pack-bitmap.h: remove magic number
[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"
4a288478 26#include "parse-options.h"
3ac68a93 27#include "help.h"
b14ed5ad 28#include "promisor-remote.h"
6973dcae 29
1510fea7
SP
30#ifdef NO_FAST_WORKING_DIRECTORY
31#define FAST_WORKING_DIRECTORY 0
32#else
33#define FAST_WORKING_DIRECTORY 1
34#endif
35
96f1e58f 36static int diff_detect_rename_default;
33de7163 37static int diff_indent_heuristic = 1;
92c57e5c 38static int diff_rename_limit_default = 400;
a624eaa7 39static int diff_suppress_blank_empty;
d2aea137 40static int diff_use_color_default = -1;
2e2d5ac1 41static int diff_color_moved_default;
626c0b5d 42static int diff_color_moved_ws_default;
6468a4e5 43static int diff_context_default = 3;
c4888677 44static int diff_interhunk_context_default;
98a4d87b 45static const char *diff_word_regex_cfg;
cbe02100 46static const char *external_diff_cmd_cfg;
6d8940b5 47static const char *diff_order_file_cfg;
aecbf914 48int diff_auto_refresh_index = 1;
a5a818ee 49static int diff_mnemonic_prefix;
f89504dd 50static int diff_no_prefix;
df44483a 51static int diff_stat_graph_width;
712d2c7d 52static int diff_dirstat_permille_default = 30;
be4f2b40 53static struct diff_options default_diff_options;
07ab4dec 54static long diff_algorithm;
a17505f2 55static unsigned ws_error_highlight_default = WSEH_NEW;
6973dcae 56
7c92fe0e 57static char diff_colors[][COLOR_MAXLEN] = {
dc6ebd4c 58 GIT_COLOR_RESET,
8dbf3eb6 59 GIT_COLOR_NORMAL, /* CONTEXT */
dc6ebd4c
AL
60 GIT_COLOR_BOLD, /* METAINFO */
61 GIT_COLOR_CYAN, /* FRAGINFO */
62 GIT_COLOR_RED, /* OLD */
63 GIT_COLOR_GREEN, /* NEW */
64 GIT_COLOR_YELLOW, /* COMMIT */
65 GIT_COLOR_BG_RED, /* WHITESPACE */
89cb73a1 66 GIT_COLOR_NORMAL, /* FUNCINFO */
86b452e2
SB
67 GIT_COLOR_BOLD_MAGENTA, /* OLD_MOVED */
68 GIT_COLOR_BOLD_BLUE, /* OLD_MOVED ALTERNATIVE */
69 GIT_COLOR_FAINT, /* OLD_MOVED_DIM */
70 GIT_COLOR_FAINT_ITALIC, /* OLD_MOVED_ALTERNATIVE_DIM */
71 GIT_COLOR_BOLD_CYAN, /* NEW_MOVED */
72 GIT_COLOR_BOLD_YELLOW, /* NEW_MOVED ALTERNATIVE */
73 GIT_COLOR_FAINT, /* NEW_MOVED_DIM */
74 GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
a7be92ac
JS
75 GIT_COLOR_FAINT, /* CONTEXT_DIM */
76 GIT_COLOR_FAINT_RED, /* OLD_DIM */
77 GIT_COLOR_FAINT_GREEN, /* NEW_DIM */
78 GIT_COLOR_BOLD, /* CONTEXT_BOLD */
79 GIT_COLOR_BOLD_RED, /* OLD_BOLD */
80 GIT_COLOR_BOLD_GREEN, /* NEW_BOLD */
cd112cef
JS
81};
82
a73b3680
NTND
83static const char *color_diff_slots[] = {
84 [DIFF_CONTEXT] = "context",
85 [DIFF_METAINFO] = "meta",
86 [DIFF_FRAGINFO] = "frag",
87 [DIFF_FILE_OLD] = "old",
88 [DIFF_FILE_NEW] = "new",
89 [DIFF_COMMIT] = "commit",
90 [DIFF_WHITESPACE] = "whitespace",
91 [DIFF_FUNCINFO] = "func",
92 [DIFF_FILE_OLD_MOVED] = "oldMoved",
93 [DIFF_FILE_OLD_MOVED_ALT] = "oldMovedAlternative",
94 [DIFF_FILE_OLD_MOVED_DIM] = "oldMovedDimmed",
95 [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
96 [DIFF_FILE_NEW_MOVED] = "newMoved",
97 [DIFF_FILE_NEW_MOVED_ALT] = "newMovedAlternative",
98 [DIFF_FILE_NEW_MOVED_DIM] = "newMovedDimmed",
99 [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
a7be92ac
JS
100 [DIFF_CONTEXT_DIM] = "contextDimmed",
101 [DIFF_FILE_OLD_DIM] = "oldDimmed",
102 [DIFF_FILE_NEW_DIM] = "newDimmed",
103 [DIFF_CONTEXT_BOLD] = "contextBold",
104 [DIFF_FILE_OLD_BOLD] = "oldBold",
105 [DIFF_FILE_NEW_BOLD] = "newBold",
a73b3680
NTND
106};
107
3ac68a93
NTND
108define_list_config_array_extra(color_diff_slots, {"plain"});
109
9e1a5ebe 110static int parse_diff_color_slot(const char *var)
801235c5 111{
a73b3680 112 if (!strcasecmp(var, "plain"))
8dbf3eb6 113 return DIFF_CONTEXT;
a73b3680 114 return LOOKUP_CONFIG(color_diff_slots, var);
801235c5
JH
115}
116
02e8ca0e 117static int parse_dirstat_params(struct diff_options *options, const char *params_string,
51670fc8 118 struct strbuf *errmsg)
333f3fb0 119{
02e8ca0e
MH
120 char *params_copy = xstrdup(params_string);
121 struct string_list params = STRING_LIST_INIT_NODUP;
122 int ret = 0;
123 int i;
51670fc8 124
02e8ca0e
MH
125 if (*params_copy)
126 string_list_split_in_place(&params, params_copy, ',', -1);
127 for (i = 0; i < params.nr; i++) {
128 const char *p = params.items[i].string;
129 if (!strcmp(p, "changes")) {
0d1e0e78
BW
130 options->flags.dirstat_by_line = 0;
131 options->flags.dirstat_by_file = 0;
02e8ca0e 132 } else if (!strcmp(p, "lines")) {
0d1e0e78
BW
133 options->flags.dirstat_by_line = 1;
134 options->flags.dirstat_by_file = 0;
02e8ca0e 135 } else if (!strcmp(p, "files")) {
0d1e0e78
BW
136 options->flags.dirstat_by_line = 0;
137 options->flags.dirstat_by_file = 1;
02e8ca0e 138 } else if (!strcmp(p, "noncumulative")) {
0d1e0e78 139 options->flags.dirstat_cumulative = 0;
02e8ca0e 140 } else if (!strcmp(p, "cumulative")) {
0d1e0e78 141 options->flags.dirstat_cumulative = 1;
333f3fb0
JH
142 } else if (isdigit(*p)) {
143 char *end;
51670fc8
JH
144 int permille = strtoul(p, &end, 10) * 10;
145 if (*end == '.' && isdigit(*++end)) {
712d2c7d 146 /* only use first digit */
51670fc8 147 permille += *end - '0';
712d2c7d 148 /* .. and ignore any further digits */
51670fc8 149 while (isdigit(*++end))
712d2c7d
JH
150 ; /* nothing */
151 }
02e8ca0e 152 if (!*end)
51670fc8
JH
153 options->dirstat_permille = permille;
154 else {
02e8ca0e
MH
155 strbuf_addf(errmsg, _(" Failed to parse dirstat cut-off percentage '%s'\n"),
156 p);
51670fc8
JH
157 ret++;
158 }
159 } else {
02e8ca0e 160 strbuf_addf(errmsg, _(" Unknown dirstat parameter '%s'\n"), p);
51670fc8 161 ret++;
333f3fb0 162 }
51670fc8 163
333f3fb0 164 }
02e8ca0e
MH
165 string_list_clear(&params, 0);
166 free(params_copy);
51670fc8 167 return ret;
333f3fb0
JH
168}
169
c47ef57c
RR
170static int parse_submodule_params(struct diff_options *options, const char *value)
171{
172 if (!strcmp(value, "log"))
61cfbc05 173 options->submodule_format = DIFF_SUBMODULE_LOG;
c47ef57c 174 else if (!strcmp(value, "short"))
61cfbc05 175 options->submodule_format = DIFF_SUBMODULE_SHORT;
fd47ae6a
JK
176 else if (!strcmp(value, "diff"))
177 options->submodule_format = DIFF_SUBMODULE_INLINE_DIFF;
5a59a230
NTND
178 /*
179 * Please update $__git_diff_submodule_formats in
180 * git-completion.bash when you add new formats.
181 */
c47ef57c
RR
182 else
183 return -1;
184 return 0;
185}
186
85b46030 187int git_config_rename(const char *var, const char *value)
cced5fbc
LT
188{
189 if (!value)
190 return DIFF_DETECT_RENAME;
191 if (!strcasecmp(value, "copies") || !strcasecmp(value, "copy"))
192 return DIFF_DETECT_COPY;
193 return git_config_bool(var,value) ? DIFF_DETECT_RENAME : 0;
194}
195
07924d4d 196long parse_algorithm_value(const char *value)
07ab4dec
MP
197{
198 if (!value)
199 return -1;
200 else if (!strcasecmp(value, "myers") || !strcasecmp(value, "default"))
201 return 0;
202 else if (!strcasecmp(value, "minimal"))
203 return XDF_NEED_MINIMAL;
204 else if (!strcasecmp(value, "patience"))
205 return XDF_PATIENCE_DIFF;
206 else if (!strcasecmp(value, "histogram"))
207 return XDF_HISTOGRAM_DIFF;
5a59a230
NTND
208 /*
209 * Please update $__git_diff_algorithms in git-completion.bash
210 * when you add new algorithms.
211 */
07ab4dec
MP
212 return -1;
213}
214
0b4b42e7
JH
215static int parse_one_token(const char **arg, const char *token)
216{
217 const char *rest;
218 if (skip_prefix(*arg, token, &rest) && (!*rest || *rest == ',')) {
219 *arg = rest;
220 return 1;
221 }
222 return 0;
223}
224
225static int parse_ws_error_highlight(const char *arg)
226{
227 const char *orig_arg = arg;
228 unsigned val = 0;
229
230 while (*arg) {
231 if (parse_one_token(&arg, "none"))
232 val = 0;
233 else if (parse_one_token(&arg, "default"))
234 val = WSEH_NEW;
235 else if (parse_one_token(&arg, "all"))
236 val = WSEH_NEW | WSEH_OLD | WSEH_CONTEXT;
237 else if (parse_one_token(&arg, "new"))
238 val |= WSEH_NEW;
239 else if (parse_one_token(&arg, "old"))
240 val |= WSEH_OLD;
241 else if (parse_one_token(&arg, "context"))
242 val |= WSEH_CONTEXT;
243 else {
244 return -1 - (int)(arg - orig_arg);
245 }
246 if (*arg)
247 arg++;
248 }
249 return val;
250}
251
83ad63cf
JH
252/*
253 * These are to give UI layer defaults.
254 * The core-level commands such as git-diff-files should
255 * never be affected by the setting of diff.renames
256 * the user happens to have in the configuration file.
257 */
5404c116
MM
258void init_diff_ui_defaults(void)
259{
06dba2b0 260 diff_detect_rename_default = DIFF_DETECT_RENAME;
5404c116
MM
261}
262
5b162879
MH
263int git_diff_heuristic_config(const char *var, const char *value, void *cb)
264{
3cde4e02 265 if (!strcmp(var, "diff.indentheuristic"))
5b162879 266 diff_indent_heuristic = git_config_bool(var, value);
5b162879
MH
267 return 0;
268}
269
2e2d5ac1
SB
270static int parse_color_moved(const char *arg)
271{
272 switch (git_parse_maybe_bool(arg)) {
273 case 0:
274 return COLOR_MOVED_NO;
275 case 1:
276 return COLOR_MOVED_DEFAULT;
277 default:
278 break;
279 }
280
281 if (!strcmp(arg, "no"))
282 return COLOR_MOVED_NO;
176841f0
SB
283 else if (!strcmp(arg, "plain"))
284 return COLOR_MOVED_PLAIN;
51da15eb
SB
285 else if (!strcmp(arg, "blocks"))
286 return COLOR_MOVED_BLOCKS;
2e2d5ac1
SB
287 else if (!strcmp(arg, "zebra"))
288 return COLOR_MOVED_ZEBRA;
289 else if (!strcmp(arg, "default"))
290 return COLOR_MOVED_DEFAULT;
e3f2f5f9
ES
291 else if (!strcmp(arg, "dimmed-zebra"))
292 return COLOR_MOVED_ZEBRA_DIM;
86b452e2
SB
293 else if (!strcmp(arg, "dimmed_zebra"))
294 return COLOR_MOVED_ZEBRA_DIM;
2e2d5ac1 295 else
706b0b5e 296 return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
2e2d5ac1
SB
297}
298
d173e799 299static unsigned parse_color_moved_ws(const char *arg)
b3095712
SB
300{
301 int ret = 0;
302 struct string_list l = STRING_LIST_INIT_DUP;
303 struct string_list_item *i;
304
305 string_list_split(&l, arg, ',', -1);
306
307 for_each_string_list_item(i, &l) {
308 struct strbuf sb = STRBUF_INIT;
309 strbuf_addstr(&sb, i->string);
310 strbuf_trim(&sb);
311
b73bcbac
PW
312 if (!strcmp(sb.buf, "no"))
313 ret = 0;
314 else if (!strcmp(sb.buf, "ignore-space-change"))
b3095712
SB
315 ret |= XDF_IGNORE_WHITESPACE_CHANGE;
316 else if (!strcmp(sb.buf, "ignore-space-at-eol"))
317 ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
318 else if (!strcmp(sb.buf, "ignore-all-space"))
319 ret |= XDF_IGNORE_WHITESPACE;
ca1f4ae4
SB
320 else if (!strcmp(sb.buf, "allow-indentation-change"))
321 ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
d173e799
SB
322 else {
323 ret |= COLOR_MOVED_WS_ERROR;
324 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);
325 }
b3095712
SB
326
327 strbuf_release(&sb);
328 }
329
ca1f4ae4 330 if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
d173e799 331 (ret & XDF_WHITESPACE_FLAGS)) {
15b07cba 332 error(_("color-moved-ws: allow-indentation-change cannot be combined with other whitespace modes"));
d173e799
SB
333 ret |= COLOR_MOVED_WS_ERROR;
334 }
ca1f4ae4 335
b3095712
SB
336 string_list_clear(&l, 0);
337
338 return ret;
2e2d5ac1
SB
339}
340
ef90d6d4 341int git_diff_ui_config(const char *var, const char *value, void *cb)
801235c5 342{
a159ca0c 343 if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
e269eb79 344 diff_use_color_default = git_config_colorbool(var, value);
801235c5
JH
345 return 0;
346 }
2e2d5ac1
SB
347 if (!strcmp(var, "diff.colormoved")) {
348 int cm = parse_color_moved(value);
349 if (cm < 0)
350 return -1;
351 diff_color_moved_default = cm;
352 return 0;
353 }
626c0b5d 354 if (!strcmp(var, "diff.colormovedws")) {
d173e799
SB
355 unsigned cm = parse_color_moved_ws(value);
356 if (cm & COLOR_MOVED_WS_ERROR)
626c0b5d
SB
357 return -1;
358 diff_color_moved_ws_default = cm;
359 return 0;
360 }
6468a4e5
JM
361 if (!strcmp(var, "diff.context")) {
362 diff_context_default = git_config_int(var, value);
363 if (diff_context_default < 0)
364 return -1;
365 return 0;
366 }
c4888677
VN
367 if (!strcmp(var, "diff.interhunkcontext")) {
368 diff_interhunk_context_default = git_config_int(var, value);
369 if (diff_interhunk_context_default < 0)
370 return -1;
371 return 0;
372 }
b68ea12e 373 if (!strcmp(var, "diff.renames")) {
cced5fbc 374 diff_detect_rename_default = git_config_rename(var, value);
b68ea12e
EW
375 return 0;
376 }
aecbf914
JH
377 if (!strcmp(var, "diff.autorefreshindex")) {
378 diff_auto_refresh_index = git_config_bool(var, value);
379 return 0;
380 }
a5a818ee
JH
381 if (!strcmp(var, "diff.mnemonicprefix")) {
382 diff_mnemonic_prefix = git_config_bool(var, value);
383 return 0;
384 }
f89504dd
EC
385 if (!strcmp(var, "diff.noprefix")) {
386 diff_no_prefix = git_config_bool(var, value);
387 return 0;
388 }
df44483a
ZJS
389 if (!strcmp(var, "diff.statgraphwidth")) {
390 diff_stat_graph_width = git_config_int(var, value);
391 return 0;
392 }
daec808c
BH
393 if (!strcmp(var, "diff.external"))
394 return git_config_string(&external_diff_cmd_cfg, var, value);
98a4d87b
BSSJ
395 if (!strcmp(var, "diff.wordregex"))
396 return git_config_string(&diff_word_regex_cfg, var, value);
6d8940b5
SB
397 if (!strcmp(var, "diff.orderfile"))
398 return git_config_pathname(&diff_order_file_cfg, var, value);
f1af60bd 399
be4f2b40
JS
400 if (!strcmp(var, "diff.ignoresubmodules"))
401 handle_ignore_submodules_arg(&default_diff_options, value);
402
c47ef57c
RR
403 if (!strcmp(var, "diff.submodule")) {
404 if (parse_submodule_params(&default_diff_options, value))
405 warning(_("Unknown value for 'diff.submodule' config variable: '%s'"),
406 value);
407 return 0;
408 }
409
07ab4dec
MP
410 if (!strcmp(var, "diff.algorithm")) {
411 diff_algorithm = parse_algorithm_value(value);
412 if (diff_algorithm < 0)
413 return -1;
414 return 0;
415 }
416
a17505f2
JH
417 if (!strcmp(var, "diff.wserrorhighlight")) {
418 int val = parse_ws_error_highlight(value);
419 if (val < 0)
420 return -1;
421 ws_error_highlight_default = val;
422 return 0;
423 }
424
33c643bb
JK
425 if (git_color_config(var, value, cb) < 0)
426 return -1;
427
ef90d6d4 428 return git_diff_basic_config(var, value, cb);
9a1805a8
JK
429}
430
ef90d6d4 431int git_diff_basic_config(const char *var, const char *value, void *cb)
9a1805a8 432{
ae021d87
JK
433 const char *name;
434
2b6ca6df
LT
435 if (!strcmp(var, "diff.renamelimit")) {
436 diff_rename_limit_default = git_config_int(var, value);
437 return 0;
438 }
439
6680a087
JK
440 if (userdiff_config(var, value) < 0)
441 return -1;
c7534ef4 442
ae021d87
JK
443 if (skip_prefix(var, "diff.color.", &name) ||
444 skip_prefix(var, "color.diff.", &name)) {
445 int slot = parse_diff_color_slot(name);
8b8e8624
JK
446 if (slot < 0)
447 return 0;
64f30e94
JH
448 if (!value)
449 return config_error_nonbool(var);
f6c5a296 450 return color_parse(value, diff_colors[slot]);
801235c5 451 }
f1af60bd 452
a624eaa7 453 /* like GNU diff's --suppress-blank-empty option */
950db879
JS
454 if (!strcmp(var, "diff.suppressblankempty") ||
455 /* for backwards compatibility */
456 !strcmp(var, "diff.suppress-blank-empty")) {
a624eaa7
JM
457 diff_suppress_blank_empty = git_config_bool(var, value);
458 return 0;
459 }
460
2d174951 461 if (!strcmp(var, "diff.dirstat")) {
51670fc8 462 struct strbuf errmsg = STRBUF_INIT;
712d2c7d 463 default_diff_options.dirstat_permille = diff_dirstat_permille_default;
51670fc8 464 if (parse_dirstat_params(&default_diff_options, value, &errmsg))
7478ac57 465 warning(_("Found errors in 'diff.dirstat' config variable:\n%s"),
51670fc8
JH
466 errmsg.buf);
467 strbuf_release(&errmsg);
712d2c7d 468 diff_dirstat_permille_default = default_diff_options.dirstat_permille;
2d174951
JH
469 return 0;
470 }
471
cf5e7722
MB
472 if (git_diff_heuristic_config(var, value, cb) < 0)
473 return -1;
474
3e1dd17a 475 return git_default_config(var, value, cb);
801235c5
JH
476}
477
6973dcae
JH
478static char *quote_two(const char *one, const char *two)
479{
480 int need_one = quote_c_style(one, NULL, NULL, 1);
481 int need_two = quote_c_style(two, NULL, NULL, 1);
f285a2d7 482 struct strbuf res = STRBUF_INIT;
6973dcae
JH
483
484 if (need_one + need_two) {
663af342
PH
485 strbuf_addch(&res, '"');
486 quote_c_style(one, &res, NULL, 1);
487 quote_c_style(two, &res, NULL, 1);
488 strbuf_addch(&res, '"');
489 } else {
490 strbuf_addstr(&res, one);
491 strbuf_addstr(&res, two);
6973dcae 492 }
b315c5c0 493 return strbuf_detach(&res, NULL);
6973dcae
JH
494}
495
496static const char *external_diff(void)
497{
498 static const char *external_diff_cmd = NULL;
499 static int done_preparing = 0;
500
501 if (done_preparing)
502 return external_diff_cmd;
6776a84d 503 external_diff_cmd = xstrdup_or_null(getenv("GIT_EXTERNAL_DIFF"));
cbe02100
JS
504 if (!external_diff_cmd)
505 external_diff_cmd = external_diff_cmd_cfg;
6973dcae
JH
506 done_preparing = 1;
507 return external_diff_cmd;
508}
509
284098f1
MH
510/*
511 * Keep track of files used for diffing. Sometimes such an entry
512 * refers to a temporary file, sometimes to an existing file, and
513 * sometimes to "/dev/null".
