convert: always initialize attr_action in convert_attrs
[git/git.git] / convert.c
CommitLineData
a33e0b2a 1#define NO_THE_INDEX_COMPATIBILITY_MACROS
6c510bee 2#include "cache.h"
b2141fc1 3#include "config.h"
35ebfd6a 4#include "attr.h"
3fed15f5 5#include "run-command.h"
a2b665de 6#include "quote.h"
6424c2ad 7#include "sigchain.h"
edcc8581 8#include "pkt-line.h"
99605d62 9#include "sub-process.h"
35ebfd6a 10
6c510bee
LT
11/*
12 * convert.c - convert a file when checking it out and checking it in.
13 *
14 * This should use the pathname to decide on whether it wants to do some
15 * more interesting conversions (automatic gzip/unzip, general format
16 * conversions etc etc), but by default it just does automatic CRLF<->LF
942e7747 17 * translation when the "text" attribute or "auto_crlf" option is set.
6c510bee
LT
18 */
19
a7630bd4
TB
20/* Stat bits: When BIN is set, the txt bits are unset */
21#define CONVERT_STAT_BITS_TXT_LF 0x1
22#define CONVERT_STAT_BITS_TXT_CRLF 0x2
23#define CONVERT_STAT_BITS_BIN 0x4
24
c61dcff9 25enum crlf_action {
df747b81
TB
26 CRLF_UNDEFINED,
27 CRLF_BINARY,
fd6cce9e 28 CRLF_TEXT,
df747b81
TB
29 CRLF_TEXT_INPUT,
30 CRLF_TEXT_CRLF,
31 CRLF_AUTO,
32 CRLF_AUTO_INPUT,
33 CRLF_AUTO_CRLF
fd6cce9e 34};
163b9591 35
6c510bee 36struct text_stat {
28624193 37 /* NUL, CR, LF and CRLF counts */
6e336a53 38 unsigned nul, lonecr, lonelf, crlf;
6c510bee
LT
39
40 /* These are just approximations! */
41 unsigned printable, nonprintable;
42};
43
44static void gather_stats(const char *buf, unsigned long size, struct text_stat *stats)
45{
46 unsigned long i;
47
48 memset(stats, 0, sizeof(*stats));
49
50 for (i = 0; i < size; i++) {
51 unsigned char c = buf[i];
52 if (c == '\r') {
6e336a53 53 if (i+1 < size && buf[i+1] == '\n') {
6c510bee 54 stats->crlf++;
6e336a53
TB
55 i++;
56 } else
57 stats->lonecr++;
6c510bee
LT
58 continue;
59 }
60 if (c == '\n') {
6e336a53 61 stats->lonelf++;
6c510bee
LT
62 continue;
63 }
64 if (c == 127)
65 /* DEL */
66 stats->nonprintable++;
67 else if (c < 32) {
68 switch (c) {
69 /* BS, HT, ESC and FF */
70 case '\b': case '\t': case '\033': case '\014':
71 stats->printable++;
72 break;
28624193
DP
73 case 0:
74 stats->nul++;
75 /* fall through */
6c510bee
LT
76 default:
77 stats->nonprintable++;
78 }
79 }
80 else
81 stats->printable++;
82 }
f9dd4bf4
DK
83
84 /* If file ends with EOF then don't count this EOF as non-printable. */
85 if (size >= 1 && buf[size-1] == '\032')
86 stats->nonprintable--;
6c510bee
LT
87}
88
89/*
90 * The same heuristics as diff.c::mmfile_is_binary()
a7630bd4 91 * We treat files with bare CR as binary
6c510bee 92 */
a7630bd4 93static int convert_is_binary(unsigned long size, const struct text_stat *stats)
6c510bee 94{
6e336a53 95 if (stats->lonecr)
a7630bd4 96 return 1;
28624193
DP
97 if (stats->nul)
98 return 1;
6c510bee
LT
99 if ((stats->printable >> 7) < stats->nonprintable)
100 return 1;
6c510bee
LT
101 return 0;
102}
103
a7630bd4
TB
104static unsigned int gather_convert_stats(const char *data, unsigned long size)
105{
106 struct text_stat stats;
6e336a53 107 int ret = 0;
a7630bd4
TB
108 if (!data || !size)
109 return 0;
110 gather_stats(data, size, &stats);
111 if (convert_is_binary(size, &stats))
6e336a53
TB
112 ret |= CONVERT_STAT_BITS_BIN;
113 if (stats.crlf)
114 ret |= CONVERT_STAT_BITS_TXT_CRLF;
115 if (stats.lonelf)
116 ret |= CONVERT_STAT_BITS_TXT_LF;
117
118 return ret;
a7630bd4
TB
119}
120
121static const char *gather_convert_stats_ascii(const char *data, unsigned long size)
122{
123 unsigned int convert_stats = gather_convert_stats(data, size);
124
125 if (convert_stats & CONVERT_STAT_BITS_BIN)
126 return "-text";
127 switch (convert_stats) {
128 case CONVERT_STAT_BITS_TXT_LF:
129 return "lf";
130 case CONVERT_STAT_BITS_TXT_CRLF:
131 return "crlf";
132 case CONVERT_STAT_BITS_TXT_LF | CONVERT_STAT_BITS_TXT_CRLF:
133 return "mixed";
134 default:
135 return "none";
136 }
137}
138
a7609c54
BW
139const char *get_cached_convert_stats_ascii(const struct index_state *istate,
140 const char *path)
a7630bd4
TB
141{
142 const char *ret;
143 unsigned long sz;
a7609c54 144 void *data = read_blob_data_from_index(istate, path, &sz);
a7630bd4
TB
145 ret = gather_convert_stats_ascii(data, sz);
146 free(data);
147 return ret;
148}
149
150const char *get_wt_convert_stats_ascii(const char *path)
151{
152 const char *ret = "";
153 struct strbuf sb = STRBUF_INIT;
154 if (strbuf_read_file(&sb, path, 0) >= 0)
155 ret = gather_convert_stats_ascii(sb.buf, sb.len);
156 strbuf_release(&sb);
157 return ret;
158}
159
4b4024f5
TB
160static int text_eol_is_crlf(void)
161{
162 if (auto_crlf == AUTO_CRLF_TRUE)
163 return 1;
164 else if (auto_crlf == AUTO_CRLF_INPUT)
165 return 0;
166 if (core_eol == EOL_CRLF)
167 return 1;
168 if (core_eol == EOL_UNSET && EOL_NATIVE == EOL_CRLF)
169 return 1;
170 return 0;
171}
172
c61dcff9 173static enum eol output_eol(enum crlf_action crlf_action)
f217f0e8 174{
c61dcff9 175 switch (crlf_action) {
942e7747
EB
176 case CRLF_BINARY:
177 return EOL_UNSET;
df747b81 178 case CRLF_TEXT_CRLF:
942e7747 179 return EOL_CRLF;
df747b81 180 case CRLF_TEXT_INPUT:
942e7747 181 return EOL_LF;
df747b81
TB
182 case CRLF_UNDEFINED:
183 case CRLF_AUTO_CRLF:
