diff: drop options parameter from diffcore_fix_diff_index()
[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
d173e799 294static unsigned parse_color_moved_ws(const char *arg)
b3095712
SB
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;
d173e799
SB
317 else {
318 ret |= COLOR_MOVED_WS_ERROR;
319 error(_("unknown color-moved-ws mode '%s', possible values are 'ignore-space-change', 'ignore-space-at-eol', 'ignore-all-space', 'allow-indentation-change'"), sb.buf);
320 }
b3095712
SB
321
322 strbuf_release(&sb);
323 }
324
ca1f4ae4 325 if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
d173e799 326 (ret & XDF_WHITESPACE_FLAGS)) {
15b07cba 327 error(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
d173e799
SB
328 ret |= COLOR_MOVED_WS_ERROR;
329 }
ca1f4ae4 330
b3095712
SB
331 string_list_clear(&l, 0);
332
333 return ret;
2e2d5ac1
SB
334}
335
ef90d6d4 336int git_diff_ui_config(const char *var, const char *value, void *cb)
801235c5 337{
a159ca0c 338 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
e269eb79 339 diff_use_color_default = git_config_colorbool(var, value);
801235c5
JH
340 return 0;
341 }
2e2d5ac1
SB
342 if (!strcmp(var, "diff.colormoved")) {
343 int cm = parse_color_moved(value);
344 if (cm < 0)
345 return -1;
346 diff_color_moved_default = cm;
347 return 0;
348 }
626c0b5d 349 if (!strcmp(var, "diff.colormovedws")) {
d173e799
SB
350 unsigned cm = parse_color_moved_ws(value);
351 if (cm & COLOR_MOVED_WS_ERROR)
626c0b5d
SB
352 return -1;
353 diff_color_moved_ws_default = cm;
354 return 0;
355 }
6468a4e5
JM
356 if (!strcmp(var, "diff.context")) {
357 diff_context_default = git_config_int(var, value);
358 if (diff_context_default < 0)
359 return -1;
360 return 0;
361 }
c4888677
VN
362 if (!strcmp(var, "diff.interhunkcontext")) {
363 diff_interhunk_context_default = git_config_int(var, value);
364 if (diff_interhunk_context_default < 0)
365 return -1;
366 return 0;
367 }
b68ea12e 368 if (!strcmp(var, "diff.renames")) {
cced5fbc 369 diff_detect_rename_default = git_config_rename(var, value);
b68ea12e
EW
370 return 0;
371 }
aecbf914
JH
372 if (!strcmp(var, "diff.autorefreshindex")) {
373 diff_auto_refresh_index = git_config_bool(var, value);
374 return 0;
375 }
a5a818ee
JH
376 if (!strcmp(var, "diff.mnemonicprefix")) {
377 diff_mnemonic_prefix = git_config_bool(var, value);
378 return 0;
379 }
f89504dd
EC
380 if (!strcmp(var, "diff.noprefix")) {
381 diff_no_prefix = git_config_bool(var, value);
382 return 0;
383 }
df44483a
ZJS
384 if (!strcmp(var, "diff.statgraphwidth")) {
385 diff_stat_graph_width = git_config_int(var, value);
386 return 0;
387 }
daec808c
BH
388 if (!strcmp(var, "diff.external"))
389 return git_config_string(&external_diff_cmd_cfg, var, value);
98a4d87b
BSSJ
390 if (!strcmp(var, "diff.wordregex"))
391 return git_config_string(&diff_word_regex_cfg, var, value);
6d8940b5
SB
392 if (!strcmp(var, "diff.orderfile"))
393 return git_config_pathname(&diff_order_file_cfg, var, value);
f1af60bd 394
be4f2b40
JS
395 if (!strcmp(var, "diff.ignoresubmodules"))
396 handle_ignore_submodules_arg(&default_diff_options, value);
397
c47ef57c
RR
398 if (!strcmp(var, "diff.submodule")) {
399 if (parse_submodule_params(&default_diff_options, value))
400 warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
401 value);
402 return 0;
403 }
404
07ab4dec
MP
405 if (!strcmp(var, "diff.algorithm")) {
406 diff_algorithm = parse_algorithm_value(value);
407 if (diff_algorithm < 0)
408 return -1;
409 return 0;
410 }
411
a17505f2
JH
412 if (!strcmp(var, "diff.wserrorhighlight")) {
413 int val = parse_ws_error_highlight(value);
414 if (val < 0)
415 return -1;
416 ws_error_highlight_default = val;
417 return 0;
418 }
419
33c643bb
JK
420 if (git_color_config(var, value, cb) < 0)
421 return -1;
422
ef90d6d4 423 return git_diff_basic_config(var, value, cb);
9a1805a8
JK
424}
425
ef90d6d4 426int git_diff_basic_config(const char *var, const char *value, void *cb)
9a1805a8 427{
ae021d87
JK
428 const char *name;
429
2b6ca6df
LT
430 if (!strcmp(var, "diff.renamelimit")) {
431 diff_rename_limit_default = git_config_int(var, value);
432 return 0;
433 }
434
6680a087
JK
435 if (userdiff_config(var, value) < 0)
436 return -1;
c7534ef4 437
ae021d87
JK
438 if (skip_prefix(var, "diff.color.", &name) ||
439 skip_prefix(var, "color.diff.", &name)) {
440 int slot = parse_diff_color_slot(name);
8b8e8624
JK
441 if (slot < 0)
442 return 0;
64f30e94
JH
443 if (!value)
444 return config_error_nonbool(var);
f6c5a296 445 return color_parse(value, diff_colors[slot]);
801235c5 446 }
f1af60bd 447
a624eaa7 448 /* like GNU diff's --suppress-blank-empty option */
950db879
JS
449 if (!strcmp(var, "diff.suppressblankempty") ||
450 /* for backwards compatibility */
451 !strcmp(var, "diff.suppress-blank-empty")) {
a624eaa7
JM
452 diff_suppress_blank_empty = git_config_bool(var, value);
453 return 0;
454 }
455
2d174951 456 if (!strcmp(var, "diff.dirstat")) {
51670fc8 457 struct strbuf errmsg = STRBUF_INIT;
712d2c7d 458 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
51670fc8 459 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
7478ac57 460 warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
51670fc8
JH
461 errmsg.buf);
462 strbuf_release(&errmsg);
712d2c7d 463 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
2d174951
JH
464 return 0;
465 }
466
cf5e7722
MB
467 if (git_diff_heuristic_config(var, value, cb) < 0)
468 return -1;
469
3e1dd17a 470 return git_default_config(var, value, cb);
801235c5
JH
471}
472
6973dcae
JH
473static char *quote_two(const char *one, const char *two)
474{
475 int need_one = quote_c_style(one, NULL, NULL, 1);
476 int need_two = quote_c_style(two, NULL, NULL, 1);
f285a2d7 477 struct strbuf res = STRBUF_INIT;
6973dcae
JH
478
479 if (need_one + need_two) {
663af342
PH
480 strbuf_addch(&res, '"');
481 quote_c_style(one, &res, NULL, 1);
482 quote_c_style(two, &res, NULL, 1);
483 strbuf_addch(&res, '"');
484 } else {
485 strbuf_addstr(&res, one);
486 strbuf_addstr(&res, two);
6973dcae 487 }
b315c5c0 488 return strbuf_detach(&res, NULL);
6973dcae
JH
489}
490
491static const char *external_diff(void)
492{
493 static const char *external_diff_cmd = NULL;
494 static int done_preparing = 0;
495
496 if (done_preparing)
497 return external_diff_cmd;
6776a84d 498 external_diff_cmd = xstrdup_or_null(getenv("GIT_EXTERNAL_DIFF"));
cbe02100
JS
499 if (!external_diff_cmd)
500 external_diff_cmd = external_diff_cmd_cfg;
6973dcae
JH
501 done_preparing = 1;
502 return external_diff_cmd;
503}
504
284098f1
MH
505/*
506 * Keep track of files used for diffing. Sometimes such an entry
507 * refers to a temporary file, sometimes to an existing file, and
508 * sometimes to "/dev/null".
509 */
6973dcae 510static struct diff_tempfile {
284098f1
MH
511 /*
512 * filename external diff should read from, or NULL if this
513 * entry is currently not in use:
514 */
515 const char *name;
516
dc01505f 517 char hex[GIT_MAX_HEXSZ + 1];
6973dcae 518 char mode[10];
284098f1
MH
519
520 /*
521 * If this diff_tempfile instance refers to a temporary file,
522 * this tempfile object is used to manage its lifetime.
523 */
076aa2cb 524 struct tempfile *tempfile;
6973dcae
JH
525} diff_temp[2];
526
6957eb9a 527struct emit_callback {
6957eb9a
JH
528 int color_diff;
529 unsigned ws_rule;
530 int blank_at_eof_in_preimage;
531 int blank_at_eof_in_postimage;
532 int lno_in_preimage;
533 int lno_in_postimage;
6957eb9a
JH
534 const char **label_path;
535 struct diff_words_data *diff_words;
a3c158d4 536 struct diff_options *opt;
3e97c7c6 537 struct strbuf *header;
6957eb9a
JH
538};
539
6973dcae
JH
540static int count_lines(const char *data, int size)
541{
542 int count, ch, completely_empty = 1, nl_just_seen = 0;
543 count = 0;
544 while (0 < size--) {
545 ch = *data++;
546 if (ch == '\n') {
547 count++;
548 nl_just_seen = 1;
549 completely_empty = 0;
550 }
551 else {
552 nl_just_seen = 0;
553 completely_empty = 0;
554 }
555 }
556 if (completely_empty)
557 return 0;
558 if (!nl_just_seen)
559 count++; /* no trailing newline */
560 return count;
561}
562
b78ea5fc
NTND
563static int fill_mmfile(struct repository *r, mmfile_t *mf,
564 struct diff_filespec *one)
6957eb9a
JH
565{
566 if (!DIFF_FILE_VALID(one)) {
567 mf->ptr = (char *)""; /* does not matter */
568 mf->size = 0;
569 return 0;
570 }
b78ea5fc 571 else if (diff_populate_filespec(r, one, 0))
6957eb9a 572 return -1;
bb35fefb 573
6957eb9a
JH
574 mf->ptr = one->data;
575 mf->size = one->size;
576 return 0;
577}
578
abb371a1 579/* like fill_mmfile, but only for size, so we can avoid retrieving blob */
b78ea5fc
NTND
580static unsigned long diff_filespec_size(struct repository *r,
581 struct diff_filespec *one)
abb371a1
JK
582{
583 if (!DIFF_FILE_VALID(one))
584 return 0;
b78ea5fc 585 diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
abb371a1
JK
586 return one->size;
587}
588
6957eb9a
JH
589static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
590{
591 char *ptr = mf->ptr;
592 long size = mf->size;
593 int cnt = 0;
594
595 if (!size)
596 return cnt;
597 ptr += size - 1; /* pointing at the very end */
598 if (*ptr != '\n')
599 ; /* incomplete line */
600 else
601 ptr--; /* skip the last LF */
602 while (mf->ptr < ptr) {
603 char *prev_eol;
604 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
605 if (*prev_eol == '\n')
606 break;
607 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
608 break;
609 cnt++;
610 ptr = prev_eol - 1;
611 }
612 return cnt;
613}
614
615static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
616 struct emit_callback *ecbdata)
617{
618 int l1, l2, at;
619 unsigned ws_rule = ecbdata->ws_rule;
620 l1 = count_trailing_blank(mf1, ws_rule);
621 l2 = count_trailing_blank(mf2, ws_rule);
622 if (l2 <= l1) {
623 ecbdata->blank_at_eof_in_preimage = 0;
624 ecbdata->blank_at_eof_in_postimage = 0;
625 return;
626 }
627 at = count_lines(mf1->ptr, mf1->size);
628 ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
629
630 at = count_lines(mf2->ptr, mf2->size);
631 ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
632}
633
f7c3b4e2 634static void emit_line_0(struct diff_options *o,
017ac45e 635 const char *set_sign, const char *set, unsigned reverse, const char *reset,
250f7993 636 int first, const char *line, int len)
6957eb9a
JH
637{
638 int has_trailing_newline, has_trailing_carriage_return;
44410679 639 int needs_reset = 0; /* at the end of the line */
a3c158d4
BY
640 FILE *file = o->file;
641
f103a6fa 642 fputs(diff_line_prefix(o), file);
6957eb9a 643
44410679
SB
644 has_trailing_newline = (len > 0 && line[len-1] == '\n');
645 if (has_trailing_newline)
646 len--;
647
648 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
649 if (has_trailing_carriage_return)
650 len--;
651
652 if (!len && !first)
653 goto end_of_line;
654
655 if (reverse && want_color(o->use_color)) {
656 fputs(GIT_COLOR_REVERSE, file);
657 needs_reset = 1;
250f7993 658 }
6957eb9a 659
44410679
SB
660 if (set_sign) {
661 fputs(set_sign, file);
662 needs_reset = 1;
06a47552 663 }
44410679
SB
664
665 if (first)
666 fputc(first, file);
667
668 if (!len)
669 goto end_of_line;
670
671 if (set) {
672 if (set_sign && set != set_sign)
673 fputs(reset, file);
06a47552 674 fputs(set, file);
44410679 675 needs_reset = 1;
06a47552 676 }
44410679
SB
677 fwrite(line, len, 1, file);
678 needs_reset = 1; /* 'line' may contain color codes. */
679
680end_of_line:
681 if (needs_reset)
682 fputs(reset, file);
6957eb9a
JH
683 if (has_trailing_carriage_return)
684 fputc('\r', file);
685 if (has_trailing_newline)
686 fputc('\n', file);
687}
688
a3c158d4 689static void emit_line(struct diff_options *o, const char *set, const char *reset,
250f7993
JH
690 const char *line, int len)
691{
44410679 692 emit_line_0(o, set, NULL, 0, reset, 0, line, len);
250f7993
JH
693}
694
36a4cefd 695enum diff_symbol {
4eed0ebd
SB
696 DIFF_SYMBOL_BINARY_DIFF_HEADER,
697 DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
698 DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
699 DIFF_SYMBOL_BINARY_DIFF_BODY,
700 DIFF_SYMBOL_BINARY_DIFF_FOOTER,
0911c475
SB
701 DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
702 DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
703 DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
704 DIFF_SYMBOL_STATS_LINE,
bd033291 705 DIFF_SYMBOL_WORD_DIFF,
30b7e1e7 706 DIFF_SYMBOL_STAT_SEP,
146fdb0d 707 DIFF_SYMBOL_SUMMARY,
f3597138
SB
708 DIFF_SYMBOL_SUBMODULE_ADD,
709 DIFF_SYMBOL_SUBMODULE_DEL,
710 DIFF_SYMBOL_SUBMODULE_UNTRACKED,
711 DIFF_SYMBOL_SUBMODULE_MODIFIED,
712 DIFF_SYMBOL_SUBMODULE_HEADER,
713 DIFF_SYMBOL_SUBMODULE_ERROR,
714 DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
5af6ea95 715 DIFF_SYMBOL_REWRITE_DIFF,
4acaaa7a 716 DIFF_SYMBOL_BINARY_FILES,
a29b0a13 717 DIFF_SYMBOL_HEADER,
3ee8b7bf
SB
718 DIFF_SYMBOL_FILEPAIR_PLUS,
719 DIFF_SYMBOL_FILEPAIR_MINUS,
ff958679
SB
720 DIFF_SYMBOL_WORDS_PORCELAIN,
721 DIFF_SYMBOL_WORDS,
091f8e28 722 DIFF_SYMBOL_CONTEXT,
f2bb1218 723 DIFF_SYMBOL_CONTEXT_INCOMPLETE,
091f8e28
SB
724 DIFF_SYMBOL_PLUS,
725 DIFF_SYMBOL_MINUS,
b9cbfde6 726 DIFF_SYMBOL_NO_LF_EOF,
68abc6f1 727 DIFF_SYMBOL_CONTEXT_FRAGINFO,
c64b420b 728 DIFF_SYMBOL_CONTEXT_MARKER,
36a4cefd
SB
729 DIFF_SYMBOL_SEPARATOR
730};
091f8e28
SB
731/*
732 * Flags for content lines:
733 * 0..12 are whitespace rules
734 * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
735 * 16 is marking if the line is blank at EOF
736 */
2e2d5ac1
SB
737#define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF (1<<16)
738#define DIFF_SYMBOL_MOVED_LINE (1<<17)
739#define DIFF_SYMBOL_MOVED_LINE_ALT (1<<18)
86b452e2 740#define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING (1<<19)
091f8e28
SB
741#define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
742
e6e045f8
SB
743/*
744 * This struct is used when we need to buffer the output of the diff output.