514 */
6973dcae 515static struct diff_tempfile {
284098f1
MH
516 /*
517 * filename external diff should read from, or NULL if this
518 * entry is currently not in use:
519 */
520 const char *name;
521
dc01505f 522 char hex[GIT_MAX_HEXSZ + 1];
6973dcae 523 char mode[10];
284098f1
MH
524
525 /*
526 * If this diff_tempfile instance refers to a temporary file,
527 * this tempfile object is used to manage its lifetime.
528 */
076aa2cb 529 struct tempfile *tempfile;
6973dcae
JH
530} diff_temp[2];
531
6957eb9a 532struct emit_callback {
6957eb9a
JH
533 int color_diff;
534 unsigned ws_rule;
535 int blank_at_eof_in_preimage;
536 int blank_at_eof_in_postimage;
537 int lno_in_preimage;
538 int lno_in_postimage;
6957eb9a
JH
539 const char **label_path;
540 struct diff_words_data *diff_words;
a3c158d4 541 struct diff_options *opt;
3e97c7c6 542 struct strbuf *header;
6957eb9a
JH
543};
544
6973dcae
JH
545static int count_lines(const char *data, int size)
546{
547 int count, ch, completely_empty = 1, nl_just_seen = 0;
548 count = 0;
549 while (0 < size--) {
550 ch = *data++;
551 if (ch == '\n') {
552 count++;
553 nl_just_seen = 1;
554 completely_empty = 0;
555 }
556 else {
557 nl_just_seen = 0;
558 completely_empty = 0;
559 }
560 }
561 if (completely_empty)
562 return 0;
563 if (!nl_just_seen)
564 count++; /* no trailing newline */
565 return count;
566}
567
b78ea5fc
NTND
568static int fill_mmfile(struct repository *r, mmfile_t *mf,
569 struct diff_filespec *one)
6957eb9a
JH
570{
571 if (!DIFF_FILE_VALID(one)) {
572 mf->ptr = (char *)""; /* does not matter */
573 mf->size = 0;
574 return 0;
575 }
b78ea5fc 576 else if (diff_populate_filespec(r, one, 0))
6957eb9a 577 return -1;
bb35fefb 578
6957eb9a
JH
579 mf->ptr = one->data;
580 mf->size = one->size;
581 return 0;
582}
583
abb371a1 584/* like fill_mmfile, but only for size, so we can avoid retrieving blob */
b78ea5fc
NTND
585static unsigned long diff_filespec_size(struct repository *r,
586 struct diff_filespec *one)
abb371a1
JK
587{
588 if (!DIFF_FILE_VALID(one))
589 return 0;
b78ea5fc 590 diff_populate_filespec(r, one, CHECK_SIZE_ONLY);
abb371a1
JK
591 return one->size;
592}
593
6957eb9a
JH
594static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
595{
596 char *ptr = mf->ptr;
597 long size = mf->size;
598 int cnt = 0;
599
600 if (!size)
601 return cnt;
602 ptr += size - 1; /* pointing at the very end */
603 if (*ptr != '\n')
604 ; /* incomplete line */
605 else
606 ptr--; /* skip the last LF */
607 while (mf->ptr < ptr) {
608 char *prev_eol;
609 for (prev_eol = ptr; mf->ptr <= prev_eol; prev_eol--)
610 if (*prev_eol == '\n')
611 break;
612 if (!ws_blank_line(prev_eol + 1, ptr - prev_eol, ws_rule))
613 break;
614 cnt++;
615 ptr = prev_eol - 1;
616 }
617 return cnt;
618}
619
620static void check_blank_at_eof(mmfile_t *mf1, mmfile_t *mf2,
621 struct emit_callback *ecbdata)
622{
623 int l1, l2, at;
624 unsigned ws_rule = ecbdata->ws_rule;
625 l1 = count_trailing_blank(mf1, ws_rule);
626 l2 = count_trailing_blank(mf2, ws_rule);
627 if (l2 <= l1) {
628 ecbdata->blank_at_eof_in_preimage = 0;
629 ecbdata->blank_at_eof_in_postimage = 0;
630 return;
631 }
632 at = count_lines(mf1->ptr, mf1->size);
633 ecbdata->blank_at_eof_in_preimage = (at - l1) + 1;
634
635 at = count_lines(mf2->ptr, mf2->size);
636 ecbdata->blank_at_eof_in_postimage = (at - l2) + 1;
637}
638
f7c3b4e2 639static void emit_line_0(struct diff_options *o,
017ac45e 640 const char *set_sign, const char *set, unsigned reverse, const char *reset,
250f7993 641 int first, const char *line, int len)
6957eb9a
JH
642{
643 int has_trailing_newline, has_trailing_carriage_return;
44410679 644 int needs_reset = 0; /* at the end of the line */
a3c158d4
BY
645 FILE *file = o->file;
646
f103a6fa 647 fputs(diff_line_prefix(o), file);
6957eb9a 648
44410679
SB
649 has_trailing_newline = (len > 0 && line[len-1] == '\n');
650 if (has_trailing_newline)
651 len--;
652
653 has_trailing_carriage_return = (len > 0 && line[len-1] == '\r');
654 if (has_trailing_carriage_return)
655 len--;
656
657 if (!len && !first)
658 goto end_of_line;
659
660 if (reverse && want_color(o->use_color)) {
661 fputs(GIT_COLOR_REVERSE, file);
662 needs_reset = 1;
250f7993 663 }
6957eb9a 664
44410679
SB
665 if (set_sign) {
666 fputs(set_sign, file);
667 needs_reset = 1;
06a47552 668 }
44410679
SB
669
670 if (first)
671 fputc(first, file);
672
673 if (!len)
674 goto end_of_line;
675
676 if (set) {
677 if (set_sign && set != set_sign)
678 fputs(reset, file);
06a47552 679 fputs(set, file);
44410679 680 needs_reset = 1;
06a47552 681 }
44410679
SB
682 fwrite(line, len, 1, file);
683 needs_reset = 1; /* 'line' may contain color codes. */
684
685end_of_line:
686 if (needs_reset)
687 fputs(reset, file);
6957eb9a
JH
688 if (has_trailing_carriage_return)
689 fputc('\r', file);
690 if (has_trailing_newline)
691 fputc('\n', file);
692}
693
a3c158d4 694static void emit_line(struct diff_options *o, const char *set, const char *reset,
250f7993
JH
695 const char *line, int len)
696{
44410679 697 emit_line_0(o, set, NULL, 0, reset, 0, line, len);
250f7993
JH
698}
699
36a4cefd 700enum diff_symbol {
4eed0ebd
SB
701 DIFF_SYMBOL_BINARY_DIFF_HEADER,
702 DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
703 DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
704 DIFF_SYMBOL_BINARY_DIFF_BODY,
705 DIFF_SYMBOL_BINARY_DIFF_FOOTER,
0911c475
SB
706 DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
707 DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
708 DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
709 DIFF_SYMBOL_STATS_LINE,
bd033291 710 DIFF_SYMBOL_WORD_DIFF,
30b7e1e7 711 DIFF_SYMBOL_STAT_SEP,
146fdb0d 712 DIFF_SYMBOL_SUMMARY,
f3597138
SB
713 DIFF_SYMBOL_SUBMODULE_ADD,
714 DIFF_SYMBOL_SUBMODULE_DEL,
715 DIFF_SYMBOL_SUBMODULE_UNTRACKED,
716 DIFF_SYMBOL_SUBMODULE_MODIFIED,
717 DIFF_SYMBOL_SUBMODULE_HEADER,
718 DIFF_SYMBOL_SUBMODULE_ERROR,
719 DIFF_SYMBOL_SUBMODULE_PIPETHROUGH,
5af6ea95 720 DIFF_SYMBOL_REWRITE_DIFF,
4acaaa7a 721 DIFF_SYMBOL_BINARY_FILES,
a29b0a13 722 DIFF_SYMBOL_HEADER,
3ee8b7bf
SB
723 DIFF_SYMBOL_FILEPAIR_PLUS,
724 DIFF_SYMBOL_FILEPAIR_MINUS,
ff958679
SB
725 DIFF_SYMBOL_WORDS_PORCELAIN,
726 DIFF_SYMBOL_WORDS,
091f8e28 727 DIFF_SYMBOL_CONTEXT,
f2bb1218 728 DIFF_SYMBOL_CONTEXT_INCOMPLETE,
091f8e28
SB
729 DIFF_SYMBOL_PLUS,
730 DIFF_SYMBOL_MINUS,
b9cbfde6 731 DIFF_SYMBOL_NO_LF_EOF,
68abc6f1 732 DIFF_SYMBOL_CONTEXT_FRAGINFO,
c64b420b 733 DIFF_SYMBOL_CONTEXT_MARKER,
36a4cefd
SB
734 DIFF_SYMBOL_SEPARATOR
735};
091f8e28
SB
736/*
737 * Flags for content lines:
738 * 0..12 are whitespace rules
739 * 13-15 are WSEH_NEW | WSEH_OLD | WSEH_CONTEXT
740 * 16 is marking if the line is blank at EOF
741 */
2e2d5ac1
SB
742#define DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF (1<<16)
743#define DIFF_SYMBOL_MOVED_LINE (1<<17)
744#define DIFF_SYMBOL_MOVED_LINE_ALT (1<<18)
86b452e2 745#define DIFF_SYMBOL_MOVED_LINE_UNINTERESTING (1<<19)
091f8e28
SB
746#define DIFF_SYMBOL_CONTENT_WS_MASK (WSEH_NEW | WSEH_OLD | WSEH_CONTEXT | WS_RULE_MASK)
747
e6e045f8
SB
748/*
749 * This struct is used when we need to buffer the output of the diff output.
750 *
751 * NEEDSWORK: Instead of storing a copy of the line, add an offset pointer
752 * into the pre/post image file. This pointer could be a union with the
753 * line pointer. By storing an offset into the file instead of the literal line,
754 * we can decrease the memory footprint for the buffered output. At first we
755 * may want to only have indirection for the content lines, but we could also
756 * enhance the state for emitting prefabricated lines, e.g. the similarity
757 * score line or hunk/file headers would only need to store a number or path
758 * and then the output can be constructed later on depending on state.
759 */
760struct emitted_diff_symbol {
761 const char *line;
762 int len;
763 int flags;
21536d07
PW
764 int indent_off; /* Offset to first non-whitespace character */
765 int indent_width; /* The visual width of the indentation */
e6e045f8
SB
766 enum diff_symbol s;
767};
768#define EMITTED_DIFF_SYMBOL_INIT {NULL}
769
770struct emitted_diff_symbols {
771 struct emitted_diff_symbol *buf;
772 int nr, alloc;
773};
774#define EMITTED_DIFF_SYMBOLS_INIT {NULL, 0, 0}
775
776static void append_emitted_diff_symbol(struct diff_options *o,
777 struct emitted_diff_symbol *e)
6957eb9a 778{
e6e045f8
SB
779 struct emitted_diff_symbol *f;
780
781 ALLOC_GROW(o->emitted_symbols->buf,
782 o->emitted_symbols->nr + 1,
783 o->emitted_symbols->alloc);
784 f = &o->emitted_symbols->buf[o->emitted_symbols->nr++];
785
786 memcpy(f, e, sizeof(struct emitted_diff_symbol));
787 f->line = e->line ? xmemdupz(e->line, e->len) : NULL;
6957eb9a
JH
788}
789
2e2d5ac1
SB
790struct moved_entry {
791 struct hashmap_entry ent;
792 const struct emitted_diff_symbol *es;
793 struct moved_entry *next_line;
794};
795
74d156f4
PW
796struct moved_block {
797 struct moved_entry *match;
21536d07 798 int wsd; /* The whitespace delta of this block */
74d156f4
PW
799};
800
801static void moved_block_clear(struct moved_block *b)
802{
21536d07 803 memset(b, 0, sizeof(*b));
74d156f4
PW
804}
805
0cd51e9d
PW
806#define INDENT_BLANKLINE INT_MIN
807
21536d07 808static void fill_es_indent_data(struct emitted_diff_symbol *es)
ca1f4ae4 809{
0cd51e9d 810 unsigned int off = 0, i;
21536d07
PW
811 int width = 0, tab_width = es->flags & WS_TAB_WIDTH_MASK;
812 const char *s = es->line;
813 const int len = es->len;
814
815 /* skip any \v \f \r at start of indentation */
816 while (s[off] == '\f' || s[off] == '\v' ||
817 (s[off] == '\r' && off < len - 1))
818 off++;
819
820 /* calculate the visual width of indentation */
821 while(1) {
822 if (s[off] == ' ') {
823 width++;
824 off++;
825 } else if (s[off] == '\t') {
826 width += tab_width - (width % tab_width);
827 while (s[++off] == '\t')
828 width += tab_width;
829 } else {
830 break;
831 }
832 }
833
0cd51e9d
PW
834 /* check if this line is blank */
835 for (i = off; i < len; i++)
836 if (!isspace(s[i]))
837 break;
ca1f4ae4 838
0cd51e9d
PW
839 if (i == len) {
840 es->indent_width = INDENT_BLANKLINE;
841 es->indent_off = len;
842 } else {
843 es->indent_off = off;
844 es->indent_width = width;
845 }
74d156f4
PW
846}
847
ca1f4ae4 848static int compute_ws_delta(const struct emitted_diff_symbol *a,
21536d07
PW
849 const struct emitted_diff_symbol *b,
850 int *out)
ca1f4ae4 851{
21536d07
PW
852 int a_len = a->len,
853 b_len = b->len,
854 a_off = a->indent_off,
855 a_width = a->indent_width,
856 b_off = b->indent_off,
857 b_width = b->indent_width;
858 int delta;
859
0cd51e9d
PW
860 if (a_width == INDENT_BLANKLINE && b_width == INDENT_BLANKLINE) {
861 *out = INDENT_BLANKLINE;
862 return 1;
863 }
864
21536d07
PW
865 if (a->s == DIFF_SYMBOL_PLUS)
866 delta = a_width - b_width;
867 else
868 delta = b_width - a_width;
ca1f4ae4 869
21536d07
PW
870 if (a_len - a_off != b_len - b_off ||
871 memcmp(a->line + a_off, b->line + b_off, a_len - a_off))
fe4516d1
PW
872 return 0;
873
21536d07 874 *out = delta;
ca1f4ae4 875
fe4516d1 876 return 1;
ca1f4ae4
SB
877}
878
879static int cmp_in_block_with_wsd(const struct diff_options *o,
880 const struct moved_entry *cur,
881 const struct moved_entry *match,
74d156f4 882 struct moved_block *pmb,
ca1f4ae4
SB
883 int n)
884{
885 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
7a4252c4 886 int al = cur->es->len, bl = match->es->len, cl = l->len;
ca1f4ae4
SB
887 const char *a = cur->es->line,
888 *b = match->es->line,
889 *c = l->line;
21536d07
PW
890 int a_off = cur->es->indent_off,
891 a_width = cur->es->indent_width,
892 c_off = l->indent_off,
893 c_width = l->indent_width;
894 int delta;
ca1f4ae4
SB
895
896 /*
7a4252c4
PW
897 * We need to check if 'cur' is equal to 'match'. As those
898 * are from the same (+/-) side, we do not need to adjust for
899 * indent changes. However these were found using fuzzy
900 * matching so we do have to check if they are equal. Here we
901 * just check the lengths. We delay calling memcmp() to check
902 * the contents until later as if the length comparison for a
903 * and c fails we can avoid the call all together.
ca1f4ae4 904 */
7a4252c4 905 if (al != bl)
ca1f4ae4
SB
906 return 1;
907
0cd51e9d
PW
908 /* If 'l' and 'cur' are both blank then they match. */
909 if (a_width == INDENT_BLANKLINE && c_width == INDENT_BLANKLINE)
910 return 0;
ca1f4ae4
SB
911
912 /*
21536d07
PW
913 * The indent changes of the block are known and stored in pmb->wsd;
914 * however we need to check if the indent changes of the current line
915 * match those of the current block and that the text of 'l' and 'cur'
916 * after the indentation match.
ca1f4ae4 917 */
21536d07
PW
918 if (cur->es->s == DIFF_SYMBOL_PLUS)
919 delta = a_width - c_width;
920 else
921 delta = c_width - a_width;
ca1f4ae4 922
0cd51e9d
PW
923 /*
924 * If the previous lines of this block were all blank then set its
925 * whitespace delta.
926 */
927 if (pmb->wsd == INDENT_BLANKLINE)
928 pmb->wsd = delta;
ca1f4ae4 929
21536d07
PW
930 return !(delta == pmb->wsd && al - a_off == cl - c_off &&
931 !memcmp(a, b, al) && !
