }
6a83d902 2224 FREE_AND_NULL(ecbdata->diff_words);
f59a59e2
JS
2225 }
2226}
2227
ce436973 2228const char *diff_get_color(int diff_use_color, enum color_diff ix)
cd112cef 2229{
daa0c3d9 2230 if (want_color(diff_use_color))
50f575fc
LT
2231 return diff_colors[ix];
2232 return "";
cd112cef
JS
2233}
2234
f1922234
JK
2235const char *diff_line_prefix(struct diff_options *opt)
2236{
2237 struct strbuf *msgbuf;
2238 if (!opt->output_prefix)
2239 return "";
2240
2241 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2242 return msgbuf->buf;
2243}
2244
23707811
JH
2245static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2246{
2247 const char *cp;
2248 unsigned long allot;
2249 size_t l = len;
2250
23707811
JH
2251 cp = line;
2252 allot = l;
2253 while (0 < l) {
2254 (void) utf8_width(&cp, &l);
2255 if (!cp)
2256 break; /* truncated in the middle? */
2257 }
2258 return allot - l;
2259}
2260
d68fe26f 2261static void find_lno(const char *line, struct emit_callback *ecbdata)
690ed843 2262{
d68fe26f
JH
2263 const char *p;
2264 ecbdata->lno_in_preimage = 0;
2265 ecbdata->lno_in_postimage = 0;
2266 p = strchr(line, '-');
690ed843 2267 if (!p)
d68fe26f
JH
2268 return; /* cannot happen */
2269 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2270 p = strchr(p, '+');
2271 if (!p)
2272 return; /* cannot happen */
2273 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
690ed843
JH
2274}
2275
cd112cef 2276static void fn_out_consume(void *priv, char *line, unsigned long len)
6973dcae 2277{
6973dcae 2278 struct emit_callback *ecbdata = priv;
ce436973 2279 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
7be57610 2280 struct diff_options *o = ecbdata->opt;
6973dcae 2281
ba16233c
SB
2282 o->found_changes = 1;
2283
3e97c7c6 2284 if (ecbdata->header) {
a29b0a13
SB
2285 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2286 ecbdata->header->buf, ecbdata->header->len, 0);
3e97c7c6
GB
2287 strbuf_reset(ecbdata->header);
2288 ecbdata->header = NULL;
2289 }
34a5e1a2 2290
6973dcae 2291 if (ecbdata->label_path[0]) {
3ee8b7bf
SB
2292 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2293 ecbdata->label_path[0],
2294 strlen(ecbdata->label_path[0]), 0);
2295 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2296 ecbdata->label_path[1],
2297 strlen(ecbdata->label_path[1]), 0);
6973dcae
JH
2298 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2299 }
cd112cef 2300
a624eaa7
JM
2301 if (diff_suppress_blank_empty
2302 && len == 2 && line[0] == ' ' && line[1] == '\n') {
2303 line[0] = '\n';
2304 len = 1;
2305 }
2306
b8d9c1a6 2307 if (line[0] == '@') {
76fd2828
JH
2308 if (ecbdata->diff_words)
2309 diff_words_flush(ecbdata);
23707811 2310 len = sane_truncate_line(ecbdata, line, len);
d68fe26f 2311 find_lno(line, ecbdata);
89cb73a1 2312 emit_hunk_header(ecbdata, line, len);
448c3ef1 2313 return;
cd112cef 2314 }
448c3ef1 2315
448c3ef1 2316 if (ecbdata->diff_words) {
ff958679
SB
2317 enum diff_symbol s =
2318 ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2319 DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
448c3ef1
JH
2320 if (line[0] == '-') {
2321 diff_words_append(line, len,
2322 &ecbdata->diff_words->minus);
2323 return;
2324 } else if (line[0] == '+') {
2325 diff_words_append(line, len,
2326 &ecbdata->diff_words->plus);
2327 return;
59556548 2328 } else if (starts_with(line, "\\ ")) {
c7c2bc0a
TR
2329 /*
2330 * Eat the "no newline at eof" marker as if we
2331 * saw a "+" or "-" line with nothing on it,
2332 * and return without diff_words_flush() to
2333 * defer processing. If this is the end of
2334 * preimage, more "+" lines may come after it.
2335 */
2336 return;
448c3ef1 2337 }
76fd2828 2338 diff_words_flush(ecbdata);
ff958679 2339 emit_diff_symbol(o, s, line, len, 0);
448c3ef1
JH
2340 return;
2341 }
448c3ef1 2342
0e383e18
JH
2343 switch (line[0]) {
2344 case '+':
d68fe26f 2345 ecbdata->lno_in_postimage++;
018cff70 2346 emit_add_line(reset, ecbdata, line + 1, len - 1);
0e383e18
JH
2347 break;
2348 case '-':
2349 ecbdata->lno_in_preimage++;
2350 emit_del_line(reset, ecbdata, line + 1, len - 1);
2351 break;
2352 case ' ':
2353 ecbdata->lno_in_postimage++;
2354 ecbdata->lno_in_preimage++;
2355 emit_context_line(reset, ecbdata, line + 1, len - 1);
2356 break;
2357 default:
2358 /* incomplete line at the end */
2359 ecbdata->lno_in_preimage++;
f2bb1218
SB
2360 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2361 line, len, 0);
0e383e18 2362 break;
448c3ef1 2363 }
6973dcae
JH
2364}
2365
c905cbc4 2366static void pprint_rename(struct strbuf *name, const char *a, const char *b)
6973dcae 2367{
63a01c3f
BW
2368 const char *old_name = a;
2369 const char *new_name = b;
6973dcae 2370 int pfx_length, sfx_length;
dd281f09 2371 int pfx_adjust_for_slash;
6973dcae
JH
2372 int len_a = strlen(a);
2373 int len_b = strlen(b);
663af342 2374 int a_midlen, b_midlen;
e5bfbf9b
AJ
2375 int qlen_a = quote_c_style(a, NULL, NULL, 0);
2376 int qlen_b = quote_c_style(b, NULL, NULL, 0);
2377
2378 if (qlen_a || qlen_b) {
c905cbc4
NTND
2379 quote_c_style(a, name, NULL, 0);
2380 strbuf_addstr(name, " => ");
2381 quote_c_style(b, name, NULL, 0);
2382 return;
e5bfbf9b 2383 }
6973dcae
JH
2384
2385 /* Find common prefix */
2386 pfx_length = 0;
63a01c3f
BW
2387 while (*old_name && *new_name && *old_name == *new_name) {
2388 if (*old_name == '/')
2389 pfx_length = old_name - a + 1;
2390 old_name++;
2391 new_name++;
6973dcae
JH
2392 }
2393
2394 /* Find common suffix */
63a01c3f
BW
2395 old_name = a + len_a;
2396 new_name = b + len_b;
6973dcae 2397 sfx_length = 0;
d020e27f 2398 /*
dd281f09
TR
2399 * If there is a common prefix, it must end in a slash. In
2400 * that case we let this loop run 1 into the prefix to see the
2401 * same slash.