745 *
746 * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
747 * into the pre/post image file. This pointer could be a union with the
748 * line pointer. By storing an offset into the file instead of the literal line,
749 * we can decrease the memory footprint for the buffered output. At first we
750 * may want to only have indirection for the content lines, but we could also
751 * enhance the state for emitting prefabricated lines, e.g. the similarity
752 * score line or hunk/file headers would only need to store a number or path
753 * and then the output can be constructed later on depending on state.
754 */
755struct emitted_diff_symbol {
756 const char *line;
757 int len;
758 int flags;
21536d07
PW
759 int indent_off; /* Offset to first non-whitespace character */
760 int indent_width; /* The visual width of the indentation */
e6e045f8
SB
761 enum diff_symbol s;
762};
763#define EMITTED_DIFF_SYMBOL_INIT {NULL}
764
765struct emitted_diff_symbols {
766 struct emitted_diff_symbol *buf;
767 int nr, alloc;
768};
769#define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
770
771static void append_emitted_diff_symbol(struct diff_options *o,
772 struct emitted_diff_symbol *e)
6957eb9a 773{
e6e045f8
SB
774 struct emitted_diff_symbol *f;
775
776 ALLOC_GROW(o->emitted_symbols->buf,
777 o->emitted_symbols->nr + 1,
778 o->emitted_symbols->alloc);
779 f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
780
781 memcpy(f, e, sizeof(struct emitted_diff_symbol));
782 f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
6957eb9a
JH
783}
784
2e2d5ac1
SB
785struct moved_entry {
786 struct hashmap_entry ent;
787 const struct emitted_diff_symbol *es;
788 struct moved_entry *next_line;
789};
790
74d156f4
PW
791struct moved_block {
792 struct moved_entry *match;
21536d07 793 int wsd; /* The whitespace delta of this block */
74d156f4
PW
794};
795
796static void moved_block_clear(struct moved_block *b)
797{
21536d07 798 memset(b, 0, sizeof(*b));
74d156f4
PW
799}
800
0cd51e9d
PW
801#define INDENT_BLANKLINE INT_MIN
802
21536d07 803static void fill_es_indent_data(struct emitted_diff_symbol *es)
ca1f4ae4 804{
0cd51e9d 805 unsigned int off = 0, i;
21536d07
PW
806 int width = 0, tab_width = es->flags & WS_TAB_WIDTH_MASK;
807 const char *s = es->line;
808 const int len = es->len;
809
810 /* skip any \v \f \r at start of indentation */
811 while (s[off] == '\f' || s[off] == '\v' ||
812 (s[off] == '\r' && off < len - 1))
813 off++;
814
815 /* calculate the visual width of indentation */
816 while(1) {
817 if (s[off] == ' ') {
818 width++;
819 off++;
820 } else if (s[off] == '\t') {
821 width += tab_width - (width % tab_width);
822 while (s[++off] == '\t')
823 width += tab_width;
824 } else {
825 break;
826 }
827 }
828
0cd51e9d
PW
829 /* check if this line is blank */
830 for (i = off; i < len; i++)
831 if (!isspace(s[i]))
832 break;
ca1f4ae4 833
0cd51e9d
PW
834 if (i == len) {
835 es->indent_width = INDENT_BLANKLINE;
836 es->indent_off = len;
837 } else {
838 es->indent_off = off;
839 es->indent_width = width;
840 }
74d156f4
PW
841}
842
ca1f4ae4 843static int compute_ws_delta(const struct emitted_diff_symbol *a,
21536d07
PW
844 const struct emitted_diff_symbol *b,
845 int *out)
ca1f4ae4 846{
21536d07
PW
847 int a_len = a->len,
848 b_len = b->len,
849 a_off = a->indent_off,
850 a_width = a->indent_width,
851 b_off = b->indent_off,
852 b_width = b->indent_width;
853 int delta;
854
0cd51e9d
PW
855 if (a_width == INDENT_BLANKLINE && b_width == INDENT_BLANKLINE) {
856 *out = INDENT_BLANKLINE;
857 return 1;
858 }
859
21536d07
PW
860 if (a->s == DIFF_SYMBOL_PLUS)
861 delta = a_width - b_width;
862 else
863 delta = b_width - a_width;
ca1f4ae4 864
21536d07
PW
865 if (a_len - a_off != b_len - b_off ||
866 memcmp(a->line + a_off, b->line + b_off, a_len - a_off))
fe4516d1
PW
867 return 0;
868
21536d07 869 *out = delta;
ca1f4ae4 870
fe4516d1 871 return 1;
ca1f4ae4
SB
872}
873
874static int cmp_in_block_with_wsd(const struct diff_options *o,
875 const struct moved_entry *cur,
876 const struct moved_entry *match,
74d156f4 877 struct moved_block *pmb,
ca1f4ae4
SB
878 int n)
879{
880 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
7a4252c4 881 int al = cur->es->len, bl = match->es->len, cl = l->len;
ca1f4ae4
SB
882 const char *a = cur->es->line,
883 *b = match->es->line,
884 *c = l->line;
21536d07
PW
885 int a_off = cur->es->indent_off,
886 a_width = cur->es->indent_width,
887 c_off = l->indent_off,
888 c_width = l->indent_width;
889 int delta;
ca1f4ae4
SB
890
891 /*
7a4252c4
PW
892 * We need to check if 'cur' is equal to 'match'. As those
893 * are from the same (+/-) side, we do not need to adjust for
894 * indent changes. However these were found using fuzzy
895 * matching so we do have to check if they are equal. Here we
896 * just check the lengths. We delay calling memcmp() to check
897 * the contents until later as if the length comparison for a
898 * and c fails we can avoid the call all together.
ca1f4ae4 899 */
7a4252c4 900 if (al != bl)
ca1f4ae4
SB
901 return 1;
902
0cd51e9d
PW
903 /* If 'l' and 'cur' are both blank then they match. */
904 if (a_width == INDENT_BLANKLINE && c_width == INDENT_BLANKLINE)
905 return 0;
ca1f4ae4
SB
906
907 /*
21536d07
PW
908 * The indent changes of the block are known and stored in pmb->wsd;
909 * however we need to check if the indent changes of the current line
910 * match those of the current block and that the text of 'l' and 'cur'
911 * after the indentation match.
ca1f4ae4 912 */
21536d07
PW
913 if (cur->es->s == DIFF_SYMBOL_PLUS)
914 delta = a_width - c_width;
915 else
916 delta = c_width - a_width;
ca1f4ae4 917
0cd51e9d
PW
918 /*
919 * If the previous lines of this block were all blank then set its
920 * whitespace delta.
921 */
922 if (pmb->wsd == INDENT_BLANKLINE)
923 pmb->wsd = delta;
ca1f4ae4 924
21536d07
PW
925 return !(delta == pmb->wsd && al - a_off == cl - c_off &&
926 !memcmp(a, b, al) && !
927 memcmp(a + a_off, c + c_off, al - a_off));
ca1f4ae4 928}
2e2d5ac1 929
ee1df66f
SB
930static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
931 const void *entry,
932 const void *entry_or_key,
2e2d5ac1
SB
933 const void *keydata)
934{
ee1df66f
SB
935 const struct diff_options *diffopt = hashmap_cmp_fn_data;
936 const struct moved_entry *a = entry;
937 const struct moved_entry *b = entry_or_key;
b3095712
SB
938 unsigned flags = diffopt->color_moved_ws_handling
939 & XDF_WHITESPACE_FLAGS;
ee1df66f 940
ca1f4ae4
SB
941 if (diffopt->color_moved_ws_handling &
942 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
943 /*
944 * As there is not specific white space config given,
945 * we'd need to check for a new block, so ignore all
946 * white space. The setup of the white space
947 * configuration for the next block is done else where
948 */
949 flags |= XDF_IGNORE_WHITESPACE;
950
01be97c2
SB
951 return !xdiff_compare_lines(a->es->line, a->es->len,
952 b->es->line, b->es->len,
b3095712 953 flags);
2e2d5ac1
SB
954}
955
956static struct moved_entry *prepare_entry(struct diff_options *o,
957 int line_no)
958{
959 struct moved_entry *ret = xmalloc(sizeof(*ret));
960 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
b3095712 961 unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
2e2d5ac1 962
b3095712 963 ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
2e2d5ac1
SB
964 ret->es = l;
965 ret->next_line = NULL;
966
967 return ret;
968}
969
970static void add_lines_to_move_detection(struct diff_options *o,
971 struct hashmap *add_lines,
972 struct hashmap *del_lines)
973{
974 struct moved_entry *prev_line = NULL;
975
976 int n;
977 for (n = 0; n < o->emitted_symbols->nr; n++) {
978 struct hashmap *hm;
979 struct moved_entry *key;
980
981 switch (o->emitted_symbols->buf[n].s) {
982 case DIFF_SYMBOL_PLUS:
983 hm = add_lines;
984 break;
985 case DIFF_SYMBOL_MINUS:
986 hm = del_lines;
987 break;
988 default:
989 prev_line = NULL;
990 continue;
991 }
992
21536d07
PW
993 if (o->color_moved_ws_handling &
994 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
995 fill_es_indent_data(&o->emitted_symbols->buf[n]);
2e2d5ac1
SB
996 key = prepare_entry(o, n);
997 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
998 prev_line->next_line = key;
999
1000 hashmap_add(hm, key);
1001 prev_line = key;
1002 }
1003}
1004
e2fe6abc
SB
1005static void pmb_advance_or_null(struct diff_options *o,
1006 struct moved_entry *match,
1007 struct hashmap *hm,
74d156f4 1008 struct moved_block *pmb,
e2fe6abc
SB
1009 int pmb_nr)
1010{
1011 int i;
1012 for (i = 0; i < pmb_nr; i++) {
74d156f4 1013 struct moved_entry *prev = pmb[i].match;
e2fe6abc
SB
1014 struct moved_entry *cur = (prev && prev->next_line) ?
1015 prev->next_line : NULL;
1016 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
74d156f4 1017 pmb[i].match = cur;
e2fe6abc 1018 } else {
74d156f4 1019 pmb[i].match = NULL;
e2fe6abc
SB
1020 }
1021 }
1022}
1023
ca1f4ae4
SB
1024static void pmb_advance_or_null_multi_match(struct diff_options *o,
1025 struct moved_entry *match,
1026 struct hashmap *hm,
74d156f4 1027 struct moved_block *pmb,
ca1f4ae4
SB
1028 int pmb_nr, int n)
1029{
1030 int i;
1031 char *got_match = xcalloc(1, pmb_nr);
1032
1033 for (; match; match = hashmap_get_next(hm, match)) {
1034 for (i = 0; i < pmb_nr; i++) {
74d156f4 1035 struct moved_entry *prev = pmb[i].match;
ca1f4ae4
SB
1036 struct moved_entry *cur = (prev && prev->next_line) ?
1037 prev->next_line : NULL;
1038 if (!cur)
1039 continue;
74d156f4 1040 if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
ca1f4ae4
SB
1041 got_match[i] |= 1;
1042 }
1043 }
1044
1045 for (i = 0; i < pmb_nr; i++) {
1046 if (got_match[i]) {
74d156f4
PW
1047 /* Advance to the next line */
1048 pmb[i].match = pmb[i].match->next_line;
fab01ec5 1049 } else {
74d156f4 1050 moved_block_clear(&pmb[i]);
fab01ec5 1051 }
ca1f4ae4 1052 }
9c1a6c2b
PW
1053
1054 free(got_match);
ca1f4ae4
SB
1055}
1056
74d156f4 1057static int shrink_potential_moved_blocks(struct moved_block *pmb,
2e2d5ac1
SB
1058 int pmb_nr)
1059{
1060 int lp, rp;
1061
1062 /* Shrink the set of potential block to the remaining running */
1063 for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
74d156f4 1064 while (lp < pmb_nr && pmb[lp].match)
2e2d5ac1
SB
1065 lp++;
1066 /* lp points at the first NULL now */
1067
74d156f4 1068 while (rp > -1 && !pmb[rp].match)
2e2d5ac1
SB
1069 rp--;
1070 /* rp points at the last non-NULL */
1071
1072 if (lp < pmb_nr && rp > -1 && lp < rp) {
1073 pmb[lp] = pmb[rp];
21536d07 1074 memset(&pmb[rp], 0, sizeof(pmb[rp]));
2e2d5ac1
SB
1075 rp--;
1076 lp++;
1077 }
1078 }
1079
1080 /* Remember the number of running sets */
1081 return rp + 1;
1082}
1083
09153277
JT
1084/*
1085 * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1086 *
f0b8fb6e
JT
1087 * Otherwise, if the last block has fewer alphanumeric characters than
1088 * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
09153277
JT
1089 * that block.
1090 *
1091 * The last block consists of the (n - block_length)'th line up to but not
1092 * including the nth line.
f0b8fb6e 1093 *
b0a2ba47
PW
1094 * Returns 0 if the last block is empty or is unset by this function, non zero
1095 * otherwise.
1096 *
f0b8fb6e
JT
1097 * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1098 * Think of a way to unify them.
09153277 1099 */
b0a2ba47 1100static int adjust_last_block(struct diff_options *o, int n, int block_length)
09153277 1101{
f0b8fb6e
JT
1102 int i, alnum_count = 0;
1103 if (o->color_moved == COLOR_MOVED_PLAIN)
b0a2ba47 1104 return block_length;
f0b8fb6e
JT
1105 for (i = 1; i < block_length + 1; i++) {
1106 const char *c = o->emitted_symbols->buf[n - i].line;
1107 for (; *c; c++) {
1108 if (!isalnum(*c))
1109 continue;
1110 alnum_count++;
1111 if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
b0a2ba47 1112 return 1;
f0b8fb6e
JT
1113 }
1114 }
09153277
JT
1115 for (i = 1; i < block_length + 1; i++)
1116 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
b0a2ba47 1117 return 0;
09153277
JT
1118}
1119
2e2d5ac1
SB
1120/* Find blocks of moved code, delegate actual coloring decision to helper */
1121static void mark_color_as_moved(struct diff_options *o,
1122 struct hashmap *add_lines,
1123 struct hashmap *del_lines)
1124{
74d156f4 1125 struct moved_block *pmb = NULL; /* potentially moved blocks */
2e2d5ac1 1126 int pmb_nr = 0, pmb_alloc = 0;
b0a2ba47 1127 int n, flipped_block = 0, block_length = 0;
2e2d5ac1
SB
1128
1129
1130 for (n = 0; n < o->emitted_symbols->nr; n++) {
1131 struct hashmap *hm = NULL;
1132 struct moved_entry *key;
1133 struct moved_entry *match = NULL;
1134 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
b0a2ba47 1135 enum diff_symbol last_symbol = 0;
2e2d5ac1
SB
1136
1137 switch (l->s) {
1138 case DIFF_SYMBOL_PLUS:
1139 hm = del_lines;
1140 key = prepare_entry(o, n);
3783aad4 1141 match = hashmap_get(hm, key, NULL);
2e2d5ac1
SB
1142 free(key);
1143 break;
1144 case DIFF_SYMBOL_MINUS:
1145 hm = add_lines;
1146 key = prepare_entry(o, n);
3783aad4 1147 match = hashmap_get(hm, key, NULL);
2e2d5ac1
SB
1148 free(key);
1149 break;
1150 default:
b0a2ba47 1151 flipped_block = 0;
2e2d5ac1
SB
1152 }
1153
1154 if (!match) {
74d156f4
PW
1155 int i;
1156
09153277 1157 adjust_last_block(o, n, block_length);
74d156f4
PW
1158 for(i = 0; i < pmb_nr; i++)
1159 moved_block_clear(&pmb[i]);
2e2d5ac1
SB
1160 pmb_nr = 0;
1161 block_length = 0;
b0a2ba47
PW
1162 flipped_block = 0;
1163 last_symbol = l->s;
2e2d5ac1
SB
1164 continue;
1165 }
1166
2034b473 1167 if (o->color_moved == COLOR_MOVED_PLAIN) {
b0a2ba47 1168 last_symbol = l->s;
2034b473 1169 l->flags |= DIFF_SYMBOL_MOVED_LINE;
176841f0 1170 continue;
2034b473 1171 }
176841f0 1172
ca1f4ae4
SB
1173 if (o->color_moved_ws_handling &
1174 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1175 pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1176 else
1177 pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
2e2d5ac1
SB
1178
1179 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1180
1181 if (pmb_nr == 0) {
1182 /*
1183 * The current line is the start of a new block.