932 memcmp(a + a_off, c + c_off, al - a_off));
ca1f4ae4 933}
2e2d5ac1 934
ee1df66f
SB
935static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
936 const void *entry,
937 const void *entry_or_key,
2e2d5ac1
SB
938 const void *keydata)
939{
ee1df66f
SB
940 const struct diff_options *diffopt = hashmap_cmp_fn_data;
941 const struct moved_entry *a = entry;
942 const struct moved_entry *b = entry_or_key;
b3095712
SB
943 unsigned flags = diffopt->color_moved_ws_handling
944 & XDF_WHITESPACE_FLAGS;
ee1df66f 945
ca1f4ae4
SB
946 if (diffopt->color_moved_ws_handling &
947 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
948 /*
949 * As there is not specific white space config given,
950 * we'd need to check for a new block, so ignore all
951 * white space. The setup of the white space
952 * configuration for the next block is done else where
953 */
954 flags |= XDF_IGNORE_WHITESPACE;
955
01be97c2
SB
956 return !xdiff_compare_lines(a->es->line, a->es->len,
957 b->es->line, b->es->len,
b3095712 958 flags);
2e2d5ac1
SB
959}
960
961static struct moved_entry *prepare_entry(struct diff_options *o,
962 int line_no)
963{
964 struct moved_entry *ret = xmalloc(sizeof(*ret));
965 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
b3095712 966 unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
2e2d5ac1 967
b3095712 968 ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
2e2d5ac1
SB
969 ret->es = l;
970 ret->next_line = NULL;
971
972 return ret;
973}
974
975static void add_lines_to_move_detection(struct diff_options *o,
976 struct hashmap *add_lines,
977 struct hashmap *del_lines)
978{
979 struct moved_entry *prev_line = NULL;
980
981 int n;
982 for (n = 0; n < o->emitted_symbols->nr; n++) {
983 struct hashmap *hm;
984 struct moved_entry *key;
985
986 switch (o->emitted_symbols->buf[n].s) {
987 case DIFF_SYMBOL_PLUS:
988 hm = add_lines;
989 break;
990 case DIFF_SYMBOL_MINUS:
991 hm = del_lines;
992 break;
993 default:
994 prev_line = NULL;
995 continue;
996 }
997
21536d07
PW
998 if (o->color_moved_ws_handling &
999 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1000 fill_es_indent_data(&o->emitted_symbols->buf[n]);
2e2d5ac1
SB
1001 key = prepare_entry(o, n);
1002 if (prev_line && prev_line->es->s == o->emitted_symbols->buf[n].s)
1003 prev_line->next_line = key;
1004
1005 hashmap_add(hm, key);
1006 prev_line = key;
1007 }
1008}
1009
e2fe6abc
SB
1010static void pmb_advance_or_null(struct diff_options *o,
1011 struct moved_entry *match,
1012 struct hashmap *hm,
74d156f4 1013 struct moved_block *pmb,
e2fe6abc
SB
1014 int pmb_nr)
1015{
1016 int i;
1017 for (i = 0; i < pmb_nr; i++) {
74d156f4 1018 struct moved_entry *prev = pmb[i].match;
e2fe6abc
SB
1019 struct moved_entry *cur = (prev && prev->next_line) ?
1020 prev->next_line : NULL;
1021 if (cur && !hm->cmpfn(o, cur, match, NULL)) {
74d156f4 1022 pmb[i].match = cur;
e2fe6abc 1023 } else {
74d156f4 1024 pmb[i].match = NULL;
e2fe6abc
SB
1025 }
1026 }
1027}
1028
ca1f4ae4
SB
1029static void pmb_advance_or_null_multi_match(struct diff_options *o,
1030 struct moved_entry *match,
1031 struct hashmap *hm,
74d156f4 1032 struct moved_block *pmb,
ca1f4ae4
SB
1033 int pmb_nr, int n)
1034{
1035 int i;
1036 char *got_match = xcalloc(1, pmb_nr);
1037
1038 for (; match; match = hashmap_get_next(hm, match)) {
1039 for (i = 0; i < pmb_nr; i++) {
74d156f4 1040 struct moved_entry *prev = pmb[i].match;
ca1f4ae4
SB
1041 struct moved_entry *cur = (prev && prev->next_line) ?
1042 prev->next_line : NULL;
1043 if (!cur)
1044 continue;
74d156f4 1045 if (!cmp_in_block_with_wsd(o, cur, match, &pmb[i], n))
ca1f4ae4
SB
1046 got_match[i] |= 1;
1047 }
1048 }
1049
1050 for (i = 0; i < pmb_nr; i++) {
1051 if (got_match[i]) {
74d156f4
PW
1052 /* Advance to the next line */
1053 pmb[i].match = pmb[i].match->next_line;
fab01ec5 1054 } else {
74d156f4 1055 moved_block_clear(&pmb[i]);
fab01ec5 1056 }
ca1f4ae4 1057 }
9c1a6c2b
PW
1058
1059 free(got_match);
ca1f4ae4
SB
1060}
1061
74d156f4 1062static int shrink_potential_moved_blocks(struct moved_block *pmb,
2e2d5ac1
SB
1063 int pmb_nr)
1064{
1065 int lp, rp;
1066
1067 /* Shrink the set of potential block to the remaining running */
1068 for (lp = 0, rp = pmb_nr - 1; lp <= rp;) {
74d156f4 1069 while (lp < pmb_nr && pmb[lp].match)
2e2d5ac1
SB
1070 lp++;
1071 /* lp points at the first NULL now */
1072
74d156f4 1073 while (rp > -1 && !pmb[rp].match)
2e2d5ac1
SB
1074 rp--;
1075 /* rp points at the last non-NULL */
1076
1077 if (lp < pmb_nr && rp > -1 && lp < rp) {
1078 pmb[lp] = pmb[rp];
21536d07 1079 memset(&pmb[rp], 0, sizeof(pmb[rp]));
2e2d5ac1
SB
1080 rp--;
1081 lp++;
1082 }
1083 }
1084
1085 /* Remember the number of running sets */
1086 return rp + 1;
1087}
1088
09153277
JT
1089/*
1090 * If o->color_moved is COLOR_MOVED_PLAIN, this function does nothing.
1091 *
f0b8fb6e
JT
1092 * Otherwise, if the last block has fewer alphanumeric characters than
1093 * COLOR_MOVED_MIN_ALNUM_COUNT, unset DIFF_SYMBOL_MOVED_LINE on all lines in
09153277
JT
1094 * that block.
1095 *
1096 * The last block consists of the (n - block_length)'th line up to but not
1097 * including the nth line.
f0b8fb6e 1098 *
b0a2ba47
PW
1099 * Returns 0 if the last block is empty or is unset by this function, non zero
1100 * otherwise.
1101 *
f0b8fb6e
JT
1102 * NEEDSWORK: This uses the same heuristic as blame_entry_score() in blame.c.
1103 * Think of a way to unify them.
09153277 1104 */
b0a2ba47 1105static int adjust_last_block(struct diff_options *o, int n, int block_length)
09153277 1106{
f0b8fb6e
JT
1107 int i, alnum_count = 0;
1108 if (o->color_moved == COLOR_MOVED_PLAIN)
b0a2ba47 1109 return block_length;
f0b8fb6e
JT
1110 for (i = 1; i < block_length + 1; i++) {
1111 const char *c = o->emitted_symbols->buf[n - i].line;
1112 for (; *c; c++) {
1113 if (!isalnum(*c))
1114 continue;
1115 alnum_count++;
1116 if (alnum_count >= COLOR_MOVED_MIN_ALNUM_COUNT)
b0a2ba47 1117 return 1;
f0b8fb6e
JT
1118 }
1119 }
09153277
JT
1120 for (i = 1; i < block_length + 1; i++)
1121 o->emitted_symbols->buf[n - i].flags &= ~DIFF_SYMBOL_MOVED_LINE;
b0a2ba47 1122 return 0;
09153277
JT
1123}
1124
2e2d5ac1
SB
1125/* Find blocks of moved code, delegate actual coloring decision to helper */
1126static void mark_color_as_moved(struct diff_options *o,
1127 struct hashmap *add_lines,
1128 struct hashmap *del_lines)
1129{
74d156f4 1130 struct moved_block *pmb = NULL; /* potentially moved blocks */
2e2d5ac1 1131 int pmb_nr = 0, pmb_alloc = 0;
b0a2ba47 1132 int n, flipped_block = 0, block_length = 0;
2e2d5ac1
SB
1133
1134
1135 for (n = 0; n < o->emitted_symbols->nr; n++) {
1136 struct hashmap *hm = NULL;
1137 struct moved_entry *key;
1138 struct moved_entry *match = NULL;
1139 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
b0a2ba47 1140 enum diff_symbol last_symbol = 0;
2e2d5ac1
SB
1141
1142 switch (l->s) {
1143 case DIFF_SYMBOL_PLUS:
1144 hm = del_lines;
1145 key = prepare_entry(o, n);
3783aad4 1146 match = hashmap_get(hm, key, NULL);
2e2d5ac1
SB
1147 free(key);
1148 break;
1149 case DIFF_SYMBOL_MINUS:
1150 hm = add_lines;
1151 key = prepare_entry(o, n);
3783aad4 1152 match = hashmap_get(hm, key, NULL);
2e2d5ac1
SB
1153 free(key);
1154 break;
1155 default:
b0a2ba47 1156 flipped_block = 0;
2e2d5ac1
SB
1157 }
1158
1159 if (!match) {
74d156f4
PW
1160 int i;
1161
09153277 1162 adjust_last_block(o, n, block_length);
74d156f4
PW
1163 for(i = 0; i < pmb_nr; i++)
1164 moved_block_clear(&pmb[i]);
2e2d5ac1
SB
1165 pmb_nr = 0;
1166 block_length = 0;
b0a2ba47
PW
1167 flipped_block = 0;
1168 last_symbol = l->s;
2e2d5ac1
SB
1169 continue;
1170 }
1171
2034b473 1172 if (o->color_moved == COLOR_MOVED_PLAIN) {
b0a2ba47 1173 last_symbol = l->s;
2034b473 1174 l->flags |= DIFF_SYMBOL_MOVED_LINE;
176841f0 1175 continue;
2034b473 1176 }
176841f0 1177
ca1f4ae4
SB
1178 if (o->color_moved_ws_handling &
1179 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
1180 pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
1181 else
1182 pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
2e2d5ac1
SB
1183
1184 pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
1185
1186 if (pmb_nr == 0) {
1187 /*
1188 * The current line is the start of a new block.
1189 * Setup the set of potential blocks.
1190 */
1191 for (; match; match = hashmap_get_next(hm, match)) {
1192 ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
ca1f4ae4
SB
1193 if (o->color_moved_ws_handling &
1194 COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
74d156f4
PW
1195 if (compute_ws_delta(l, match->es,
1196 &pmb[pmb_nr].wsd))
1197 pmb[pmb_nr++].match = match;
ca1f4ae4 1198 } else {
21536d07 1199 pmb[pmb_nr].wsd = 0;
74d156f4 1200 pmb[pmb_nr++].match = match;
ca1f4ae4 1201 }
2e2d5ac1
SB
1202 }
1203
b0a2ba47
PW
1204 if (adjust_last_block(o, n, block_length) &&
1205 pmb_nr && last_symbol != l->s)
1206 flipped_block = (flipped_block + 1) % 2;
1207 else
1208 flipped_block = 0;
f0b8fb6e 1209
f0b8fb6e 1210 block_length = 0;
2e2d5ac1
SB
1211 }
1212
2034b473
PW
1213 if (pmb_nr) {
1214 block_length++;
2034b473
PW
1215 l->flags |= DIFF_SYMBOL_MOVED_LINE;
1216 if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
1217 l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
1218 }
b0a2ba47 1219 last_symbol = l->s;
2e2d5ac1 1220 }
09153277 1221 adjust_last_block(o, n, block_length);
2e2d5ac1 1222
74d156f4
PW
1223 for(n = 0; n < pmb_nr; n++)
1224 moved_block_clear(&pmb[n]);
2e2d5ac1
SB
1225 free(pmb);
1226}
e6e045f8 1227
86b452e2
SB
1228#define DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK \
1229 (DIFF_SYMBOL_MOVED_LINE | DIFF_SYMBOL_MOVED_LINE_ALT)
1230static void dim_moved_lines(struct diff_options *o)
1231{
1232 int n;
1233 for (n = 0; n < o->emitted_symbols->nr; n++) {
1234 struct emitted_diff_symbol *prev = (n != 0) ?
1235 &o->emitted_symbols->buf[n - 1] : NULL;
1236 struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
1237 struct emitted_diff_symbol *next =
1238 (n < o->emitted_symbols->nr - 1) ?
1239 &o->emitted_symbols->buf[n + 1] : NULL;
1240
1241 /* Not a plus or minus line? */
1242 if (l->s != DIFF_SYMBOL_PLUS && l->s != DIFF_SYMBOL_MINUS)
1243 continue;
1244
1245 /* Not a moved line? */
1246 if (!(l->flags & DIFF_SYMBOL_MOVED_LINE))
1247 continue;
1248
1249 /*
1250 * If prev or next are not a plus or minus line,
1251 * pretend they don't exist
1252 */
1253 if (prev && prev->s != DIFF_SYMBOL_PLUS &&
1254 prev->s != DIFF_SYMBOL_MINUS)
1255 prev = NULL;
1256 if (next && next->s != DIFF_SYMBOL_PLUS &&
1257 next->s != DIFF_SYMBOL_MINUS)
1258 next = NULL;
1259
1260 /* Inside a block? */
1261 if ((prev &&
1262 (prev->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1263 (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK)) &&
1264 (next &&
1265 (next->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK) ==
1266 (l->flags & DIFF_SYMBOL_MOVED_LINE_ZEBRA_MASK))) {
1267 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1268 continue;
1269 }
1270
1271 /* Check if we are at an interesting bound: */
1272 if (prev && (prev->flags & DIFF_SYMBOL_MOVED_LINE) &&
1273 (prev->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1274 (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1275 continue;
1276 if (next && (next->flags & DIFF_SYMBOL_MOVED_LINE) &&
1277 (next->flags & DIFF_SYMBOL_MOVED_LINE_ALT) !=
1278 (l->flags & DIFF_SYMBOL_MOVED_LINE_ALT))
1279 continue;
1280
1281 /*
1282 * The boundary to prev and next are not interesting,
1283 * so this line is not interesting as a whole
1284 */
1285 l->flags |= DIFF_SYMBOL_MOVED_LINE_UNINTERESTING;
1286 }
1287}
1288
091f8e28 1289static void emit_line_ws_markup(struct diff_options *o,
9d1e16bf
SB
1290 const char *set_sign, const char *set,
1291 const char *reset,
bc9feb05 1292 int sign_index, const char *line, int len,
091f8e28 1293 unsigned ws_rule, int blank_at_eof)
6957eb9a 1294{
b8767f79 1295 const char *ws = NULL;
bc9feb05 1296 int sign = o->output_indicators[sign_index];
6957eb9a 1297
091f8e28
SB
1298 if (o->ws_error_highlight & ws_rule) {
1299 ws = diff_get_color_opt(o, DIFF_WHITESPACE);
b8767f79
JH
1300 if (!*ws)
1301 ws = NULL;
1302 }
1303
f7c3b4e2 1304 if (!ws && !set_sign)
017ac45e 1305 emit_line_0(o, set, NULL, 0, reset, sign, line, len);
f7c3b4e2 1306 else if (!ws) {
29ef759d 1307 emit_line_0(o, set_sign, set, !!set_sign, reset, sign, line, len);
f7c3b4e2 1308 } else if (blank_at_eof)
6957eb9a 1309 /* Blank line at EOF - paint '+' as well */
017ac45e 1310 emit_line_0(o, ws, NULL, 0, reset, sign, line, len);
6957eb9a
JH
1311 else {
1312 /* Emit just the prefix, then the rest. */
017ac45e 1313 emit_line_0(o, set_sign ? set_sign : set, NULL, !!set_sign, reset,
f7c3b4e2 1314 sign, "", 0);
091f8e28
SB
1315 ws_check_emit(line, len, ws_rule,
1316 o->file, set, reset, ws);
6957eb9a
JH
1317 }
1318}
1319
e6e045f8
SB
1320static void emit_diff_symbol_from_struct(struct diff_options *o,
1321 struct emitted_diff_symbol *eds)
36a4cefd 1322{
b9cbfde6 1323 static const char *nneof = " No newline at end of file\n";
f7c3b4e2 1324 const char *context, *reset, *set, *set_sign, *meta, *fraginfo;
0911c475 1325 struct strbuf sb = STRBUF_INIT;
e6e045f8
SB
1326
1327 enum diff_symbol s = eds->s;
1328 const char *line = eds->line;
1329 int len = eds->len;
1330 unsigned flags = eds->flags;
1331
36a4cefd 1332 switch (s) {
b9cbfde6
SB
1333 case DIFF_SYMBOL_NO_LF_EOF:
1334 context = diff_get_color_opt(o, DIFF_CONTEXT);
1335 reset = diff_get_color_opt(o, DIFF_RESET);
1336 putc('\n', o->file);
017ac45e 1337 emit_line_0(o, context, NULL, 0, reset, '\\',
b9cbfde6
SB
1338 nneof, strlen(nneof));
1339 break;
f3597138
SB
1340 case DIFF_SYMBOL_SUBMODULE_HEADER:
1341 case DIFF_SYMBOL_SUBMODULE_ERROR:
1342 case DIFF_SYMBOL_SUBMODULE_PIPETHROUGH:
0911c475 1343 case DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES:
146fdb0d 1344 case DIFF_SYMBOL_SUMMARY:
0911c475 1345 case DIFF_SYMBOL_STATS_LINE:
4eed0ebd 1346 case DIFF_SYMBOL_BINARY_DIFF_BODY:
68abc6f1
SB
1347 case DIFF_SYMBOL_CONTEXT_FRAGINFO:
1348 emit_line(o, "", "", line, len);
1349 break;
f2bb1218 1350 case DIFF_SYMBOL_CONTEXT_INCOMPLETE:
c64b420b
SB
1351 case DIFF_SYMBOL_CONTEXT_MARKER:
1352 context = diff_get_color_opt(o, DIFF_CONTEXT);
1353 reset = diff_get_color_opt(o, DIFF_RESET);
1354 emit_line(o, context, reset, line, len);
1355 break;
36a4cefd
SB
1356 case DIFF_SYMBOL_SEPARATOR:
1357 fprintf(o->file, "%s%c",
1358 diff_line_prefix(o),
1359 o->line_termination);
1360 break;
091f8e28
SB
1361 case DIFF_SYMBOL_CONTEXT:
1362 set = diff_get_color_opt(o, DIFF_CONTEXT);
1363 reset = diff_get_color_opt(o, DIFF_RESET);
f7c3b4e2
JS
1364 set_sign = NULL;
1365 if (o->flags.dual_color_diffed_diffs) {
1366 char c = !len ? 0 : line[0];
1367
1368 if (c == '+')
1369 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1370 else if (c == '@')
1371 set = diff_get_color_opt(o, DIFF_FRAGINFO);
1372 else if (c == '-')
1373 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1374 }
7648b79e 1375 emit_line_ws_markup(o, set_sign, set, reset,
bc9feb05 1376 OUTPUT_INDICATOR_CONTEXT, line, len,
091f8e28
SB
1377 flags & (DIFF_SYMBOL_CONTENT_WS_MASK), 0);
1378 break;
1379 case DIFF_SYMBOL_PLUS:
86b452e2
SB
1380 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1381 DIFF_SYMBOL_MOVED_LINE_ALT |
1382 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1383 case DIFF_SYMBOL_MOVED_LINE |
1384 DIFF_SYMBOL_MOVED_LINE_ALT |
1385 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1386 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT_DIM);
1387 break;
1388 case DIFF_SYMBOL_MOVED_LINE |
1389 DIFF_SYMBOL_MOVED_LINE_ALT:
2e2d5ac1 1390 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_ALT);
86b452e2
SB
1391 break;
1392 case DIFF_SYMBOL_MOVED_LINE |
1393 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1394 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED_DIM);
1395 break;
1396 case DIFF_SYMBOL_MOVED_LINE:
2e2d5ac1 1397 set = diff_get_color_opt(o, DIFF_FILE_NEW_MOVED);
86b452e2
SB
1398 break;
1399 default:
2e2d5ac1 1400 set = diff_get_color_opt(o, DIFF_FILE_NEW);
86b452e2 1401 }
091f8e28 1402 reset = diff_get_color_opt(o, DIFF_RESET);
f7c3b4e2
JS
1403 if (!o->flags.dual_color_diffed_diffs)
1404 set_sign = NULL;
1405 else {
1406 char c = !len ? 0 : line[0];
1407
1408 set_sign = set;
1409 if (c == '-')
a7be92ac 1410 set = diff_get_color_opt(o, DIFF_FILE_OLD_BOLD);
f7c3b4e2
JS
1411 else if (c == '@')
1412 set = diff_get_color_opt(o, DIFF_FRAGINFO);
a7be92ac
JS
1413 else if (c == '+')
1414 set = diff_get_color_opt(o, DIFF_FILE_NEW_BOLD);
1415 else
1416 set = diff_get_color_opt(o, DIFF_CONTEXT_BOLD);
0b91faa0 1417 flags &= ~DIFF_SYMBOL_CONTENT_WS_MASK;
f7c3b4e2 1418 }
7648b79e 1419 emit_line_ws_markup(o, set_sign, set, reset,
bc9feb05 1420 OUTPUT_INDICATOR_NEW, line, len,
091f8e28
SB
1421 flags & DIFF_SYMBOL_CONTENT_WS_MASK,
1422 flags & DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF);
1423 break;
1424 case DIFF_SYMBOL_MINUS:
86b452e2
SB
1425 switch (flags & (DIFF_SYMBOL_MOVED_LINE |
1426 DIFF_SYMBOL_MOVED_LINE_ALT |
1427 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING)) {
1428 case DIFF_SYMBOL_MOVED_LINE |
1429 DIFF_SYMBOL_MOVED_LINE_ALT |
1430 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1431 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT_DIM);
1432 break;
1433 case DIFF_SYMBOL_MOVED_LINE |
1434 DIFF_SYMBOL_MOVED_LINE_ALT:
2e2d5ac1 1435 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_ALT);
86b452e2
SB
1436 break;
1437 case DIFF_SYMBOL_MOVED_LINE |
1438 DIFF_SYMBOL_MOVED_LINE_UNINTERESTING:
1439 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED_DIM);
1440 break;
1441 case DIFF_SYMBOL_MOVED_LINE:
2e2d5ac1 1442 set = diff_get_color_opt(o, DIFF_FILE_OLD_MOVED);
86b452e2
SB
1443 break;
1444 default:
2e2d5ac1 1445 set = diff_get_color_opt(o, DIFF_FILE_OLD);
86b452e2 1446 }
091f8e28 1447 reset = diff_get_color_opt(o, DIFF_RESET);
f7c3b4e2
JS
1448 if (!o->flags.dual_color_diffed_diffs)
1449 set_sign = NULL;
1450 else {
1451 char c = !len ? 0 : line[0];
1452
1453 set_sign = set;
1454 if (c == '+')
a7be92ac 1455 set = diff_get_color_opt(o, DIFF_FILE_NEW_DIM);
f7c3b4e2
JS
1456 else if (c == '@')
1457 set = diff_get_color_opt(o, DIFF_FRAGINFO);
a7be92ac
JS
1458 else if (c == '-')
1459 set = diff_get_color_opt(o, DIFF_FILE_OLD_DIM);
1460 else
1461 set = diff_get_color_opt(o, DIFF_CONTEXT_DIM);
f7c3b4e2 1462 }
7648b79e 1463 emit_line_ws_markup(o, set_sign, set, reset,
bc9feb05 1464 OUTPUT_INDICATOR_OLD, line, len,
091f8e28
SB
1465 flags & DIFF_SYMBOL_CONTENT_WS_MASK, 0);
1466 break;
ff958679
SB
1467 case DIFF_SYMBOL_WORDS_PORCELAIN:
1468 context = diff_get_color_opt(o, DIFF_CONTEXT);
1469 reset = diff_get_color_opt(o, DIFF_RESET);
1470 emit_line(o, context, reset, line, len);
1471 fputs("~\n", o->file);
1472 break;
1473 case DIFF_SYMBOL_WORDS:
1474 context = diff_get_color_opt(o, DIFF_CONTEXT);
1475 reset = diff_get_color_opt(o, DIFF_RESET);
1476 /*
1477 * Skip the prefix character, if any. With
1478 * diff_suppress_blank_empty, there may be
1479 * none.