65237284 184 return EOL_CRLF;
df747b81 185 case CRLF_AUTO_INPUT:
65237284 186 return EOL_LF;
942e7747
EB
187 case CRLF_TEXT:
188 case CRLF_AUTO:
df747b81 189 /* fall through */
4b4024f5 190 return text_eol_is_crlf() ? EOL_CRLF : EOL_LF;
942e7747 191 }
df747b81 192 warning("Illegal crlf_action %d\n", (int)crlf_action);
ec70f52f 193 return core_eol;
942e7747
EB
194}
195
c61dcff9 196static void check_safe_crlf(const char *path, enum crlf_action crlf_action,
a0ad53c1
TB
197 struct text_stat *old_stats, struct text_stat *new_stats,
198 enum safe_crlf checksafe)
21e5ad50 199{
a0ad53c1 200 if (old_stats->crlf && !new_stats->crlf ) {
21e5ad50 201 /*
a0ad53c1 202 * CRLFs would not be restored by checkout
21e5ad50 203 */
a0ad53c1 204 if (checksafe == SAFE_CRLF_WARN)
87cb7845
VA
205 warning(_("CRLF will be replaced by LF in %s.\n"
206 "The file will have its original line"
207 " endings in your working directory."), path);
a0ad53c1 208 else /* i.e. SAFE_CRLF_FAIL */
87cb7845 209 die(_("CRLF would be replaced by LF in %s."), path);
a0ad53c1 210 } else if (old_stats->lonelf && !new_stats->lonelf ) {
21e5ad50 211 /*
a0ad53c1 212 * CRLFs would be added by checkout
21e5ad50 213 */
a0ad53c1 214 if (checksafe == SAFE_CRLF_WARN)
87cb7845
VA
215 warning(_("LF will be replaced by CRLF in %s.\n"
216 "The file will have its original line"
217 " endings in your working directory."), path);
a0ad53c1 218 else /* i.e. SAFE_CRLF_FAIL */
87cb7845 219 die(_("LF would be replaced by CRLF in %s"), path);
21e5ad50
SP
220 }
221}
222
49a6d31f 223static int has_cr_in_index(const struct index_state *istate, const char *path)
c4805393 224{
c4805393 225 unsigned long sz;
c4805393
FAG
226 void *data;
227 int has_cr;
c4805393 228
49a6d31f 229 data = read_blob_data_from_index(istate, path, &sz);
4982fd78 230 if (!data)
c4805393 231 return 0;
c4805393
FAG
232 has_cr = memchr(data, '\r', sz) != NULL;
233 free(data);
234 return has_cr;
235}
236
a0ad53c1
TB
237static int will_convert_lf_to_crlf(size_t len, struct text_stat *stats,
238 enum crlf_action crlf_action)
239{
240 if (output_eol(crlf_action) != EOL_CRLF)
241 return 0;
242 /* No "naked" LF? Nothing to convert, regardless. */
243 if (!stats->lonelf)
244 return 0;
245
246 if (crlf_action == CRLF_AUTO || crlf_action == CRLF_AUTO_INPUT || crlf_action == CRLF_AUTO_CRLF) {
247 /* If we have any CR or CRLF line endings, we do not touch it */
248 /* This is the new safer autocrlf-handling */
249 if (stats->lonecr || stats->crlf)
250 return 0;
251
252 if (convert_is_binary(len, stats))
253 return 0;
254 }
255 return 1;
256
257}
258
49a6d31f
BW
259static int crlf_to_git(const struct index_state *istate,
260 const char *path, const char *src, size_t len,
c61dcff9
JH
261 struct strbuf *buf,
262 enum crlf_action crlf_action, enum safe_crlf checksafe)
6c510bee 263{
6c510bee 264 struct text_stat stats;
5ecd293d 265 char *dst;
a0ad53c1 266 int convert_crlf_into_lf;
6c510bee 267
c61dcff9 268 if (crlf_action == CRLF_BINARY ||
4c3b57b9 269 (src && !len))
5ecd293d 270 return 0;
6c510bee 271
4c3b57b9
JK
272 /*
273 * If we are doing a dry-run and have no source buffer, there is
274 * nothing to analyze; we must assume we would convert.
275 */
276 if (!buf && !src)
277 return 1;
278
5ecd293d 279 gather_stats(src, len, &stats);
a0ad53c1
TB
280 /* Optimization: No CRLF? Nothing to convert, regardless. */
281 convert_crlf_into_lf = !!stats.crlf;
6c510bee 282
df747b81 283 if (crlf_action == CRLF_AUTO || crlf_action == CRLF_AUTO_INPUT || crlf_action == CRLF_AUTO_CRLF) {
a7630bd4 284 if (convert_is_binary(len, &stats))
5ecd293d 285 return 0;
65237284 286 /*
1c25d2d8
TB
287 * If the file in the index has any CR in it, do not
288 * convert. This is the new safer autocrlf handling,
289 * unless we want to renormalize in a merge or
290 * cherry-pick.
65237284 291 */
49a6d31f
BW
292 if ((checksafe != SAFE_CRLF_RENORMALIZE) &&
293 has_cr_in_index(istate, path))
a0ad53c1 294 convert_crlf_into_lf = 0;
201ac8ef 295 }
1c25d2d8
TB
296 if ((checksafe == SAFE_CRLF_WARN ||
297 (checksafe == SAFE_CRLF_FAIL)) && len) {
a0ad53c1
TB
298 struct text_stat new_stats;
299 memcpy(&new_stats, &stats, sizeof(new_stats));
300 /* simulate "git add" */
301 if (convert_crlf_into_lf) {
302 new_stats.lonelf += new_stats.crlf;
303 new_stats.crlf = 0;
304 }
305 /* simulate "git checkout" */
306 if (will_convert_lf_to_crlf(len, &new_stats, crlf_action)) {
307 new_stats.crlf += new_stats.lonelf;
308 new_stats.lonelf = 0;
309 }
310 check_safe_crlf(path, crlf_action, &stats, &new_stats, checksafe);
311 }
312 if (!convert_crlf_into_lf)
21e5ad50
SP
313 return 0;
314
92ac3197
JK
315 /*
316 * At this point all of our source analysis is done, and we are sure we
317 * would convert. If we are in dry-run mode, we can give an answer.
318 */
319 if (!buf)
320 return 1;
321
90d16ec0
PH
322 /* only grow if not in place */
323 if (strbuf_avail(buf) + buf->len < len)
324 strbuf_grow(buf, len - buf->len);
5ecd293d 325 dst = buf->buf;
df747b81 326 if (crlf_action == CRLF_AUTO || crlf_action == CRLF_AUTO_INPUT || crlf_action == CRLF_AUTO_CRLF) {
163b9591
JH
327 /*
328 * If we guessed, we already know we rejected a file with
329 * lone CR, and we can strip a CR without looking at what
330 * follow it.