1184 * Setup the set of potential blocks.
1185 */
1186 for (; match; match = hashmap_get_next(hm, match)) {
1187 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
ca1f4ae4
SB
1188 if (o->color_moved_ws_handling &
1189 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
74d156f4
PW
1190 if (compute_ws_delta(l, match->es,
1191 &pmb[pmb_nr].wsd))
1192 pmb[pmb_nr++].match = match;
ca1f4ae4 1193 } else {
21536d07 1194 pmb[pmb_nr].wsd = 0;
74d156f4 1195 pmb[pmb_nr++].match = match;
ca1f4ae4 1196 }
2e2d5ac1
SB
1197 }
1198
b0a2ba47
PW
1199 if (adjust_last_block(o, n, block_length) &&
1200 pmb_nr && last_symbol != l->s)
1201 flipped_block = (flipped_block + 1) % 2;
1202 else
1203 flipped_block = 0;
f0b8fb6e 1204
f0b8fb6e 1205 block_length = 0;
2e2d5ac1
SB
1206 }
1207
2034b473
PW
1208 if (pmb_nr) {
1209 block_length++;
2034b473
PW
1210 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1211 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1212 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1213 }
b0a2ba47 1214 last_symbol = l->s;
2e2d5ac1 1215 }
09153277 1216 adjust_last_block(o, n, block_length);
2e2d5ac1 1217
74d156f4
PW
1218 for(n = 0; n < pmb_nr; n++)
1219 moved_block_clear(&pmb[n]);
2e2d5ac1
SB
1220 free(pmb);
1221}
e6e045f8 1222
86b452e2
SB
1223#define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1224 (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1225static void dim_moved_lines(struct diff_options *o)
1226{
1227 int n;
1228 for (n = 0; n < o->emitted_symbols->nr; n++) {
1229 struct emitted_diff_symbol *prev = (n != 0) ?
1230 &o->emitted_symbols->buf[n - 1] : NULL;
1231 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1232 struct emitted_diff_symbol *next =
1233 (n < o->emitted_symbols->nr - 1) ?
1234 &o->emitted_symbols->buf[n + 1] : NULL;
1235
1236 /* Not a plus or minus line? */
1237 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1238 continue;
1239
1240 /* Not a moved line? */
1241 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1242 continue;
1243
1244 /*
1245 * If prev or next are not a plus or minus line,
1246 * pretend they don't exist
1247 */
1248 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1249 prev->s != DIFF_SYMBOL_MINUS)
1250 prev = NULL;
1251 if (next && next->s != DIFF_SYMBOL_PLUS &&
1252 next->s != DIFF_SYMBOL_MINUS)
1253 next = NULL;
1254
1255 /* Inside a block? */
1256 if ((prev &&
1257 (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1258 (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1259 (next &&
1260 (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1261 (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1262 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1263 continue;
1264 }
1265
1266 /* Check if we are at an interesting bound: */
1267 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1268 (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1269 (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1270 continue;
1271 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1272 (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1273 (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1274 continue;
1275
1276 /*
1277 * The boundary to prev and next are not interesting,
1278 * so this line is not interesting as a whole
1279 */
1280 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1281 }
1282}
1283
091f8e28 1284static void emit_line_ws_markup(struct diff_options *o,
9d1e16bf
SB
1285 const char *set_sign, const char *set,
1286 const char *reset,
bc9feb05 1287 int sign_index, const char *line, int len,
091f8e28 1288 unsigned ws_rule, int blank_at_eof)
6957eb9a 1289{
b8767f79 1290 const char *ws = NULL;
bc9feb05 1291 int sign = o->output_indicators[sign_index];
6957eb9a 1292
091f8e28
SB
1293 if (o->ws_error_highlight & ws_rule) {
1294 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
b8767f79
JH
1295 if (!*ws)
1296 ws = NULL;
1297 }
1298
f7c3b4e2 1299 if (!ws && !set_sign)
017ac45e 1300 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
f7c3b4e2 1301 else if (!ws) {
29ef759d 1302 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
f7c3b4e2 1303 } else if (blank_at_eof)
6957eb9a 1304 /* Blank line at EOF - paint '+' as well */
017ac45e 1305 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
6957eb9a
JH
1306 else {
1307 /* Emit just the prefix, then the rest. */
017ac45e 1308 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
f7c3b4e2 1309 sign, "", 0);
091f8e28
SB
1310 ws_check_emit(line, len, ws_rule,
1311 o->file, set, reset, ws);
6957eb9a
JH
1312 }
1313}
1314
e6e045f8
SB
1315static void emit_diff_symbol_from_struct(struct diff_options *o,
1316 struct emitted_diff_symbol *eds)
36a4cefd 1317{
b9cbfde6 1318 static const char *nneof = " No newline at end of file\n";
f7c3b4e2 1319 const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
0911c475 1320 struct strbuf sb = STRBUF_INIT;
e6e045f8
SB
1321
1322 enum diff_symbol s = eds->s;
1323 const char *line = eds->line;
1324 int len = eds->len;
1325 unsigned flags = eds->flags;
1326
36a4cefd 1327 switch (s) {
b9cbfde6
SB
1328 case DIFF_SYMBOL_NO_LF_EOF:
1329 context = diff_get_color_opt(o, DIFF_CONTEXT);
1330 reset = diff_get_color_opt(o, DIFF_RESET);
1331 putc('\n', o->file);
017ac45e 1332 emit_line_0(o, context, NULL, 0, reset, '\\',
b9cbfde6
SB
1333 nneof, strlen(nneof));
1334 break;
f3597138
SB
1335 case DIFF_SYMBOL_SUBMODULE_HEADER:
1336 case DIFF_SYMBOL_SUBMODULE_ERROR:
1337 case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
0911c475 1338 case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
146fdb0d 1339 case DIFF_SYMBOL_SUMMARY:
0911c475 1340 case DIFF_SYMBOL_STATS_LINE:
4eed0ebd 1341 case DIFF_SYMBOL_BINARY_DIFF_BODY:
68abc6f1
SB
1342 case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1343 emit_line(o, "", "", line, len);
1344 break;
f2bb1218 1345 case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
c64b420b
SB
1346 case DIFF_SYMBOL_CONTEXT_MARKER:
1347 context = diff_get_color_opt(o, DIFF_CONTEXT);
1348 reset = diff_get_color_opt(o, DIFF_RESET);
1349 emit_line(o, context, reset, line, len);
1350 break;
36a4cefd
SB
1351 case DIFF_SYMBOL_SEPARATOR:
1352 fprintf(o->file, "%s%c",
1353 diff_line_prefix(o),
1354 o->line_termination);
1355 break;
091f8e28
SB
1356 case DIFF_SYMBOL_CONTEXT:
1357 set = diff_get_color_opt(o, DIFF_CONTEXT);
1358 reset = diff_get_color_opt(o, DIFF_RESET);
f7c3b4e2
JS
1359 set_sign = NULL;
1360 if (o->flags.dual_color_diffed_diffs) {
1361 char c = !len ? 0 : line[0];
1362
1363 if (c == '+')
1364 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1365 else if (c == '@')
1366 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1367 else if (c == '-')
1368 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1369 }
7648b79e 1370 emit_line_ws_markup(o, set_sign, set, reset,
bc9feb05 1371 OUTPUT_INDICATOR_CONTEXT, line, len,
091f8e28
SB
1372 flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1373 break;
1374 case DIFF_SYMBOL_PLUS:
86b452e2
SB
1375 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1376 DIFF_SYMBOL_MOVED_LINE_ALT |
1377 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1378 case DIFF_SYMBOL_MOVED_LINE |
1379 DIFF_SYMBOL_MOVED_LINE_ALT |
1380 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1381 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1382 break;
1383 case DIFF_SYMBOL_MOVED_LINE |
1384 DIFF_SYMBOL_MOVED_LINE_ALT:
2e2d5ac1 1385 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
86b452e2
SB
1386 break;
1387 case DIFF_SYMBOL_MOVED_LINE |
1388 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1389 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1390 break;
1391 case DIFF_SYMBOL_MOVED_LINE:
2e2d5ac1 1392 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
86b452e2
SB
1393 break;
1394 default:
2e2d5ac1 1395 set = diff_get_color_opt(o, DIFF_FILE_NEW);
86b452e2 1396 }
091f8e28 1397 reset = diff_get_color_opt(o, DIFF_RESET);
f7c3b4e2
JS
1398 if (!o->flags.dual_color_diffed_diffs)
1399 set_sign = NULL;
1400 else {
1401 char c = !len ? 0 : line[0];
1402
1403 set_sign = set;
1404 if (c == '-')
a7be92ac 1405 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
f7c3b4e2
JS
1406 else if (c == '@')
1407 set = diff_get_color_opt(o, DIFF_FRAGINFO);
a7be92ac
JS
1408 else if (c == '+')
1409 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1410 else
1411 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
0b91faa0 1412 flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
f7c3b4e2 1413 }
7648b79e 1414 emit_line_ws_markup(o, set_sign, set, reset,
bc9feb05 1415 OUTPUT_INDICATOR_NEW, line, len,
091f8e28
SB
1416 flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1417 flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1418 break;
1419 case DIFF_SYMBOL_MINUS:
86b452e2
SB
1420 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1421 DIFF_SYMBOL_MOVED_LINE_ALT |
1422 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1423 case DIFF_SYMBOL_MOVED_LINE |
1424 DIFF_SYMBOL_MOVED_LINE_ALT |
1425 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1426 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1427 break;
1428 case DIFF_SYMBOL_MOVED_LINE |
1429 DIFF_SYMBOL_MOVED_LINE_ALT:
2e2d5ac1 1430 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
86b452e2
SB
1431 break;
1432 case DIFF_SYMBOL_MOVED_LINE |
1433 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1434 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1435 break;
1436 case DIFF_SYMBOL_MOVED_LINE:
2e2d5ac1 1437 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
86b452e2
SB
1438 break;
1439 default:
2e2d5ac1 1440 set = diff_get_color_opt(o, DIFF_FILE_OLD);
86b452e2 1441 }
091f8e28 1442 reset = diff_get_color_opt(o, DIFF_RESET);
f7c3b4e2
JS
1443 if (!o->flags.dual_color_diffed_diffs)
1444 set_sign = NULL;
1445 else {
1446 char c = !len ? 0 : line[0];
1447
1448 set_sign = set;
1449 if (c == '+')
a7be92ac 1450 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
f7c3b4e2
JS
1451 else if (c == '@')
1452 set = diff_get_color_opt(o, DIFF_FRAGINFO);
a7be92ac
JS
1453 else if (c == '-')
1454 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1455 else
1456 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
f7c3b4e2 1457 }
7648b79e 1458 emit_line_ws_markup(o, set_sign, set, reset,
bc9feb05 1459 OUTPUT_INDICATOR_OLD, line, len,
091f8e28
SB
1460 flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1461 break;
ff958679
SB
1462 case DIFF_SYMBOL_WORDS_PORCELAIN:
1463 context = diff_get_color_opt(o, DIFF_CONTEXT);
1464 reset = diff_get_color_opt(o, DIFF_RESET);
1465 emit_line(o, context, reset, line, len);
1466 fputs("~\n", o->file);
1467 break;
1468 case DIFF_SYMBOL_WORDS:
1469 context = diff_get_color_opt(o, DIFF_CONTEXT);
1470 reset = diff_get_color_opt(o, DIFF_RESET);
1471 /*
1472 * Skip the prefix character, if any. With
1473 * diff_suppress_blank_empty, there may be
1474 * none.
1475 */
1476 if (line[0] != '\n') {
1477 line++;
1478 len--;
1479 }
1480 emit_line(o, context, reset, line, len);
1481 break;
3ee8b7bf
SB
1482 case DIFF_SYMBOL_FILEPAIR_PLUS:
1483 meta = diff_get_color_opt(o, DIFF_METAINFO);
1484 reset = diff_get_color_opt(o, DIFF_RESET);
1485 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1486 line, reset,
1487 strchr(line, ' ') ? "\t" : "");
1488 break;
1489 case DIFF_SYMBOL_FILEPAIR_MINUS:
1490 meta = diff_get_color_opt(o, DIFF_METAINFO);
1491 reset = diff_get_color_opt(o, DIFF_RESET);
1492 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1493 line, reset,
1494 strchr(line, ' ') ? "\t" : "");
1495 break;
4acaaa7a 1496 case DIFF_SYMBOL_BINARY_FILES:
a29b0a13
SB
1497 case DIFF_SYMBOL_HEADER:
1498 fprintf(o->file, "%s", line);
1499 break;
4eed0ebd
SB
1500 case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1501 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1502 break;
1503 case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1504 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1505 break;
1506 case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1507 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1508 break;
1509 case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1510 fputs(diff_line_prefix(o), o->file);
1511 fputc('\n', o->file);
1512 break;
5af6ea95
SB
1513 case DIFF_SYMBOL_REWRITE_DIFF:
1514 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1515 reset = diff_get_color_opt(o, DIFF_RESET);
1516 emit_line(o, fraginfo, reset, line, len);
1517 break;
f3597138
SB
1518 case DIFF_SYMBOL_SUBMODULE_ADD:
1519 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1520 reset = diff_get_color_opt(o, DIFF_RESET);
1521 emit_line(o, set, reset, line, len);
1522 break;
1523 case DIFF_SYMBOL_SUBMODULE_DEL:
1524 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1525 reset = diff_get_color_opt(o, DIFF_RESET);
1526 emit_line(o, set, reset, line, len);
1527 break;
1528 case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1529 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1530 diff_line_prefix(o), line);
1531 break;
1532 case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1533 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1534 diff_line_prefix(o), line);
1535 break;
0911c475
SB
1536 case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1537 emit_line(o, "", "", " 0 files changed\n",
1538 strlen(" 0 files changed\n"));
1539 break;
1540 case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1541 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1542 break;
bd033291
SB
1543 case DIFF_SYMBOL_WORD_DIFF:
1544 fprintf(o->file, "%.*s", len, line);
1545 break;
30b7e1e7
SB
1546 case DIFF_SYMBOL_STAT_SEP:
1547 fputs(o->stat_sep, o->file);
1548 break;
36a4cefd 1549 default:
033abf97 1550 BUG("unknown diff symbol");
36a4cefd 1551 }
0911c475 1552 strbuf_release(&sb);
36a4cefd
SB
1553}
1554
e6e045f8
SB
1555static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1556 const char *line, int len, unsigned flags)
1557{
21536d07 1558 struct emitted_diff_symbol e = {line, len, flags, 0, 0, s};
e6e045f8
SB
1559
1560 if (o->emitted_symbols)
1561 append_emitted_diff_symbol(o, &e);
1562 else
1563 emit_diff_symbol_from_struct(o, &e);
1564}
1565
f3597138
SB
1566void diff_emit_submodule_del(struct diff_options *o, const char *line)
1567{
1568 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1569}
1570
1571void diff_emit_submodule_add(struct diff_options *o, const char *line)
1572{
1573 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1574}
1575
1576void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1577{
1578 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1579 path, strlen(path), 0);
1580}
1581
1582void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1583{
1584 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1585 path, strlen(path), 0);
1586}
1587
1588void diff_emit_submodule_header(struct diff_options *o, const char *header)
1589{
1590 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1591 header, strlen(header), 0);
1592}
1593
1594void diff_emit_submodule_error(struct diff_options *o, const char *err)
1595{
1596 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1597}
1598
1599void diff_emit_submodule_pipethrough(struct diff_options *o,
1600 const char *line, int len)
1601{
1602 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1603}
1604
6957eb9a
JH
1605static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1606{
1607 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1608 ecbdata->blank_at_eof_in_preimage &&
1609 ecbdata->blank_at_eof_in_postimage &&
1610 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1611 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1612 return 0;
018cff70 1613 return ws_blank_line(line, len, ecbdata->ws_rule);
6957eb9a
JH
1614}
1615
b8767f79 1616static void emit_add_line(const char *reset,
0e383e18
JH
1617 struct emit_callback *ecbdata,
1618 const char *line, int len)
1619{
091f8e28
SB
1620 unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1621 if (new_blank_line_at_eof(ecbdata, line, len))
1622 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1623
1624 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
b8767f79 1625}
0e383e18 1626
b8767f79
JH
1627static void emit_del_line(const char *reset,
1628 struct emit_callback *ecbdata,
1629 const char *line, int len)
1630{
091f8e28
SB
1631 unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1632 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
0e383e18
JH
1633}
1634
1635static void emit_context_line(const char *reset,
1636 struct emit_callback *ecbdata,
1637 const char *line, int len)
1638{
091f8e28
SB
1639 unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1640 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
0e383e18
JH
1641}
1642
89cb73a1
BW
1643static void emit_hunk_header(struct emit_callback *ecbdata,
1644 const char *line, int len)
1645{
8dbf3eb6 1646 const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
89cb73a1
BW
1647 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1648 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1649 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
f7c3b4e2 1650 const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
89cb73a1
BW
1651 static const char atat[2] = { '@', '@' };
1652 const char *cp, *ep;
2efcc977
BY
1653 struct strbuf msgbuf = STRBUF_INIT;
1654 int org_len = len;
1655 int i = 1;
89cb73a1
BW
1656
1657 /*
1658 * As a hunk header must begin with "@@ -<old>, +<new> @@",
1659 * it always is at least 10 bytes long.