1480 */
1481 if (line[0] != '\n') {
1482 line++;
1483 len--;
1484 }
1485 emit_line(o, context, reset, line, len);
1486 break;
3ee8b7bf
SB
1487 case DIFF_SYMBOL_FILEPAIR_PLUS:
1488 meta = diff_get_color_opt(o, DIFF_METAINFO);
1489 reset = diff_get_color_opt(o, DIFF_RESET);
1490 fprintf(o->file, "%s%s+++ %s%s%s\n", diff_line_prefix(o), meta,
1491 line, reset,
1492 strchr(line, ' ') ? "\t" : "");
1493 break;
1494 case DIFF_SYMBOL_FILEPAIR_MINUS:
1495 meta = diff_get_color_opt(o, DIFF_METAINFO);
1496 reset = diff_get_color_opt(o, DIFF_RESET);
1497 fprintf(o->file, "%s%s--- %s%s%s\n", diff_line_prefix(o), meta,
1498 line, reset,
1499 strchr(line, ' ') ? "\t" : "");
1500 break;
4acaaa7a 1501 case DIFF_SYMBOL_BINARY_FILES:
a29b0a13
SB
1502 case DIFF_SYMBOL_HEADER:
1503 fprintf(o->file, "%s", line);
1504 break;
4eed0ebd
SB
1505 case DIFF_SYMBOL_BINARY_DIFF_HEADER:
1506 fprintf(o->file, "%sGIT binary patch\n", diff_line_prefix(o));
1507 break;
1508 case DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA:
1509 fprintf(o->file, "%sdelta %s\n", diff_line_prefix(o), line);
1510 break;
1511 case DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL:
1512 fprintf(o->file, "%sliteral %s\n", diff_line_prefix(o), line);
1513 break;
1514 case DIFF_SYMBOL_BINARY_DIFF_FOOTER:
1515 fputs(diff_line_prefix(o), o->file);
1516 fputc('\n', o->file);
1517 break;
5af6ea95
SB
1518 case DIFF_SYMBOL_REWRITE_DIFF:
1519 fraginfo = diff_get_color(o->use_color, DIFF_FRAGINFO);
1520 reset = diff_get_color_opt(o, DIFF_RESET);
1521 emit_line(o, fraginfo, reset, line, len);
1522 break;
f3597138
SB
1523 case DIFF_SYMBOL_SUBMODULE_ADD:
1524 set = diff_get_color_opt(o, DIFF_FILE_NEW);
1525 reset = diff_get_color_opt(o, DIFF_RESET);
1526 emit_line(o, set, reset, line, len);
1527 break;
1528 case DIFF_SYMBOL_SUBMODULE_DEL:
1529 set = diff_get_color_opt(o, DIFF_FILE_OLD);
1530 reset = diff_get_color_opt(o, DIFF_RESET);
1531 emit_line(o, set, reset, line, len);
1532 break;
1533 case DIFF_SYMBOL_SUBMODULE_UNTRACKED:
1534 fprintf(o->file, "%sSubmodule %s contains untracked content\n",
1535 diff_line_prefix(o), line);
1536 break;
1537 case DIFF_SYMBOL_SUBMODULE_MODIFIED:
1538 fprintf(o->file, "%sSubmodule %s contains modified content\n",
1539 diff_line_prefix(o), line);
1540 break;
0911c475
SB
1541 case DIFF_SYMBOL_STATS_SUMMARY_NO_FILES:
1542 emit_line(o, "", "", " 0 files changed\n",
1543 strlen(" 0 files changed\n"));
1544 break;
1545 case DIFF_SYMBOL_STATS_SUMMARY_ABBREV:
1546 emit_line(o, "", "", " ...\n", strlen(" ...\n"));
1547 break;
bd033291
SB
1548 case DIFF_SYMBOL_WORD_DIFF:
1549 fprintf(o->file, "%.*s", len, line);
1550 break;
30b7e1e7
SB
1551 case DIFF_SYMBOL_STAT_SEP:
1552 fputs(o->stat_sep, o->file);
1553 break;
36a4cefd 1554 default:
033abf97 1555 BUG("unknown diff symbol");
36a4cefd 1556 }
0911c475 1557 strbuf_release(&sb);
36a4cefd
SB
1558}
1559
e6e045f8
SB
1560static void emit_diff_symbol(struct diff_options *o, enum diff_symbol s,
1561 const char *line, int len, unsigned flags)
1562{
21536d07 1563 struct emitted_diff_symbol e = {line, len, flags, 0, 0, s};
e6e045f8
SB
1564
1565 if (o->emitted_symbols)
1566 append_emitted_diff_symbol(o, &e);
1567 else
1568 emit_diff_symbol_from_struct(o, &e);
1569}
1570
f3597138
SB
1571void diff_emit_submodule_del(struct diff_options *o, const char *line)
1572{
1573 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_DEL, line, strlen(line), 0);
1574}
1575
1576void diff_emit_submodule_add(struct diff_options *o, const char *line)
1577{
1578 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ADD, line, strlen(line), 0);
1579}
1580
1581void diff_emit_submodule_untracked(struct diff_options *o, const char *path)
1582{
1583 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_UNTRACKED,
1584 path, strlen(path), 0);
1585}
1586
1587void diff_emit_submodule_modified(struct diff_options *o, const char *path)
1588{
1589 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_MODIFIED,
1590 path, strlen(path), 0);
1591}
1592
1593void diff_emit_submodule_header(struct diff_options *o, const char *header)
1594{
1595 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_HEADER,
1596 header, strlen(header), 0);
1597}
1598
1599void diff_emit_submodule_error(struct diff_options *o, const char *err)
1600{
1601 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_ERROR, err, strlen(err), 0);
1602}
1603
1604void diff_emit_submodule_pipethrough(struct diff_options *o,
1605 const char *line, int len)
1606{
1607 emit_diff_symbol(o, DIFF_SYMBOL_SUBMODULE_PIPETHROUGH, line, len, 0);
1608}
1609
6957eb9a
JH
1610static int new_blank_line_at_eof(struct emit_callback *ecbdata, const char *line, int len)
1611{
1612 if (!((ecbdata->ws_rule & WS_BLANK_AT_EOF) &&
1613 ecbdata->blank_at_eof_in_preimage &&
1614 ecbdata->blank_at_eof_in_postimage &&
1615 ecbdata->blank_at_eof_in_preimage <= ecbdata->lno_in_preimage &&
1616 ecbdata->blank_at_eof_in_postimage <= ecbdata->lno_in_postimage))
1617 return 0;
018cff70 1618 return ws_blank_line(line, len, ecbdata->ws_rule);
6957eb9a
JH
1619}
1620
e04df612 1621static void emit_add_line(struct emit_callback *ecbdata,
0e383e18
JH
1622 const char *line, int len)
1623{
091f8e28
SB
1624 unsigned flags = WSEH_NEW | ecbdata->ws_rule;
1625 if (new_blank_line_at_eof(ecbdata, line, len))
1626 flags |= DIFF_SYMBOL_CONTENT_BLANK_LINE_EOF;
1627
1628 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_PLUS, line, len, flags);
b8767f79 1629}
0e383e18 1630
e04df612 1631static void emit_del_line(struct emit_callback *ecbdata,
b8767f79
JH
1632 const char *line, int len)
1633{
091f8e28
SB
1634 unsigned flags = WSEH_OLD | ecbdata->ws_rule;
1635 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_MINUS, line, len, flags);
0e383e18
JH
1636}
1637
e04df612 1638static void emit_context_line(struct emit_callback *ecbdata,
0e383e18
JH
1639 const char *line, int len)
1640{
091f8e28
SB
1641 unsigned flags = WSEH_CONTEXT | ecbdata->ws_rule;
1642 emit_diff_symbol(ecbdata->opt, DIFF_SYMBOL_CONTEXT, line, len, flags);
0e383e18
JH
1643}
1644
89cb73a1
BW
1645static void emit_hunk_header(struct emit_callback *ecbdata,
1646 const char *line, int len)
1647{
8dbf3eb6 1648 const char *context = diff_get_color(ecbdata->color_diff, DIFF_CONTEXT);
89cb73a1
BW
1649 const char *frag = diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO);
1650 const char *func = diff_get_color(ecbdata->color_diff, DIFF_FUNCINFO);
1651 const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
f7c3b4e2 1652 const char *reverse = ecbdata->color_diff ? GIT_COLOR_REVERSE : "";
89cb73a1
BW
1653 static const char atat[2] = { '@', '@' };
1654 const char *cp, *ep;
2efcc977
BY
1655 struct strbuf msgbuf = STRBUF_INIT;
1656 int org_len = len;
1657 int i = 1;
89cb73a1
BW
1658
1659 /*
1660 * As a hunk header must begin with "@@ -<old>, +<new> @@",
1661 * it always is at least 10 bytes long.