331 */
201ac8ef 332 do {
ac78e548 333 unsigned char c = *src++;
201ac8ef 334 if (c != '\r')
ac78e548 335 *dst++ = c;
5ecd293d 336 } while (--len);
201ac8ef
JH
337 } else {
338 do {
ac78e548 339 unsigned char c = *src++;
5ecd293d 340 if (! (c == '\r' && (1 < len && *src == '\n')))
ac78e548 341 *dst++ = c;
5ecd293d 342 } while (--len);
201ac8ef 343 }
5ecd293d
PH
344 strbuf_setlen(buf, dst - buf->buf);
345 return 1;
6c510bee
LT
346}
347
5ecd293d 348static int crlf_to_worktree(const char *path, const char *src, size_t len,
c61dcff9 349 struct strbuf *buf, enum crlf_action crlf_action)
6c510bee 350{
5ecd293d 351 char *to_free = NULL;
6c510bee 352 struct text_stat stats;
6c510bee 353
c61dcff9 354 if (!len || output_eol(crlf_action) != EOL_CRLF)
5ecd293d 355 return 0;
6c510bee 356
5ecd293d 357 gather_stats(src, len, &stats);
a0ad53c1 358 if (!will_convert_lf_to_crlf(len, &stats, crlf_action))
5ecd293d 359 return 0;
6c510bee 360
5ecd293d
PH
361 /* are we "faking" in place editing ? */
362 if (src == buf->buf)
b315c5c0 363 to_free = strbuf_detach(buf, NULL);
5ecd293d 364
6e336a53 365 strbuf_grow(buf, len + stats.lonelf);
5ecd293d
PH
366 for (;;) {
367 const char *nl = memchr(src, '\n', len);
368 if (!nl)
369 break;
370 if (nl > src && nl[-1] == '\r') {
371 strbuf_add(buf, src, nl + 1 - src);
372 } else {
373 strbuf_add(buf, src, nl - src);
374 strbuf_addstr(buf, "\r\n");
375 }
376 len -= nl + 1 - src;
377 src = nl + 1;
378 }
379 strbuf_add(buf, src, len);
380
381 free(to_free);
382 return 1;
6c510bee 383}
35ebfd6a 384
546bb582
JS
385struct filter_params {
386 const char *src;
387 unsigned long size;
9035d75a 388 int fd;
546bb582 389 const char *cmd;
a2b665de 390 const char *path;
546bb582
JS
391};
392
9035d75a 393static int filter_buffer_or_fd(int in, int out, void *data)
aa4ed402
JH
394{
395 /*
396 * Spawn cmd and feed the buffer contents through its stdin.
397 */
d3180279 398 struct child_process child_process = CHILD_PROCESS_INIT;
546bb582 399 struct filter_params *params = (struct filter_params *)data;
aa4ed402 400 int write_err, status;
66dbfd55
GV
401 const char *argv[] = { NULL, NULL };
402
a2b665de
PW
403 /* apply % substitution to cmd */
404 struct strbuf cmd = STRBUF_INIT;
405 struct strbuf path = STRBUF_INIT;
406 struct strbuf_expand_dict_entry dict[] = {
407 { "f", NULL, },
408 { NULL, NULL, },
409 };
410
411 /* quote the path to preserve spaces, etc. */
412 sq_quote_buf(&path, params->path);
413 dict[0].value = path.buf;
414
415 /* expand all %f with the quoted path */
416 strbuf_expand(&cmd, params->cmd, strbuf_expand_dict_cb, &dict);
417 strbuf_release(&path);
418
419 argv[0] = cmd.buf;
aa4ed402 420
dc1bfdcd 421 child_process.argv = argv;
ac0ba18d 422 child_process.use_shell = 1;
dc1bfdcd 423 child_process.in = -1;
ae6a5609 424 child_process.out = out;
aa4ed402 425
dc1bfdcd 426 if (start_command(&child_process))
255f04d6 427 return error("cannot fork to run external filter '%s'", params->cmd);
aa4ed402 428
6424c2ad
JB
429 sigchain_push(SIGPIPE, SIG_IGN);
430
9035d75a 431 if (params->src) {
0c4dd67a
JH
432 write_err = (write_in_full(child_process.in,
433 params->src, params->size) < 0);
434 if (errno == EPIPE)
435 write_err = 0;
9035d75a
SP
436 } else {
437 write_err = copy_fd(params->fd, child_process.in);
0c4dd67a
JH
438 if (write_err == COPY_WRITE_ERROR && errno == EPIPE)
439 write_err = 0;
9035d75a
SP
440 }
441
dc1bfdcd 442 if (close(child_process.in))
aa4ed402
JH
443 write_err = 1;
444 if (write_err)
255f04d6 445 error("cannot feed the input to external filter '%s'", params->cmd);
aa4ed402 446
6424c2ad
JB
447 sigchain_pop(SIGPIPE);
448
aa4ed402
JH
449 status = finish_command(&child_process);
450 if (status)
255f04d6 451 error("external filter '%s' failed %d", params->cmd, status);
a2b665de
PW
452
453 strbuf_release(&cmd);
aa4ed402
JH
454 return (write_err || status);
455}
456
234fa07e 457static int apply_single_file_filter(const char *path, const char *src, size_t len, int fd,
5ecd293d 458 struct strbuf *dst, const char *cmd)
aa4ed402
JH
459{
460 /*
461 * Create a pipeline to have the command filter the buffer's
462 * contents.
463 *
464 * (child --> cmd) --> us
465 */
b84be553 466 int err = 0;
f285a2d7 467 struct strbuf nbuf = STRBUF_INIT;
546bb582
JS
468 struct async async;
469 struct filter_params params;
aa4ed402 470
546bb582 471 memset(&async, 0, sizeof(async));
9035d75a 472 async.proc = filter_buffer_or_fd;
546bb582 473 async.data = &params;
ae6a5609 474 async.out = -1;
546bb582
JS
475 params.src = src;
476 params.size = len;
9035d75a 477 params.fd = fd;
546bb582 478 params.cmd = cmd;
a2b665de 479 params.path = path;
aa4ed402
JH
480
481 fflush(NULL);
546bb582
JS
482 if (start_async(&async))
483 return 0; /* error was already reported */
aa4ed402 484
546bb582 485 if (strbuf_read(&nbuf, async.out, len) < 0) {
b84be553 486 err = error("read from external filter '%s' failed", cmd);
aa4ed402 487 }
546bb582 488 if (close(async.out)) {
b84be553 489 err = error("read from external filter '%s' failed", cmd);
aa4ed402 490 }
546bb582 491 if (finish_async(&async)) {
b84be553 492 err = error("external filter '%s' failed", cmd);
aa4ed402
JH
493 }
494
b84be553 495 if (!err) {
90d16ec0 496 strbuf_swap(dst, &nbuf);
5ecd293d 497 }
90d16ec0 498 strbuf_release(&nbuf);
b84be553 499 return !err;
aa4ed402
JH
500}
501
234fa07e
LS
502#define CAP_CLEAN (1u<<0)
503#define CAP_SMUDGE (1u<<1)
504
1b0b46ee
BP
505struct cmd2process {
506 struct subprocess_entry subprocess; /* must be the first member! */
507 unsigned int supported_capabilities;
508};
509
f514d7d1
BP
510static int subprocess_map_initialized;
511static struct hashmap subprocess_map;
edcc8581 512
7ddb9b2c 513static int start_multi_file_filter_fn(struct subprocess_entry *subprocess)
edcc8581
LS
514{
515 int err;
7ddb9b2c 516 struct cmd2process *entry = (struct cmd2process *)subprocess;
edcc8581
LS
517 struct string_list cap_list = STRING_LIST_INIT_NODUP;
518 char *cap_buf;
519 const char *cap_name;
7ddb9b2c
BP
520 struct child_process *process = &subprocess->process;
521 const char *cmd = subprocess->cmd;
edcc8581
LS
522
523 sigchain_push(SIGPIPE, SIG_IGN);
524