1660 */
1661 if (len < 10 ||
1662 memcmp(line, atat, 2) ||
1663 !(ep = memmem(line + 2, len - 2, atat, 2))) {
c64b420b 1664 emit_diff_symbol(ecbdata->opt,
091f8e28 1665 DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
89cb73a1
BW
1666 return;
1667 }
1668 ep += 2; /* skip over @@ */
1669
1670 /* The hunk header in fraginfo color */
f7c3b4e2
JS
1671 if (ecbdata->opt->flags.dual_color_diffed_diffs)
1672 strbuf_addstr(&msgbuf, reverse);
cedc61a9 1673 strbuf_addstr(&msgbuf, frag);
2efcc977 1674 strbuf_add(&msgbuf, line, ep - line);
cedc61a9 1675 strbuf_addstr(&msgbuf, reset);
2efcc977
BY
1676
1677 /*
1678 * trailing "\r\n"
1679 */
1680 for ( ; i < 3; i++)
1681 if (line[len - i] == '\r' || line[len - i] == '\n')
1682 len--;
89cb73a1
BW
1683
1684 /* blank before the func header */
1685 for (cp = ep; ep - line < len; ep++)
1686 if (*ep != ' ' && *ep != '\t')
1687 break;
2efcc977 1688 if (ep != cp) {
8dbf3eb6 1689 strbuf_addstr(&msgbuf, context);
2efcc977 1690 strbuf_add(&msgbuf, cp, ep - cp);
cedc61a9 1691 strbuf_addstr(&msgbuf, reset);
2efcc977
BY
1692 }
1693
1694 if (ep < line + len) {
cedc61a9 1695 strbuf_addstr(&msgbuf, func);
2efcc977 1696 strbuf_add(&msgbuf, ep, line + len - ep);
cedc61a9 1697 strbuf_addstr(&msgbuf, reset);
2efcc977 1698 }
89cb73a1 1699
2efcc977 1700 strbuf_add(&msgbuf, line + len, org_len - len);
dfb7728f 1701 strbuf_complete_line(&msgbuf);
68abc6f1 1702 emit_diff_symbol(ecbdata->opt,
091f8e28 1703 DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
2efcc977 1704 strbuf_release(&msgbuf);
89cb73a1
BW
1705}
1706
3b335762
NTND
1707static struct diff_tempfile *claim_diff_tempfile(void)
1708{
479b0ae8
JK
1709 int i;
1710 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1711 if (!diff_temp[i].name)
1712 return diff_temp + i;
033abf97 1713 BUG("diff is failing to clean up its tempfiles");
479b0ae8
JK
1714}
1715
479b0ae8
JK
1716static void remove_tempfile(void)
1717{
1718 int i;
a8344abe 1719 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
076aa2cb 1720 if (is_tempfile_active(diff_temp[i].tempfile))
284098f1 1721 delete_tempfile(&diff_temp[i].tempfile);
a8344abe
NR
1722 diff_temp[i].name = NULL;
1723 }
479b0ae8
JK
1724}
1725
5af6ea95 1726static void add_line_count(struct strbuf *out, int count)
6973dcae
JH
1727{
1728 switch (count) {
1729 case 0:
5af6ea95 1730 strbuf_addstr(out, "0,0");
6973dcae
JH
1731 break;
1732 case 1:
5af6ea95 1733 strbuf_addstr(out, "1");
6973dcae
JH
1734 break;
1735 default:
5af6ea95 1736 strbuf_addf(out, "1,%d", count);
6973dcae
JH
1737 break;
1738 }
1739}
1740
7f7ee2ff
JH
1741static void emit_rewrite_lines(struct emit_callback *ecb,
1742 int prefix, const char *data, int size)
6973dcae 1743{
7f7ee2ff 1744 const char *endp = NULL;
7f7ee2ff
JH
1745 const char *reset = diff_get_color(ecb->color_diff, DIFF_RESET);
1746
1747 while (0 < size) {
1748 int len;
1749
1750 endp = memchr(data, '\n', size);
1751 len = endp ? (endp - data + 1) : size;
1752 if (prefix != '+') {
1753 ecb->lno_in_preimage++;
0e383e18 1754 emit_del_line(reset, ecb, data, len);
7f7ee2ff
JH
1755 } else {
1756 ecb->lno_in_postimage++;
1757 emit_add_line(reset, ecb, data, len);
13e36ec5 1758 }
7f7ee2ff
JH
1759 size -= len;
1760 data += len;
1761 }
b9cbfde6 1762 if (!endp)
091f8e28 1763 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
6973dcae
JH
1764}
1765
1766static void emit_rewrite_diff(const char *name_a,
1767 const char *name_b,
1768 struct diff_filespec *one,
13e36ec5 1769 struct diff_filespec *two,
d9bae1a1
JK
1770 struct userdiff_driver *textconv_one,
1771 struct userdiff_driver *textconv_two,
eab9a40b 1772 struct diff_options *o)
6973dcae
JH
1773{
1774 int lc_a, lc_b;
d5625091 1775 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
a5a818ee 1776 const char *a_prefix, *b_prefix;
840383b2 1777 char *data_one, *data_two;
3aa1f7ca 1778 size_t size_one, size_two;
7f7ee2ff 1779 struct emit_callback ecbdata;
5af6ea95 1780 struct strbuf out = STRBUF_INIT;
a5a818ee 1781
0d1e0e78 1782 if (diff_mnemonic_prefix && o->flags.reverse_diff) {
a5a818ee
JH
1783 a_prefix = o->b_prefix;
1784 b_prefix = o->a_prefix;
1785 } else {
1786 a_prefix = o->a_prefix;
1787 b_prefix = o->b_prefix;
1788 }
1a9eb3b9 1789
8a13becc
JH
1790 name_a += (*name_a == '/');
1791 name_b += (*name_b == '/');
1a9eb3b9 1792
d5625091
JH
1793 strbuf_reset(&a_name);
1794 strbuf_reset(&b_name);
a5a818ee
JH
1795 quote_two_c_style(&a_name, a_prefix, name_a, 0);
1796 quote_two_c_style(&b_name, b_prefix, name_b, 0);
d5625091 1797
6afaf807
NTND
1798 size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1799 size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
3aa1f7ca 1800
d91ba8fa 1801 memset(&ecbdata, 0, sizeof(ecbdata));
daa0c3d9 1802 ecbdata.color_diff = want_color(o->use_color);
26d024ec 1803 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
a3c158d4 1804 ecbdata.opt = o;
d91ba8fa
JH
1805 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1806 mmfile_t mf1, mf2;
1807 mf1.ptr = (char *)data_one;
1808 mf2.ptr = (char *)data_two;
1809 mf1.size = size_one;
1810 mf2.size = size_two;
1811 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1812 }
1813 ecbdata.lno_in_preimage = 1;
1814 ecbdata.lno_in_postimage = 1;
1815
3aa1f7ca
JK
1816 lc_a = count_lines(data_one, size_one);
1817 lc_b = count_lines(data_two, size_two);
3ee8b7bf
SB
1818
1819 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1820 a_name.buf, a_name.len, 0);
1821 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1822 b_name.buf, b_name.len, 0);
1823
5af6ea95 1824 strbuf_addstr(&out, "@@ -");
467ddc14 1825 if (!o->irreversible_delete)
5af6ea95 1826 add_line_count(&out, lc_a);
467ddc14 1827 else
5af6ea95
SB
1828 strbuf_addstr(&out, "?,?");
1829 strbuf_addstr(&out, " +");
1830 add_line_count(&out, lc_b);
1831 strbuf_addstr(&out, " @@\n");
1832 emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1833 strbuf_release(&out);
1834
467ddc14 1835 if (lc_a && !o->irreversible_delete)
d91ba8fa 1836 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
6973dcae 1837 if (lc_b)
d91ba8fa 1838 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
b76c056b 1839 if (textconv_one)
aed6ca52 1840 free((char *)data_one);
b76c056b 1841 if (textconv_two)
aed6ca52 1842 free((char *)data_two);
6973dcae
JH
1843}
1844
f59a59e2
JS
1845struct diff_words_buffer {
1846 mmfile_t text;
071bcaab 1847 unsigned long alloc;
2e5d2003
JS
1848 struct diff_words_orig {
1849 const char *begin, *end;
1850 } *orig;
1851 int orig_nr, orig_alloc;
f59a59e2
JS
1852};
1853
1854static void diff_words_append(char *line, unsigned long len,
1855 struct diff_words_buffer *buffer)
1856{
23c1575f 1857 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
f59a59e2
JS
1858 line++;
1859 len--;
1860 memcpy(buffer->text.ptr + buffer->text.size, line, len);
1861 buffer->text.size += len;
2b6a5417 1862 buffer->text.ptr[buffer->text.size] = '\0';
f59a59e2
JS
1863}
1864
9cba13ca 1865struct diff_words_style_elem {
882749a0
TR
1866 const char *prefix;
1867 const char *suffix;
1868 const char *color; /* NULL; filled in by the setup code if
1869 * color is enabled */
1870};
1871
9cba13ca 1872struct diff_words_style {
882749a0 1873 enum diff_words_type type;
63a01c3f 1874 struct diff_words_style_elem new_word, old_word, ctx;
882749a0
TR
1875 const char *newline;
1876};
1877
c2e86add 1878static struct diff_words_style diff_words_styles[] = {
882749a0
TR
1879 { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1880 { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1881 { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1882};
1883
f59a59e2 1884struct diff_words_data {
f59a59e2 1885 struct diff_words_buffer minus, plus;
2e5d2003 1886 const char *current_plus;
4297c0ae
BY
1887 int last_minus;
1888 struct diff_options *opt;
2b6a5417 1889 regex_t *word_regex;
882749a0
TR
1890 enum diff_words_type type;
1891 struct diff_words_style *style;
f59a59e2
JS
1892};
1893
bd033291 1894static int fn_out_diff_words_write_helper(struct diff_options *o,
882749a0
TR
1895 struct diff_words_style_elem *st_el,
1896 const char *newline,
bd033291 1897 size_t count, const char *buf)
882749a0 1898{
4297c0ae 1899 int print = 0;
bd033291 1900 struct strbuf sb = STRBUF_INIT;
4297c0ae 1901
882749a0
TR
1902 while (count) {
1903 char *p = memchr(buf, '\n', count);
4297c0ae 1904 if (print)
bd033291
SB
1905 strbuf_addstr(&sb, diff_line_prefix(o));
1906
882749a0 1907 if (p != buf) {
bd033291
SB
1908 const char *reset = st_el->color && *st_el->color ?
1909 GIT_COLOR_RESET : NULL;
1910 if (st_el->color && *st_el->color)
1911 strbuf_addstr(&sb, st_el->color);
1912 strbuf_addstr(&sb, st_el->prefix);
1913 strbuf_add(&sb, buf, p ? p - buf : count);
1914 strbuf_addstr(&sb, st_el->suffix);
1915 if (reset)
1916 strbuf_addstr(&sb, reset);
882749a0
TR
1917 }
1918 if (!p)
bd033291
SB
1919 goto out;
1920
1921 strbuf_addstr(&sb, newline);
882749a0
TR
1922 count -= p + 1 - buf;
1923 buf = p + 1;
4297c0ae 1924 print = 1;
bd033291
SB
1925 if (count) {
1926 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1927 sb.buf, sb.len, 0);
1928 strbuf_reset(&sb);
1929 }
882749a0 1930 }
bd033291
SB
1931
1932out:
1933 if (sb.len)
1934 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1935 sb.buf, sb.len, 0);
1936 strbuf_release(&sb);
882749a0
TR
1937 return 0;
1938}
1939
4297c0ae
BY
1940/*
1941 * '--color-words' algorithm can be described as:
1942 *
5621760f 1943 * 1. collect the minus/plus lines of a diff hunk, divided into
4297c0ae
BY
1944 * minus-lines and plus-lines;
1945 *
1946 * 2. break both minus-lines and plus-lines into words and
1947 * place them into two mmfile_t with one word for each line;
1948 *
1949 * 3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1950 *
1951 * And for the common parts of the both file, we output the plus side text.
1952 * diff_words->current_plus is used to trace the current position of the plus file
1953 * which printed. diff_words->last_minus is used to trace the last minus word
1954 * printed.
1955 *
1956 * For '--graph' to work with '--color-words', we need to output the graph prefix
1957 * on each line of color words output. Generally, there are two conditions on
1958 * which we should output the prefix.
1959 *
1960 * 1. diff_words->last_minus == 0 &&
1961 * diff_words->current_plus == diff_words->plus.text.ptr
1962 *
1963 * that is: the plus text must start as a new line, and if there is no minus
1964 * word printed, a graph prefix must be printed.