1662 */
1663 if (len < 10 ||
1664 memcmp(line, atat, 2) ||
1665 !(ep = memmem(line + 2, len - 2, atat, 2))) {
c64b420b 1666 emit_diff_symbol(ecbdata->opt,
091f8e28 1667 DIFF_SYMBOL_CONTEXT_MARKER, line, len, 0);
89cb73a1
BW
1668 return;
1669 }
1670 ep += 2; /* skip over @@ */
1671
1672 /* The hunk header in fraginfo color */
f7c3b4e2
JS
1673 if (ecbdata->opt->flags.dual_color_diffed_diffs)
1674 strbuf_addstr(&msgbuf, reverse);
cedc61a9 1675 strbuf_addstr(&msgbuf, frag);
430be36e
TG
1676 if (ecbdata->opt->flags.suppress_hunk_header_line_count)
1677 strbuf_add(&msgbuf, atat, sizeof(atat));
1678 else
1679 strbuf_add(&msgbuf, line, ep - line);
cedc61a9 1680 strbuf_addstr(&msgbuf, reset);
2efcc977
BY
1681
1682 /*
1683 * trailing "\r\n"
1684 */
1685 for ( ; i < 3; i++)
1686 if (line[len - i] == '\r' || line[len - i] == '\n')
1687 len--;
89cb73a1
BW
1688
1689 /* blank before the func header */
1690 for (cp = ep; ep - line < len; ep++)
1691 if (*ep != ' ' && *ep != '\t')
1692 break;
2efcc977 1693 if (ep != cp) {
8dbf3eb6 1694 strbuf_addstr(&msgbuf, context);
2efcc977 1695 strbuf_add(&msgbuf, cp, ep - cp);
cedc61a9 1696 strbuf_addstr(&msgbuf, reset);
2efcc977
BY
1697 }
1698
1699 if (ep < line + len) {
cedc61a9 1700 strbuf_addstr(&msgbuf, func);
2efcc977 1701 strbuf_add(&msgbuf, ep, line + len - ep);
cedc61a9 1702 strbuf_addstr(&msgbuf, reset);
2efcc977 1703 }
89cb73a1 1704
2efcc977 1705 strbuf_add(&msgbuf, line + len, org_len - len);
dfb7728f 1706 strbuf_complete_line(&msgbuf);
68abc6f1 1707 emit_diff_symbol(ecbdata->opt,
091f8e28 1708 DIFF_SYMBOL_CONTEXT_FRAGINFO, msgbuf.buf, msgbuf.len, 0);
2efcc977 1709 strbuf_release(&msgbuf);
89cb73a1
BW
1710}
1711
3b335762
NTND
1712static struct diff_tempfile *claim_diff_tempfile(void)
1713{
479b0ae8
JK
1714 int i;
1715 for (i = 0; i < ARRAY_SIZE(diff_temp); i++)
1716 if (!diff_temp[i].name)
1717 return diff_temp + i;
033abf97 1718 BUG("diff is failing to clean up its tempfiles");
479b0ae8
JK
1719}
1720
479b0ae8
JK
1721static void remove_tempfile(void)
1722{
1723 int i;
a8344abe 1724 for (i = 0; i < ARRAY_SIZE(diff_temp); i++) {
076aa2cb 1725 if (is_tempfile_active(diff_temp[i].tempfile))
284098f1 1726 delete_tempfile(&diff_temp[i].tempfile);
a8344abe
NR
1727 diff_temp[i].name = NULL;
1728 }
479b0ae8
JK
1729}
1730
5af6ea95 1731static void add_line_count(struct strbuf *out, int count)
6973dcae
JH
1732{
1733 switch (count) {
1734 case 0:
5af6ea95 1735 strbuf_addstr(out, "0,0");
6973dcae
JH
1736 break;
1737 case 1:
5af6ea95 1738 strbuf_addstr(out, "1");
6973dcae
JH
1739 break;
1740 default:
5af6ea95 1741 strbuf_addf(out, "1,%d", count);
6973dcae
JH
1742 break;
1743 }
1744}
1745
7f7ee2ff
JH
1746static void emit_rewrite_lines(struct emit_callback *ecb,
1747 int prefix, const char *data, int size)
6973dcae 1748{
7f7ee2ff 1749 const char *endp = NULL;
7f7ee2ff
JH
1750
1751 while (0 < size) {
1752 int len;
1753
1754 endp = memchr(data, '\n', size);
1755 len = endp ? (endp - data + 1) : size;
1756 if (prefix != '+') {
1757 ecb->lno_in_preimage++;
e04df612 1758 emit_del_line(ecb, data, len);
7f7ee2ff
JH
1759 } else {
1760 ecb->lno_in_postimage++;
e04df612 1761 emit_add_line(ecb, data, len);
13e36ec5 1762 }
7f7ee2ff
JH
1763 size -= len;
1764 data += len;
1765 }
b9cbfde6 1766 if (!endp)
091f8e28 1767 emit_diff_symbol(ecb->opt, DIFF_SYMBOL_NO_LF_EOF, NULL, 0, 0);
6973dcae
JH
1768}
1769
1770static void emit_rewrite_diff(const char *name_a,
1771 const char *name_b,
1772 struct diff_filespec *one,
13e36ec5 1773 struct diff_filespec *two,
d9bae1a1
JK
1774 struct userdiff_driver *textconv_one,
1775 struct userdiff_driver *textconv_two,
eab9a40b 1776 struct diff_options *o)
6973dcae
JH
1777{
1778 int lc_a, lc_b;
d5625091 1779 static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
a5a818ee 1780 const char *a_prefix, *b_prefix;
840383b2 1781 char *data_one, *data_two;
3aa1f7ca 1782 size_t size_one, size_two;
7f7ee2ff 1783 struct emit_callback ecbdata;
5af6ea95 1784 struct strbuf out = STRBUF_INIT;
a5a818ee 1785
0d1e0e78 1786 if (diff_mnemonic_prefix && o->flags.reverse_diff) {
a5a818ee
JH
1787 a_prefix = o->b_prefix;
1788 b_prefix = o->a_prefix;
1789 } else {
1790 a_prefix = o->a_prefix;
1791 b_prefix = o->b_prefix;
1792 }
1a9eb3b9 1793
8a13becc
JH
1794 name_a += (*name_a == '/');
1795 name_b += (*name_b == '/');
1a9eb3b9 1796
d5625091
JH
1797 strbuf_reset(&a_name);
1798 strbuf_reset(&b_name);
a5a818ee
JH
1799 quote_two_c_style(&a_name, a_prefix, name_a, 0);
1800 quote_two_c_style(&b_name, b_prefix, name_b, 0);
d5625091 1801
6afaf807
NTND
1802 size_one = fill_textconv(o->repo, textconv_one, one, &data_one);
1803 size_two = fill_textconv(o->repo, textconv_two, two, &data_two);
3aa1f7ca 1804
d91ba8fa 1805 memset(&ecbdata, 0, sizeof(ecbdata));
daa0c3d9 1806 ecbdata.color_diff = want_color(o->use_color);
26d024ec 1807 ecbdata.ws_rule = whitespace_rule(o->repo->index, name_b);
a3c158d4 1808 ecbdata.opt = o;
d91ba8fa
JH
1809 if (ecbdata.ws_rule & WS_BLANK_AT_EOF) {
1810 mmfile_t mf1, mf2;
1811 mf1.ptr = (char *)data_one;
1812 mf2.ptr = (char *)data_two;
1813 mf1.size = size_one;
1814 mf2.size = size_two;
1815 check_blank_at_eof(&mf1, &mf2, &ecbdata);
1816 }
1817 ecbdata.lno_in_preimage = 1;
1818 ecbdata.lno_in_postimage = 1;
1819
3aa1f7ca
JK
1820 lc_a = count_lines(data_one, size_one);
1821 lc_b = count_lines(data_two, size_two);
3ee8b7bf
SB
1822
1823 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
1824 a_name.buf, a_name.len, 0);
1825 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
1826 b_name.buf, b_name.len, 0);
1827
5af6ea95 1828 strbuf_addstr(&out, "@@ -");
467ddc14 1829 if (!o->irreversible_delete)
5af6ea95 1830 add_line_count(&out, lc_a);
467ddc14 1831 else
5af6ea95
SB
1832 strbuf_addstr(&out, "?,?");
1833 strbuf_addstr(&out, " +");
1834 add_line_count(&out, lc_b);
1835 strbuf_addstr(&out, " @@\n");
1836 emit_diff_symbol(o, DIFF_SYMBOL_REWRITE_DIFF, out.buf, out.len, 0);
1837 strbuf_release(&out);
1838
467ddc14 1839 if (lc_a && !o->irreversible_delete)
d91ba8fa 1840 emit_rewrite_lines(&ecbdata, '-', data_one, size_one);
6973dcae 1841 if (lc_b)
d91ba8fa 1842 emit_rewrite_lines(&ecbdata, '+', data_two, size_two);
b76c056b 1843 if (textconv_one)
aed6ca52 1844 free((char *)data_one);
b76c056b 1845 if (textconv_two)
aed6ca52 1846 free((char *)data_two);
6973dcae
JH
1847}
1848
f59a59e2
JS
1849struct diff_words_buffer {
1850 mmfile_t text;
071bcaab 1851 unsigned long alloc;
2e5d2003
JS
1852 struct diff_words_orig {
1853 const char *begin, *end;
1854 } *orig;
1855 int orig_nr, orig_alloc;
f59a59e2
JS
1856};
1857
1858static void diff_words_append(char *line, unsigned long len,
1859 struct diff_words_buffer *buffer)
1860{
23c1575f 1861 ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
f59a59e2
JS
1862 line++;
1863 len--;
1864 memcpy(buffer->text.ptr + buffer->text.size, line, len);
1865 buffer->text.size += len;
2b6a5417 1866 buffer->text.ptr[buffer->text.size] = '\0';
f59a59e2
JS
1867}
1868
9cba13ca 1869struct diff_words_style_elem {
882749a0
TR
1870 const char *prefix;
1871 const char *suffix;
1872 const char *color; /* NULL; filled in by the setup code if
1873 * color is enabled */
1874};
1875
9cba13ca 1876struct diff_words_style {
882749a0 1877 enum diff_words_type type;
63a01c3f 1878 struct diff_words_style_elem new_word, old_word, ctx;
882749a0
TR
1879 const char *newline;
1880};
1881
c2e86add 1882static struct diff_words_style diff_words_styles[] = {
882749a0
TR
1883 { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
1884 { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
1885 { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
1886};
1887
f59a59e2 1888struct diff_words_data {
f59a59e2 1889 struct diff_words_buffer minus, plus;
2e5d2003 1890 const char *current_plus;
4297c0ae
BY
1891 int last_minus;
1892 struct diff_options *opt;
2b6a5417 1893 regex_t *word_regex;
882749a0
TR
1894 enum diff_words_type type;
1895 struct diff_words_style *style;
f59a59e2
JS
1896};
1897
bd033291 1898static int fn_out_diff_words_write_helper(struct diff_options *o,
882749a0
TR
1899 struct diff_words_style_elem *st_el,
1900 const char *newline,
bd033291 1901 size_t count, const char *buf)
882749a0 1902{
4297c0ae 1903 int print = 0;
bd033291 1904 struct strbuf sb = STRBUF_INIT;
4297c0ae 1905
882749a0
TR
1906 while (count) {
1907 char *p = memchr(buf, '\n', count);
4297c0ae 1908 if (print)
bd033291
SB
1909 strbuf_addstr(&sb, diff_line_prefix(o));
1910
882749a0 1911 if (p != buf) {
bd033291
SB
1912 const char *reset = st_el->color && *st_el->color ?
1913 GIT_COLOR_RESET : NULL;
1914 if (st_el->color && *st_el->color)
1915 strbuf_addstr(&sb, st_el->color);
1916 strbuf_addstr(&sb, st_el->prefix);
1917 strbuf_add(&sb, buf, p ? p - buf : count);
1918 strbuf_addstr(&sb, st_el->suffix);
1919 if (reset)
1920 strbuf_addstr(&sb, reset);
882749a0
TR
1921 }
1922 if (!p)
bd033291
SB
1923 goto out;
1924
1925 strbuf_addstr(&sb, newline);
882749a0
TR
1926 count -= p + 1 - buf;
1927 buf = p + 1;
4297c0ae 1928 print = 1;
bd033291
SB
1929 if (count) {
1930 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1931 sb.buf, sb.len, 0);
1932 strbuf_reset(&sb);
1933 }
882749a0 1934 }
bd033291
SB
1935
1936out:
1937 if (sb.len)
1938 emit_diff_symbol(o, DIFF_SYMBOL_WORD_DIFF,
1939 sb.buf, sb.len, 0);
1940 strbuf_release(&sb);
882749a0
TR
1941 return 0;
1942}
1943
4297c0ae
BY
1944/*
1945 * '--color-words' algorithm can be described as:
1946 *
5621760f 1947 * 1. collect the minus/plus lines of a diff hunk, divided into
4297c0ae
BY
1948 * minus-lines and plus-lines;
1949 *
1950 * 2. break both minus-lines and plus-lines into words and
1951 * place them into two mmfile_t with one word for each line;
1952 *
1953 * 3. use xdiff to run diff on the two mmfile_t to get the words level diff;
1954 *
1955 * And for the common parts of the both file, we output the plus side text.
1956 * diff_words->current_plus is used to trace the current position of the plus file
1957 * which printed. diff_words->last_minus is used to trace the last minus word
1958 * printed.
1959 *
1960 * For '--graph' to work with '--color-words', we need to output the graph prefix
1961 * on each line of color words output. Generally, there are two conditions on
1962 * which we should output the prefix.
1963 *
1964 * 1. diff_words->last_minus == 0 &&
1965 * diff_words->current_plus == diff_words->plus.text.ptr
1966 *
1967 * that is: the plus text must start as a new line, and if there is no minus
1968 * word printed, a graph prefix must be printed.
1969 *
1970 * 2. diff_words->current_plus > diff_words->plus.text.ptr &&
1971 * *(diff_words->current_plus - 1) == '\n'
1972 *
1973 * that is: a graph prefix must be printed following a '\n'
1974 */
1975static int color_words_output_graph_prefix(struct diff_words_data *diff_words)
1976{
1977 if ((diff_words->last_minus == 0 &&
1978 diff_words->current_plus == diff_words->plus.text.ptr) ||
1979 (diff_words->current_plus > diff_words->plus.text.ptr &&
1980 *(diff_words->current_plus - 1) == '\n')) {
1981 return 1;
1982 } else {
1983 return 0;
1984 }
1985}
1986
7c61e25f
JK
1987static void fn_out_diff_words_aux(void *priv,
1988 long minus_first, long minus_len,
1989 long plus_first, long plus_len,
1990 const char *func, long funclen)
f59a59e2 1991{
f59a59e2 1992 struct diff_words_data *diff_words = priv;
882749a0 1993 struct diff_words_style *style = diff_words->style;
2e5d2003 1994 const char *minus_begin, *minus_end, *plus_begin, *plus_end;
4297c0ae 1995 struct diff_options *opt = diff_words->opt;
30997bb8 1996 const char *line_prefix;
f59a59e2 1997
4297c0ae 1998 assert(opt);
30997bb8 1999 line_prefix = diff_line_prefix(opt);
4297c0ae 2000
2e5d2003
JS
2001 /* POSIX requires that first be decremented by one if len == 0... */
2002 if (minus_len) {
2003 minus_begin = diff_words->minus.orig[minus_first].begin;
2004 minus_end =
2005 diff_words->minus.orig[minus_first + minus_len - 1].end;
2006 } else
2007 minus_begin = minus_end =
2008 diff_words->minus.orig[minus_first].end;
2009
2010 if (plus_len) {
2011 plus_begin = diff_words->plus.orig[plus_first].begin;
2012 plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
2013 } else
2014 plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
2015
4297c0ae
BY
2016 if (color_words_output_graph_prefix(diff_words)) {
2017 fputs(line_prefix, diff_words->opt->file);
2018 }
2019 if (diff_words->current_plus != plus_begin) {
bd033291 2020 fn_out_diff_words_write_helper(diff_words->opt,
882749a0
TR
2021 &style->ctx, style->newline,
2022 plus_begin - diff_words->current_plus,
bd033291 2023 diff_words->current_plus);
4297c0ae
BY
2024 }
2025 if (minus_begin != minus_end) {
bd033291 2026 fn_out_diff_words_write_helper(diff_words->opt,
63a01c3f 2027 &style->old_word, style->newline,
bd033291 2028 minus_end - minus_begin, minus_begin);
4297c0ae
BY
2029 }
2030 if (plus_begin != plus_end) {
bd033291 2031 fn_out_diff_words_write_helper(diff_words->opt,
63a01c3f 2032 &style->new_word, style->newline,
bd033291 2033 plus_end - plus_begin, plus_begin);
4297c0ae 2034 }
2e5d2003
JS
2035
2036 diff_words->current_plus = plus_end;
4297c0ae 2037 diff_words->last_minus = minus_first;
f59a59e2
JS
2038}
2039
2b6a5417
JS
2040/* This function starts looking at *begin, and returns 0 iff a word was found. */
2041static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
2042 int *begin, int *end)
2043{
2044 if (word_regex && *begin < buffer->size) {
2045 regmatch_t match[1];
b7d36ffc
JS
2046 if (!regexec_buf(word_regex, buffer->ptr + *begin,
2047 buffer->size - *begin, 1, match, 0)) {
2b6a5417
JS
2048 char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
2049 '\n', match[0].rm_eo - match[0].rm_so);
2050 *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
2051 *begin += match[0].rm_so;
2052 return *begin >= *end;
2053 }
2054 return -1;
f59a59e2
JS
2055 }
2056
2b6a5417
JS
2057 /* find the next word */
2058 while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
2059 (*begin)++;
2060 if (*begin >= buffer->size)
2061 return -1;
f59a59e2 2062
2b6a5417
JS
2063 /* find the end of the word */
2064 *end = *begin + 1;
2065 while (*end < buffer->size && !isspace(buffer->ptr[*end]))
2066 (*end)++;
2067
2068 return 0;
f59a59e2
JS
2069}
2070
23c1575f 2071/*
2e5d2003
JS
2072 * This function splits the words in buffer->text, stores the list with
2073 * newline separator into out, and saves the offsets of the original words
2074 * in buffer->orig.
23c1575f 2075 */
2b6a5417
JS
2076static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
2077 regex_t *word_regex)
f59a59e2 2078{
2e5d2003 2079 int i, j;
2b6a5417 2080 long alloc = 0;
f59a59e2 2081
2e5d2003 2082 out->size = 0;
2b6a5417 2083 out->ptr = NULL;
f59a59e2 2084
2e5d2003
JS
2085 /* fake an empty "0th" word */
2086 ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
2087 buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
2088 buffer->orig_nr = 1;
2089
2090 for (i = 0; i < buffer->text.size; i++) {
2b6a5417
JS
2091 if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
2092 return;
2e5d2003
JS
2093
2094 /* store original boundaries */
2095 ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
2096 buffer->orig_alloc);
2097 buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
2098 buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
2099 buffer->orig_nr++;
2100
2101 /* store one word */
2b6a5417 2102 ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
2e5d2003
JS
2103 memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
2104 out->ptr[out->size + j - i] = '\n';
2105 out->size += j - i + 1;
2106
2107 i = j - 1;
f59a59e2
JS
2108 }
2109}
2110
2111/* this executes the word diff on the accumulated buffers */
2112static void diff_words_show(struct diff_words_data *diff_words)
2113{
2114 xpparam_t xpp;
2115 xdemitconf_t xecfg;
f59a59e2 2116 mmfile_t minus, plus;
882749a0 2117 struct diff_words_style *style = diff_words->style;
f59a59e2 2118
4297c0ae 2119 struct diff_options *opt = diff_words->opt;
30997bb8 2120 const char *line_prefix;
4297c0ae
BY
2121
2122 assert(opt);
30997bb8 2123 line_prefix = diff_line_prefix(opt);
4297c0ae 2124
2e5d2003
JS
2125 /* special case: only removal */
2126 if (!diff_words->plus.text.size) {
bd033291
SB
2127 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2128 line_prefix, strlen(line_prefix), 0);
2129 fn_out_diff_words_write_helper(diff_words->opt,
63a01c3f 2130 &style->old_word, style->newline,
4297c0ae 2131 diff_words->minus.text.size,
bd033291 2132 diff_words->minus.text.ptr);
2e5d2003
JS
2133 diff_words->minus.text.size = 0;
2134 return;
2135 }
2136
2137 diff_words->current_plus = diff_words->plus.text.ptr;
4297c0ae 2138 diff_words->last_minus = 0;
f59a59e2 2139
9ccd0a88 2140 memset(&xpp, 0, sizeof(xpp));
30b25010 2141 memset(&xecfg, 0, sizeof(xecfg));
2b6a5417
JS
2142 diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
2143 diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
582aa00b 2144 xpp.flags = 0;
2b6a5417 2145 /* as only the hunk header will be parsed, we need a 0-context */
2e5d2003 2146 xecfg.ctxlen = 0;
7c61e25f 2147 if (xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, NULL,
9346d6d1 2148 diff_words, &xpp, &xecfg))
3efb9880 2149 die("unable to generate word diff");
f59a59e2
JS
2150 free(minus.ptr);
2151 free(plus.ptr);
2e5d2003 2152 if (diff_words->current_plus != diff_words->plus.text.ptr +
4297c0ae
BY
2153 diff_words->plus.text.size) {
2154 if (color_words_output_graph_prefix(diff_words))
bd033291
SB
2155 emit_diff_symbol(diff_words->opt, DIFF_SYMBOL_WORD_DIFF,
2156 line_prefix, strlen(line_prefix), 0);
2157 fn_out_diff_words_write_helper(diff_words->opt,
882749a0 2158 &style->ctx, style->newline,
2e5d2003 2159 diff_words->plus.text.ptr + diff_words->plus.text.size
bd033291 2160 - diff_words->current_plus, diff_words->current_plus);
4297c0ae 2161 }
f59a59e2 2162 diff_words->minus.text.size = diff_words->plus.text.size = 0;
f59a59e2
JS
2163}
2164
76fd2828
JH
2165/* In "color-words" mode, show word-diff of words accumulated in the buffer */
2166static void diff_words_flush(struct emit_callback *ecbdata)
2167{
e6e045f8
SB
2168 struct diff_options *wo = ecbdata->diff_words->opt;
2169
76fd2828
JH
2170 if (ecbdata->diff_words->minus.text.size ||
2171 ecbdata->diff_words->plus.text.size)
2172 diff_words_show(ecbdata->diff_words);
e6e045f8
SB
2173
2174 if (wo->emitted_symbols) {
2175 struct diff_options *o = ecbdata->opt;
2176 struct emitted_diff_symbols *wol = wo->emitted_symbols;
2177 int i;
2178
2179 /*
2180 * NEEDSWORK:
2181 * Instead of appending each, concat all words to a line?