c0c70f7a 525 err = packet_writel(process->in, "git-filter-client", "version=2", NULL);
edcc8581
LS
526 if (err)
527 goto done;
528
529 err = strcmp(packet_read_line(process->out, NULL), "git-filter-server");
530 if (err) {
531 error("external filter '%s' does not support filter protocol version 2", cmd);
532 goto done;
533 }
534 err = strcmp(packet_read_line(process->out, NULL), "version=2");
535 if (err)
536 goto done;
537 err = packet_read_line(process->out, NULL) != NULL;
538 if (err)
539 goto done;
540
c0c70f7a 541 err = packet_writel(process->in, "capability=clean", "capability=smudge", NULL);
edcc8581
LS
542
543 for (;;) {
544 cap_buf = packet_read_line(process->out, NULL);
545 if (!cap_buf)
546 break;
547 string_list_split_in_place(&cap_list, cap_buf, '=', 1);
548
549 if (cap_list.nr != 2 || strcmp(cap_list.items[0].string, "capability"))
550 continue;
551
552 cap_name = cap_list.items[1].string;
553 if (!strcmp(cap_name, "clean")) {
554 entry->supported_capabilities |= CAP_CLEAN;
555 } else if (!strcmp(cap_name, "smudge")) {
556 entry->supported_capabilities |= CAP_SMUDGE;
557 } else {
558 warning(
559 "external filter '%s' requested unsupported filter capability '%s'",
560 cmd, cap_name
561 );
562 }
563
564 string_list_clear(&cap_list, 0);
565 }
566
567done:
568 sigchain_pop(SIGPIPE);
569
a810ea99
BP
570 return err;
571}
572
edcc8581
LS
573static int apply_multi_file_filter(const char *path, const char *src, size_t len,
574 int fd, struct strbuf *dst, const char *cmd,
575 const unsigned int wanted_capability)
576{
577 int err;
578 struct cmd2process *entry;
579 struct child_process *process;
580 struct strbuf nbuf = STRBUF_INIT;
581 struct strbuf filter_status = STRBUF_INIT;
582 const char *filter_type;
583
f514d7d1
BP
584 if (!subprocess_map_initialized) {
585 subprocess_map_initialized = 1;
7663cdc8
SB
586 hashmap_init(&subprocess_map, (hashmap_cmp_fn) cmd2process_cmp,
587 NULL, 0);
edcc8581
LS
588 entry = NULL;
589 } else {
f514d7d1 590 entry = (struct cmd2process *)subprocess_find_entry(&subprocess_map, cmd);
edcc8581
LS
591 }
592
593 fflush(NULL);
594
595 if (!entry) {
7ddb9b2c
BP
596 entry = xmalloc(sizeof(*entry));
597 entry->supported_capabilities = 0;
598
f514d7d1 599 if (subprocess_start(&subprocess_map, &entry->subprocess, cmd, start_multi_file_filter_fn)) {
7ddb9b2c 600 free(entry);
edcc8581 601 return 0;
7ddb9b2c 602 }
edcc8581 603 }
1b0b46ee 604 process = &entry->subprocess.process;
edcc8581
LS
605
606 if (!(wanted_capability & entry->supported_capabilities))
607 return 0;
608
609 if (CAP_CLEAN & wanted_capability)
610 filter_type = "clean";
611 else if (CAP_SMUDGE & wanted_capability)
612 filter_type = "smudge";
613 else
614 die("unexpected filter type");
615
616 sigchain_push(SIGPIPE, SIG_IGN);
617
618 assert(strlen(filter_type) < LARGE_PACKET_DATA_MAX - strlen("command=\n"));
619 err = packet_write_fmt_gently(process->in, "command=%s\n", filter_type);
620 if (err)
621 goto done;
622
623 err = strlen(path) > LARGE_PACKET_DATA_MAX - strlen("pathname=\n");
624 if (err) {
625 error("path name too long for external filter");
626 goto done;
627 }
628
629 err = packet_write_fmt_gently(process->in, "pathname=%s\n", path);
630 if (err)
631 goto done;
632
633 err = packet_flush_gently(process->in);
634 if (err)
635 goto done;
636
637 if (fd >= 0)
638 err = write_packetized_from_fd(fd, process->in);
639 else
640 err = write_packetized_from_buf(src, len, process->in);
641 if (err)
642 goto done;
643
4f2a2e9f
BP
644 err = subprocess_read_status(process->out, &filter_status);
645 if (err)
646 goto done;
647
edcc8581
LS
648 err = strcmp(filter_status.buf, "success");
649 if (err)
650 goto done;
651
652 err = read_packetized_to_strbuf(process->out, &nbuf) < 0;
653 if (err)
654 goto done;
655
4f2a2e9f
BP
656 err = subprocess_read_status(process->out, &filter_status);
657 if (err)
658 goto done;
659
edcc8581
LS
660 err = strcmp(filter_status.buf, "success");
661
662done:
663 sigchain_pop(SIGPIPE);
664
070e5f72 665 if (err) {
edcc8581
LS
666 if (!strcmp(filter_status.buf, "error")) {
667 /* The filter signaled a problem with the file. */
668 } else if (!strcmp(filter_status.buf, "abort")) {
669 /*
670 * The filter signaled a permanent problem. Don't try to filter
671 * files with the same command for the lifetime of the current
672 * Git process.
673 */
674 entry->supported_capabilities &= ~wanted_capability;
675 } else {
676 /*
677 * Something went wrong with the protocol filter.
678 * Force shutdown and restart if another blob requires filtering.
679 */
680 error("external filter '%s' failed", cmd);
f514d7d1 681 subprocess_stop(&subprocess_map, &entry->subprocess);
7ddb9b2c 682 free(entry);
edcc8581
LS
683 }
684 } else {
685 strbuf_swap(dst, &nbuf);
686 }
687 strbuf_release(&nbuf);
688 return !err;
aa4ed402
JH
689}
690
691static struct convert_driver {
692 const char *name;
693 struct convert_driver *next;
cd8be6c9
BH
694 const char *smudge;
695 const char *clean;
edcc8581 696 const char *process;
36daaaca 697 int required;
aa4ed402
JH
698} *user_convert, **user_convert_tail;
699
234fa07e
LS
700static int apply_filter(const char *path, const char *src, size_t len,
701 int fd, struct strbuf *dst, struct convert_driver *drv,
702 const unsigned int wanted_capability)
703{
704 const char *cmd = NULL;
705
706 if (!drv)
707 return 0;
708
709 if (!dst)
710 return 1;
711
edcc8581 712 if ((CAP_CLEAN & wanted_capability) && !drv->process && drv->clean)
234fa07e 713 cmd = drv->clean;
edcc8581 714 else if ((CAP_SMUDGE & wanted_capability) && !drv->process && drv->smudge)
234fa07e
LS
715 cmd = drv->smudge;
716
717 if (cmd && *cmd)
718 return apply_single_file_filter(path, src, len, fd, dst, cmd);
edcc8581
LS
719 else if (drv->process && *drv->process)
720 return apply_multi_file_filter(path, src, len, fd, dst, drv->process, wanted_capability);
234fa07e
LS
721
722 return 0;
723}
724
ef90d6d4 725static int read_convert_config(const char *var, const char *value, void *cb)
aa4ed402 726{
d731f0ad 727 const char *key, *name;
aa4ed402
JH
728 int namelen;
729 struct convert_driver *drv;
730
731 /*
732 * External conversion drivers are configured using
733 * "filter.<name>.variable".