1965 *
1966 * 2. diff_words->current_plus > diff_words->plus.text.ptr &&
1967 * *(diff_words->current_plus - 1) == '\n'
1968 *
1969 * that is: a graph prefix must be printed following a '\n'
1970 */
1971static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1972{
1973 if ((diff_words->last_minus == 0 &&
1974 diff_words->current_plus == diff_words->plus.text.ptr) ||
1975 (diff_words->current_plus > diff_words->plus.text.ptr &&
1976 *(diff_words->current_plus - 1) == '\n')) {
1977 return 1;
1978 } else {
1979 return 0;
1980 }
1981}
1982
7c61e25f
JK
1983static void fn_out_diff_words_aux(void *priv,
1984 long minus_first, long minus_len,
1985 long plus_first, long plus_len,
1986 const char *func, long funclen)
f59a59e2 1987{
f59a59e2 1988 struct diff_words_data *diff_words = priv;
882749a0 1989 struct diff_words_style *style = diff_words->style;
2e5d2003 1990 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
4297c0ae 1991 struct diff_options *opt = diff_words->opt;
30997bb8 1992 const char *line_prefix;
f59a59e2 1993
4297c0ae 1994 assert(opt);
30997bb8 1995 line_prefix = diff_line_prefix(opt);
4297c0ae 1996
2e5d2003
JS
1997 /* POSIX requires that first be decremented by one if len == 0... */
1998 if (minus_len) {
1999 minus_begin = diff_words->minus.orig[minus_first].begin;
2000 minus_end =
2001 diff_words->minus.orig[minus_first + minus_len - 1].end;
2002 } else
2003 minus_begin = minus_end =
2004 diff_words->minus.orig[minus_first].end;
2005
2006 if (plus_len) {
2007 plus_begin = diff_words->plus.orig[plus_first].begin;
2008 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
2009 } else
2010 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
2011
4297c0ae
BY
2012 if (color_words_output_graph_prefix(diff_words)) {
2013 fputs(line_prefix, diff_words->opt->file);
2014 }
2015 if (diff_words->current_plus != plus_begin) {
bd033291 2016 fn_out_diff_words_write_helper(diff_words->opt,
882749a0
TR
2017 &style->ctx, style->newline,
2018 plus_begin - diff_words->current_plus,
bd033291 2019 diff_words->current_plus);
4297c0ae
BY
2020 }
2021 if (minus_begin != minus_end) {
bd033291 2022 fn_out_diff_words_write_helper(diff_words->opt,
63a01c3f 2023 &style->old_word, style->newline,
bd033291 2024 minus_end - minus_begin, minus_begin);
4297c0ae
BY
2025 }
2026 if (plus_begin != plus_end) {
bd033291 2027 fn_out_diff_words_write_helper(diff_words->opt,
63a01c3f 2028 &style->new_word, style->newline,
bd033291 2029 plus_end - plus_begin, plus_begin);
4297c0ae 2030 }
2e5d2003
JS
2031
2032 diff_words->current_plus = plus_end;
4297c0ae 2033 diff_words->last_minus = minus_first;
f59a59e2
JS
2034}
2035
2b6a5417
JS
2036/* This function starts looking at *begin, and returns 0 iff a word was found. */
2037static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2038 int *begin, int *end)
2039{
2040 if (word_regex && *begin < buffer->size) {
2041 regmatch_t match[1];
b7d36ffc
JS
2042 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2043 buffer->size - *begin, 1, match, 0)) {
2b6a5417
JS
2044 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2045 '\n', match[0].rm_eo - match[0].rm_so);
2046 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2047 *begin += match[0].rm_so;
2048 return *begin >= *end;
2049 }
2050 return -1;
f59a59e2
JS
2051 }
2052
2b6a5417
JS
2053 /* find the next word */
2054 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2055 (*begin)++;
2056 if (*begin >= buffer->size)
2057 return -1;
f59a59e2 2058
2b6a5417
JS
2059 /* find the end of the word */
2060 *end = *begin + 1;
2061 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2062 (*end)++;
2063
2064 return 0;
f59a59e2
JS
2065}
2066
23c1575f 2067/*
2e5d2003
JS
2068 * This function splits the words in buffer->text, stores the list with
2069 * newline separator into out, and saves the offsets of the original words
2070 * in buffer->orig.
23c1575f 2071 */
2b6a5417
JS
2072static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2073 regex_t *word_regex)
f59a59e2 2074{
2e5d2003 2075 int i, j;
2b6a5417 2076 long alloc = 0;
f59a59e2 2077
2e5d2003 2078 out->size = 0;
2b6a5417 2079 out->ptr = NULL;
f59a59e2 2080
2e5d2003
JS
2081 /* fake an empty "0th" word */
2082 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2083 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2084 buffer->orig_nr = 1;
2085
2086 for (i = 0; i < buffer->text.size; i++) {
2b6a5417
JS
2087 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2088 return;
2e5d2003
JS
2089
2090 /* store original boundaries */
2091 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2092 buffer->orig_alloc);
2093 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2094 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2095 buffer->orig_nr++;
2096
2097 /* store one word */
2b6a5417 2098 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2e5d2003
JS
2099 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2100 out->ptr[out->size + j - i] = '\n';
2101 out->size += j - i + 1;
2102
2103 i = j - 1;
f59a59e2
JS
2104 }
2105}
2106
2107/* this executes the word diff on the accumulated buffers */
2108static void diff_words_show(struct diff_words_data *diff_words)
2109{
2110 xpparam_t xpp;
2111 xdemitconf_t xecfg;
f59a59e2 2112 mmfile_t minus, plus;
882749a0 2113 struct diff_words_style *style = diff_words->style;
f59a59e2 2114
4297c0ae 2115 struct diff_options *opt = diff_words->opt;
30997bb8 2116 const char *line_prefix;
4297c0ae
BY
2117
2118 assert(opt);
30997bb8 2119 line_prefix = diff_line_prefix(opt);
4297c0ae 2120
2e5d2003
JS
2121 /* special case: only removal */
2122 if (!diff_words->plus.text.size) {
bd033291
SB
2123 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2124 line_prefix, strlen(line_prefix), 0);
2125 fn_out_diff_words_write_helper(diff_words->opt,
63a01c3f 2126 &style->old_word, style->newline,
4297c0ae 2127 diff_words->minus.text.size,
bd033291 2128 diff_words->minus.text.ptr);
2e5d2003
JS
2129 diff_words->minus.text.size = 0;
2130 return;
2131 }
2132
2133 diff_words->current_plus = diff_words->plus.text.ptr;
4297c0ae 2134 diff_words->last_minus = 0;
f59a59e2 2135
9ccd0a88 2136 memset(&xpp, 0, sizeof(xpp));
30b25010 2137 memset(&xecfg, 0, sizeof(xecfg));
2b6a5417
JS
2138 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2139 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
582aa00b 2140 xpp.flags = 0;
2b6a5417 2141 /* as only the hunk header will be parsed, we need a 0-context */
2e5d2003 2142 xecfg.ctxlen = 0;
7c61e25f 2143 if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
9346d6d1 2144 diff_words, &xpp, &xecfg))
3efb9880 2145 die("unable to generate word diff");
f59a59e2
JS
2146 free(minus.ptr);
2147 free(plus.ptr);
2e5d2003 2148 if (diff_words->current_plus != diff_words->plus.text.ptr +
4297c0ae
BY
2149 diff_words->plus.text.size) {
2150 if (color_words_output_graph_prefix(diff_words))
bd033291
SB
2151 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2152 line_prefix, strlen(line_prefix), 0);
2153 fn_out_diff_words_write_helper(diff_words->opt,
882749a0 2154 &style->ctx, style->newline,
2e5d2003 2155 diff_words->plus.text.ptr + diff_words->plus.text.size
bd033291 2156 - diff_words->current_plus, diff_words->current_plus);
4297c0ae 2157 }
f59a59e2 2158 diff_words->minus.text.size = diff_words->plus.text.size = 0;
f59a59e2
JS
2159}
2160
76fd2828
JH
2161/* In "color-words" mode, show word-diff of words accumulated in the buffer */
2162static void diff_words_flush(struct emit_callback *ecbdata)
2163{
e6e045f8
SB
2164 struct diff_options *wo = ecbdata->diff_words->opt;
2165
76fd2828
JH
2166 if (ecbdata->diff_words->minus.text.size ||
2167 ecbdata->diff_words->plus.text.size)
2168 diff_words_show(ecbdata->diff_words);
e6e045f8
SB
2169
2170 if (wo->emitted_symbols) {
2171 struct diff_options *o = ecbdata->opt;
2172 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2173 int i;
2174
2175 /*
2176 * NEEDSWORK:
2177 * Instead of appending each, concat all words to a line?
2178 */
2179 for (i = 0; i < wol->nr; i++)
2180 append_emitted_diff_symbol(o, &wol->buf[i]);
2181
2182 for (i = 0; i < wol->nr; i++)
2183 free((void *)wol->buf[i].line);
2184
2185 wol->nr = 0;
2186 }
76fd2828
JH
2187}
2188
acd00ea0
NTND
2189static void diff_filespec_load_driver(struct diff_filespec *one,
2190 struct index_state *istate)
77d1a520
TR
2191{
2192 /* Use already-loaded driver */
2193 if (one->driver)
2194 return;
2195
2196 if (S_ISREG(one->mode))
acd00ea0 2197 one->driver = userdiff_find_by_path(istate, one->path);
77d1a520
TR
2198
2199 /* Fallback to default settings */
2200 if (!one->driver)
2201 one->driver = userdiff_find_by_name("default");
2202}
2203
acd00ea0
NTND
2204static const char *userdiff_word_regex(struct diff_filespec *one,
2205 struct index_state *istate)
77d1a520 2206{
acd00ea0 2207 diff_filespec_load_driver(one, istate);
77d1a520
TR
2208 return one->driver->word_regex;
2209}
2210
2211static void init_diff_words_data(struct emit_callback *ecbdata,
6440d341 2212 struct diff_options *orig_opts,
77d1a520
TR
2213 struct diff_filespec *one,
2214 struct diff_filespec *two)
2215{
2216 int i;
6440d341
TR
2217 struct diff_options *o = xmalloc(sizeof(struct diff_options));
2218 memcpy(o, orig_opts, sizeof(struct diff_options));
77d1a520
TR
2219
2220 ecbdata->diff_words =
2221 xcalloc(1, sizeof(struct diff_words_data));
2222 ecbdata->diff_words->type = o->word_diff;
2223 ecbdata->diff_words->opt = o;
e6e045f8
SB
2224
2225 if (orig_opts->emitted_symbols)
2226 o->emitted_symbols =
2227 xcalloc(1, sizeof(struct emitted_diff_symbols));
2228
77d1a520 2229 if (!o->word_regex)
acd00ea0 2230 o->word_regex = userdiff_word_regex(one, o->repo->index);
77d1a520 2231 if (!o->word_regex)
acd00ea0 2232 o->word_regex = userdiff_word_regex(two, o->repo->index);
77d1a520
TR
2233 if (!o->word_regex)
2234 o->word_regex = diff_word_regex_cfg;
2235 if (o->word_regex) {
2236 ecbdata->diff_words->word_regex = (regex_t *)
2237 xmalloc(sizeof(regex_t));
2238 if (regcomp(ecbdata->diff_words->word_regex,
2239 o->word_regex,
2240 REG_EXTENDED | REG_NEWLINE))
1a07e59c
NTND
2241 die("invalid regular expression: %s",
2242 o->word_regex);
77d1a520
TR
2243 }
2244 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2245 if (o->word_diff == diff_words_styles[i].type) {
2246 ecbdata->diff_words->style =
2247 &diff_words_styles[i];
2248 break;
2249 }
2250 }
2251 if (want_color(o->use_color)) {
2252 struct diff_words_style *st = ecbdata->diff_words->style;
63a01c3f
BW
2253 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2254 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
8dbf3eb6 2255 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
77d1a520
TR
2256 }
2257}
2258
f59a59e2
JS
2259static void free_diff_words_data(struct emit_callback *ecbdata)
2260{
2261 if (ecbdata->diff_words) {
76fd2828 2262 diff_words_flush(ecbdata);
e6e045f8 2263 free (ecbdata->diff_words->opt->emitted_symbols);
6440d341 2264 free (ecbdata->diff_words->opt);
8e0f7003 2265 free (ecbdata->diff_words->minus.text.ptr);
2e5d2003 2266 free (ecbdata->diff_words->minus.orig);
8e0f7003 2267 free (ecbdata->diff_words->plus.text.ptr);
2e5d2003 2268 free (ecbdata->diff_words->plus.orig);
ef5644ea
BC
2269 if (ecbdata->diff_words->word_regex) {
2270 regfree(ecbdata->diff_words->word_regex);
2271 free(ecbdata->diff_words->word_regex);
2272 }
6a83d902 2273 FREE_AND_NULL(ecbdata->diff_words);
f59a59e2
JS
2274 }
2275}
2276
ce436973 2277const char *diff_get_color(int diff_use_color, enum color_diff ix)
cd112cef 2278{
daa0c3d9 2279 if (want_color(diff_use_color))
50f575fc
LT
2280 return diff_colors[ix];
2281 return "";
cd112cef
JS
2282}
2283
f1922234
JK
2284const char *diff_line_prefix(struct diff_options *opt)
2285{
2286 struct strbuf *msgbuf;
2287 if (!opt->output_prefix)
2288 return "";
2289
2290 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2291 return msgbuf->buf;
2292}
2293
23707811
JH
2294static unsigned long sane_truncate_line(struct emit_callback *ecb, char *line, unsigned long len)
2295{
2296 const char *cp;
2297 unsigned long allot;
2298 size_t l = len;
2299
23707811
JH
2300 cp = line;
2301 allot = l;
2302 while (0 < l) {
2303 (void) utf8_width(&cp, &l);
2304 if (!cp)
2305 break; /* truncated in the middle? */
2306 }
2307 return allot - l;
2308}
2309
d68fe26f 2310static void find_lno(const char *line, struct emit_callback *ecbdata)
690ed843 2311{
d68fe26f
JH
2312 const char *p;
2313 ecbdata->lno_in_preimage = 0;
2314 ecbdata->lno_in_postimage = 0;
2315 p = strchr(line, '-');
690ed843 2316 if (!p)
d68fe26f
JH
2317 return; /* cannot happen */
2318 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2319 p = strchr(p, '+');
2320 if (!p)
2321 return; /* cannot happen */
2322 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
690ed843
JH
2323}
2324
cd112cef 2325static void fn_out_consume(void *priv, char *line, unsigned long len)
6973dcae 2326{
6973dcae 2327 struct emit_callback *ecbdata = priv;
ce436973 2328 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
7be57610 2329 struct diff_options *o = ecbdata->opt;
6973dcae 2330
ba16233c
SB
2331 o->found_changes = 1;
2332
3e97c7c6 2333 if (ecbdata->header) {
a29b0a13
SB
2334 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2335 ecbdata->header->buf, ecbdata->header->len, 0);
3e97c7c6
GB
2336 strbuf_reset(ecbdata->header);
2337 ecbdata->header = NULL;
2338 }
34a5e1a2 2339
6973dcae 2340 if (ecbdata->label_path[0]) {
3ee8b7bf
SB
2341 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2342 ecbdata->label_path[0],
2343 strlen(ecbdata->label_path[0]), 0);
2344 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2345 ecbdata->label_path[1],
2346 strlen(ecbdata->label_path[1]), 0);
6973dcae
JH
2347 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2348 }
cd112cef 2349
a624eaa7
JM
2350 if (diff_suppress_blank_empty
2351 && len == 2 && line[0] == ' ' && line[1] == '\n') {
2352 line[0] = '\n';
2353 len = 1;
2354 }
2355
b8d9c1a6 2356 if (line[0] == '@') {
76fd2828
JH
2357 if (ecbdata->diff_words)
2358 diff_words_flush(ecbdata);
23707811 2359 len = sane_truncate_line(ecbdata, line, len);
d68fe26f 2360 find_lno(line, ecbdata);
89cb73a1 2361 emit_hunk_header(ecbdata, line, len);
448c3ef1 2362 return;
cd112cef 2363 }
448c3ef1 2364
448c3ef1 2365 if (ecbdata->diff_words) {
ff958679
SB
2366 enum diff_symbol s =
2367 ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2368 DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
448c3ef1
JH
2369 if (line[0] == '-') {
2370 diff_words_append(line, len,
2371 &ecbdata->diff_words->minus);
2372 return;
2373 } else if (line[0] == '+') {
2374 diff_words_append(line, len,
2375 &ecbdata->diff_words->plus);
2376 return;
59556548 2377 } else if (starts_with(line, "\\ ")) {
c7c2bc0a
TR
2378 /*
2379 * Eat the "no newline at eof" marker as if we
2380 * saw a "+" or "-" line with nothing on it,
2381 * and return without diff_words_flush() to
2382 * defer processing. If this is the end of
2383 * preimage, more "+" lines may come after it.