2402 *
2403 * If there is no common prefix, we cannot do this as it would
2404 * underrun the input strings.
d020e27f 2405 */
dd281f09 2406 pfx_adjust_for_slash = (pfx_length ? 1 : 0);
63a01c3f
BW
2407 while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2408 b + pfx_length - pfx_adjust_for_slash <= new_name &&
2409 *old_name == *new_name) {
2410 if (*old_name == '/')
2411 sfx_length = len_a - (old_name - a);
2412 old_name--;
2413 new_name--;
6973dcae
JH
2414 }
2415
2416 /*
2417 * pfx{mid-a => mid-b}sfx
2418 * {pfx-a => pfx-b}sfx
2419 * pfx{sfx-a => sfx-b}
2420 * name-a => name-b
2421 */
663af342
PH
2422 a_midlen = len_a - pfx_length - sfx_length;
2423 b_midlen = len_b - pfx_length - sfx_length;
2424 if (a_midlen < 0)
2425 a_midlen = 0;
2426 if (b_midlen < 0)
2427 b_midlen = 0;
2428
c905cbc4 2429 strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
6973dcae 2430 if (pfx_length + sfx_length) {
c905cbc4
NTND
2431 strbuf_add(name, a, pfx_length);
2432 strbuf_addch(name, '{');
6973dcae 2433 }
c905cbc4
NTND
2434 strbuf_add(name, a + pfx_length, a_midlen);
2435 strbuf_addstr(name, " => ");
2436 strbuf_add(name, b + pfx_length, b_midlen);
663af342 2437 if (pfx_length + sfx_length) {
c905cbc4
NTND
2438 strbuf_addch(name, '}');
2439 strbuf_add(name, a + len_a - sfx_length, sfx_length);
6973dcae 2440 }
6973dcae
JH
2441}
2442
2443struct diffstat_t {
6973dcae
JH
2444 int nr;
2445 int alloc;
2446 struct diffstat_file {
f604652e 2447 char *from_name;
6973dcae 2448 char *name;
f604652e 2449 char *print_name;
ddf88fa6 2450 const char *comments;
6973dcae
JH
2451 unsigned is_unmerged:1;
2452 unsigned is_binary:1;
2453 unsigned is_renamed:1;
74faaa16 2454 unsigned is_interesting:1;
0974c117 2455 uintmax_t added, deleted;
6973dcae
JH
2456 } **files;
2457};
2458
2459static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2460 const char *name_a,
2461 const char *name_b)
2462{
2463 struct diffstat_file *x;
1a4927c5 2464 x = xcalloc(1, sizeof(*x));
4c960a43 2465 ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
6973dcae
JH
2466 diffstat->files[diffstat->nr++] = x;
2467 if (name_b) {
f604652e
JH
2468 x->from_name = xstrdup(name_a);
2469 x->name = xstrdup(name_b);
6973dcae
JH
2470 x->is_renamed = 1;
2471 }
f604652e
JH
2472 else {
2473 x->from_name = NULL;
9befac47 2474 x->name = xstrdup(name_a);
f604652e 2475 }
6973dcae
JH
2476 return x;
2477}
2478
2479static void diffstat_consume(void *priv, char *line, unsigned long len)
2480{
2481 struct diffstat_t *diffstat = priv;
2482 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2483
2484 if (line[0] == '+')
2485 x->added++;
2486 else if (line[0] == '-')
2487 x->deleted++;
2488}
2489
698ce6f8 2490const char mime_boundary_leader[] = "------------";
6973dcae 2491
a2540023
JH
2492static int scale_linear(int it, int width, int max_change)
2493{
2eeeef24
JH
2494 if (!it)
2495 return 0;
a2540023 2496 /*
2eeeef24
JH
2497 * make sure that at least one '-' or '+' is printed if
2498 * there is any change to this path. The easiest way is to
2499 * scale linearly as if the alloted width is one column shorter
2500 * than it is, and then add 1 to the result.
a2540023 2501 */
2eeeef24 2502 return 1 + (it * (width - 1) / max_change);
a2540023
JH
2503}
2504
0911c475
SB
2505static void show_graph(struct strbuf *out, char ch, int cnt,
2506 const char *set, const char *reset)
a2540023
JH
2507{
2508 if (cnt <= 0)
2509 return;
0911c475
SB
2510 strbuf_addstr(out, set);
2511 strbuf_addchars(out, ch, cnt);
2512 strbuf_addstr(out, reset);
a2540023
JH
2513}
2514
f604652e
JH
2515static void fill_print_name(struct diffstat_file *file)
2516{
c905cbc4 2517 struct strbuf pname = STRBUF_INIT;
f604652e
JH
2518
2519 if (file->print_name)
2520 return;
2521
c905cbc4
NTND
2522 if (file->is_renamed)
2523 pprint_rename(&pname, file->from_name, file->name);
2524 else
2525 quote_c_style(file->name, &pname, NULL, 0);
2526
ddf88fa6
NTND
2527 if (file->comments)
2528 strbuf_addf(&pname, " (%s)", file->comments);
2529
c905cbc4 2530 file->print_name = strbuf_detach(&pname, NULL);
f604652e
JH
2531}
2532
0911c475
SB
2533static void print_stat_summary_inserts_deletes(struct diff_options *options,
2534 int files, int insertions, int deletions)
7f814632
NTND
2535{
2536 struct strbuf sb = STRBUF_INIT;
7f814632
NTND
2537
2538 if (!files) {
2539 assert(insertions == 0 && deletions == 0);
0911c475
SB
2540 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2541 NULL, 0, 0);
2542 return;
7f814632
NTND
2543 }
2544
2545 strbuf_addf(&sb,
218adaaa 2546 (files == 1) ? " %d file changed" : " %d files changed",
7f814632
NTND
2547 files);
2548
2549 /*
2550 * For binary diff, the caller may want to print "x files
2551 * changed" with insertions == 0 && deletions == 0.
2552 *
2553 * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2554 * is probably less confusing (i.e skip over "2 files changed
2555 * but nothing about added/removed lines? Is this a bug in Git?").