734 */
d731f0ad 735 if (parse_config_key(var, "filter", &name, &namelen, &key) < 0 || !name)
aa4ed402 736 return 0;
aa4ed402
JH
737 for (drv = user_convert; drv; drv = drv->next)
738 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
739 break;
740 if (!drv) {
aa4ed402 741 drv = xcalloc(1, sizeof(struct convert_driver));
182af834 742 drv->name = xmemdupz(name, namelen);
aa4ed402
JH
743 *user_convert_tail = drv;
744 user_convert_tail = &(drv->next);
745 }
746
aa4ed402
JH
747 /*
748 * filter.<name>.smudge and filter.<name>.clean specifies
749 * the command line:
750 *
751 * command-line
752 *
753 * The command-line will not be interpolated in any way.
754 */
755
d731f0ad 756 if (!strcmp("smudge", key))
cd8be6c9
BH
757 return git_config_string(&drv->smudge, var, value);
758
d731f0ad 759 if (!strcmp("clean", key))
cd8be6c9 760 return git_config_string(&drv->clean, var, value);
aa4ed402 761
edcc8581
LS
762 if (!strcmp("process", key))
763 return git_config_string(&drv->process, var, value);
764
d731f0ad 765 if (!strcmp("required", key)) {
36daaaca
JB
766 drv->required = git_config_bool(var, value);
767 return 0;
768 }
769
aa4ed402
JH
770 return 0;
771}
772
3fed15f5
JH
773static int count_ident(const char *cp, unsigned long size)
774{
775 /*
af9b54bb 776 * "$Id: 0000000000000000000000000000000000000000 $" <=> "$Id$"
3fed15f5
JH
777 */
778 int cnt = 0;
779 char ch;
780
781 while (size) {
782 ch = *cp++;
783 size--;
784 if (ch != '$')
785 continue;
af9b54bb 786 if (size < 3)
3fed15f5 787 break;
af9b54bb 788 if (memcmp("Id", cp, 2))
3fed15f5 789 continue;
af9b54bb
AP
790 ch = cp[2];
791 cp += 3;
792 size -= 3;
3fed15f5 793 if (ch == '$')
af9b54bb 794 cnt++; /* $Id$ */
3fed15f5
JH
795 if (ch != ':')
796 continue;
797
798 /*
af9b54bb 799 * "$Id: ... "; scan up to the closing dollar sign and discard.
3fed15f5
JH
800 */
801 while (size) {
802 ch = *cp++;
803 size--;
804 if (ch == '$') {
805 cnt++;
806 break;
807 }
a9f3049f
HG
808 if (ch == '\n')
809 break;
3fed15f5
JH
810 }
811 }
812 return cnt;
813}
814
5ecd293d
PH
815static int ident_to_git(const char *path, const char *src, size_t len,
816 struct strbuf *buf, int ident)
3fed15f5 817{
5ecd293d 818 char *dst, *dollar;
3fed15f5 819
4c3b57b9 820 if (!ident || (src && !count_ident(src, len)))
5ecd293d
PH
821 return 0;
822
92ac3197
JK
823 if (!buf)
824 return 1;
825
90d16ec0
PH
826 /* only grow if not in place */
827 if (strbuf_avail(buf) + buf->len < len)
828 strbuf_grow(buf, len - buf->len);
5ecd293d
PH
829 dst = buf->buf;
830 for (;;) {
831 dollar = memchr(src, '$', len);
832 if (!dollar)
833 break;
77321184 834 memmove(dst, src, dollar + 1 - src);
5ecd293d
PH
835 dst += dollar + 1 - src;
836 len -= dollar + 1 - src;
837 src = dollar + 1;
838
839 if (len > 3 && !memcmp(src, "Id:", 3)) {
840 dollar = memchr(src + 3, '$', len - 3);
841 if (!dollar)
842 break;
a9f3049f
HG
843 if (memchr(src + 3, '\n', dollar - src - 3)) {
844 /* Line break before the next dollar. */
845 continue;
846 }
847
af9b54bb
AP
848 memcpy(dst, "Id$", 3);
849 dst += 3;
5ecd293d
PH
850 len -= dollar + 1 - src;
851 src = dollar + 1;
3fed15f5
JH
852 }
853 }
77321184 854 memmove(dst, src, len);
5ecd293d
PH
855 strbuf_setlen(buf, dst + len - buf->buf);
856 return 1;
3fed15f5
JH
857}
858
5ecd293d
PH
859static int ident_to_worktree(const char *path, const char *src, size_t len,
860 struct strbuf *buf, int ident)
3fed15f5 861{
3fed15f5 862 unsigned char sha1[20];
07814d90 863 char *to_free = NULL, *dollar, *spc;
5ecd293d 864 int cnt;
3fed15f5
JH
865
866 if (!ident)
5ecd293d 867 return 0;
3fed15f5 868
5ecd293d 869 cnt = count_ident(src, len);
3fed15f5 870 if (!cnt)
5ecd293d 871 return 0;
3fed15f5 872
5ecd293d
PH
873 /* are we "faking" in place editing ? */
874 if (src == buf->buf)
b315c5c0 875 to_free = strbuf_detach(buf, NULL);
5ecd293d 876 hash_sha1_file(src, len, "blob", sha1);
3fed15f5 877
5ecd293d
PH
878 strbuf_grow(buf, len + cnt * 43);
879 for (;;) {
880 /* step 1: run to the next '$' */
881 dollar = memchr(src, '$', len);
882 if (!dollar)
883 break;
884 strbuf_add(buf, src, dollar + 1 - src);
885 len -= dollar + 1 - src;
886 src = dollar + 1;
c23290d5 887
5ecd293d
PH
888 /* step 2: does it looks like a bit like Id:xxx$ or Id$ ? */
889 if (len < 3 || memcmp("Id", src, 2))
3fed15f5
JH
890 continue;
891
5ecd293d
PH
892 /* step 3: skip over Id$ or Id:xxxxx$ */
893 if (src[2] == '$') {
894 src += 3;
895 len -= 3;
896 } else if (src[2] == ':') {
897 /*
898 * It's possible that an expanded Id has crept its way into the
07814d90
HG
899 * repository, we cope with that by stripping the expansion out.
900 * This is probably not a good idea, since it will cause changes
901 * on checkout, which won't go away by stash, but let's keep it
902 * for git-style ids.
5ecd293d
PH
903 */
904 dollar = memchr(src + 3, '$', len - 3);
905 if (!dollar) {
906 /* incomplete keyword, no more '$', so just quit the loop */
907 break;
908 }
c23290d5 909
a9f3049f
HG
910 if (memchr(src + 3, '\n', dollar - src - 3)) {
911 /* Line break before the next dollar. */
912 continue;
913 }
914
07814d90
HG
915 spc = memchr(src + 4, ' ', dollar - src - 4);
916 if (spc && spc < dollar-1) {
917 /* There are spaces in unexpected places.
918 * This is probably an id from some other
919 * versioning system. Keep it for now.