2384 */
2385 return;
448c3ef1 2386 }
76fd2828 2387 diff_words_flush(ecbdata);
ff958679 2388 emit_diff_symbol(o, s, line, len, 0);
448c3ef1
JH
2389 return;
2390 }
448c3ef1 2391
0e383e18
JH
2392 switch (line[0]) {
2393 case '+':
d68fe26f 2394 ecbdata->lno_in_postimage++;
018cff70 2395 emit_add_line(reset, ecbdata, line + 1, len - 1);
0e383e18
JH
2396 break;
2397 case '-':
2398 ecbdata->lno_in_preimage++;
2399 emit_del_line(reset, ecbdata, line + 1, len - 1);
2400 break;
2401 case ' ':
2402 ecbdata->lno_in_postimage++;
2403 ecbdata->lno_in_preimage++;
2404 emit_context_line(reset, ecbdata, line + 1, len - 1);
2405 break;
2406 default:
2407 /* incomplete line at the end */
2408 ecbdata->lno_in_preimage++;
f2bb1218
SB
2409 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2410 line, len, 0);
0e383e18 2411 break;
448c3ef1 2412 }
6973dcae
JH
2413}
2414
c905cbc4 2415static void pprint_rename(struct strbuf *name, const char *a, const char *b)
6973dcae 2416{
63a01c3f
BW
2417 const char *old_name = a;
2418 const char *new_name = b;
6973dcae 2419 int pfx_length, sfx_length;
dd281f09 2420 int pfx_adjust_for_slash;
6973dcae
JH
2421 int len_a = strlen(a);
2422 int len_b = strlen(b);
663af342 2423 int a_midlen, b_midlen;
e5bfbf9b
AJ
2424 int qlen_a = quote_c_style(a, NULL, NULL, 0);
2425 int qlen_b = quote_c_style(b, NULL, NULL, 0);
2426
2427 if (qlen_a || qlen_b) {
c905cbc4
NTND
2428 quote_c_style(a, name, NULL, 0);
2429 strbuf_addstr(name, " => ");
2430 quote_c_style(b, name, NULL, 0);
2431 return;
e5bfbf9b 2432 }
6973dcae
JH
2433
2434 /* Find common prefix */
2435 pfx_length = 0;
63a01c3f
BW
2436 while (*old_name && *new_name && *old_name == *new_name) {
2437 if (*old_name == '/')
2438 pfx_length = old_name - a + 1;
2439 old_name++;
2440 new_name++;
6973dcae
JH
2441 }
2442
2443 /* Find common suffix */
63a01c3f
BW
2444 old_name = a + len_a;
2445 new_name = b + len_b;
6973dcae 2446 sfx_length = 0;
d020e27f 2447 /*
dd281f09
TR
2448 * If there is a common prefix, it must end in a slash. In
2449 * that case we let this loop run 1 into the prefix to see the
2450 * same slash.
2451 *
2452 * If there is no common prefix, we cannot do this as it would
2453 * underrun the input strings.
d020e27f 2454 */
dd281f09 2455 pfx_adjust_for_slash = (pfx_length ? 1 : 0);
63a01c3f
BW
2456 while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2457 b + pfx_length - pfx_adjust_for_slash <= new_name &&
2458 *old_name == *new_name) {
2459 if (*old_name == '/')
2460 sfx_length = len_a - (old_name - a);
2461 old_name--;
2462 new_name--;
6973dcae
JH
2463 }
2464
2465 /*
2466 * pfx{mid-a => mid-b}sfx
2467 * {pfx-a => pfx-b}sfx
2468 * pfx{sfx-a => sfx-b}
2469 * name-a => name-b
2470 */
663af342
PH
2471 a_midlen = len_a - pfx_length - sfx_length;
2472 b_midlen = len_b - pfx_length - sfx_length;
2473 if (a_midlen < 0)
2474 a_midlen = 0;
2475 if (b_midlen < 0)
2476 b_midlen = 0;
2477
c905cbc4 2478 strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
6973dcae 2479 if (pfx_length + sfx_length) {
c905cbc4
NTND
2480 strbuf_add(name, a, pfx_length);
2481 strbuf_addch(name, '{');
6973dcae 2482 }
c905cbc4
NTND
2483 strbuf_add(name, a + pfx_length, a_midlen);
2484 strbuf_addstr(name, " => ");
2485 strbuf_add(name, b + pfx_length, b_midlen);
663af342 2486 if (pfx_length + sfx_length) {
c905cbc4
NTND
2487 strbuf_addch(name, '}');
2488 strbuf_add(name, a + len_a - sfx_length, sfx_length);
6973dcae 2489 }
6973dcae
JH
2490}
2491
2492struct diffstat_t {
6973dcae
JH
2493 int nr;
2494 int alloc;
2495 struct diffstat_file {
f604652e 2496 char *from_name;
6973dcae 2497 char *name;
f604652e 2498 char *print_name;
ddf88fa6 2499 const char *comments;
6973dcae
JH
2500 unsigned is_unmerged:1;
2501 unsigned is_binary:1;
2502 unsigned is_renamed:1;
74faaa16 2503 unsigned is_interesting:1;
0974c117 2504 uintmax_t added, deleted;
6973dcae
JH
2505 } **files;
2506};
2507
2508static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2509 const char *name_a,
2510 const char *name_b)
2511{
2512 struct diffstat_file *x;
1a4927c5 2513 x = xcalloc(1, sizeof(*x));
4c960a43 2514 ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
6973dcae
JH
2515 diffstat->files[diffstat->nr++] = x;
2516 if (name_b) {
f604652e
JH
2517 x->from_name = xstrdup(name_a);
2518 x->name = xstrdup(name_b);
6973dcae
JH
2519 x->is_renamed = 1;
2520 }
f604652e
JH
2521 else {
2522 x->from_name = NULL;
9befac47 2523 x->name = xstrdup(name_a);
f604652e 2524 }
6973dcae
JH
2525 return x;
2526}
2527
2528static void diffstat_consume(void *priv, char *line, unsigned long len)
2529{
2530 struct diffstat_t *diffstat = priv;
2531 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2532
2533 if (line[0] == '+')
2534 x->added++;
2535 else if (line[0] == '-')
2536 x->deleted++;
2537}
2538
698ce6f8 2539const char mime_boundary_leader[] = "------------";
6973dcae 2540
a2540023
JH
2541static int scale_linear(int it, int width, int max_change)
2542{
2eeeef24
JH
2543 if (!it)
2544 return 0;
a2540023 2545 /*
2eeeef24
JH
2546 * make sure that at least one '-' or '+' is printed if
2547 * there is any change to this path. The easiest way is to
2548 * scale linearly as if the alloted width is one column shorter
2549 * than it is, and then add 1 to the result.
a2540023 2550 */
2eeeef24 2551 return 1 + (it * (width - 1) / max_change);
a2540023
JH
2552}
2553
0911c475
SB
2554static void show_graph(struct strbuf *out, char ch, int cnt,
2555 const char *set, const char *reset)
a2540023
JH
2556{
2557 if (cnt <= 0)
2558 return;
0911c475
SB
2559 strbuf_addstr(out, set);
2560 strbuf_addchars(out, ch, cnt);
2561 strbuf_addstr(out, reset);
a2540023
JH
2562}
2563
f604652e
JH
2564static void fill_print_name(struct diffstat_file *file)
2565{
c905cbc4 2566 struct strbuf pname = STRBUF_INIT;
f604652e
JH
2567
2568 if (file->print_name)
2569 return;
2570
c905cbc4
NTND
2571 if (file->is_renamed)
2572 pprint_rename(&pname, file->from_name, file->name);
2573 else
2574 quote_c_style(file->name, &pname, NULL, 0);
2575
ddf88fa6
NTND
2576 if (file->comments)
2577 strbuf_addf(&pname, " (%s)", file->comments);
2578
c905cbc4 2579 file->print_name = strbuf_detach(&pname, NULL);
f604652e
JH
2580}
2581
0911c475
SB
2582static void print_stat_summary_inserts_deletes(struct diff_options *options,
2583 int files, int insertions, int deletions)
7f814632
NTND
2584{
2585 struct strbuf sb = STRBUF_INIT;
7f814632
NTND
2586
2587 if (!files) {
2588 assert(insertions == 0 && deletions == 0);
0911c475
SB
2589 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2590 NULL, 0, 0);
2591 return;
7f814632
NTND
2592 }
2593
2594 strbuf_addf(&sb,
218adaaa 2595 (files == 1) ? " %d file changed" : " %d files changed",
7f814632
NTND
2596 files);
2597
2598 /*
2599 * For binary diff, the caller may want to print "x files
2600 * changed" with insertions == 0 && deletions == 0.
2601 *
2602 * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2603 * is probably less confusing (i.e skip over "2 files changed
2604 * but nothing about added/removed lines? Is this a bug in Git?").
2605 */
2606 if (insertions || deletions == 0) {
7f814632 2607 strbuf_addf(&sb,
218adaaa 2608 (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
7f814632
NTND
2609 insertions);
2610 }
2611
2612 if (deletions || insertions == 0) {
7f814632 2613 strbuf_addf(&sb,
218adaaa 2614 (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
7f814632
NTND
2615 deletions);
2616 }
2617 strbuf_addch(&sb, '\n');
0911c475
SB
2618 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2619 sb.buf, sb.len, 0);
7f814632 2620 strbuf_release(&sb);
0911c475
SB
2621}
2622
2623void print_stat_summary(FILE *fp, int files,
2624 int insertions, int deletions)
2625{
2626 struct diff_options o;
2627 memset(&o, 0, sizeof(o));
2628 o.file = fp;
2629
2630 print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
7f814632
NTND
2631}
2632
4b25d091 2633static void show_stats(struct diffstat_t *data, struct diff_options *options)
6973dcae 2634{
eb3a9dd3 2635 int i, len, add, del, adds = 0, dels = 0;
0974c117 2636 uintmax_t max_change = 0, max_len = 0;
dc801e71
ZJS
2637 int total_files = data->nr, count;
2638 int width, name_width, graph_width, number_width = 0, bin_width = 0;
c0aa335c 2639 const char *reset, *add_c, *del_c;
e5f85df8 2640 int extra_shown = 0;
0911c475
SB
2641 const char *line_prefix = diff_line_prefix(options);
2642 struct strbuf out = STRBUF_INIT;
6973dcae
JH
2643
2644 if (data->nr == 0)
2645 return;
2646
808e1db2 2647 count = options->stat_count ? options->stat_count : data->nr;
a2540023 2648
8f67f8ae 2649 reset = diff_get_color_opt(options, DIFF_RESET);
8f67f8ae
PH
2650 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2651 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
785f7432 2652
1b058bc3
ZJS
2653 /*
2654 * Find the longest filename and max number of changes
2655 */
808e1db2 2656 for (i = 0; (i < count) && (i < data->nr); i++) {
6973dcae 2657 struct diffstat_file *file = data->files[i];
0974c117 2658 uintmax_t change = file->added + file->deleted;
af0ed819
JH
2659
2660 if (!file->is_interesting && (change == 0)) {
808e1db2 2661 count++; /* not shown == room for one more */
358e460e
MG
2662 continue;
2663 }
f604652e
JH
2664 fill_print_name(file);
2665 len = strlen(file->print_name);
6973dcae
JH
2666 if (max_len < len)
2667 max_len = len;
2668
dc801e71
ZJS
2669 if (file->is_unmerged) {
2670 /* "Unmerged" is 8 characters */
2671 bin_width = bin_width < 8 ? 8 : bin_width;
6973dcae 2672 continue;
dc801e71
ZJS
2673 }
2674 if (file->is_binary) {
2675 /* "Bin XXX -> YYY bytes" */
2676 int w = 14 + decimal_width(file->added)
2677 + decimal_width(file->deleted);
2678 bin_width = bin_width < w ? w : bin_width;
2679 /* Display change counts aligned with "Bin" */
2680 number_width = 3;
2681 continue;
2682 }
2683
a2540023
JH
2684 if (max_change < change)
2685 max_change = change;
6973dcae 2686 }
a20d3c0d 2687 count = i; /* where we can stop scanning in data->files[] */
6973dcae 2688
1b058bc3
ZJS
2689 /*
2690 * We have width = stat_width or term_columns() columns total.
2691 * We want a maximum of min(max_len, stat_name_width) for the name part.
969fe57b 2692 * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1b058bc3
ZJS
2693 * We also need 1 for " " and 4 + decimal_width(max_change)
2694 * for " | NNNN " and one the empty column at the end, altogether
2695 * 6 + decimal_width(max_change).
2696 *
2697 * If there's not enough space, we will use the smaller of
2698 * stat_name_width (if set) and 5/8*width for the filename,
969fe57b
ZJS
2699 * and the rest for constant elements + graph part, but no more
2700 * than stat_graph_width for the graph part.
1b058bc3
ZJS
2701 * (5/8 gives 50 for filename and 30 for the constant parts + graph
2702 * for the standard terminal size).
a2540023 2703 *
1b058bc3
ZJS
2704 * In other words: stat_width limits the maximum width, and
2705 * stat_name_width fixes the maximum width of the filename,
2706 * and is also used to divide available columns if there
2707 * aren't enough.
dc801e71
ZJS
2708 *
2709 * Binary files are displayed with "Bin XXX -> YYY bytes"
2710 * instead of the change count and graph. This part is treated
2711 * similarly to the graph part, except that it is not
41ccfdd9 2712 * "scaled". If total width is too small to accommodate the
dc801e71
ZJS
2713 * guaranteed minimum width of the filename part and the
2714 * separators and this message, this message will "overflow"
2715 * making the line longer than the maximum width.
a2540023 2716 */
1b058bc3
ZJS
2717
2718 if (options->stat_width == -1)
cd48dadb 2719 width = term_columns() - strlen(line_prefix);
a2540023 2720 else
1b058bc3 2721 width = options->stat_width ? options->stat_width : 80;
dc801e71
ZJS
2722 number_width = decimal_width(max_change) > number_width ?
2723 decimal_width(max_change) : number_width;
a2540023 2724
df44483a
ZJS
2725 if (options->stat_graph_width == -1)
2726 options->stat_graph_width = diff_stat_graph_width;
a2540023 2727
1b058bc3
ZJS
2728 /*
2729 * Guarantee 3/8*16==6 for the graph part
2730 * and 5/8*16==10 for the filename part
2731 */
2732 if (width < 16 + 6 + number_width)
2733 width = 16 + 6 + number_width;
2734
2735 /*
2736 * First assign sizes that are wanted, ignoring available width.
dc801e71
ZJS
2737 * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2738 * starting from "XXX" should fit in graph_width.
1b058bc3 2739 */
dc801e71
ZJS
2740 graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2741 if (options->stat_graph_width &&
2742 options->stat_graph_width < graph_width)
2743 graph_width = options->stat_graph_width;
2744
1b058bc3
ZJS
2745 name_width = (options->stat_name_width > 0 &&
2746 options->stat_name_width < max_len) ?