2556 */
2557 if (insertions || deletions == 0) {
7f814632 2558 strbuf_addf(&sb,
218adaaa 2559 (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
7f814632
NTND
2560 insertions);
2561 }
2562
2563 if (deletions || insertions == 0) {
7f814632 2564 strbuf_addf(&sb,
218adaaa 2565 (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
7f814632
NTND
2566 deletions);
2567 }
2568 strbuf_addch(&sb, '\n');
0911c475
SB
2569 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2570 sb.buf, sb.len, 0);
7f814632 2571 strbuf_release(&sb);
0911c475
SB
2572}
2573
2574void print_stat_summary(FILE *fp, int files,
2575 int insertions, int deletions)
2576{
2577 struct diff_options o;
2578 memset(&o, 0, sizeof(o));
2579 o.file = fp;
2580
2581 print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
7f814632
NTND
2582}
2583
4b25d091 2584static void show_stats(struct diffstat_t *data, struct diff_options *options)
6973dcae 2585{
eb3a9dd3 2586 int i, len, add, del, adds = 0, dels = 0;
0974c117 2587 uintmax_t max_change = 0, max_len = 0;
dc801e71
ZJS
2588 int total_files = data->nr, count;
2589 int width, name_width, graph_width, number_width = 0, bin_width = 0;
c0aa335c 2590 const char *reset, *add_c, *del_c;
e5f85df8 2591 int extra_shown = 0;
0911c475
SB
2592 const char *line_prefix = diff_line_prefix(options);
2593 struct strbuf out = STRBUF_INIT;
6973dcae
JH
2594
2595 if (data->nr == 0)
2596 return;
2597
808e1db2 2598 count = options->stat_count ? options->stat_count : data->nr;
a2540023 2599
8f67f8ae 2600 reset = diff_get_color_opt(options, DIFF_RESET);
8f67f8ae
PH
2601 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2602 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
785f7432 2603
1b058bc3
ZJS
2604 /*
2605 * Find the longest filename and max number of changes
2606 */
808e1db2 2607 for (i = 0; (i < count) && (i < data->nr); i++) {
6973dcae 2608 struct diffstat_file *file = data->files[i];
0974c117 2609 uintmax_t change = file->added + file->deleted;
af0ed819
JH
2610
2611 if (!file->is_interesting && (change == 0)) {
808e1db2 2612 count++; /* not shown == room for one more */
358e460e
MG
2613 continue;
2614 }
f604652e
JH
2615 fill_print_name(file);
2616 len = strlen(file->print_name);
6973dcae
JH
2617 if (max_len < len)
2618 max_len = len;
2619
dc801e71
ZJS
2620 if (file->is_unmerged) {
2621 /* "Unmerged" is 8 characters */
2622 bin_width = bin_width < 8 ? 8 : bin_width;
6973dcae 2623 continue;
dc801e71
ZJS
2624 }
2625 if (file->is_binary) {
2626 /* "Bin XXX -> YYY bytes" */
2627 int w = 14 + decimal_width(file->added)
2628 + decimal_width(file->deleted);
2629 bin_width = bin_width < w ? w : bin_width;
2630 /* Display change counts aligned with "Bin" */
2631 number_width = 3;
2632 continue;
2633 }
2634
a2540023
JH
2635 if (max_change < change)
2636 max_change = change;
6973dcae 2637 }
a20d3c0d 2638 count = i; /* where we can stop scanning in data->files[] */
6973dcae 2639
1b058bc3
ZJS
2640 /*
2641 * We have width = stat_width or term_columns() columns total.
2642 * We want a maximum of min(max_len, stat_name_width) for the name part.
969fe57b 2643 * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1b058bc3
ZJS
2644 * We also need 1 for " " and 4 + decimal_width(max_change)
2645 * for " | NNNN " and one the empty column at the end, altogether
2646 * 6 + decimal_width(max_change).
2647 *
2648 * If there's not enough space, we will use the smaller of
2649 * stat_name_width (if set) and 5/8*width for the filename,
969fe57b
ZJS
2650 * and the rest for constant elements + graph part, but no more
2651 * than stat_graph_width for the graph part.
1b058bc3
ZJS
2652 * (5/8 gives 50 for filename and 30 for the constant parts + graph
2653 * for the standard terminal size).
a2540023 2654 *
1b058bc3
ZJS
2655 * In other words: stat_width limits the maximum width, and
2656 * stat_name_width fixes the maximum width of the filename,
2657 * and is also used to divide available columns if there
2658 * aren't enough.
dc801e71
ZJS
2659 *
2660 * Binary files are displayed with "Bin XXX -> YYY bytes"
2661 * instead of the change count and graph. This part is treated
2662 * similarly to the graph part, except that it is not
41ccfdd9 2663 * "scaled". If total width is too small to accommodate the
dc801e71
ZJS
2664 * guaranteed minimum width of the filename part and the
2665 * separators and this message, this message will "overflow"
2666 * making the line longer than the maximum width.
a2540023 2667 */
1b058bc3
ZJS
2668
2669 if (options->stat_width == -1)
cd48dadb 2670 width = term_columns() - strlen(line_prefix);
a2540023 2671 else
1b058bc3 2672 width = options->stat_width ? options->stat_width : 80;
dc801e71
ZJS
2673 number_width = decimal_width(max_change) > number_width ?
2674 decimal_width(max_change) : number_width;
a2540023 2675
df44483a
ZJS
2676 if (options->stat_graph_width == -1)
2677 options->stat_graph_width = diff_stat_graph_width;
a2540023 2678
1b058bc3
ZJS
2679 /*
2680 * Guarantee 3/8*16==6 for the graph part
2681 * and 5/8*16==10 for the filename part
2682 */
2683 if (width < 16 + 6 + number_width)
2684 width = 16 + 6 + number_width;
2685
2686 /*
2687 * First assign sizes that are wanted, ignoring available width.
dc801e71
ZJS
2688 * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2689 * starting from "XXX" should fit in graph_width.
1b058bc3 2690 */
dc801e71
ZJS
2691 graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2692 if (options->stat_graph_width &&
2693 options->stat_graph_width < graph_width)
2694 graph_width = options->stat_graph_width;
2695
1b058bc3
ZJS
2696 name_width = (options->stat_name_width > 0 &&
2697 options->stat_name_width < max_len) ?
2698 options->stat_name_width : max_len;
2699
2700 /*
2701 * Adjust adjustable widths not to exceed maximum width
2702 */
2703 if (name_width + number_width + 6 + graph_width > width) {
678c5741 2704 if (graph_width > width * 3/8 - number_width - 6) {
1b058bc3 2705 graph_width = width * 3/8 - number_width - 6;
678c5741
LP
2706 if (graph_width < 6)
2707 graph_width = 6;
2708 }
2709
969fe57b
ZJS
2710 if (options->stat_graph_width &&
2711 graph_width > options->stat_graph_width)
2712 graph_width = options->stat_graph_width;
1b058bc3
ZJS
2713 if (name_width > width - number_width - 6 - graph_width)
2714 name_width = width - number_width - 6 - graph_width;
2715 else
2716 graph_width = width - number_width - 6 - name_width;
2717 }
2718
2719 /*
2720 * From here name_width is the width of the name area,
2721 * and graph_width is the width of the graph area.
2722 * max_change is used to scale graph properly.