920 */
921 continue;
922 }
923
5ecd293d
PH
924 len -= dollar + 1 - src;
925 src = dollar + 1;
926 } else {
927 /* it wasn't a "Id$" or "Id:xxxx$" */
928 continue;
929 }
c23290d5 930
5ecd293d
PH
931 /* step 4: substitute */
932 strbuf_addstr(buf, "Id: ");
933 strbuf_add(buf, sha1_to_hex(sha1), 40);
934 strbuf_addstr(buf, " $");
3fed15f5 935 }
5ecd293d 936 strbuf_add(buf, src, len);
3fed15f5 937
5ecd293d
PH
938 free(to_free);
939 return 1;
35ebfd6a
JH
940}
941
7bd18054 942static enum crlf_action git_path_check_crlf(struct attr_check_item *check)
35ebfd6a 943{
6073ee85
JH
944 const char *value = check->value;
945
946 if (ATTR_TRUE(value))
947 return CRLF_TEXT;
948 else if (ATTR_FALSE(value))
949 return CRLF_BINARY;
950 else if (ATTR_UNSET(value))
951 ;
952 else if (!strcmp(value, "input"))
df747b81 953 return CRLF_TEXT_INPUT;
fd6cce9e
EB
954 else if (!strcmp(value, "auto"))
955 return CRLF_AUTO;
df747b81 956 return CRLF_UNDEFINED;
35ebfd6a
JH
957}
958
7bd18054 959static enum eol git_path_check_eol(struct attr_check_item *check)
fd6cce9e
EB
960{
961 const char *value = check->value;
962
963 if (ATTR_UNSET(value))
964 ;
965 else if (!strcmp(value, "lf"))
966 return EOL_LF;
967 else if (!strcmp(value, "crlf"))
968 return EOL_CRLF;
969 return EOL_UNSET;
970}
971
7bd18054 972static struct convert_driver *git_path_check_convert(struct attr_check_item *check)
aa4ed402
JH
973{
974 const char *value = check->value;
975 struct convert_driver *drv;
976
977 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
978 return NULL;
979 for (drv = user_convert; drv; drv = drv->next)
980 if (!strcmp(value, drv->name))
981 return drv;
982 return NULL;
983}
984
7bd18054 985static int git_path_check_ident(struct attr_check_item *check)
3fed15f5
JH
986{
987 const char *value = check->value;
988
989 return !!ATTR_TRUE(value);
990}
991
3bfba20d
JH
992struct conv_attrs {
993 struct convert_driver *drv;
bb211b4d
TB
994 enum crlf_action attr_action; /* What attr says */
995 enum crlf_action crlf_action; /* When no attr is set, use core.autocrlf */
3bfba20d
JH
996 int ident;
997};
998
3bfba20d 999static void convert_attrs(struct conv_attrs *ca, const char *path)
83295964 1000{
2aef63d3 1001 static struct attr_check *check;
83295964 1002
2aef63d3
JH
1003 if (!check) {
1004 check = attr_check_initl("crlf", "ident", "filter",
1005 "eol", "text", NULL);
83295964
JH
1006 user_convert_tail = &user_convert;
1007 git_config(read_convert_config, NULL);
1008 }
3bfba20d 1009
2aef63d3
JH
1010 if (!git_check_attr(path, check)) {
1011 struct attr_check_item *ccheck = check->items;
bb211b4d 1012 ca->crlf_action = git_path_check_crlf(ccheck + 4);
df747b81 1013 if (ca->crlf_action == CRLF_UNDEFINED)
92cce135
TB
1014 ca->crlf_action = git_path_check_crlf(ccheck + 0);
1015 ca->ident = git_path_check_ident(ccheck + 1);
1016 ca->drv = git_path_check_convert(ccheck + 2);
817a0c79
TB
1017 if (ca->crlf_action != CRLF_BINARY) {
1018 enum eol eol_attr = git_path_check_eol(ccheck + 3);
65237284
TB
1019 if (ca->crlf_action == CRLF_AUTO && eol_attr == EOL_LF)
1020 ca->crlf_action = CRLF_AUTO_INPUT;
1021 else if (ca->crlf_action == CRLF_AUTO && eol_attr == EOL_CRLF)
1022 ca->crlf_action = CRLF_AUTO_CRLF;
1023 else if (eol_attr == EOL_LF)
817a0c79
TB
1024 ca->crlf_action = CRLF_TEXT_INPUT;
1025 else if (eol_attr == EOL_CRLF)
1026 ca->crlf_action = CRLF_TEXT_CRLF;
1027 }
3bfba20d
JH
1028 } else {
1029 ca->drv = NULL;
df747b81 1030 ca->crlf_action = CRLF_UNDEFINED;
3bfba20d
JH
1031 ca->ident = 0;
1032 }
5c94c93d
1033
1034 /* Save attr and make a decision for action */
1035 ca->attr_action = ca->crlf_action;
df747b81
TB
1036 if (ca->crlf_action == CRLF_TEXT)
1037 ca->crlf_action = text_eol_is_crlf() ? CRLF_TEXT_CRLF : CRLF_TEXT_INPUT;
1038 if (ca->crlf_action == CRLF_UNDEFINED && auto_crlf == AUTO_CRLF_FALSE)
1039 ca->crlf_action = CRLF_BINARY;
1040 if (ca->crlf_action == CRLF_UNDEFINED && auto_crlf == AUTO_CRLF_TRUE)
1041 ca->crlf_action = CRLF_AUTO_CRLF;
1042 if (ca->crlf_action == CRLF_UNDEFINED && auto_crlf == AUTO_CRLF_INPUT)
1043 ca->crlf_action = CRLF_AUTO_INPUT;
83295964
JH
1044}
1045
9035d75a
SP
1046int would_convert_to_git_filter_fd(const char *path)
1047{
1048 struct conv_attrs ca;
1049
1050 convert_attrs(&ca, path);
1051 if (!ca.drv)
1052 return 0;
1053
1054 /*
1055 * Apply a filter to an fd only if the filter is required to succeed.
1056 * We must die if the filter fails, because the original data before
1057 * filtering is not available.