2182 */
2183 for (i = 0; i < wol->nr; i++)
2184 append_emitted_diff_symbol(o, &wol->buf[i]);
2185
2186 for (i = 0; i < wol->nr; i++)
2187 free((void *)wol->buf[i].line);
2188
2189 wol->nr = 0;
2190 }
76fd2828
JH
2191}
2192
acd00ea0
NTND
2193static void diff_filespec_load_driver(struct diff_filespec *one,
2194 struct index_state *istate)
77d1a520
TR
2195{
2196 /* Use already-loaded driver */
2197 if (one->driver)
2198 return;
2199
2200 if (S_ISREG(one->mode))
acd00ea0 2201 one->driver = userdiff_find_by_path(istate, one->path);
77d1a520
TR
2202
2203 /* Fallback to default settings */
2204 if (!one->driver)
2205 one->driver = userdiff_find_by_name("default");
2206}
2207
acd00ea0
NTND
2208static const char *userdiff_word_regex(struct diff_filespec *one,
2209 struct index_state *istate)
77d1a520 2210{
acd00ea0 2211 diff_filespec_load_driver(one, istate);
77d1a520
TR
2212 return one->driver->word_regex;
2213}
2214
2215static void init_diff_words_data(struct emit_callback *ecbdata,
6440d341 2216 struct diff_options *orig_opts,
77d1a520
TR
2217 struct diff_filespec *one,
2218 struct diff_filespec *two)
2219{
2220 int i;
6440d341
TR
2221 struct diff_options *o = xmalloc(sizeof(struct diff_options));
2222 memcpy(o, orig_opts, sizeof(struct diff_options));
77d1a520
TR
2223
2224 ecbdata->diff_words =
2225 xcalloc(1, sizeof(struct diff_words_data));
2226 ecbdata->diff_words->type = o->word_diff;
2227 ecbdata->diff_words->opt = o;
e6e045f8
SB
2228
2229 if (orig_opts->emitted_symbols)
2230 o->emitted_symbols =
2231 xcalloc(1, sizeof(struct emitted_diff_symbols));
2232
77d1a520 2233 if (!o->word_regex)
acd00ea0 2234 o->word_regex = userdiff_word_regex(one, o->repo->index);
77d1a520 2235 if (!o->word_regex)
acd00ea0 2236 o->word_regex = userdiff_word_regex(two, o->repo->index);
77d1a520
TR
2237 if (!o->word_regex)
2238 o->word_regex = diff_word_regex_cfg;
2239 if (o->word_regex) {
2240 ecbdata->diff_words->word_regex = (regex_t *)
2241 xmalloc(sizeof(regex_t));
2242 if (regcomp(ecbdata->diff_words->word_regex,
2243 o->word_regex,
2244 REG_EXTENDED | REG_NEWLINE))
1a07e59c
NTND
2245 die("invalid regular expression: %s",
2246 o->word_regex);
77d1a520
TR
2247 }
2248 for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
2249 if (o->word_diff == diff_words_styles[i].type) {
2250 ecbdata->diff_words->style =
2251 &diff_words_styles[i];
2252 break;
2253 }
2254 }
2255 if (want_color(o->use_color)) {
2256 struct diff_words_style *st = ecbdata->diff_words->style;
63a01c3f
BW
2257 st->old_word.color = diff_get_color_opt(o, DIFF_FILE_OLD);
2258 st->new_word.color = diff_get_color_opt(o, DIFF_FILE_NEW);
8dbf3eb6 2259 st->ctx.color = diff_get_color_opt(o, DIFF_CONTEXT);
77d1a520
TR
2260 }
2261}
2262
f59a59e2
JS
2263static void free_diff_words_data(struct emit_callback *ecbdata)
2264{
2265 if (ecbdata->diff_words) {
76fd2828 2266 diff_words_flush(ecbdata);
e6e045f8 2267 free (ecbdata->diff_words->opt->emitted_symbols);
6440d341 2268 free (ecbdata->diff_words->opt);
8e0f7003 2269 free (ecbdata->diff_words->minus.text.ptr);
2e5d2003 2270 free (ecbdata->diff_words->minus.orig);
8e0f7003 2271 free (ecbdata->diff_words->plus.text.ptr);
2e5d2003 2272 free (ecbdata->diff_words->plus.orig);
ef5644ea
BC
2273 if (ecbdata->diff_words->word_regex) {
2274 regfree(ecbdata->diff_words->word_regex);
2275 free(ecbdata->diff_words->word_regex);
2276 }
6a83d902 2277 FREE_AND_NULL(ecbdata->diff_words);
f59a59e2
JS
2278 }
2279}
2280
ce436973 2281const char *diff_get_color(int diff_use_color, enum color_diff ix)
cd112cef 2282{
daa0c3d9 2283 if (want_color(diff_use_color))
50f575fc
LT
2284 return diff_colors[ix];
2285 return "";
cd112cef
JS
2286}
2287
f1922234
JK
2288const char *diff_line_prefix(struct diff_options *opt)
2289{
2290 struct strbuf *msgbuf;
2291 if (!opt->output_prefix)
2292 return "";
2293
2294 msgbuf = opt->output_prefix(opt, opt->output_prefix_data);
2295 return msgbuf->buf;
2296}
2297
19b9046e 2298static unsigned long sane_truncate_line(char *line, unsigned long len)
23707811
JH
2299{
2300 const char *cp;
2301 unsigned long allot;
2302 size_t l = len;
2303
23707811
JH
2304 cp = line;
2305 allot = l;
2306 while (0 < l) {
2307 (void) utf8_width(&cp, &l);
2308 if (!cp)
2309 break; /* truncated in the middle? */
2310 }
2311 return allot - l;
2312}
2313
d68fe26f 2314static void find_lno(const char *line, struct emit_callback *ecbdata)
690ed843 2315{
d68fe26f
JH
2316 const char *p;
2317 ecbdata->lno_in_preimage = 0;
2318 ecbdata->lno_in_postimage = 0;
2319 p = strchr(line, '-');
690ed843 2320 if (!p)
d68fe26f
JH
2321 return; /* cannot happen */
2322 ecbdata->lno_in_preimage = strtol(p + 1, NULL, 10);
2323 p = strchr(p, '+');
2324 if (!p)
2325 return; /* cannot happen */
2326 ecbdata->lno_in_postimage = strtol(p + 1, NULL, 10);
690ed843
JH
2327}
2328
cd112cef 2329static void fn_out_consume(void *priv, char *line, unsigned long len)
6973dcae 2330{
6973dcae 2331 struct emit_callback *ecbdata = priv;
7be57610 2332 struct diff_options *o = ecbdata->opt;
6973dcae 2333
ba16233c
SB
2334 o->found_changes = 1;
2335
3e97c7c6 2336 if (ecbdata->header) {
a29b0a13
SB
2337 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
2338 ecbdata->header->buf, ecbdata->header->len, 0);
3e97c7c6
GB
2339 strbuf_reset(ecbdata->header);
2340 ecbdata->header = NULL;
2341 }
34a5e1a2 2342
6973dcae 2343 if (ecbdata->label_path[0]) {
3ee8b7bf
SB
2344 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_MINUS,
2345 ecbdata->label_path[0],
2346 strlen(ecbdata->label_path[0]), 0);
2347 emit_diff_symbol(o, DIFF_SYMBOL_FILEPAIR_PLUS,
2348 ecbdata->label_path[1],
2349 strlen(ecbdata->label_path[1]), 0);
6973dcae
JH
2350 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
2351 }
cd112cef 2352
a624eaa7
JM
2353 if (diff_suppress_blank_empty
2354 && len == 2 && line[0] == ' ' && line[1] == '\n') {
2355 line[0] = '\n';
2356 len = 1;
2357 }
2358
b8d9c1a6 2359 if (line[0] == '@') {
76fd2828
JH
2360 if (ecbdata->diff_words)
2361 diff_words_flush(ecbdata);
19b9046e 2362 len = sane_truncate_line(line, len);
d68fe26f 2363 find_lno(line, ecbdata);
89cb73a1 2364 emit_hunk_header(ecbdata, line, len);
448c3ef1 2365 return;
cd112cef 2366 }
448c3ef1 2367
448c3ef1 2368 if (ecbdata->diff_words) {
ff958679
SB
2369 enum diff_symbol s =
2370 ecbdata->diff_words->type == DIFF_WORDS_PORCELAIN ?
2371 DIFF_SYMBOL_WORDS_PORCELAIN : DIFF_SYMBOL_WORDS;
448c3ef1
JH
2372 if (line[0] == '-') {
2373 diff_words_append(line, len,
2374 &ecbdata->diff_words->minus);
2375 return;
2376 } else if (line[0] == '+') {
2377 diff_words_append(line, len,
2378 &ecbdata->diff_words->plus);
2379 return;
59556548 2380 } else if (starts_with(line, "\\ ")) {
c7c2bc0a
TR
2381 /*
2382 * Eat the "no newline at eof" marker as if we
2383 * saw a "+" or "-" line with nothing on it,
2384 * and return without diff_words_flush() to
2385 * defer processing. If this is the end of
2386 * preimage, more "+" lines may come after it.
2387 */
2388 return;
448c3ef1 2389 }
76fd2828 2390 diff_words_flush(ecbdata);
ff958679 2391 emit_diff_symbol(o, s, line, len, 0);
448c3ef1
JH
2392 return;
2393 }
448c3ef1 2394
0e383e18
JH
2395 switch (line[0]) {
2396 case '+':
d68fe26f 2397 ecbdata->lno_in_postimage++;
e04df612 2398 emit_add_line(ecbdata, line + 1, len - 1);
0e383e18
JH
2399 break;
2400 case '-':
2401 ecbdata->lno_in_preimage++;
e04df612 2402 emit_del_line(ecbdata, line + 1, len - 1);
0e383e18
JH
2403 break;
2404 case ' ':
2405 ecbdata->lno_in_postimage++;
2406 ecbdata->lno_in_preimage++;
e04df612 2407 emit_context_line(ecbdata, line + 1, len - 1);
0e383e18
JH
2408 break;
2409 default:
2410 /* incomplete line at the end */
2411 ecbdata->lno_in_preimage++;
f2bb1218
SB
2412 emit_diff_symbol(o, DIFF_SYMBOL_CONTEXT_INCOMPLETE,
2413 line, len, 0);
0e383e18 2414 break;
448c3ef1 2415 }
6973dcae
JH
2416}
2417
c905cbc4 2418static void pprint_rename(struct strbuf *name, const char *a, const char *b)
6973dcae 2419{
63a01c3f
BW
2420 const char *old_name = a;
2421 const char *new_name = b;
6973dcae 2422 int pfx_length, sfx_length;
dd281f09 2423 int pfx_adjust_for_slash;
6973dcae
JH
2424 int len_a = strlen(a);
2425 int len_b = strlen(b);
663af342 2426 int a_midlen, b_midlen;
e5bfbf9b
AJ
2427 int qlen_a = quote_c_style(a, NULL, NULL, 0);
2428 int qlen_b = quote_c_style(b, NULL, NULL, 0);
2429
2430 if (qlen_a || qlen_b) {
c905cbc4
NTND
2431 quote_c_style(a, name, NULL, 0);
2432 strbuf_addstr(name, " => ");
2433 quote_c_style(b, name, NULL, 0);
2434 return;
e5bfbf9b 2435 }
6973dcae
JH
2436
2437 /* Find common prefix */
2438 pfx_length = 0;
63a01c3f
BW
2439 while (*old_name && *new_name && *old_name == *new_name) {
2440 if (*old_name == '/')
2441 pfx_length = old_name - a + 1;
2442 old_name++;
2443 new_name++;
6973dcae
JH
2444 }
2445
2446 /* Find common suffix */
63a01c3f
BW
2447 old_name = a + len_a;
2448 new_name = b + len_b;
6973dcae 2449 sfx_length = 0;
d020e27f 2450 /*
dd281f09
TR
2451 * If there is a common prefix, it must end in a slash. In
2452 * that case we let this loop run 1 into the prefix to see the
2453 * same slash.
2454 *
2455 * If there is no common prefix, we cannot do this as it would
2456 * underrun the input strings.
d020e27f 2457 */
dd281f09 2458 pfx_adjust_for_slash = (pfx_length ? 1 : 0);
63a01c3f
BW
2459 while (a + pfx_length - pfx_adjust_for_slash <= old_name &&
2460 b + pfx_length - pfx_adjust_for_slash <= new_name &&
2461 *old_name == *new_name) {
2462 if (*old_name == '/')
2463 sfx_length = len_a - (old_name - a);
2464 old_name--;
2465 new_name--;
6973dcae
JH
2466 }
2467
2468 /*
2469 * pfx{mid-a => mid-b}sfx
2470 * {pfx-a => pfx-b}sfx
2471 * pfx{sfx-a => sfx-b}
2472 * name-a => name-b
2473 */
663af342
PH
2474 a_midlen = len_a - pfx_length - sfx_length;
2475 b_midlen = len_b - pfx_length - sfx_length;
2476 if (a_midlen < 0)
2477 a_midlen = 0;
2478 if (b_midlen < 0)
2479 b_midlen = 0;
2480
c905cbc4 2481 strbuf_grow(name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
6973dcae 2482 if (pfx_length + sfx_length) {
c905cbc4
NTND
2483 strbuf_add(name, a, pfx_length);
2484 strbuf_addch(name, '{');
6973dcae 2485 }
c905cbc4
NTND
2486 strbuf_add(name, a + pfx_length, a_midlen);
2487 strbuf_addstr(name, " => ");
2488 strbuf_add(name, b + pfx_length, b_midlen);
663af342 2489 if (pfx_length + sfx_length) {
c905cbc4
NTND
2490 strbuf_addch(name, '}');
2491 strbuf_add(name, a + len_a - sfx_length, sfx_length);
6973dcae 2492 }
6973dcae
JH
2493}
2494
2495struct diffstat_t {
6973dcae
JH
2496 int nr;
2497 int alloc;
2498 struct diffstat_file {
f604652e 2499 char *from_name;
6973dcae 2500 char *name;
f604652e 2501 char *print_name;
ddf88fa6 2502 const char *comments;
6973dcae
JH
2503 unsigned is_unmerged:1;
2504 unsigned is_binary:1;
2505 unsigned is_renamed:1;
74faaa16 2506 unsigned is_interesting:1;
0974c117 2507 uintmax_t added, deleted;
6973dcae
JH
2508 } **files;
2509};
2510
2511static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
2512 const char *name_a,
2513 const char *name_b)
2514{
2515 struct diffstat_file *x;
1a4927c5 2516 x = xcalloc(1, sizeof(*x));
4c960a43 2517 ALLOC_GROW(diffstat->files, diffstat->nr + 1, diffstat->alloc);
6973dcae
JH
2518 diffstat->files[diffstat->nr++] = x;
2519 if (name_b) {
f604652e
JH
2520 x->from_name = xstrdup(name_a);
2521 x->name = xstrdup(name_b);
6973dcae
JH
2522 x->is_renamed = 1;
2523 }
f604652e
JH
2524 else {
2525 x->from_name = NULL;
9befac47 2526 x->name = xstrdup(name_a);
f604652e 2527 }
6973dcae
JH
2528 return x;
2529}
2530
2531static void diffstat_consume(void *priv, char *line, unsigned long len)
2532{
2533 struct diffstat_t *diffstat = priv;
2534 struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
2535
2536 if (line[0] == '+')
2537 x->added++;
2538 else if (line[0] == '-')
2539 x->deleted++;
2540}
2541
698ce6f8 2542const char mime_boundary_leader[] = "------------";
6973dcae 2543
a2540023
JH
2544static int scale_linear(int it, int width, int max_change)
2545{
2eeeef24
JH
2546 if (!it)
2547 return 0;
a2540023 2548 /*
2eeeef24
JH
2549 * make sure that at least one '-' or '+' is printed if
2550 * there is any change to this path. The easiest way is to
2551 * scale linearly as if the alloted width is one column shorter
2552 * than it is, and then add 1 to the result.
a2540023 2553 */
2eeeef24 2554 return 1 + (it * (width - 1) / max_change);
a2540023
JH
2555}
2556
0911c475
SB
2557static void show_graph(struct strbuf *out, char ch, int cnt,
2558 const char *set, const char *reset)
a2540023
JH
2559{
2560 if (cnt <= 0)
2561 return;
0911c475
SB
2562 strbuf_addstr(out, set);
2563 strbuf_addchars(out, ch, cnt);
2564 strbuf_addstr(out, reset);
a2540023
JH
2565}
2566
f604652e
JH
2567static void fill_print_name(struct diffstat_file *file)
2568{
c905cbc4 2569 struct strbuf pname = STRBUF_INIT;
f604652e
JH
2570
2571 if (file->print_name)
2572 return;
2573
c905cbc4
NTND
2574 if (file->is_renamed)
2575 pprint_rename(&pname, file->from_name, file->name);
2576 else
2577 quote_c_style(file->name, &pname, NULL, 0);
2578
ddf88fa6
NTND
2579 if (file->comments)
2580 strbuf_addf(&pname, " (%s)", file->comments);
2581
c905cbc4 2582 file->print_name = strbuf_detach(&pname, NULL);
f604652e
JH
2583}
2584
0911c475
SB
2585static void print_stat_summary_inserts_deletes(struct diff_options *options,
2586 int files, int insertions, int deletions)
7f814632
NTND
2587{
2588 struct strbuf sb = STRBUF_INIT;
7f814632
NTND
2589
2590 if (!files) {
2591 assert(insertions == 0 && deletions == 0);
0911c475
SB
2592 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_NO_FILES,
2593 NULL, 0, 0);
2594 return;
7f814632
NTND
2595 }
2596
2597 strbuf_addf(&sb,
218adaaa 2598 (files == 1) ? " %d file changed" : " %d files changed",
7f814632
NTND
2599 files);
2600
2601 /*
2602 * For binary diff, the caller may want to print "x files
2603 * changed" with insertions == 0 && deletions == 0.
2604 *
2605 * Not omitting "0 insertions(+), 0 deletions(-)" in this case
2606 * is probably less confusing (i.e skip over "2 files changed
2607 * but nothing about added/removed lines? Is this a bug in Git?").
2608 */
2609 if (insertions || deletions == 0) {
7f814632 2610 strbuf_addf(&sb,
218adaaa 2611 (insertions == 1) ? ", %d insertion(+)" : ", %d insertions(+)",
7f814632
NTND
2612 insertions);
2613 }
2614
2615 if (deletions || insertions == 0) {
7f814632 2616 strbuf_addf(&sb,
218adaaa 2617 (deletions == 1) ? ", %d deletion(-)" : ", %d deletions(-)",
7f814632
NTND
2618 deletions);
2619 }
2620 strbuf_addch(&sb, '\n');
0911c475
SB
2621 emit_diff_symbol(options, DIFF_SYMBOL_STATS_SUMMARY_INSERTS_DELETES,
2622 sb.buf, sb.len, 0);
7f814632 2623 strbuf_release(&sb);
0911c475
SB
2624}
2625
2626void print_stat_summary(FILE *fp, int files,
2627 int insertions, int deletions)
2628{
2629 struct diff_options o;
2630 memset(&o, 0, sizeof(o));
2631 o.file = fp;
2632
2633 print_stat_summary_inserts_deletes(&o, files, insertions, deletions);
7f814632
NTND
2634}
2635
4b25d091 2636static void show_stats(struct diffstat_t *data, struct diff_options *options)
6973dcae 2637{
eb3a9dd3 2638 int i, len, add, del, adds = 0, dels = 0;
0974c117 2639 uintmax_t max_change = 0, max_len = 0;
dc801e71
ZJS
2640 int total_files = data->nr, count;
2641 int width, name_width, graph_width, number_width = 0, bin_width = 0;
c0aa335c 2642 const char *reset, *add_c, *del_c;
e5f85df8 2643 int extra_shown = 0;
0911c475
SB
2644 const char *line_prefix = diff_line_prefix(options);
2645 struct strbuf out = STRBUF_INIT;
6973dcae
JH
2646
2647 if (data->nr == 0)
2648 return;
2649
808e1db2 2650 count = options->stat_count ? options->stat_count : data->nr;
a2540023 2651
8f67f8ae 2652 reset = diff_get_color_opt(options, DIFF_RESET);
8f67f8ae
PH
2653 add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
2654 del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
785f7432 2655
1b058bc3
ZJS
2656 /*
2657 * Find the longest filename and max number of changes
2658 */
808e1db2 2659 for (i = 0; (i < count) && (i < data->nr); i++) {
6973dcae 2660 struct diffstat_file *file = data->files[i];
0974c117 2661 uintmax_t change = file->added + file->deleted;
af0ed819
JH
2662
2663 if (!file->is_interesting && (change == 0)) {
808e1db2 2664 count++; /* not shown == room for one more */
358e460e
MG
2665 continue;
2666 }
f604652e
JH
2667 fill_print_name(file);
2668 len = strlen(file->print_name);
6973dcae
JH
2669 if (max_len < len)
2670 max_len = len;
2671
dc801e71
ZJS
2672 if (file->is_unmerged) {
2673 /* "Unmerged" is 8 characters */
2674 bin_width = bin_width < 8 ? 8 : bin_width;
6973dcae 2675 continue;
dc801e71
ZJS
2676 }
2677 if (file->is_binary) {
2678 /* "Bin XXX -> YYY bytes" */
2679 int w = 14 + decimal_width(file->added)
2680 + decimal_width(file->deleted);
2681 bin_width = bin_width < w ? w : bin_width;
2682 /* Display change counts aligned with "Bin" */
2683 number_width = 3;
2684 continue;
2685 }
2686
a2540023
JH
2687 if (max_change < change)
2688 max_change = change;
6973dcae 2689 }
a20d3c0d 2690 count = i; /* where we can stop scanning in data->files[] */
6973dcae 2691
1b058bc3
ZJS
2692 /*
2693 * We have width = stat_width or term_columns() columns total.