2723 */
808e1db2 2724 for (i = 0; i < count; i++) {
d2543b8e 2725 const char *prefix = "";
af0ed819
JH
2726 struct diffstat_file *file = data->files[i];
2727 char *name = file->print_name;
2728 uintmax_t added = file->added;
2729 uintmax_t deleted = file->deleted;
a2540023 2730 int name_len;
6973dcae 2731
a20d3c0d 2732 if (!file->is_interesting && (added + deleted == 0))
358e460e 2733 continue;
a20d3c0d 2734
6973dcae
JH
2735 /*
2736 * "scale" the filename
2737 */
a2540023
JH
2738 len = name_width;
2739 name_len = strlen(name);
2740 if (name_width < name_len) {
6973dcae
JH
2741 char *slash;
2742 prefix = "...";
a2540023
JH
2743 len -= 3;
2744 name += name_len - len;
6973dcae
JH
2745 slash = strchr(name, '/');
2746 if (slash)
2747 name = slash;
2748 }
6973dcae 2749
af0ed819 2750 if (file->is_binary) {
0911c475
SB
2751 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2752 strbuf_addf(&out, " %*s", number_width, "Bin");
e18872b2 2753 if (!added && !deleted) {
0911c475
SB
2754 strbuf_addch(&out, '\n');
2755 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2756 out.buf, out.len, 0);
2757 strbuf_reset(&out);
e18872b2
ZJS
2758 continue;
2759 }
0911c475 2760 strbuf_addf(&out, " %s%"PRIuMAX"%s",
0974c117 2761 del_c, deleted, reset);
0911c475
SB
2762 strbuf_addstr(&out, " -> ");
2763 strbuf_addf(&out, "%s%"PRIuMAX"%s",
0974c117 2764 add_c, added, reset);
0911c475
SB
2765 strbuf_addstr(&out, " bytes\n");
2766 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2767 out.buf, out.len, 0);
2768 strbuf_reset(&out);
f604652e 2769 continue;
6973dcae 2770 }
af0ed819 2771 else if (file->is_unmerged) {
0911c475
SB
2772 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2773 strbuf_addstr(&out, " Unmerged\n");
2774 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2775 out.buf, out.len, 0);
2776 strbuf_reset(&out);
f604652e 2777 continue;
6973dcae 2778 }
6973dcae 2779
a2540023
JH
2780 /*
2781 * scale the add/delete
2782 */
6973dcae
JH
2783 add = added;
2784 del = deleted;
6973dcae 2785
1b058bc3 2786 if (graph_width <= max_change) {
d3c9cf32 2787 int total = scale_linear(add + del, graph_width, max_change);
2eeeef24
JH
2788 if (total < 2 && add && del)
2789 /* width >= 2 due to the sanity check */
2790 total = 2;
2791 if (add < del) {
1b058bc3 2792 add = scale_linear(add, graph_width, max_change);
2eeeef24
JH
2793 del = total - add;
2794 } else {
1b058bc3 2795 del = scale_linear(del, graph_width, max_change);
2eeeef24
JH
2796 add = total - del;
2797 }
6973dcae 2798 }
0911c475
SB
2799 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2800 strbuf_addf(&out, " %*"PRIuMAX"%s",
dc801e71
ZJS
2801 number_width, added + deleted,
2802 added + deleted ? " " : "");
0911c475
SB
2803 show_graph(&out, '+', add, add_c, reset);
2804 show_graph(&out, '-', del, del_c, reset);
2805 strbuf_addch(&out, '\n');
2806 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2807 out.buf, out.len, 0);
2808 strbuf_reset(&out);
c0c77734 2809 }
a20d3c0d
JH
2810
2811 for (i = 0; i < data->nr; i++) {
af0ed819
JH
2812 struct diffstat_file *file = data->files[i];
2813 uintmax_t added = file->added;
2814 uintmax_t deleted = file->deleted;
82dfc2c4
JH
2815
2816 if (file->is_unmerged ||
2817 (!file->is_interesting && (added + deleted == 0))) {
808e1db2
MG
2818 total_files--;
2819 continue;
2820 }
a20d3c0d 2821
82dfc2c4 2822 if (!file->is_binary) {
a20d3c0d
JH
2823 adds += added;
2824 dels += deleted;
2825 }
2826 if (i < count)
2827 continue;
e5f85df8 2828 if (!extra_shown)
0911c475
SB
2829 emit_diff_symbol(options,
2830 DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2831 NULL, 0, 0);
e5f85df8 2832 extra_shown = 1;
808e1db2 2833 }
0911c475
SB
2834
2835 print_stat_summary_inserts_deletes(options, total_files, adds, dels);
5a612017 2836 strbuf_release(&out);
6973dcae
JH
2837}
2838
2775d92c 2839static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
ebd124c6
NP
2840{
2841 int i, adds = 0, dels = 0, total_files = data->nr;
2842
2843 if (data->nr == 0)
2844 return;
2845
2846 for (i = 0; i < data->nr; i++) {
e18872b2 2847 int added = data->files[i]->added;
0911c475 2848 int deleted = data->files[i]->deleted;
e18872b2 2849
20c8cde4
JH
2850 if (data->files[i]->is_unmerged ||
2851 (!data->files[i]->is_interesting && (added + deleted == 0))) {
e18872b2 2852 total_files--;
de9658b5 2853 } else if (!data->files[i]->is_binary) { /* don't count bytes */
e18872b2
ZJS
2854 adds += added;
2855 dels += deleted;
ebd124c6 2856 }
ebd124c6 2857 }
0911c475 2858 print_stat_summary_inserts_deletes(options, total_files, adds, dels);
ebd124c6
NP
2859}
2860
4b25d091 2861static void show_numstat(struct diffstat_t *data, struct diff_options *options)
74e2abe5
JH
2862{
2863 int i;
2864
f604652e
JH
2865 if (data->nr == 0)
2866 return;
2867
74e2abe5
JH
2868 for (i = 0; i < data->nr; i++) {
2869 struct diffstat_file *file = data->files[i];
2870
30997bb8 2871 fprintf(options->file, "%s", diff_line_prefix(options));
7be57610 2872
bfddbc5e 2873 if (file->is_binary)
c0c77734 2874 fprintf(options->file, "-\t-\t");
bfddbc5e 2875 else
c0c77734 2876 fprintf(options->file,
0974c117
JK
2877 "%"PRIuMAX"\t%"PRIuMAX"\t",
2878 file->added, file->deleted);
f604652e
JH
2879 if (options->line_termination) {
2880 fill_print_name(file);
2881 if (!file->is_renamed)
c0c77734 2882 write_name_quoted(file->name, options->file,
f604652e
JH
2883 options->line_termination);
2884 else {
c0c77734
DB
2885 fputs(file->print_name, options->file);
2886 putc(options->line_termination, options->file);
f604652e 2887 }
663af342 2888 } else {
f604652e 2889 if (file->is_renamed) {
c0c77734
DB
2890 putc('\0', options->file);
2891 write_name_quoted(file->from_name, options->file, '\0');
f604652e 2892 }
c0c77734 2893 write_name_quoted(file->name, options->file, '\0');
663af342 2894 }
74e2abe5
JH
2895 }
2896}
2897
c04a7155
JH
2898struct dirstat_file {
2899 const char *name;
2900 unsigned long changed;
7df7c019
LT
2901};
2902
c04a7155
JH
2903struct dirstat_dir {
2904 struct dirstat_file *files;
712d2c7d 2905 int alloc, nr, permille, cumulative;
c04a7155
JH
2906};
2907
7be57610
BY
2908static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2909 unsigned long changed, const char *base, int baselen)
7df7c019 2910{
585c0e2e 2911 unsigned long sum_changes = 0;
7df7c019 2912 unsigned int sources = 0;
30997bb8 2913 const char *line_prefix = diff_line_prefix(opt);
7df7c019
LT
2914
2915 while (dir->nr) {
c04a7155 2916 struct dirstat_file *f = dir->files;
7df7c019 2917 int namelen = strlen(f->name);
585c0e2e 2918 unsigned long changes;
7df7c019
LT
2919 char *slash;
2920
2921 if (namelen < baselen)
2922 break;
2923 if (memcmp(f->name, base, baselen))
2924 break;
2925 slash = strchr(f->name + baselen, '/');
2926 if (slash) {
2927 int newbaselen = slash + 1 - f->name;
585c0e2e 2928 changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
7df7c019
LT
2929 sources++;
2930 } else {
585c0e2e 2931 changes = f->changed;
7df7c019
LT
2932 dir->files++;
2933 dir->nr--;
2934 sources += 2;
2935 }
585c0e2e 2936 sum_changes += changes;
7df7c019
LT
2937 }
2938
2939 /*
2940 * We don't report dirstat's for
2941 * - the top level
2942 * - or cases where everything came from a single directory
2943 * under this directory (sources == 1).