1058 */
1059 if (!ca.drv->required)
1060 return 0;
1061
234fa07e 1062 return apply_filter(path, NULL, 0, -1, NULL, ca.drv, CAP_CLEAN);
9035d75a
SP
1063}
1064
a7630bd4
TB
1065const char *get_convert_attr_ascii(const char *path)
1066{
1067 struct conv_attrs ca;
a7630bd4
TB
1068
1069 convert_attrs(&ca, path);
bb211b4d 1070 switch (ca.attr_action) {
df747b81 1071 case CRLF_UNDEFINED:
a7630bd4
TB
1072 return "";
1073 case CRLF_BINARY:
1074 return "-text";
1075 case CRLF_TEXT:
1076 return "text";
df747b81 1077 case CRLF_TEXT_INPUT:
a7630bd4 1078 return "text eol=lf";
df747b81
TB
1079 case CRLF_TEXT_CRLF:
1080 return "text eol=crlf";
a7630bd4
TB
1081 case CRLF_AUTO:
1082 return "text=auto";
df747b81 1083 case CRLF_AUTO_CRLF:
65237284 1084 return "text=auto eol=crlf";
df747b81 1085 case CRLF_AUTO_INPUT:
65237284 1086 return "text=auto eol=lf";
a7630bd4
TB
1087 }
1088 return "";
1089}
1090
82b474e0
BW
1091int convert_to_git(const struct index_state *istate,
1092 const char *path, const char *src, size_t len,
21e5ad50 1093 struct strbuf *dst, enum safe_crlf checksafe)
35ebfd6a 1094{
3bfba20d 1095 int ret = 0;
3bfba20d 1096 struct conv_attrs ca;
6073ee85 1097
3bfba20d 1098 convert_attrs(&ca, path);
3fed15f5 1099
234fa07e
LS
1100 ret |= apply_filter(path, src, len, -1, dst, ca.drv, CAP_CLEAN);
1101 if (!ret && ca.drv && ca.drv->required)
36daaaca
JB
1102 die("%s: clean filter '%s' failed", path, ca.drv->name);
1103
92ac3197 1104 if (ret && dst) {
5ecd293d
PH
1105 src = dst->buf;
1106 len = dst->len;
aa4ed402 1107 }
82b474e0 1108 ret |= crlf_to_git(istate, path, src, len, dst, ca.crlf_action, checksafe);
92ac3197 1109 if (ret && dst) {
5ecd293d
PH
1110 src = dst->buf;
1111 len = dst->len;
6073ee85 1112 }
3bfba20d 1113 return ret | ident_to_git(path, src, len, dst, ca.ident);
35ebfd6a
JH
1114}
1115
d6c41c20
BW
1116void convert_to_git_filter_fd(const struct index_state *istate,
1117 const char *path, int fd, struct strbuf *dst,
9035d75a
SP
1118 enum safe_crlf checksafe)
1119{
1120 struct conv_attrs ca;
1121 convert_attrs(&ca, path);
1122
1123 assert(ca.drv);
edcc8581 1124 assert(ca.drv->clean || ca.drv->process);
9035d75a 1125
234fa07e 1126 if (!apply_filter(path, NULL, 0, fd, dst, ca.drv, CAP_CLEAN))
9035d75a
SP
1127 die("%s: clean filter '%s' failed", path, ca.drv->name);
1128
d6c41c20 1129 crlf_to_git(istate, path, dst->buf, dst->len, dst, ca.crlf_action, checksafe);
9035d75a
SP
1130 ident_to_git(path, dst->buf, dst->len, dst, ca.ident);
1131}
1132
43dd2332
EB
1133static int convert_to_working_tree_internal(const char *path, const char *src,
1134 size_t len, struct strbuf *dst,
1135 int normalizing)
35ebfd6a 1136{
36daaaca 1137 int ret = 0, ret_filter = 0;
3bfba20d 1138 struct conv_attrs ca;
6073ee85 1139
3bfba20d 1140 convert_attrs(&ca, path);
3fed15f5 1141
3bfba20d 1142 ret |= ident_to_worktree(path, src, len, dst, ca.ident);
5ecd293d
PH
1143 if (ret) {
1144 src = dst->buf;
1145 len = dst->len;
3fed15f5 1146 }
43dd2332
EB
1147 /*
1148 * CRLF conversion can be skipped if normalizing, unless there
edcc8581
LS
1149 * is a smudge or process filter (even if the process filter doesn't
1150 * support smudge). The filters might expect CRLFs.
43dd2332 1151 */
edcc8581 1152 if ((ca.drv && (ca.drv->smudge || ca.drv->process)) || !normalizing) {
3bfba20d 1153 ret |= crlf_to_worktree(path, src, len, dst, ca.crlf_action);
43dd2332
EB
1154 if (ret) {
1155 src = dst->buf;
1156 len = dst->len;
1157 }
aa4ed402 1158 }
36daaaca 1159
234fa07e
LS
1160 ret_filter = apply_filter(path, src, len, -1, dst, ca.drv, CAP_SMUDGE);
1161 if (!ret_filter && ca.drv && ca.drv->required)
36daaaca
JB
1162 die("%s: smudge filter %s failed", path, ca.drv->name);
1163
1164 return ret | ret_filter;
35ebfd6a 1165}
f217f0e8 1166
43dd2332
EB
1167int convert_to_working_tree(const char *path, const char *src, size_t len, struct strbuf *dst)
1168{
1169 return convert_to_working_tree_internal(path, src, len, dst, 0);
1170}
1171
a33e0b2a
BW
1172int renormalize_buffer(const struct index_state *istate, const char *path,
1173 const char *src, size_t len, struct strbuf *dst)
f217f0e8 1174{
43dd2332 1175 int ret = convert_to_working_tree_internal(path, src, len, dst, 1);
f217f0e8
EB
1176 if (ret) {
1177 src = dst->buf;
1178 len = dst->len;
1179 }
a33e0b2a 1180 return ret | convert_to_git(istate, path, src, len, dst, SAFE_CRLF_RENORMALIZE);
f217f0e8 1181}
dd8e9121 1182
b6691092
JH
1183/*****************************************************************
1184 *
749f763d 1185 * Streaming conversion support
b6691092
JH
1186 *
1187 *****************************************************************/
1188
1189typedef int (*filter_fn)(struct stream_filter *,
1190 const char *input, size_t *isize_p,
1191 char *output, size_t *osize_p);
1192typedef void (*free_fn)(struct stream_filter *);
1193
1194struct stream_filter_vtbl {
1195 filter_fn filter;
1196 free_fn free;
1197};
1198
1199struct stream_filter {
1200 struct stream_filter_vtbl *vtbl;
1201};
1202
1203static int null_filter_fn(struct stream_filter *filter,
1204 const char *input, size_t *isize_p,
1205 char *output, size_t *osize_p)
1206{
4ae66704
JH
1207 size_t count;
1208
1209 if (!input)
1210 return 0; /* we do not keep any states */
1211 count = *isize_p;
b6691092
JH
1212 if (*osize_p < count)
1213 count = *osize_p;
1214 if (count) {
1215 memmove(output, input, count);
1216 *isize_p -= count;
1217 *osize_p -= count;
1218 }
1219 return 0;
1220}
1221
1222static void null_free_fn(struct stream_filter *filter)
1223{
1224 ; /* nothing -- null instances are shared */
1225}
1226
1227static struct stream_filter_vtbl null_vtbl = {
1228 null_filter_fn,
1229 null_free_fn,
1230};
1231
1232static struct stream_filter null_filter_singleton = {
1233 &null_vtbl,
1234};
1235
1236int is_null_stream_filter(struct stream_filter *filter)
1237{
1238 return filter == &null_filter_singleton;
1239}
1240
b84c7839
JH
1241
1242/*
1243 * LF-to-CRLF filter
1244 */
284e3d28
CMN
1245
1246struct lf_to_crlf_filter {
1247 struct stream_filter filter;
8496f568
JH
1248 unsigned has_held:1;
1249 char held;
284e3d28
CMN
1250};
1251
e322ee38
JH
1252static int lf_to_crlf_filter_fn(struct stream_filter *filter,
1253 const char *input, size_t *isize_p,
1254 char *output, size_t *osize_p)
1255{
284e3d28
CMN
1256 size_t count, o = 0;
1257 struct lf_to_crlf_filter *lf_to_crlf = (struct lf_to_crlf_filter *)filter;
1258
8496f568
JH
1259 /*
1260 * We may be holding onto the CR to see if it is followed by a
1261 * LF, in which case we would need to go to the main loop.
1262 * Otherwise, just emit it to the output stream.