2747 options->stat_name_width : max_len;
2748
2749 /*
2750 * Adjust adjustable widths not to exceed maximum width
2751 */
2752 if (name_width + number_width + 6 + graph_width > width) {
678c5741 2753 if (graph_width > width * 3/8 - number_width - 6) {
1b058bc3 2754 graph_width = width * 3/8 - number_width - 6;
678c5741
LP
2755 if (graph_width < 6)
2756 graph_width = 6;
2757 }
2758
969fe57b
ZJS
2759 if (options->stat_graph_width &&
2760 graph_width > options->stat_graph_width)
2761 graph_width = options->stat_graph_width;
1b058bc3
ZJS
2762 if (name_width > width - number_width - 6 - graph_width)
2763 name_width = width - number_width - 6 - graph_width;
2764 else
2765 graph_width = width - number_width - 6 - name_width;
2766 }
2767
2768 /*
2769 * From here name_width is the width of the name area,
2770 * and graph_width is the width of the graph area.
2771 * max_change is used to scale graph properly.
2772 */
808e1db2 2773 for (i = 0; i < count; i++) {
d2543b8e 2774 const char *prefix = "";
af0ed819
JH
2775 struct diffstat_file *file = data->files[i];
2776 char *name = file->print_name;
2777 uintmax_t added = file->added;
2778 uintmax_t deleted = file->deleted;
a2540023 2779 int name_len;
6973dcae 2780
a20d3c0d 2781 if (!file->is_interesting && (added + deleted == 0))
358e460e 2782 continue;
a20d3c0d 2783
6973dcae
JH
2784 /*
2785 * "scale" the filename
2786 */
a2540023
JH
2787 len = name_width;
2788 name_len = strlen(name);
2789 if (name_width < name_len) {
6973dcae
JH
2790 char *slash;
2791 prefix = "...";
a2540023
JH
2792 len -= 3;
2793 name += name_len - len;
6973dcae
JH
2794 slash = strchr(name, '/');
2795 if (slash)
2796 name = slash;
2797 }
6973dcae 2798
af0ed819 2799 if (file->is_binary) {
0911c475
SB
2800 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2801 strbuf_addf(&out, " %*s", number_width, "Bin");
e18872b2 2802 if (!added && !deleted) {
0911c475
SB
2803 strbuf_addch(&out, '\n');
2804 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2805 out.buf, out.len, 0);
2806 strbuf_reset(&out);
e18872b2
ZJS
2807 continue;
2808 }
0911c475 2809 strbuf_addf(&out, " %s%"PRIuMAX"%s",
0974c117 2810 del_c, deleted, reset);
0911c475
SB
2811 strbuf_addstr(&out, " -> ");
2812 strbuf_addf(&out, "%s%"PRIuMAX"%s",
0974c117 2813 add_c, added, reset);
0911c475
SB
2814 strbuf_addstr(&out, " bytes\n");
2815 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2816 out.buf, out.len, 0);
2817 strbuf_reset(&out);
f604652e 2818 continue;
6973dcae 2819 }
af0ed819 2820 else if (file->is_unmerged) {
0911c475
SB
2821 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2822 strbuf_addstr(&out, " Unmerged\n");
2823 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2824 out.buf, out.len, 0);
2825 strbuf_reset(&out);
f604652e 2826 continue;
6973dcae 2827 }
6973dcae 2828
a2540023
JH
2829 /*
2830 * scale the add/delete
2831 */
6973dcae
JH
2832 add = added;
2833 del = deleted;
6973dcae 2834
1b058bc3 2835 if (graph_width <= max_change) {
d3c9cf32 2836 int total = scale_linear(add + del, graph_width, max_change);
2eeeef24
JH
2837 if (total < 2 && add && del)
2838 /* width >= 2 due to the sanity check */
2839 total = 2;
2840 if (add < del) {
1b058bc3 2841 add = scale_linear(add, graph_width, max_change);
2eeeef24
JH
2842 del = total - add;
2843 } else {
1b058bc3 2844 del = scale_linear(del, graph_width, max_change);
2eeeef24
JH
2845 add = total - del;
2846 }
6973dcae 2847 }
0911c475
SB
2848 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2849 strbuf_addf(&out, " %*"PRIuMAX"%s",
dc801e71
ZJS
2850 number_width, added + deleted,
2851 added + deleted ? " " : "");
0911c475
SB
2852 show_graph(&out, '+', add, add_c, reset);
2853 show_graph(&out, '-', del, del_c, reset);
2854 strbuf_addch(&out, '\n');
2855 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2856 out.buf, out.len, 0);
2857 strbuf_reset(&out);
c0c77734 2858 }
a20d3c0d
JH
2859
2860 for (i = 0; i < data->nr; i++) {
af0ed819
JH
2861 struct diffstat_file *file = data->files[i];
2862 uintmax_t added = file->added;
2863 uintmax_t deleted = file->deleted;
82dfc2c4
JH
2864
2865 if (file->is_unmerged ||
2866 (!file->is_interesting && (added + deleted == 0))) {
808e1db2
MG
2867 total_files--;
2868 continue;
2869 }
a20d3c0d 2870
82dfc2c4 2871 if (!file->is_binary) {
a20d3c0d
JH
2872 adds += added;
2873 dels += deleted;
2874 }
2875 if (i < count)
2876 continue;
e5f85df8 2877 if (!extra_shown)
0911c475
SB
2878 emit_diff_symbol(options,
2879 DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2880 NULL, 0, 0);
e5f85df8 2881 extra_shown = 1;
808e1db2 2882 }
0911c475
SB
2883
2884 print_stat_summary_inserts_deletes(options, total_files, adds, dels);
5a612017 2885 strbuf_release(&out);
6973dcae
JH
2886}
2887
2775d92c 2888static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
ebd124c6
NP
2889{
2890 int i, adds = 0, dels = 0, total_files = data->nr;
2891
2892 if (data->nr == 0)
2893 return;
2894
2895 for (i = 0; i < data->nr; i++) {
e18872b2 2896 int added = data->files[i]->added;
0911c475 2897 int deleted = data->files[i]->deleted;
e18872b2 2898
20c8cde4
JH
2899 if (data->files[i]->is_unmerged ||
2900 (!data->files[i]->is_interesting && (added + deleted == 0))) {
e18872b2 2901 total_files--;
de9658b5 2902 } else if (!data->files[i]->is_binary) { /* don't count bytes */
e18872b2
ZJS
2903 adds += added;
2904 dels += deleted;
ebd124c6 2905 }
ebd124c6 2906 }
0911c475 2907 print_stat_summary_inserts_deletes(options, total_files, adds, dels);
ebd124c6
NP
2908}
2909
4b25d091 2910static void show_numstat(struct diffstat_t *data, struct diff_options *options)
74e2abe5
JH
2911{
2912 int i;
2913
f604652e
JH
2914 if (data->nr == 0)
2915 return;
2916
74e2abe5
JH
2917 for (i = 0; i < data->nr; i++) {
2918 struct diffstat_file *file = data->files[i];
2919
30997bb8 2920 fprintf(options->file, "%s", diff_line_prefix(options));
7be57610 2921
bfddbc5e 2922 if (file->is_binary)
c0c77734 2923 fprintf(options->file, "-\t-\t");
bfddbc5e 2924 else
c0c77734 2925 fprintf(options->file,
0974c117
JK
2926 "%"PRIuMAX"\t%"PRIuMAX"\t",
2927 file->added, file->deleted);
f604652e
JH
2928 if (options->line_termination) {
2929 fill_print_name(file);
2930 if (!file->is_renamed)
c0c77734 2931 write_name_quoted(file->name, options->file,
f604652e
JH
2932 options->line_termination);
2933 else {
c0c77734
DB
2934 fputs(file->print_name, options->file);
2935 putc(options->line_termination, options->file);
f604652e 2936 }
663af342 2937 } else {
f604652e 2938 if (file->is_renamed) {
c0c77734
DB
2939 putc('\0', options->file);
2940 write_name_quoted(file->from_name, options->file, '\0');
f604652e 2941 }
c0c77734 2942 write_name_quoted(file->name, options->file, '\0');
663af342 2943 }
74e2abe5
JH
2944 }
2945}
2946
c04a7155
JH
2947struct dirstat_file {
2948 const char *name;
2949 unsigned long changed;
7df7c019
LT
2950};
2951
c04a7155
JH
2952struct dirstat_dir {
2953 struct dirstat_file *files;
712d2c7d 2954 int alloc, nr, permille, cumulative;
c04a7155
JH
2955};
2956
7be57610
BY
2957static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2958 unsigned long changed, const char *base, int baselen)
7df7c019 2959{
585c0e2e 2960 unsigned long sum_changes = 0;
7df7c019 2961 unsigned int sources = 0;
30997bb8 2962 const char *line_prefix = diff_line_prefix(opt);
7df7c019
LT
2963
2964 while (dir->nr) {
c04a7155 2965 struct dirstat_file *f = dir->files;
7df7c019 2966 int namelen = strlen(f->name);
585c0e2e 2967 unsigned long changes;
7df7c019
LT
2968 char *slash;
2969
2970 if (namelen < baselen)
2971 break;
2972 if (memcmp(f->name, base, baselen))
2973 break;
2974 slash = strchr(f->name + baselen, '/');
2975 if (slash) {
2976 int newbaselen = slash + 1 - f->name;
585c0e2e 2977 changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
7df7c019
LT
2978 sources++;
2979 } else {
585c0e2e 2980 changes = f->changed;
7df7c019
LT
2981 dir->files++;
2982 dir->nr--;
2983 sources += 2;
2984 }
585c0e2e 2985 sum_changes += changes;
7df7c019
LT
2986 }
2987
2988 /*
2989 * We don't report dirstat's for
2990 * - the top level
2991 * - or cases where everything came from a single directory
2992 * under this directory (sources == 1).
2993 */
2994 if (baselen && sources != 1) {
585c0e2e
BW
2995 if (sum_changes) {
2996 int permille = sum_changes * 1000 / changed;
712d2c7d 2997 if (permille >= dir->permille) {
7be57610 2998 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
712d2c7d 2999 permille / 10, permille % 10, baselen, base);
7df7c019
LT
3000 if (!dir->cumulative)
3001 return 0;
3002 }
3003 }
3004 }
585c0e2e 3005 return sum_changes;
7df7c019
LT
3006}
3007
441bca0b
LT
3008static int dirstat_compare(const void *_a, const void *_b)
3009{
3010 const struct dirstat_file *a = _a;
3011 const struct dirstat_file *b = _b;
3012 return strcmp(a->name, b->name);
3013}
3014
c04a7155 3015static void show_dirstat(struct diff_options *options)
7df7c019
LT
3016{
3017 int i;
3018 unsigned long changed;
c04a7155
JH
3019 struct dirstat_dir dir;
3020 struct diff_queue_struct *q = &diff_queued_diff;
3021
3022 dir.files = NULL;
3023 dir.alloc = 0;
3024 dir.nr = 0;
712d2c7d 3025 dir.permille = options->dirstat_permille;
0d1e0e78 3026 dir.cumulative = options->flags.dirstat_cumulative;
7df7c019 3027
7df7c019 3028 changed = 0;
c04a7155
JH
3029 for (i = 0; i < q->nr; i++) {
3030 struct diff_filepair *p = q->queue[i];
3031 const char *name;
3032 unsigned long copied, added, damage;
3033
2ca86714 3034 name = p->two->path ? p->two->path : p->one->path;
c04a7155 3035
d9f62dfa
JK
3036 if (p->one->oid_valid && p->two->oid_valid &&
3037 oideq(&p->one->oid, &p->two->oid)) {
2ff3a803
JH
3038 /*
3039 * The SHA1 has not changed, so pre-/post-content is
3040 * identical. We can therefore skip looking at the
3041 * file contents altogether.
3042 */
3043 damage = 0;
3044 goto found_damage;
3045 }
3046
0d1e0e78 3047 if (options->flags.dirstat_by_file) {
0133dab7
JH
3048 /*
3049 * In --dirstat-by-file mode, we don't really need to
3050 * look at the actual file contents at all.
3051 * The fact that the SHA1 changed is enough for us to
3052 * add this file to the list of results
3053 * (with each file contributing equal damage).
3054 */
2ff3a803 3055 damage = 1;
0133dab7
JH
3056 goto found_damage;
3057 }
c04a7155
JH
3058
3059 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
b78ea5fc
NTND
3060 diff_populate_filespec(options->repo, p->one, 0);
3061 diff_populate_filespec(options->repo, p->two, 0);
3062 diffcore_count_changes(options->repo,
3063 p->one, p->two, NULL, NULL,
c04a7155
JH
3064 &copied, &added);
3065 diff_free_filespec_data(p->one);
3066 diff_free_filespec_data(p->two);
3067 } else if (DIFF_FILE_VALID(p->one)) {
b78ea5fc 3068 diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
c04a7155
JH
3069 copied = added = 0;
3070 diff_free_filespec_data(p->one);
3071 } else if (DIFF_FILE_VALID(p->two)) {
b78ea5fc 3072 diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
c04a7155
JH
3073 copied = 0;
3074 added = p->two->size;
3075 diff_free_filespec_data(p->two);
3076 } else
2b0b551d 3077 continue;
c04a7155
JH
3078
3079 /*
3080 * Original minus copied is the removed material,
3081 * added is the new material. They are both damages
0133dab7 3082 * made to the preimage.
2ff3a803
JH
3083 * If the resulting damage is zero, we know that
3084 * diffcore_count_changes() considers the two entries to
d9f62dfa 3085 * be identical, but since the oid changed, we
2ff3a803
JH
3086 * know that there must have been _some_ kind of change,
3087 * so we force all entries to have damage > 0.
c04a7155
JH
3088 */
3089 damage = (p->one->size - copied) + added;
2ff3a803 3090 if (!damage)
fd33777b 3091 damage = 1;
c04a7155 3092
0133dab7 3093found_damage:
c04a7155
JH
3094 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3095 dir.files[dir.nr].name = name;
3096 dir.files[dir.nr].changed = damage;
3097 changed += damage;
3098 dir.nr++;
7df7c019
LT
3099 }
3100
3101 /* This can happen even with many files, if everything was renames */
3102 if (!changed)
3103 return;
3104
3105 /* Show all directories with more than x% of the changes */
9ed0d8d6 3106 QSORT(dir.files, dir.nr, dirstat_compare);
7be57610 3107 gather_dirstat(options, &dir, changed, "", 0);
7df7c019
LT
3108}
3109
1c57a627
JH
3110static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3111{
3112 int i;
3113 unsigned long changed;
3114 struct dirstat_dir dir;
3115
3116 if (data->nr == 0)
3117 return;
3118
3119 dir.files = NULL;
3120 dir.alloc = 0;
3121 dir.nr = 0;
3122 dir.permille = options->dirstat_permille;
0d1e0e78 3123 dir.cumulative = options->flags.dirstat_cumulative;
1c57a627
JH
3124
3125 changed = 0;
3126 for (i = 0; i < data->nr; i++) {
3127 struct diffstat_file *file = data->files[i];
3128 unsigned long damage = file->added + file->deleted;
3129 if (file->is_binary)
3130 /*
3131 * binary files counts bytes, not lines. Must find some
3132 * way to normalize binary bytes vs. textual lines.
3133 * The following heuristic assumes that there are 64
3134 * bytes per "line".
3135 * This is stupid and ugly, but very cheap...