2944 */
2945 if (baselen && sources != 1) {
585c0e2e
BW
2946 if (sum_changes) {
2947 int permille = sum_changes * 1000 / changed;
712d2c7d 2948 if (permille >= dir->permille) {
7be57610 2949 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
712d2c7d 2950 permille / 10, permille % 10, baselen, base);
7df7c019
LT
2951 if (!dir->cumulative)
2952 return 0;
2953 }
2954 }
2955 }
585c0e2e 2956 return sum_changes;
7df7c019
LT
2957}
2958
441bca0b
LT
2959static int dirstat_compare(const void *_a, const void *_b)
2960{
2961 const struct dirstat_file *a = _a;
2962 const struct dirstat_file *b = _b;
2963 return strcmp(a->name, b->name);
2964}
2965
c04a7155 2966static void show_dirstat(struct diff_options *options)
7df7c019
LT
2967{
2968 int i;
2969 unsigned long changed;
c04a7155
JH
2970 struct dirstat_dir dir;
2971 struct diff_queue_struct *q = &diff_queued_diff;
2972
2973 dir.files = NULL;
2974 dir.alloc = 0;
2975 dir.nr = 0;
712d2c7d 2976 dir.permille = options->dirstat_permille;
0d1e0e78 2977 dir.cumulative = options->flags.dirstat_cumulative;
7df7c019 2978
7df7c019 2979 changed = 0;
c04a7155
JH
2980 for (i = 0; i < q->nr; i++) {
2981 struct diff_filepair *p = q->queue[i];
2982 const char *name;
2983 unsigned long copied, added, damage;
2984
2ca86714 2985 name = p->two->path ? p->two->path : p->one->path;
c04a7155 2986
d9f62dfa
JK
2987 if (p->one->oid_valid && p->two->oid_valid &&
2988 oideq(&p->one->oid, &p->two->oid)) {
2ff3a803
JH
2989 /*
2990 * The SHA1 has not changed, so pre-/post-content is
2991 * identical. We can therefore skip looking at the
2992 * file contents altogether.
2993 */
2994 damage = 0;
2995 goto found_damage;
2996 }
2997
0d1e0e78 2998 if (options->flags.dirstat_by_file) {
0133dab7
JH
2999 /*
3000 * In --dirstat-by-file mode, we don't really need to
3001 * look at the actual file contents at all.
3002 * The fact that the SHA1 changed is enough for us to
3003 * add this file to the list of results
3004 * (with each file contributing equal damage).
3005 */
2ff3a803 3006 damage = 1;
0133dab7
JH
3007 goto found_damage;
3008 }
c04a7155
JH
3009
3010 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
b78ea5fc
NTND
3011 diff_populate_filespec(options->repo, p->one, 0);
3012 diff_populate_filespec(options->repo, p->two, 0);
3013 diffcore_count_changes(options->repo,
3014 p->one, p->two, NULL, NULL,
c04a7155
JH
3015 &copied, &added);
3016 diff_free_filespec_data(p->one);
3017 diff_free_filespec_data(p->two);
3018 } else if (DIFF_FILE_VALID(p->one)) {
b78ea5fc 3019 diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
c04a7155
JH
3020 copied = added = 0;
3021 diff_free_filespec_data(p->one);
3022 } else if (DIFF_FILE_VALID(p->two)) {
b78ea5fc 3023 diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
c04a7155
JH
3024 copied = 0;
3025 added = p->two->size;
3026 diff_free_filespec_data(p->two);
3027 } else
2b0b551d 3028 continue;
c04a7155
JH
3029
3030 /*
3031 * Original minus copied is the removed material,
3032 * added is the new material. They are both damages
0133dab7 3033 * made to the preimage.
2ff3a803
JH
3034 * If the resulting damage is zero, we know that
3035 * diffcore_count_changes() considers the two entries to
d9f62dfa 3036 * be identical, but since the oid changed, we
2ff3a803
JH
3037 * know that there must have been _some_ kind of change,
3038 * so we force all entries to have damage > 0.
c04a7155
JH
3039 */
3040 damage = (p->one->size - copied) + added;
2ff3a803 3041 if (!damage)
fd33777b 3042 damage = 1;
c04a7155 3043
0133dab7 3044found_damage:
c04a7155
JH
3045 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3046 dir.files[dir.nr].name = name;
3047 dir.files[dir.nr].changed = damage;
3048 changed += damage;
3049 dir.nr++;
7df7c019
LT
3050 }
3051
3052 /* This can happen even with many files, if everything was renames */
3053 if (!changed)
3054 return;
3055
3056 /* Show all directories with more than x% of the changes */
9ed0d8d6 3057 QSORT(dir.files, dir.nr, dirstat_compare);
7be57610 3058 gather_dirstat(options, &dir, changed, "", 0);
7df7c019
LT
3059}
3060
1c57a627
JH
3061static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3062{
3063 int i;
3064 unsigned long changed;
3065 struct dirstat_dir dir;
3066
3067 if (data->nr == 0)
3068 return;
3069
3070 dir.files = NULL;
3071 dir.alloc = 0;
3072 dir.nr = 0;
3073 dir.permille = options->dirstat_permille;
0d1e0e78 3074 dir.cumulative = options->flags.dirstat_cumulative;
1c57a627
JH
3075
3076 changed = 0;
3077 for (i = 0; i < data->nr; i++) {
3078 struct diffstat_file *file = data->files[i];
3079 unsigned long damage = file->added + file->deleted;
3080 if (file->is_binary)
3081 /*
3082 * binary files counts bytes, not lines. Must find some
3083 * way to normalize binary bytes vs. textual lines.