2694 * We want a maximum of min(max_len, stat_name_width) for the name part.
969fe57b 2695 * We want a maximum of min(max_change, stat_graph_width) for the +- part.
1b058bc3
ZJS
2696 * We also need 1 for " " and 4 + decimal_width(max_change)
2697 * for " | NNNN " and one the empty column at the end, altogether
2698 * 6 + decimal_width(max_change).
2699 *
2700 * If there's not enough space, we will use the smaller of
2701 * stat_name_width (if set) and 5/8*width for the filename,
969fe57b
ZJS
2702 * and the rest for constant elements + graph part, but no more
2703 * than stat_graph_width for the graph part.
1b058bc3
ZJS
2704 * (5/8 gives 50 for filename and 30 for the constant parts + graph
2705 * for the standard terminal size).
a2540023 2706 *
1b058bc3
ZJS
2707 * In other words: stat_width limits the maximum width, and
2708 * stat_name_width fixes the maximum width of the filename,
2709 * and is also used to divide available columns if there
2710 * aren't enough.
dc801e71
ZJS
2711 *
2712 * Binary files are displayed with "Bin XXX -> YYY bytes"
2713 * instead of the change count and graph. This part is treated
2714 * similarly to the graph part, except that it is not
41ccfdd9 2715 * "scaled". If total width is too small to accommodate the
dc801e71
ZJS
2716 * guaranteed minimum width of the filename part and the
2717 * separators and this message, this message will "overflow"
2718 * making the line longer than the maximum width.
a2540023 2719 */
1b058bc3
ZJS
2720
2721 if (options->stat_width == -1)
cd48dadb 2722 width = term_columns() - strlen(line_prefix);
a2540023 2723 else
1b058bc3 2724 width = options->stat_width ? options->stat_width : 80;
dc801e71
ZJS
2725 number_width = decimal_width(max_change) > number_width ?
2726 decimal_width(max_change) : number_width;
a2540023 2727
df44483a
ZJS
2728 if (options->stat_graph_width == -1)
2729 options->stat_graph_width = diff_stat_graph_width;
a2540023 2730
1b058bc3
ZJS
2731 /*
2732 * Guarantee 3/8*16==6 for the graph part
2733 * and 5/8*16==10 for the filename part
2734 */
2735 if (width < 16 + 6 + number_width)
2736 width = 16 + 6 + number_width;
2737
2738 /*
2739 * First assign sizes that are wanted, ignoring available width.
dc801e71
ZJS
2740 * strlen("Bin XXX -> YYY bytes") == bin_width, and the part
2741 * starting from "XXX" should fit in graph_width.
1b058bc3 2742 */
dc801e71
ZJS
2743 graph_width = max_change + 4 > bin_width ? max_change : bin_width - 4;
2744 if (options->stat_graph_width &&
2745 options->stat_graph_width < graph_width)
2746 graph_width = options->stat_graph_width;
2747
1b058bc3
ZJS
2748 name_width = (options->stat_name_width > 0 &&
2749 options->stat_name_width < max_len) ?
2750 options->stat_name_width : max_len;
2751
2752 /*
2753 * Adjust adjustable widths not to exceed maximum width
2754 */
2755 if (name_width + number_width + 6 + graph_width > width) {
678c5741 2756 if (graph_width > width * 3/8 - number_width - 6) {
1b058bc3 2757 graph_width = width * 3/8 - number_width - 6;
678c5741
LP
2758 if (graph_width < 6)
2759 graph_width = 6;
2760 }
2761
969fe57b
ZJS
2762 if (options->stat_graph_width &&
2763 graph_width > options->stat_graph_width)
2764 graph_width = options->stat_graph_width;
1b058bc3
ZJS
2765 if (name_width > width - number_width - 6 - graph_width)
2766 name_width = width - number_width - 6 - graph_width;
2767 else
2768 graph_width = width - number_width - 6 - name_width;
2769 }
2770
2771 /*
2772 * From here name_width is the width of the name area,
2773 * and graph_width is the width of the graph area.
2774 * max_change is used to scale graph properly.
2775 */
808e1db2 2776 for (i = 0; i < count; i++) {
d2543b8e 2777 const char *prefix = "";
af0ed819
JH
2778 struct diffstat_file *file = data->files[i];
2779 char *name = file->print_name;
2780 uintmax_t added = file->added;
2781 uintmax_t deleted = file->deleted;
a2540023 2782 int name_len;
6973dcae 2783
a20d3c0d 2784 if (!file->is_interesting && (added + deleted == 0))
358e460e 2785 continue;
a20d3c0d 2786
6973dcae
JH
2787 /*
2788 * "scale" the filename
2789 */
a2540023
JH
2790 len = name_width;
2791 name_len = strlen(name);
2792 if (name_width < name_len) {
6973dcae
JH
2793 char *slash;
2794 prefix = "...";
a2540023
JH
2795 len -= 3;
2796 name += name_len - len;
6973dcae
JH
2797 slash = strchr(name, '/');
2798 if (slash)
2799 name = slash;
2800 }
6973dcae 2801
af0ed819 2802 if (file->is_binary) {
0911c475
SB
2803 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2804 strbuf_addf(&out, " %*s", number_width, "Bin");
e18872b2 2805 if (!added && !deleted) {
0911c475
SB
2806 strbuf_addch(&out, '\n');
2807 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2808 out.buf, out.len, 0);
2809 strbuf_reset(&out);
e18872b2
ZJS
2810 continue;
2811 }
0911c475 2812 strbuf_addf(&out, " %s%"PRIuMAX"%s",
0974c117 2813 del_c, deleted, reset);
0911c475
SB
2814 strbuf_addstr(&out, " -> ");
2815 strbuf_addf(&out, "%s%"PRIuMAX"%s",
0974c117 2816 add_c, added, reset);
0911c475
SB
2817 strbuf_addstr(&out, " bytes\n");
2818 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2819 out.buf, out.len, 0);
2820 strbuf_reset(&out);
f604652e 2821 continue;
6973dcae 2822 }
af0ed819 2823 else if (file->is_unmerged) {
0911c475
SB
2824 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2825 strbuf_addstr(&out, " Unmerged\n");
2826 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2827 out.buf, out.len, 0);
2828 strbuf_reset(&out);
f604652e 2829 continue;
6973dcae 2830 }
6973dcae 2831
a2540023
JH
2832 /*
2833 * scale the add/delete
2834 */
6973dcae
JH
2835 add = added;
2836 del = deleted;
6973dcae 2837
1b058bc3 2838 if (graph_width <= max_change) {
d3c9cf32 2839 int total = scale_linear(add + del, graph_width, max_change);
2eeeef24
JH
2840 if (total < 2 && add && del)
2841 /* width >= 2 due to the sanity check */
2842 total = 2;
2843 if (add < del) {
1b058bc3 2844 add = scale_linear(add, graph_width, max_change);
2eeeef24
JH
2845 del = total - add;
2846 } else {
1b058bc3 2847 del = scale_linear(del, graph_width, max_change);
2eeeef24
JH
2848 add = total - del;
2849 }
6973dcae 2850 }
0911c475
SB
2851 strbuf_addf(&out, " %s%-*s |", prefix, len, name);
2852 strbuf_addf(&out, " %*"PRIuMAX"%s",
dc801e71
ZJS
2853 number_width, added + deleted,
2854 added + deleted ? " " : "");
0911c475
SB
2855 show_graph(&out, '+', add, add_c, reset);
2856 show_graph(&out, '-', del, del_c, reset);
2857 strbuf_addch(&out, '\n');
2858 emit_diff_symbol(options, DIFF_SYMBOL_STATS_LINE,
2859 out.buf, out.len, 0);
2860 strbuf_reset(&out);
c0c77734 2861 }
a20d3c0d
JH
2862
2863 for (i = 0; i < data->nr; i++) {
af0ed819
JH
2864 struct diffstat_file *file = data->files[i];
2865 uintmax_t added = file->added;
2866 uintmax_t deleted = file->deleted;
82dfc2c4
JH
2867
2868 if (file->is_unmerged ||
2869 (!file->is_interesting && (added + deleted == 0))) {
808e1db2
MG
2870 total_files--;
2871 continue;
2872 }
a20d3c0d 2873
82dfc2c4 2874 if (!file->is_binary) {
a20d3c0d
JH
2875 adds += added;
2876 dels += deleted;
2877 }
2878 if (i < count)
2879 continue;
e5f85df8 2880 if (!extra_shown)
0911c475
SB
2881 emit_diff_symbol(options,
2882 DIFF_SYMBOL_STATS_SUMMARY_ABBREV,
2883 NULL, 0, 0);
e5f85df8 2884 extra_shown = 1;
808e1db2 2885 }
0911c475
SB
2886
2887 print_stat_summary_inserts_deletes(options, total_files, adds, dels);
5a612017 2888 strbuf_release(&out);
6973dcae
JH
2889}
2890
2775d92c 2891static void show_shortstats(struct diffstat_t *data, struct diff_options *options)
ebd124c6
NP
2892{
2893 int i, adds = 0, dels = 0, total_files = data->nr;
2894
2895 if (data->nr == 0)
2896 return;
2897
2898 for (i = 0; i < data->nr; i++) {
e18872b2 2899 int added = data->files[i]->added;
0911c475 2900 int deleted = data->files[i]->deleted;
e18872b2 2901
20c8cde4
JH
2902 if (data->files[i]->is_unmerged ||
2903 (!data->files[i]->is_interesting && (added + deleted == 0))) {
e18872b2 2904 total_files--;
de9658b5 2905 } else if (!data->files[i]->is_binary) { /* don't count bytes */
e18872b2
ZJS
2906 adds += added;
2907 dels += deleted;
ebd124c6 2908 }
ebd124c6 2909 }
0911c475 2910 print_stat_summary_inserts_deletes(options, total_files, adds, dels);
ebd124c6
NP
2911}
2912
4b25d091 2913static void show_numstat(struct diffstat_t *data, struct diff_options *options)
74e2abe5
JH
2914{
2915 int i;
2916
f604652e
JH
2917 if (data->nr == 0)
2918 return;
2919
74e2abe5
JH
2920 for (i = 0; i < data->nr; i++) {
2921 struct diffstat_file *file = data->files[i];
2922
30997bb8 2923 fprintf(options->file, "%s", diff_line_prefix(options));
7be57610 2924
bfddbc5e 2925 if (file->is_binary)
c0c77734 2926 fprintf(options->file, "-\t-\t");
bfddbc5e 2927 else
c0c77734 2928 fprintf(options->file,
0974c117
JK
2929 "%"PRIuMAX"\t%"PRIuMAX"\t",
2930 file->added, file->deleted);
f604652e
JH
2931 if (options->line_termination) {
2932 fill_print_name(file);
2933 if (!file->is_renamed)
c0c77734 2934 write_name_quoted(file->name, options->file,
f604652e
JH
2935 options->line_termination);
2936 else {
c0c77734
DB
2937 fputs(file->print_name, options->file);
2938 putc(options->line_termination, options->file);
f604652e 2939 }
663af342 2940 } else {
f604652e 2941 if (file->is_renamed) {
c0c77734
DB
2942 putc('\0', options->file);
2943 write_name_quoted(file->from_name, options->file, '\0');
f604652e 2944 }
c0c77734 2945 write_name_quoted(file->name, options->file, '\0');
663af342 2946 }
74e2abe5
JH
2947 }
2948}
2949
c04a7155
JH
2950struct dirstat_file {
2951 const char *name;
2952 unsigned long changed;
7df7c019
LT
2953};
2954
c04a7155
JH
2955struct dirstat_dir {
2956 struct dirstat_file *files;
712d2c7d 2957 int alloc, nr, permille, cumulative;
c04a7155
JH
2958};
2959
7be57610
BY
2960static long gather_dirstat(struct diff_options *opt, struct dirstat_dir *dir,
2961 unsigned long changed, const char *base, int baselen)
7df7c019 2962{
585c0e2e 2963 unsigned long sum_changes = 0;
7df7c019 2964 unsigned int sources = 0;
30997bb8 2965 const char *line_prefix = diff_line_prefix(opt);
7df7c019
LT
2966
2967 while (dir->nr) {
c04a7155 2968 struct dirstat_file *f = dir->files;
7df7c019 2969 int namelen = strlen(f->name);
585c0e2e 2970 unsigned long changes;
7df7c019
LT
2971 char *slash;
2972
2973 if (namelen < baselen)
2974 break;
2975 if (memcmp(f->name, base, baselen))
2976 break;
2977 slash = strchr(f->name + baselen, '/');
2978 if (slash) {
2979 int newbaselen = slash + 1 - f->name;
585c0e2e 2980 changes = gather_dirstat(opt, dir, changed, f->name, newbaselen);
7df7c019
LT
2981 sources++;
2982 } else {
585c0e2e 2983 changes = f->changed;
7df7c019
LT
2984 dir->files++;
2985 dir->nr--;
2986 sources += 2;
2987 }
585c0e2e 2988 sum_changes += changes;
7df7c019
LT
2989 }
2990
2991 /*
2992 * We don't report dirstat's for
2993 * - the top level
2994 * - or cases where everything came from a single directory
2995 * under this directory (sources == 1).
2996 */
2997 if (baselen && sources != 1) {
585c0e2e
BW
2998 if (sum_changes) {
2999 int permille = sum_changes * 1000 / changed;
712d2c7d 3000 if (permille >= dir->permille) {
7be57610 3001 fprintf(opt->file, "%s%4d.%01d%% %.*s\n", line_prefix,
712d2c7d 3002 permille / 10, permille % 10, baselen, base);
7df7c019
LT
3003 if (!dir->cumulative)
3004 return 0;
3005 }
3006 }
3007 }
585c0e2e 3008 return sum_changes;
7df7c019
LT
3009}
3010
441bca0b
LT
3011static int dirstat_compare(const void *_a, const void *_b)
3012{
3013 const struct dirstat_file *a = _a;
3014 const struct dirstat_file *b = _b;
3015 return strcmp(a->name, b->name);
3016}
3017
c04a7155 3018static void show_dirstat(struct diff_options *options)
7df7c019
LT
3019{
3020 int i;
3021 unsigned long changed;
c04a7155
JH
3022 struct dirstat_dir dir;
3023 struct diff_queue_struct *q = &diff_queued_diff;
3024
3025 dir.files = NULL;
3026 dir.alloc = 0;
3027 dir.nr = 0;
712d2c7d 3028 dir.permille = options->dirstat_permille;
0d1e0e78 3029 dir.cumulative = options->flags.dirstat_cumulative;
7df7c019 3030
7df7c019 3031 changed = 0;
c04a7155
JH
3032 for (i = 0; i < q->nr; i++) {
3033 struct diff_filepair *p = q->queue[i];
3034 const char *name;
3035 unsigned long copied, added, damage;
3036
2ca86714 3037 name = p->two->path ? p->two->path : p->one->path;
c04a7155 3038
d9f62dfa
JK
3039 if (p->one->oid_valid && p->two->oid_valid &&
3040 oideq(&p->one->oid, &p->two->oid)) {
2ff3a803
JH
3041 /*
3042 * The SHA1 has not changed, so pre-/post-content is
3043 * identical. We can therefore skip looking at the
3044 * file contents altogether.
3045 */
3046 damage = 0;
3047 goto found_damage;
3048 }
3049
0d1e0e78 3050 if (options->flags.dirstat_by_file) {
0133dab7
JH
3051 /*
3052 * In --dirstat-by-file mode, we don't really need to
3053 * look at the actual file contents at all.
3054 * The fact that the SHA1 changed is enough for us to
3055 * add this file to the list of results
3056 * (with each file contributing equal damage).
3057 */
2ff3a803 3058 damage = 1;
0133dab7
JH
3059 goto found_damage;
3060 }
c04a7155
JH
3061
3062 if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
b78ea5fc
NTND
3063 diff_populate_filespec(options->repo, p->one, 0);
3064 diff_populate_filespec(options->repo, p->two, 0);
3065 diffcore_count_changes(options->repo,
3066 p->one, p->two, NULL, NULL,
c04a7155
JH
3067 &copied, &added);
3068 diff_free_filespec_data(p->one);
3069 diff_free_filespec_data(p->two);
3070 } else if (DIFF_FILE_VALID(p->one)) {
b78ea5fc 3071 diff_populate_filespec(options->repo, p->one, CHECK_SIZE_ONLY);
c04a7155
JH
3072 copied = added = 0;
3073 diff_free_filespec_data(p->one);
3074 } else if (DIFF_FILE_VALID(p->two)) {
b78ea5fc 3075 diff_populate_filespec(options->repo, p->two, CHECK_SIZE_ONLY);
c04a7155
JH
3076 copied = 0;
3077 added = p->two->size;
3078 diff_free_filespec_data(p->two);
3079 } else
2b0b551d 3080 continue;
c04a7155
JH
3081
3082 /*
3083 * Original minus copied is the removed material,
3084 * added is the new material. They are both damages
0133dab7 3085 * made to the preimage.
2ff3a803
JH
3086 * If the resulting damage is zero, we know that
3087 * diffcore_count_changes() considers the two entries to
d9f62dfa 3088 * be identical, but since the oid changed, we
2ff3a803
JH
3089 * know that there must have been _some_ kind of change,
3090 * so we force all entries to have damage > 0.
c04a7155
JH
3091 */
3092 damage = (p->one->size - copied) + added;
2ff3a803 3093 if (!damage)
fd33777b 3094 damage = 1;
c04a7155 3095
0133dab7 3096found_damage:
c04a7155
JH
3097 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3098 dir.files[dir.nr].name = name;
3099 dir.files[dir.nr].changed = damage;
3100 changed += damage;
3101 dir.nr++;
7df7c019
LT
3102 }
3103
3104 /* This can happen even with many files, if everything was renames */
3105 if (!changed)
3106 return;
3107
3108 /* Show all directories with more than x% of the changes */
9ed0d8d6 3109 QSORT(dir.files, dir.nr, dirstat_compare);
7be57610 3110 gather_dirstat(options, &dir, changed, "", 0);
7df7c019
LT
3111}
3112
1c57a627
JH
3113static void show_dirstat_by_line(struct diffstat_t *data, struct diff_options *options)
3114{
3115 int i;
3116 unsigned long changed;
3117 struct dirstat_dir dir;
3118
3119 if (data->nr == 0)
3120 return;
3121
3122 dir.files = NULL;
3123 dir.alloc = 0;
3124 dir.nr = 0;
3125 dir.permille = options->dirstat_permille;
0d1e0e78 3126 dir.cumulative = options->flags.dirstat_cumulative;
1c57a627
JH
3127
3128 changed = 0;
3129 for (i = 0; i < data->nr; i++) {
3130 struct diffstat_file *file = data->files[i];
3131 unsigned long damage = file->added + file->deleted;
3132 if (file->is_binary)
3133 /*
3134 * binary files counts bytes, not lines. Must find some
3135 * way to normalize binary bytes vs. textual lines.