3136 */
42c78a21 3137 damage = DIV_ROUND_UP(damage, 64);
1c57a627
JH
3138 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3139 dir.files[dir.nr].name = file->name;
3140 dir.files[dir.nr].changed = damage;
3141 changed += damage;
3142 dir.nr++;
3143 }
3144
3145 /* This can happen even with many files, if everything was renames */
3146 if (!changed)
3147 return;
3148
3149 /* Show all directories with more than x% of the changes */
9ed0d8d6 3150 QSORT(dir.files, dir.nr, dirstat_compare);
1c57a627
JH
3151 gather_dirstat(options, &dir, changed, "", 0);
3152}
3153
f604652e
JH
3154static void free_diffstat_info(struct diffstat_t *diffstat)
3155{
3156 int i;
3157 for (i = 0; i < diffstat->nr; i++) {
3158 struct diffstat_file *f = diffstat->files[i];
c905cbc4 3159 free(f->print_name);
f604652e
JH
3160 free(f->name);
3161 free(f->from_name);
3162 free(f);
3163 }
3164 free(diffstat->files);
3165}
3166
88246898 3167struct checkdiff_t {
88246898 3168 const char *filename;
1ba111d1 3169 int lineno;
a757c646 3170 int conflict_marker_size;
1ba111d1 3171 struct diff_options *o;
cf1b7869 3172 unsigned ws_rule;
62c64895 3173 unsigned status;
88246898
JS
3174};
3175
a757c646 3176static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
04954043
JH
3177{
3178 char firstchar;
3179 int cnt;
3180
a757c646 3181 if (len < marker_size + 1)
04954043
JH
3182 return 0;
3183 firstchar = line[0];
3184 switch (firstchar) {
a757c646 3185 case '=': case '>': case '<': case '|':
04954043
JH
3186 break;
3187 default:
3188 return 0;
3189 }
a757c646 3190 for (cnt = 1; cnt < marker_size; cnt++)
04954043
JH
3191 if (line[cnt] != firstchar)
3192 return 0;
a757c646
JH
3193 /* line[1] thru line[marker_size-1] are same as firstchar */
3194 if (len < marker_size + 1 || !isspace(line[marker_size]))
04954043 3195 return 0;
04954043
JH
3196 return 1;
3197}
3198
75ab7630
JK
3199static void checkdiff_consume_hunk(void *priv,
3200 long ob, long on, long nb, long nn,
3201 const char *func, long funclen)
3202
3203{
3204 struct checkdiff_t *data = priv;
3205 data->lineno = nb - 1;
3206}
3207
88246898
JS
3208static void checkdiff_consume(void *priv, char *line, unsigned long len)
3209{
3210 struct checkdiff_t *data = priv;
a757c646 3211 int marker_size = data->conflict_marker_size;
f1c96261
JK
3212 const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3213 const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3214 const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
c1795bb0 3215 char *err;
30997bb8 3216 const char *line_prefix;
7be57610
BY
3217
3218 assert(data->o);
30997bb8 3219 line_prefix = diff_line_prefix(data->o);
88246898
JS
3220
3221 if (line[0] == '+') {
18374e58 3222 unsigned bad;
0ef617f4 3223 data->lineno++;
a757c646 3224 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
04954043
JH
3225 data->status |= 1;
3226 fprintf(data->o->file,
7be57610
BY
3227 "%s%s:%d: leftover conflict marker\n",
3228 line_prefix, data->filename, data->lineno);
04954043 3229 }
8f8841e9 3230 bad = ws_check(line + 1, len - 1, data->ws_rule);
18374e58 3231 if (!bad)
c1795bb0 3232 return;
18374e58
JH
3233 data->status |= bad;
3234 err = whitespace_error_string(bad);
7be57610
BY
3235 fprintf(data->o->file, "%s%s:%d: %s.\n",
3236 line_prefix, data->filename, data->lineno, err);
c1795bb0 3237 free(err);
a3c158d4 3238 emit_line(data->o, set, reset, line, 1);
8f8841e9 3239 ws_check_emit(line + 1, len - 1, data->ws_rule,
1ba111d1 3240 data->o->file, set, reset, ws);
877f23cc 3241 } else if (line[0] == ' ') {
88246898 3242 data->lineno++;
88246898
JS
3243 }
3244}
3245
0660626c
JH
3246static unsigned char *deflate_it(char *data,
3247 unsigned long size,
3248 unsigned long *result_size)
051308f6 3249{
0660626c
JH
3250 int bound;
3251 unsigned char *deflated;
ef49a7a0 3252 git_zstream stream;
0660626c 3253
55bb5c91 3254 git_deflate_init(&stream, zlib_compression_level);
225a6f10 3255 bound = git_deflate_bound(&stream, size);
0660626c
JH
3256 deflated = xmalloc(bound);
3257 stream.next_out = deflated;
3258 stream.avail_out = bound;
3259
3260 stream.next_in = (unsigned char *)data;
3261 stream.avail_in = size;
55bb5c91 3262 while (git_deflate(&stream, Z_FINISH) == Z_OK)
0660626c 3263 ; /* nothing */
55bb5c91 3264 git_deflate_end(&stream);
0660626c
JH
3265 *result_size = stream.total_out;
3266 return deflated;
051308f6
JH
3267}
3268
4eed0ebd
SB
3269static void emit_binary_diff_body(struct diff_options *o,
3270 mmfile_t *one, mmfile_t *two)
051308f6 3271{
0660626c
JH
3272 void *cp;
3273 void *delta;
3274 void *deflated;
3275 void *data;
3276 unsigned long orig_size;
3277 unsigned long delta_size;
3278 unsigned long deflate_size;
3279 unsigned long data_size;
051308f6 3280
0660626c
JH
3281 /* We could do deflated delta, or we could do just deflated two,
3282 * whichever is smaller.
051308f6 3283 */
0660626c
JH
3284 delta = NULL;
3285 deflated = deflate_it(two->ptr, two->size, &deflate_size);
3286 if (one->size && two->size) {
3287 delta = diff_delta(one->ptr, one->size,
3288 two->ptr, two->size,
3289 &delta_size, deflate_size);
3290 if (delta) {
3291 void *to_free = delta;
3292 orig_size = delta_size;
3293 delta = deflate_it(delta, delta_size, &delta_size);
3294 free(to_free);
051308f6
JH
3295 }
3296 }
051308f6 3297
0660626c 3298 if (delta && delta_size < deflate_size) {
ca473cef 3299 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
4eed0ebd
SB
3300 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3301 s, strlen(s), 0);
3302 free(s);
0660626c
JH
3303 free(deflated);
3304 data = delta;
3305 data_size = delta_size;
4eed0ebd
SB
3306 } else {
3307 char *s = xstrfmt("%lu", two->size);
3308 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3309 s, strlen(s), 0);
3310 free(s);
0660626c
JH
3311 free(delta);
3312 data = deflated;
3313 data_size = deflate_size;
3314 }
051308f6 3315
0660626c
JH
3316 /* emit data encoded in base85 */
3317 cp = data;
3318 while (data_size) {
4eed0ebd 3319 int len;
0660626c 3320 int bytes = (52 < data_size) ? 52 : data_size;
4eed0ebd 3321 char line[71];
0660626c 3322 data_size -= bytes;
051308f6
JH
3323 if (bytes <= 26)
3324 line[0] = bytes + 'A' - 1;
3325 else
3326 line[0] = bytes - 26 + 'a' - 1;
3327 encode_85(line + 1, cp, bytes);
1d7f171c 3328 cp = (char *) cp + bytes;
4eed0ebd
SB
3329
3330 len = strlen(line);
3331 line[len++] = '\n';
3332 line[len] = '\0';
3333
3334 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3335 line, len, 0);
051308f6 3336 }
4eed0ebd 3337 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
0660626c 3338 free(data);
051308f6
JH
3339}
3340
4eed0ebd
SB
3341static void emit_binary_diff(struct diff_options *o,
3342 mmfile_t *one, mmfile_t *two)
d4c452f0 3343{
4eed0ebd
SB
3344 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3345 emit_binary_diff_body(o, one, two);
3346 emit_binary_diff_body(o, two, one);
d4c452f0
JH
3347}
3348
b78ea5fc
NTND
3349int diff_filespec_is_binary(struct repository *r,
3350 struct diff_filespec *one)
29a3eefd 3351{
122aa6f9 3352 if (one->is_binary == -1) {
acd00ea0 3353 diff_filespec_load_driver(one, r->index);
122aa6f9
JK
3354 if (one->driver->binary != -1)
3355 one->is_binary = one->driver->binary;
3356 else {
3357 if (!one->data && DIFF_FILE_VALID(one))
b78ea5fc 3358 diff_populate_filespec(r, one, CHECK_BINARY);
6bf3b813 3359 if (one->is_binary == -1 && one->data)
122aa6f9
JK
3360 one->is_binary = buffer_is_binary(one->data,
3361 one->size);
3362 if (one->is_binary == -1)
3363 one->is_binary = 0;
3364 }
3365 }
29a3eefd 3366 return one->is_binary;
6973dcae
JH
3367}
3368
acd00ea0
NTND
3369static const struct userdiff_funcname *
3370diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
f258475a 3371{
acd00ea0 3372 diff_filespec_load_driver(one, o->repo->index);
122aa6f9 3373 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
f258475a
JH
3374}
3375
a5a818ee
JH
3376void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3377{
3378 if (!options->a_prefix)
3379 options->a_prefix = a;
3380 if (!options->b_prefix)
3381 options->b_prefix = b;
3382}
3383
bd7ad45b 3384struct userdiff_driver *get_textconv(struct repository *r,
acd00ea0 3385 struct diff_filespec *one)
04427ac8
JK
3386{
3387 if (!DIFF_FILE_VALID(one))
3388 return NULL;
d391c0ff 3389
bd7ad45b
NTND
3390 diff_filespec_load_driver(one, r->index);
3391 return userdiff_get_textconv(r, one->driver);
04427ac8
JK
3392}
3393
6973dcae
JH
3394static void builtin_diff(const char *name_a,
3395 const char *name_b,
3396 struct diff_filespec *one,
3397 struct diff_filespec *two,
3398 const char *xfrm_msg,
296c6bb2 3399 int must_show_header,
051308f6 3400 struct diff_options *o,
6973dcae
JH
3401 int complete_rewrite)
3402{
3403 mmfile_t mf1, mf2;
3404 const char *lbl[2];
3405 char *a_one, *b_two;
d9c552f1 3406 const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
8f67f8ae 3407 const char *reset = diff_get_color_opt(o, DIFF_RESET);
a5a818ee 3408 const char *a_prefix, *b_prefix;
d9bae1a1
JK
3409 struct userdiff_driver *textconv_one = NULL;
3410 struct userdiff_driver *textconv_two = NULL;
3e97c7c6 3411 struct strbuf header = STRBUF_INIT;
30997bb8 3412 const char *line_prefix = diff_line_prefix(o);
a5a818ee 3413
fd47ae6a 3414 diff_set_mnemonic_prefix(o, "a/", "b/");
0d1e0e78 3415 if (o->flags.reverse_diff) {
fd47ae6a
JK
3416 a_prefix = o->b_prefix;
3417 b_prefix = o->a_prefix;
3418 } else {
3419 a_prefix = o->a_prefix;
3420 b_prefix = o->b_prefix;
3421 }
3422
61cfbc05
JK
3423 if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3424 (!one->mode || S_ISGITLINK(one->mode)) &&
3425 (!two->mode || S_ISGITLINK(two->mode))) {
f3597138 3426 show_submodule_summary(o, one->path ? one->path : two->path,
602a283a 3427 &one->oid, &two->oid,
f3597138 3428 two->dirty_submodule);
752c0c24 3429 return;
fd47ae6a
JK
3430 } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3431 (!one->mode || S_ISGITLINK(one->mode)) &&
3432 (!two->mode || S_ISGITLINK(two->mode))) {
f3597138 3433 show_submodule_inline_diff(o, one->path ? one->path : two->path,
fd47ae6a 3434 &one->oid, &two->oid,
f3597138 3435 two->dirty_submodule);
fd47ae6a 3436 return;
752c0c24
JS
3437 }
3438
0d1e0e78 3439 if (o->flags.allow_textconv) {
bd7ad45b
NTND
3440 textconv_one = get_textconv(o->repo, one);
3441 textconv_two = get_textconv(o->repo, two);
3aa1f7ca
JK
3442 }
3443
71b989e7
LT
3444 /* Never use a non-valid filename anywhere if at all possible */
3445 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3446 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3447
a5a818ee
JH
3448 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3449 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
6973dcae
JH
3450 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3451 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
d9c552f1 3452 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
6973dcae
JH
3453 if (lbl[0][0] == '/') {
3454 /* /dev/null */
d9c552f1 3455 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
37466447
BW
3456 if (xfrm_msg)
3457 strbuf_addstr(&header, xfrm_msg);
296c6bb2 3458 must_show_header = 1;
6973dcae
JH
3459 }
3460 else if (lbl[1][0] == '/') {
d9c552f1 3461 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
37466447
BW
3462 if (xfrm_msg)
3463 strbuf_addstr(&header, xfrm_msg);
296c6bb2 3464 must_show_header = 1;
6973dcae
JH
3465 }
3466 else {
3467 if (one->mode != two->mode) {
d9c552f1
JK
3468 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3469 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
296c6bb2 3470 must_show_header = 1;
cd112cef 3471 }
37466447
BW
3472 if (xfrm_msg)
3473 strbuf_addstr(&header, xfrm_msg);
3e97c7c6 3474
6973dcae
JH
3475 /*
3476 * we do not run diff between different kind
3477 * of objects.
3478 */
3479 if ((one->mode ^ two->mode) & S_IFMT)
3480 goto free_ab_and_return;
0c01857d 3481 if (complete_rewrite &&
b78ea5fc
NTND
3482 (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3483 (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
a29b0a13
SB
3484 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3485 header.buf, header.len, 0);
3e97c7c6 3486 strbuf_reset(&header);
3aa1f7ca 3487 emit_rewrite_diff(name_a, name_b, one, two,
b78ea5fc 3488 textconv_one, textconv_two, o);
34a5e1a2 3489 o->found_changes = 1;
6973dcae
JH
3490 goto free_ab_and_return;
3491 }
3492 }
3493
467ddc14 3494 if (o->irreversible_delete && lbl[1][0] == '/') {
a29b0a13
SB
3495 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3496 header.len, 0);
467ddc14
JH
3497 strbuf_reset(&header);
3498 goto free_ab_and_return;
0d1e0e78 3499 } else if (!o->flags.text &&
b78ea5fc
NTND
3500 ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3501 (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
4acaaa7a 3502 struct strbuf sb = STRBUF_INIT;
1aaf69e6
NTND
3503 if (!one->data && !two->data &&
3504 S_ISREG(one->mode) && S_ISREG(two->mode) &&
0d1e0e78 3505 !o->flags.binary) {
4a7e27e9 3506 if (oideq(&one->oid, &two->oid)) {
1aaf69e6 3507 if (must_show_header)
a29b0a13
SB
3508 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3509 header.buf, header.len,
3510 0);
1aaf69e6
NTND
3511 goto free_ab_and_return;
3512 }
a29b0a13
SB
3513 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3514 header.buf, header.len, 0);
4acaaa7a
SB
3515 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3516 diff_line_prefix(o), lbl[0], lbl[1]);
3517 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3518 sb.buf, sb.len, 0);
3519 strbuf_release(&sb);
1aaf69e6
NTND
3520 goto free_ab_and_return;
3521 }
b78ea5fc
NTND
3522 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3523 fill_mmfile(o->repo, &mf2, two) < 0)
b3373982 3524 die("unable to read files to diff");
0660626c
JH
3525 /* Quite common confusing case */
3526 if (mf1.size == mf2.size &&
296c6bb2
CC
3527 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3528 if (must_show_header)
a29b0a13
SB
3529 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3530 header.buf, header.len, 0);
0660626c 3531 goto free_ab_and_return;
296c6bb2 3532 }
a29b0a13 3533 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3e97c7c6 3534 strbuf_reset(&header);
0d1e0e78 3535 if (o->flags.binary)
4eed0ebd 3536 emit_binary_diff(o, &mf1, &mf2);
4acaaa7a
SB
3537 else {
3538 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3539 diff_line_prefix(o), lbl[0], lbl[1]);
3540 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3541 sb.buf, sb.len, 0);
3542 strbuf_release(&sb);
3543 }
34a5e1a2 3544 o->found_changes = 1;
467ddc14 3545 } else {
6973dcae 3546 /* Crazy xdl interfaces.. */
0da0e926 3547 const char *diffopts;