3084 * The following heuristic assumes that there are 64
3085 * bytes per "line".
3086 * This is stupid and ugly, but very cheap...
3087 */
42c78a21 3088 damage = DIV_ROUND_UP(damage, 64);
1c57a627
JH
3089 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3090 dir.files[dir.nr].name = file->name;
3091 dir.files[dir.nr].changed = damage;
3092 changed += damage;
3093 dir.nr++;
3094 }
3095
3096 /* This can happen even with many files, if everything was renames */
3097 if (!changed)
3098 return;
3099
3100 /* Show all directories with more than x% of the changes */
9ed0d8d6 3101 QSORT(dir.files, dir.nr, dirstat_compare);
1c57a627
JH
3102 gather_dirstat(options, &dir, changed, "", 0);
3103}
3104
f604652e
JH
3105static void free_diffstat_info(struct diffstat_t *diffstat)
3106{
3107 int i;
3108 for (i = 0; i < diffstat->nr; i++) {
3109 struct diffstat_file *f = diffstat->files[i];
c905cbc4 3110 free(f->print_name);
f604652e
JH
3111 free(f->name);
3112 free(f->from_name);
3113 free(f);
3114 }
3115 free(diffstat->files);
3116}
3117
88246898 3118struct checkdiff_t {
88246898 3119 const char *filename;
1ba111d1 3120 int lineno;
a757c646 3121 int conflict_marker_size;
1ba111d1 3122 struct diff_options *o;
cf1b7869 3123 unsigned ws_rule;
62c64895 3124 unsigned status;
88246898
JS
3125};
3126
a757c646 3127static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
04954043
JH
3128{
3129 char firstchar;
3130 int cnt;
3131
a757c646 3132 if (len < marker_size + 1)
04954043
JH
3133 return 0;
3134 firstchar = line[0];
3135 switch (firstchar) {
a757c646 3136 case '=': case '>': case '<': case '|':
04954043
JH
3137 break;
3138 default:
3139 return 0;
3140 }
a757c646 3141 for (cnt = 1; cnt < marker_size; cnt++)
04954043
JH
3142 if (line[cnt] != firstchar)
3143 return 0;
a757c646
JH
3144 /* line[1] thru line[marker_size-1] are same as firstchar */
3145 if (len < marker_size + 1 || !isspace(line[marker_size]))
04954043 3146 return 0;
04954043
JH
3147 return 1;
3148}
3149
75ab7630
JK
3150static void checkdiff_consume_hunk(void *priv,
3151 long ob, long on, long nb, long nn,
3152 const char *func, long funclen)
3153
3154{
3155 struct checkdiff_t *data = priv;
3156 data->lineno = nb - 1;
3157}
3158
88246898
JS
3159static void checkdiff_consume(void *priv, char *line, unsigned long len)
3160{
3161 struct checkdiff_t *data = priv;
a757c646 3162 int marker_size = data->conflict_marker_size;
f1c96261
JK
3163 const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3164 const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3165 const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
c1795bb0 3166 char *err;
30997bb8 3167 const char *line_prefix;
7be57610
BY
3168
3169 assert(data->o);
30997bb8 3170 line_prefix = diff_line_prefix(data->o);
88246898
JS
3171
3172 if (line[0] == '+') {
18374e58 3173 unsigned bad;
0ef617f4 3174 data->lineno++;
a757c646 3175 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
04954043
JH
3176 data->status |= 1;
3177 fprintf(data->o->file,
7be57610
BY
3178 "%s%s:%d: leftover conflict marker\n",
3179 line_prefix, data->filename, data->lineno);
04954043 3180 }
8f8841e9 3181 bad = ws_check(line + 1, len - 1, data->ws_rule);
18374e58 3182 if (!bad)
c1795bb0 3183 return;
18374e58
JH
3184 data->status |= bad;
3185 err = whitespace_error_string(bad);
7be57610
BY
3186 fprintf(data->o->file, "%s%s:%d: %s.\n",
3187 line_prefix, data->filename, data->lineno, err);
c1795bb0 3188 free(err);
a3c158d4 3189 emit_line(data->o, set, reset, line, 1);
8f8841e9 3190 ws_check_emit(line + 1, len - 1, data->ws_rule,
1ba111d1 3191 data->o->file, set, reset, ws);
877f23cc 3192 } else if (line[0] == ' ') {
88246898 3193 data->lineno++;
88246898
JS
3194 }
3195}
3196
0660626c
JH
3197static unsigned char *deflate_it(char *data,
3198 unsigned long size,
3199 unsigned long *result_size)
051308f6 3200{
0660626c
JH
3201 int bound;
3202 unsigned char *deflated;
ef49a7a0 3203 git_zstream stream;
0660626c 3204
55bb5c91 3205 git_deflate_init(&stream, zlib_compression_level);
225a6f10 3206 bound = git_deflate_bound(&stream, size);
0660626c
JH
3207 deflated = xmalloc(bound);
3208 stream.next_out = deflated;
3209 stream.avail_out = bound;
3210
3211 stream.next_in = (unsigned char *)data;
3212 stream.avail_in = size;
55bb5c91 3213 while (git_deflate(&stream, Z_FINISH) == Z_OK)
0660626c 3214 ; /* nothing */
55bb5c91 3215 git_deflate_end(&stream);
0660626c
JH
3216 *result_size = stream.total_out;
3217 return deflated;
051308f6
JH
3218}
3219
4eed0ebd
SB
3220static void emit_binary_diff_body(struct diff_options *o,
3221 mmfile_t *one, mmfile_t *two)
051308f6 3222{
0660626c
JH
3223 void *cp;
3224 void *delta;
3225 void *deflated;
3226 void *data;
3227 unsigned long orig_size;
3228 unsigned long delta_size;
3229 unsigned long deflate_size;
3230 unsigned long data_size;
051308f6 3231
0660626c
JH
3232 /* We could do deflated delta, or we could do just deflated two,
3233 * whichever is smaller.