3136 * The following heuristic assumes that there are 64
3137 * bytes per "line".
3138 * This is stupid and ugly, but very cheap...
3139 */
42c78a21 3140 damage = DIV_ROUND_UP(damage, 64);
1c57a627
JH
3141 ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
3142 dir.files[dir.nr].name = file->name;
3143 dir.files[dir.nr].changed = damage;
3144 changed += damage;
3145 dir.nr++;
3146 }
3147
3148 /* This can happen even with many files, if everything was renames */
3149 if (!changed)
3150 return;
3151
3152 /* Show all directories with more than x% of the changes */
9ed0d8d6 3153 QSORT(dir.files, dir.nr, dirstat_compare);
1c57a627
JH
3154 gather_dirstat(options, &dir, changed, "", 0);
3155}
3156
f604652e
JH
3157static void free_diffstat_info(struct diffstat_t *diffstat)
3158{
3159 int i;
3160 for (i = 0; i < diffstat->nr; i++) {
3161 struct diffstat_file *f = diffstat->files[i];
c905cbc4 3162 free(f->print_name);
f604652e
JH
3163 free(f->name);
3164 free(f->from_name);
3165 free(f);
3166 }
3167 free(diffstat->files);
3168}
3169
88246898 3170struct checkdiff_t {
88246898 3171 const char *filename;
1ba111d1 3172 int lineno;
a757c646 3173 int conflict_marker_size;
1ba111d1 3174 struct diff_options *o;
cf1b7869 3175 unsigned ws_rule;
62c64895 3176 unsigned status;
88246898
JS
3177};
3178
a757c646 3179static int is_conflict_marker(const char *line, int marker_size, unsigned long len)
04954043
JH
3180{
3181 char firstchar;
3182 int cnt;
3183
a757c646 3184 if (len < marker_size + 1)
04954043
JH
3185 return 0;
3186 firstchar = line[0];
3187 switch (firstchar) {
a757c646 3188 case '=': case '>': case '<': case '|':
04954043
JH
3189 break;
3190 default:
3191 return 0;
3192 }
a757c646 3193 for (cnt = 1; cnt < marker_size; cnt++)
04954043
JH
3194 if (line[cnt] != firstchar)
3195 return 0;
a757c646
JH
3196 /* line[1] thru line[marker_size-1] are same as firstchar */
3197 if (len < marker_size + 1 || !isspace(line[marker_size]))
04954043 3198 return 0;
04954043
JH
3199 return 1;
3200}
3201
75ab7630
JK
3202static void checkdiff_consume_hunk(void *priv,
3203 long ob, long on, long nb, long nn,
3204 const char *func, long funclen)
3205
3206{
3207 struct checkdiff_t *data = priv;
3208 data->lineno = nb - 1;
3209}
3210
88246898
JS
3211static void checkdiff_consume(void *priv, char *line, unsigned long len)
3212{
3213 struct checkdiff_t *data = priv;
a757c646 3214 int marker_size = data->conflict_marker_size;
f1c96261
JK
3215 const char *ws = diff_get_color(data->o->use_color, DIFF_WHITESPACE);
3216 const char *reset = diff_get_color(data->o->use_color, DIFF_RESET);
3217 const char *set = diff_get_color(data->o->use_color, DIFF_FILE_NEW);
c1795bb0 3218 char *err;
30997bb8 3219 const char *line_prefix;
7be57610
BY
3220
3221 assert(data->o);
30997bb8 3222 line_prefix = diff_line_prefix(data->o);
88246898
JS
3223
3224 if (line[0] == '+') {
18374e58 3225 unsigned bad;
0ef617f4 3226 data->lineno++;
a757c646 3227 if (is_conflict_marker(line + 1, marker_size, len - 1)) {
04954043
JH
3228 data->status |= 1;
3229 fprintf(data->o->file,
7be57610
BY
3230 "%s%s:%d: leftover conflict marker\n",
3231 line_prefix, data->filename, data->lineno);
04954043 3232 }
8f8841e9 3233 bad = ws_check(line + 1, len - 1, data->ws_rule);
18374e58 3234 if (!bad)
c1795bb0 3235 return;
18374e58
JH
3236 data->status |= bad;
3237 err = whitespace_error_string(bad);
7be57610
BY
3238 fprintf(data->o->file, "%s%s:%d: %s.\n",
3239 line_prefix, data->filename, data->lineno, err);
c1795bb0 3240 free(err);
a3c158d4 3241 emit_line(data->o, set, reset, line, 1);
8f8841e9 3242 ws_check_emit(line + 1, len - 1, data->ws_rule,
1ba111d1 3243 data->o->file, set, reset, ws);
877f23cc 3244 } else if (line[0] == ' ') {
88246898 3245 data->lineno++;
88246898
JS
3246 }
3247}
3248
0660626c
JH
3249static unsigned char *deflate_it(char *data,
3250 unsigned long size,
3251 unsigned long *result_size)
051308f6 3252{
0660626c
JH
3253 int bound;
3254 unsigned char *deflated;
ef49a7a0 3255 git_zstream stream;
0660626c 3256
55bb5c91 3257 git_deflate_init(&stream, zlib_compression_level);
225a6f10 3258 bound = git_deflate_bound(&stream, size);
0660626c
JH
3259 deflated = xmalloc(bound);
3260 stream.next_out = deflated;
3261 stream.avail_out = bound;
3262
3263 stream.next_in = (unsigned char *)data;
3264 stream.avail_in = size;
55bb5c91 3265 while (git_deflate(&stream, Z_FINISH) == Z_OK)
0660626c 3266 ; /* nothing */
55bb5c91 3267 git_deflate_end(&stream);
0660626c
JH
3268 *result_size = stream.total_out;
3269 return deflated;
051308f6
JH
3270}
3271
4eed0ebd
SB
3272static void emit_binary_diff_body(struct diff_options *o,
3273 mmfile_t *one, mmfile_t *two)
051308f6 3274{
0660626c
JH
3275 void *cp;
3276 void *delta;
3277 void *deflated;
3278 void *data;
3279 unsigned long orig_size;
3280 unsigned long delta_size;
3281 unsigned long deflate_size;
3282 unsigned long data_size;
051308f6 3283
0660626c
JH
3284 /* We could do deflated delta, or we could do just deflated two,
3285 * whichever is smaller.
051308f6 3286 */
0660626c
JH
3287 delta = NULL;
3288 deflated = deflate_it(two->ptr, two->size, &deflate_size);
3289 if (one->size && two->size) {
3290 delta = diff_delta(one->ptr, one->size,
3291 two->ptr, two->size,
3292 &delta_size, deflate_size);
3293 if (delta) {
3294 void *to_free = delta;
3295 orig_size = delta_size;
3296 delta = deflate_it(delta, delta_size, &delta_size);
3297 free(to_free);
051308f6
JH
3298 }
3299 }
051308f6 3300
0660626c 3301 if (delta && delta_size < deflate_size) {
ca473cef 3302 char *s = xstrfmt("%"PRIuMAX , (uintmax_t)orig_size);
4eed0ebd
SB
3303 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_DELTA,
3304 s, strlen(s), 0);
3305 free(s);
0660626c
JH
3306 free(deflated);
3307 data = delta;
3308 data_size = delta_size;
4eed0ebd
SB
3309 } else {
3310 char *s = xstrfmt("%lu", two->size);
3311 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER_LITERAL,
3312 s, strlen(s), 0);
3313 free(s);
0660626c
JH
3314 free(delta);
3315 data = deflated;
3316 data_size = deflate_size;
3317 }
051308f6 3318
0660626c
JH
3319 /* emit data encoded in base85 */
3320 cp = data;
3321 while (data_size) {
4eed0ebd 3322 int len;
0660626c 3323 int bytes = (52 < data_size) ? 52 : data_size;
4eed0ebd 3324 char line[71];
0660626c 3325 data_size -= bytes;
051308f6
JH
3326 if (bytes <= 26)
3327 line[0] = bytes + 'A' - 1;
3328 else
3329 line[0] = bytes - 26 + 'a' - 1;
3330 encode_85(line + 1, cp, bytes);
1d7f171c 3331 cp = (char *) cp + bytes;
4eed0ebd
SB
3332
3333 len = strlen(line);
3334 line[len++] = '\n';
3335 line[len] = '\0';
3336
3337 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_BODY,
3338 line, len, 0);
051308f6 3339 }
4eed0ebd 3340 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_FOOTER, NULL, 0, 0);
0660626c 3341 free(data);
051308f6
JH
3342}
3343
4eed0ebd
SB
3344static void emit_binary_diff(struct diff_options *o,
3345 mmfile_t *one, mmfile_t *two)
d4c452f0 3346{
4eed0ebd
SB
3347 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_DIFF_HEADER, NULL, 0, 0);
3348 emit_binary_diff_body(o, one, two);
3349 emit_binary_diff_body(o, two, one);
d4c452f0
JH
3350}
3351
b78ea5fc
NTND
3352int diff_filespec_is_binary(struct repository *r,
3353 struct diff_filespec *one)
29a3eefd 3354{
122aa6f9 3355 if (one->is_binary == -1) {
acd00ea0 3356 diff_filespec_load_driver(one, r->index);
122aa6f9
JK
3357 if (one->driver->binary != -1)
3358 one->is_binary = one->driver->binary;
3359 else {
3360 if (!one->data && DIFF_FILE_VALID(one))
b78ea5fc 3361 diff_populate_filespec(r, one, CHECK_BINARY);
6bf3b813 3362 if (one->is_binary == -1 && one->data)
122aa6f9
JK
3363 one->is_binary = buffer_is_binary(one->data,
3364 one->size);
3365 if (one->is_binary == -1)
3366 one->is_binary = 0;
3367 }
3368 }
29a3eefd 3369 return one->is_binary;
6973dcae
JH
3370}
3371
acd00ea0
NTND
3372static const struct userdiff_funcname *
3373diff_funcname_pattern(struct diff_options *o, struct diff_filespec *one)
f258475a 3374{
acd00ea0 3375 diff_filespec_load_driver(one, o->repo->index);
122aa6f9 3376 return one->driver->funcname.pattern ? &one->driver->funcname : NULL;
f258475a
JH
3377}
3378
a5a818ee
JH
3379void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const char *b)
3380{
3381 if (!options->a_prefix)
3382 options->a_prefix = a;
3383 if (!options->b_prefix)
3384 options->b_prefix = b;
3385}
3386
bd7ad45b 3387struct userdiff_driver *get_textconv(struct repository *r,
acd00ea0 3388 struct diff_filespec *one)
04427ac8
JK
3389{
3390 if (!DIFF_FILE_VALID(one))
3391 return NULL;
d391c0ff 3392
bd7ad45b
NTND
3393 diff_filespec_load_driver(one, r->index);
3394 return userdiff_get_textconv(r, one->driver);
04427ac8
JK
3395}
3396
6973dcae
JH
3397static void builtin_diff(const char *name_a,
3398 const char *name_b,
3399 struct diff_filespec *one,
3400 struct diff_filespec *two,
3401 const char *xfrm_msg,
296c6bb2 3402 int must_show_header,
051308f6 3403 struct diff_options *o,
6973dcae
JH
3404 int complete_rewrite)
3405{
3406 mmfile_t mf1, mf2;
3407 const char *lbl[2];
3408 char *a_one, *b_two;
d9c552f1 3409 const char *meta = diff_get_color_opt(o, DIFF_METAINFO);
8f67f8ae 3410 const char *reset = diff_get_color_opt(o, DIFF_RESET);
a5a818ee 3411 const char *a_prefix, *b_prefix;
d9bae1a1
JK
3412 struct userdiff_driver *textconv_one = NULL;
3413 struct userdiff_driver *textconv_two = NULL;
3e97c7c6 3414 struct strbuf header = STRBUF_INIT;
30997bb8 3415 const char *line_prefix = diff_line_prefix(o);
a5a818ee 3416
fd47ae6a 3417 diff_set_mnemonic_prefix(o, "a/", "b/");
0d1e0e78 3418 if (o->flags.reverse_diff) {
fd47ae6a
JK
3419 a_prefix = o->b_prefix;
3420 b_prefix = o->a_prefix;
3421 } else {
3422 a_prefix = o->a_prefix;
3423 b_prefix = o->b_prefix;
3424 }
3425
61cfbc05
JK
3426 if (o->submodule_format == DIFF_SUBMODULE_LOG &&
3427 (!one->mode || S_ISGITLINK(one->mode)) &&
3428 (!two->mode || S_ISGITLINK(two->mode))) {
f3597138 3429 show_submodule_summary(o, one->path ? one->path : two->path,
602a283a 3430 &one->oid, &two->oid,
f3597138 3431 two->dirty_submodule);
752c0c24 3432 return;
fd47ae6a
JK
3433 } else if (o->submodule_format == DIFF_SUBMODULE_INLINE_DIFF &&
3434 (!one->mode || S_ISGITLINK(one->mode)) &&
3435 (!two->mode || S_ISGITLINK(two->mode))) {
f3597138 3436 show_submodule_inline_diff(o, one->path ? one->path : two->path,
fd47ae6a 3437 &one->oid, &two->oid,
f3597138 3438 two->dirty_submodule);
fd47ae6a 3439 return;
752c0c24
JS
3440 }
3441
0d1e0e78 3442 if (o->flags.allow_textconv) {
bd7ad45b
NTND
3443 textconv_one = get_textconv(o->repo, one);
3444 textconv_two = get_textconv(o->repo, two);
3aa1f7ca
JK
3445 }
3446
71b989e7
LT
3447 /* Never use a non-valid filename anywhere if at all possible */
3448 name_a = DIFF_FILE_VALID(one) ? name_a : name_b;
3449 name_b = DIFF_FILE_VALID(two) ? name_b : name_a;
3450
a5a818ee
JH
3451 a_one = quote_two(a_prefix, name_a + (*name_a == '/'));
3452 b_two = quote_two(b_prefix, name_b + (*name_b == '/'));
6973dcae
JH
3453 lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
3454 lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
d9c552f1 3455 strbuf_addf(&header, "%s%sdiff --git %s %s%s\n", line_prefix, meta, a_one, b_two, reset);
6973dcae
JH
3456 if (lbl[0][0] == '/') {
3457 /* /dev/null */
d9c552f1 3458 strbuf_addf(&header, "%s%snew file mode %06o%s\n", line_prefix, meta, two->mode, reset);
37466447
BW
3459 if (xfrm_msg)
3460 strbuf_addstr(&header, xfrm_msg);
296c6bb2 3461 must_show_header = 1;
6973dcae
JH
3462 }
3463 else if (lbl[1][0] == '/') {
d9c552f1 3464 strbuf_addf(&header, "%s%sdeleted file mode %06o%s\n", line_prefix, meta, one->mode, reset);
37466447
BW
3465 if (xfrm_msg)
3466 strbuf_addstr(&header, xfrm_msg);
296c6bb2 3467 must_show_header = 1;
6973dcae
JH
3468 }
3469 else {
3470 if (one->mode != two->mode) {
d9c552f1
JK
3471 strbuf_addf(&header, "%s%sold mode %06o%s\n", line_prefix, meta, one->mode, reset);
3472 strbuf_addf(&header, "%s%snew mode %06o%s\n", line_prefix, meta, two->mode, reset);
296c6bb2 3473 must_show_header = 1;
cd112cef 3474 }
37466447
BW
3475 if (xfrm_msg)
3476 strbuf_addstr(&header, xfrm_msg);
3e97c7c6 3477
6973dcae
JH
3478 /*
3479 * we do not run diff between different kind
3480 * of objects.
3481 */
3482 if ((one->mode ^ two->mode) & S_IFMT)
3483 goto free_ab_and_return;
0c01857d 3484 if (complete_rewrite &&
b78ea5fc
NTND
3485 (textconv_one || !diff_filespec_is_binary(o->repo, one)) &&
3486 (textconv_two || !diff_filespec_is_binary(o->repo, two))) {
a29b0a13
SB
3487 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3488 header.buf, header.len, 0);
3e97c7c6 3489 strbuf_reset(&header);
3aa1f7ca 3490 emit_rewrite_diff(name_a, name_b, one, two,
b78ea5fc 3491 textconv_one, textconv_two, o);
34a5e1a2 3492 o->found_changes = 1;
6973dcae
JH
3493 goto free_ab_and_return;
3494 }
3495 }
3496
467ddc14 3497 if (o->irreversible_delete && lbl[1][0] == '/') {
a29b0a13
SB
3498 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf,
3499 header.len, 0);
467ddc14
JH
3500 strbuf_reset(&header);
3501 goto free_ab_and_return;
0d1e0e78 3502 } else if (!o->flags.text &&
b78ea5fc
NTND
3503 ( (!textconv_one && diff_filespec_is_binary(o->repo, one)) ||
3504 (!textconv_two && diff_filespec_is_binary(o->repo, two)) )) {
4acaaa7a 3505 struct strbuf sb = STRBUF_INIT;
1aaf69e6
NTND
3506 if (!one->data && !two->data &&
3507 S_ISREG(one->mode) && S_ISREG(two->mode) &&
0d1e0e78 3508 !o->flags.binary) {
4a7e27e9 3509 if (oideq(&one->oid, &two->oid)) {
1aaf69e6 3510 if (must_show_header)
a29b0a13
SB
3511 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3512 header.buf, header.len,
3513 0);
1aaf69e6
NTND
3514 goto free_ab_and_return;
3515 }
a29b0a13
SB
3516 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3517 header.buf, header.len, 0);
4acaaa7a
SB
3518 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",
3519 diff_line_prefix(o), lbl[0], lbl[1]);
3520 emit_diff_symbol(o, DIFF_SYMBOL_BINARY_FILES,
3521 sb.buf, sb.len, 0);
3522 strbuf_release(&sb);
1aaf69e6
NTND
3523 goto free_ab_and_return;
3524 }
b78ea5fc
NTND
3525 if (fill_mmfile(o->repo, &mf1, one) < 0 ||
3526 fill_mmfile(o->repo, &mf2, two) < 0)
b3373982 3527 die("unable to read files to diff");
0660626c
JH
3528 /* Quite common confusing case */
3529 if (mf1.size == mf2.size &&
296c6bb2
CC
3530 !memcmp(mf1.ptr, mf2.ptr, mf1.size)) {
3531 if (must_show_header)
a29b0a13
SB
3532 emit_diff_symbol(o, DIFF_SYMBOL_HEADER,
3533 header.buf, header.len, 0);
0660626c 3534 goto free_ab_and_return;
296c6bb2 3535 }
a29b0a13 3536 emit_diff_symbol(o, DIFF_SYMBOL_HEADER, header.buf, header.len, 0);
3e97c7c6 3537 strbuf_reset(&header);
0d1e0e78 3538 if (o->flags.binary)
4eed0ebd 3539 emit_binary_diff(o, &mf1, &mf2);
4acaaa7a
SB
3540 else {
3541 strbuf_addf(&sb, "%sBinary files %s and %s differ\n",