051308f6 3234 */
0660626c
JH
3235 delta = NULL;
3236 deflated = deflate_it(two->ptr, two->size, &deflate_size);
3237 if (one->size && two->size) {
3238 delta = diff_delta(one->ptr, one->size,
3239 two->ptr, two->size,
3240 &delta_size, deflate_size);
3241 if (delta) {
3242 void *to_free = delta;
3243 orig_size = delta_size;
3244 delta = deflate_it(delta, delta_size, &delta_size);
3245 free(to_free);
051308f6
JH
3246 }
3247 }
051308f6 3248
0660626c 3249 if (delta && delta_size < deflate_size) {
ca473cef 3250 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
4eed0ebd
SB
3251 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3252 s, strlen(s), 0);
3253 free(s);
0660626c
JH
3254 free(deflated);
3255 data = delta;
3256 data_size = delta_size;
4eed0ebd
SB
3257 } else {
3258 char *s = xstrfmt("%lu", two->size);
3259 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3260 s, strlen(s), 0);
3261 free(s);
0660626c
JH
3262 free(delta);
3263 data = deflated;
3264 data_size = deflate_size;
3265 }
051308f6 3266
0660626c
JH
3267 /* emit data encoded in base85 */
3268 cp = data;
3269 while (data_size) {
4eed0ebd 3270 int len;
0660626c 3271 int bytes = (52 < data_size) ? 52 : data_size;
4eed0ebd 3272 char line[71];
0660626c 3273 data_size -= bytes;
051308f6
JH
3274 if (bytes <= 26)
3275 line[0] = bytes + 'A' - 1;
3276 else
3277 line[0] = bytes - 26 + 'a' - 1;
3278 encode_85(line + 1, cp, bytes);
1d7f171c 3279 cp = (char *) cp + bytes;
4eed0ebd
SB
3280
3281 len = strlen(line);
3282 line[len++] = '\n';
3283 line[len] = '\0';
3284
3285 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3286 line, len, 0);
051308f6 3287 }
4eed0ebd 3288 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
0660626c 3289 free(data);
051308f6
JH
3290}
3291
4eed0ebd
SB
3292static void emit_binary_diff(struct diff_options *o,
3293 mmfile_t *one, mmfile_t *two)
d4c452f0 3294{
4eed0ebd
SB
3295 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3296 emit_binary_diff_body(o, one, two);
3297 emit_binary_diff_body(o, two, one);
d4c452f0
JH
3298}
3299
b78ea5fc
NTND
3300int diff_filespec_is_binary(struct repository *r,
3301 struct diff_filespec *one)
29a3eefd 3302{
122aa6f9 3303 if (one->is_binary == -1) {
acd00ea0 3304 diff_filespec_load_driver(one, r->index);
122aa6f9
JK
3305 if (one->driver->binary != -1)
3306 one->is_binary = one->driver->binary;
3307 else {
3308 if (!one->data && DIFF_FILE_VALID(one))
b78ea5fc 3309 diff_populate_filespec(r, one, CHECK_BINARY);
6bf3b813 3310 if (one->is_binary == -1 && one->data)
122aa6f9
JK
3311 one->is_binary = buffer_is_binary(one->data,
3312 one->size);
3313 if (one->is_binary == -1)
3314 one->is_binary = 0;
3315 }
3316 }
29a3eefd 3317 return one->is_binary;
6973dcae
JH
3318}
3319
acd00ea0
NTND
3320static const struct userdiff_funcname *
3321diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
f258475a 3322{
acd00ea0 3323 diff_filespec_load_driver(one, o->repo->index);
122aa6f9 3324 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
f258475a
JH
3325}
3326
a5a818ee
JH
3327void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3328{
3329 if (!options->a_prefix)
3330 options->a_prefix = a;
3331 if (!options->b_prefix)
3332 options->b_prefix = b;
3333}
3334
acd00ea0
NTND
3335struct userdiff_driver *get_textconv(struct index_state *istate,
3336 struct diff_filespec *one)
04427ac8
JK
3337{
3338 if (!DIFF_FILE_VALID(one))
3339 return NULL;
d391c0ff 3340
acd00ea0 3341 diff_filespec_load_driver(one, istate);
3813e690 3342 return userdiff_get_textconv(one->driver);
04427ac8
JK
3343}
3344
6973dcae
JH
3345static void builtin_diff(const char *name_a,
3346 const char *name_b,
3347 struct diff_filespec *one,
3348 struct diff_filespec *two,
3349 const char *xfrm_msg,
296c6bb2 3350 int must_show_header,
051308f6 3351 struct diff_options *o,
6973dcae
JH
3352 int complete_rewrite)
3353{
3354 mmfile_t mf1, mf2;
3355 const char *lbl[2];
3356 char *a_one, *b_two;
d9c552f1 3357 const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
8f67f8ae 3358 const char *reset = diff_get_color_opt(o, DIFF_RESET);
a5a818ee 3359 const char *a_prefix, *b_prefix;
d9bae1a1
JK
3360 struct userdiff_driver *textconv_one = NULL;
3361 struct userdiff_driver *textconv_two = NULL;
3e97c7c6 3362 struct strbuf header = STRBUF_INIT;
30997bb8 3363 const char *line_prefix = diff_line_prefix(o);
a5a818ee 3364
fd47ae6a 3365 diff_set_mnemonic_prefix(o, "a/", "b/");
0d1e0e78 3366 if (o->flags.reverse_diff) {
fd47ae6a
JK
3367 a_prefix = o->b_prefix;
3368 b_prefix = o->a_prefix;
3369 } else {
3370 a_prefix = o->a_prefix;
3371 b_prefix = o->b_prefix;
3372 }
3373
61cfbc05
JK
3374 if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3375 (!one->mode || S_ISGITLINK(one->mode)) &&
3376 (!two->mode || S_ISGITLINK(two->mode))) {
f3597138 3377 show_submodule_summary(o, one->path ? one->path : two->path,
602a283a 3378 &one->oid, &two->oid,
f3597138 3379 two->dirty_submodule);
752c0c24 3380 return;
fd47ae6a
JK
3381 } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3382 (!one->mode || S_ISGITLINK(one->mode)) &&
3383 (!two->mode || S_ISGITLINK(two->mode))) {
f3597138 3384 show_submodule_inline_diff(o, one->path ? one->path : two->path,
fd47ae6a 3385 &one->oid, &two->oid,
f3597138 3386 two->dirty_submodule);
fd47ae6a 3387 return;
752c0c24
JS
3388 }
3389
0d1e0e78 3390 if (o->flags.allow_textconv) {
acd00ea0
NTND
3391 textconv_one = get_textconv(o->repo->index, one);
3392 textconv_two = get_textconv(o->repo->index, two);
3aa1f7ca
JK
3393 }
3394
71b989e7
LT
3395 /* Never use a non-valid filename anywhere if at all possible */
3396 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3397 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3398
a5a818ee
JH
3399 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3400 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
6973dcae
JH
3401 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3402 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
d9c552f1 3403 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
6973dcae
JH
3404 if (lbl[0][0] == '/') {
3405 /* /dev/null */
d9c552f1 3406 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
37466447
BW
3407 if (xfrm_msg)
3408 strbuf_addstr(&header, xfrm_msg);
296c6bb2 3409 must_show_header = 1;
6973dcae
JH
3410 }
3411 else if (lbl[1][0] == '/') {
d9c552f1 3412 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
37466447
BW
3413 if (xfrm_msg)
3414 strbuf_addstr(&header, xfrm_msg);
296c6bb2 3415 must_show_header = 1;
6973dcae
JH
3416 }
3417 else {
3418 if (one->mode != two->mode) {
d9c552f1
JK
3419 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3420 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
296c6bb2 3421 must_show_header = 1;
cd112cef 3422 }
37466447
BW
3423 if (xfrm_msg)
3424 strbuf_addstr(&header, xfrm_msg);
3e97c7c6 3425
6973dcae
JH
3426 /*
3427 * we do not run diff between different kind
3428 * of objects.
3429 */
3430 if ((one->mode ^ two->mode) & S_IFMT)
3431 goto free_ab_and_return;
0c01857d 3432 if (complete_rewrite &&
b78ea5fc
NTND
3433 (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3434 (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
a29b0a13
SB
3435 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3436 header.buf, header.len, 0);
3e97c7c6 3437 strbuf_reset(&header);
3aa1f7ca 3438 emit_rewrite_diff(name_a, name_b, one, two,
b78ea5fc 3439 textconv_one, textconv_two, o);
34a5e1a2 3440 o->found_changes = 1;
6973dcae
JH
3441 goto free_ab_and_return;
3442 }
3443 }
3444
467ddc14 3445 if (o->irreversible_delete && lbl[1][0] == '/') {
a29b0a13
SB
3446 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3447 header.len, 0);
467ddc14
JH
3448 strbuf_reset(&header);
3449 goto free_ab_and_return;
0d1e0e78 3450 } else if (!o->flags.text &&
b78ea5fc
NTND
3451 ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3452 (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
4acaaa7a 3453 struct strbuf sb = STRBUF_INIT;
1aaf69e6
NTND
3454 if (!one->data && !two->data &&
3455 S_ISREG(one->mode) && S_ISREG(two->mode) &&
0d1e0e78 3456 !o->flags.binary) {
4a7e27e9 3457 if (oideq(&one->oid, &two->oid)) {
1aaf69e6 3458 if (must_show_header)
a29b0a13
SB
3459 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3460 header.buf, header.len,
3461 0);
1aaf69e6
NTND
3462 goto free_ab_and_return;
3463 }
a29b0a13
SB
3464 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3465 header.buf, header.len, 0);
4acaaa7a
SB
3466 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3467 diff_line_prefix(o), lbl[0], lbl[1]);
3468 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3469 sb.buf, sb.len, 0);
3470 strbuf_release(&sb);
1aaf69e6
NTND
3471 goto free_ab_and_return;
3472 }
b78ea5fc
NTND
3473 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3474 fill_mmfile(o->repo, &mf2, two) < 0)
b3373982 3475 die("unable to read files to diff");
0660626c
JH
3476 /* Quite common confusing case */
3477 if (mf1.size == mf2.size &&
296c6bb2
CC
3478 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3479 if (must_show_header)
a29b0a13
SB
3480 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3481 header.buf, header.len, 0);
0660626c 3482 goto free_ab_and_return;
296c6bb2 3483 }
a29b0a13 3484 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3e97c7c6 3485 strbuf_reset(&header);
0d1e0e78 3486 if (o->flags.binary)
4eed0ebd 3487 emit_binary_diff(o, &mf1, &mf2);
4acaaa7a
SB
3488 else {
3489 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3490 diff_line_prefix(o), lbl[0], lbl[1]);
3491 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3492 sb.buf, sb.len, 0);
3493 strbuf_release(&sb);
3494 }
34a5e1a2 3495 o->found_changes = 1;
467ddc14 3496 } else {
6973dcae
JH
3497 /* Crazy xdl interfaces.. */
3498 const char *diffopts = getenv("GIT_DIFF_OPTS");
ae021d87 3499 const char *v;
6973dcae
JH
3500 xpparam_t xpp;
3501 xdemitconf_t xecfg;
6973dcae 3502 struct emit_callback ecbdata;
be58e70d 3503 const struct userdiff_funcname *pe;
f258475a 3504
b3f01ff2 3505 if (must_show_header) {
a29b0a13
SB
3506 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3507 header.buf, header.len, 0);
3e97c7c6
GB
3508 strbuf_reset(&header);
3509 }
3510
6afaf807
NTND
3511 mf1.size = fill_textconv(o->repo, textconv_one, one, &mf1.ptr);
3512 mf2.size = fill_textconv(o->repo, textconv_two, two, &mf2.ptr);
04427ac8 3513
acd00ea0 3514 pe = diff_funcname_pattern(o, one);
45e7ca0f 3515 if (!pe)
acd00ea0 3516 pe = diff_funcname_pattern(o, two);
6973dcae 3517
9ccd0a88 3518 memset(&xpp, 0, sizeof(xpp));
30b25010 3519 memset(&xecfg, 0, sizeof(xecfg));
cd112cef 3520 memset(&ecbdata, 0, sizeof(ecbdata));
1cdde296
JS
3521 if (o->flags.suppress_diff_headers)
3522 lbl[0] = NULL;
6973dcae 3523 ecbdata.label_path = lbl;
daa0c3d9 3524 ecbdata.color_diff = want_color(o->use_color);
26d024ec 3525 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
690ed843 3526 if (ecbdata.ws_rule & WS_BLANK_AT_EOF)
d68fe26f 3527 check_blank_at_eof(&mf1, &mf2, &ecbdata);
a3c158d4 3528 ecbdata.opt = o;
1cdde296
JS
3529 if (header.len && !o->flags.suppress_diff_headers)
3530 ecbdata.header = &header;
582aa00b 3531 xpp.flags = o->xdl_opts;
2477ab2e
JT
3532 xpp.anchors = o->anchors;
3533 xpp.anchors_nr = o->anchors_nr;
ee1e5412 3534 xecfg.ctxlen = o->context;
6d0e674a 3535 xecfg.interhunkctxlen = o->interhunkcontext;
6973dcae 3536 xecfg.flags = XDL_EMIT_FUNCNAMES;
0d1e0e78 3537 if (o->flags.funccontext)
14937c2c 3538 xecfg.flags |= XDL_EMIT_FUNCCONTEXT;
45e7ca0f 3539 if (pe)
a013585b 3540 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
6973dcae
JH
3541 if (!diffopts)
3542 ;
ae021d87
JK
3543 else if (skip_prefix(diffopts, "--unified=", &v))
3544 xecfg.ctxlen = strtoul(v, NULL, 10);