1263 */
1264 if (lf_to_crlf->has_held && (lf_to_crlf->held != '\r' || !input)) {
1265 output[o++] = lf_to_crlf->held;
1266 lf_to_crlf->has_held = 0;
284e3d28 1267 }
e322ee38 1268
87afe9a5
JH
1269 /* We are told to drain */
1270 if (!input) {
1271 *osize_p -= o;
1272 return 0;
1273 }
e322ee38 1274
e322ee38 1275 count = *isize_p;
8496f568 1276 if (count || lf_to_crlf->has_held) {
284e3d28 1277 size_t i;
8496f568
JH
1278 int was_cr = 0;
1279
1280 if (lf_to_crlf->has_held) {
1281 was_cr = 1;
1282 lf_to_crlf->has_held = 0;
1283 }
1284
284e3d28 1285 for (i = 0; o < *osize_p && i < count; i++) {
e322ee38 1286 char ch = input[i];
8496f568 1287
e322ee38 1288 if (ch == '\n') {
284e3d28 1289 output[o++] = '\r';
8496f568
JH
1290 } else if (was_cr) {
1291 /*
1292 * Previous round saw CR and it is not followed
1293 * by a LF; emit the CR before processing the
1294 * current character.
1295 */
1296 output[o++] = '\r';
e322ee38 1297 }
8496f568
JH
1298
1299 /*
1300 * We may have consumed the last output slot,
1301 * in which case we need to break out of this
1302 * loop; hold the current character before
1303 * returning.
1304 */
1305 if (*osize_p <= o) {
1306 lf_to_crlf->has_held = 1;
1307 lf_to_crlf->held = ch;
1308 continue; /* break but increment i */
1309 }
1310
1311 if (ch == '\r') {
1312 was_cr = 1;
1313 continue;
1314 }
1315
1316 was_cr = 0;
e322ee38
JH
1317 output[o++] = ch;
1318 }
1319
1320 *osize_p -= o;
1321 *isize_p -= i;
8496f568
JH
1322
1323 if (!lf_to_crlf->has_held && was_cr) {
1324 lf_to_crlf->has_held = 1;
1325 lf_to_crlf->held = '\r';
1326 }
e322ee38
JH
1327 }
1328 return 0;
1329}
1330
284e3d28
CMN
1331static void lf_to_crlf_free_fn(struct stream_filter *filter)
1332{
1333 free(filter);
1334}
1335
e322ee38
JH
1336static struct stream_filter_vtbl lf_to_crlf_vtbl = {
1337 lf_to_crlf_filter_fn,
284e3d28 1338 lf_to_crlf_free_fn,
e322ee38
JH
1339};
1340
284e3d28
CMN
1341static struct stream_filter *lf_to_crlf_filter(void)
1342{
87afe9a5 1343 struct lf_to_crlf_filter *lf_to_crlf = xcalloc(1, sizeof(*lf_to_crlf));
e322ee38 1344
284e3d28 1345 lf_to_crlf->filter.vtbl = &lf_to_crlf_vtbl;
284e3d28
CMN
1346 return (struct stream_filter *)lf_to_crlf;
1347}
b84c7839 1348
a265a7f9
JH
1349/*
1350 * Cascade filter
1351 */
1352#define FILTER_BUFFER 1024
1353struct cascade_filter {
1354 struct stream_filter filter;
1355 struct stream_filter *one;
1356 struct stream_filter *two;
1357 char buf[FILTER_BUFFER];
1358 int end, ptr;
1359};
1360
1361static int cascade_filter_fn(struct stream_filter *filter,
1362 const char *input, size_t *isize_p,
1363 char *output, size_t *osize_p)
1364{
1365 struct cascade_filter *cas = (struct cascade_filter *) filter;
1366 size_t filled = 0;
1367 size_t sz = *osize_p;
1368 size_t to_feed, remaining;
1369
1370 /*
1371 * input -- (one) --> buf -- (two) --> output
1372 */
1373 while (filled < sz) {
1374 remaining = sz - filled;
1375
1376 /* do we already have something to feed two with? */
1377 if (cas->ptr < cas->end) {
1378 to_feed = cas->end - cas->ptr;
1379 if (stream_filter(cas->two,
1380 cas->buf + cas->ptr, &to_feed,
1381 output + filled, &remaining))
1382 return -1;
1383 cas->ptr += (cas->end - cas->ptr) - to_feed;
1384 filled = sz - remaining;
1385 continue;
1386 }
1387
1388 /* feed one from upstream and have it emit into our buffer */
1389 to_feed = input ? *isize_p : 0;
1390 if (input && !to_feed)
1391 break;
1392 remaining = sizeof(cas->buf);
1393 if (stream_filter(cas->one,
1394 input, &to_feed,
1395 cas->buf, &remaining))
1396 return -1;
1397 cas->end = sizeof(cas->buf) - remaining;
1398 cas->ptr = 0;
1399 if (input) {
1400 size_t fed = *isize_p - to_feed;
1401 *isize_p -= fed;
1402 input += fed;
1403 }
1404
1405 /* do we know that we drained one completely? */
1406 if (input || cas->end)
1407 continue;
1408
1409 /* tell two to drain; we have nothing more to give it */
1410 to_feed = 0;
1411 remaining = sz - filled;
1412 if (stream_filter(cas->two,
1413 NULL, &to_feed,
1414 output + filled, &remaining))
1415 return -1;
1416 if (remaining == (sz - filled))
1417 break; /* completely drained two */
1418 filled = sz - remaining;
1419 }
1420 *osize_p -= filled;
1421 return 0;
1422}
1423
1424static void cascade_free_fn(struct stream_filter *filter)
1425{
1426 struct cascade_filter *cas = (struct cascade_filter *)filter;
1427 free_stream_filter(cas->one);
1428 free_stream_filter(cas->two);
1429 free(filter);
1430}
1431
1432static struct stream_filter_vtbl cascade_vtbl = {
1433 cascade_filter_fn,
1434 cascade_free_fn,
1435};
1436
1437static struct stream_filter *cascade_filter(struct stream_filter *one,
1438 struct stream_filter *two)
1439{
1440 struct cascade_filter *cascade;
1441
1442 if (!one || is_null_stream_filter(one))
1443 return two;
1444 if (!two || is_null_stream_filter(two))
1445 return one;
1446
1447 cascade = xmalloc(sizeof(*cascade));
1448 cascade->one = one;
1449 cascade->two = two;
1450 cascade->end = cascade->ptr = 0;
1451 cascade->filter.vtbl = &cascade_vtbl;
1452 return (struct stream_filter *)cascade;
1453}
1454
b84c7839
JH
1455/*
1456 * ident filter
1457 */
1458#define IDENT_DRAINING (-1)
1459#define IDENT_SKIPPING (-2)
1460struct ident_filter {
1461 struct stream_filter filter;
1462 struct strbuf left;
1463 int state;
1464 char ident[45]; /* ": x40 $" */
1465};
1466
1467static int is_foreign_ident(const char *str)
1468{
1469 int i;
1470
ae021d87 1471 if (!skip_prefix(str, "$Id: ", &str))
b84c7839 1472 return 0;
ae021d87 1473 for (i = 0; str[i]; i++) {
b84c7839
JH
1474 if (isspace(str[i]) && str[i+1] != '$')
1475 return 1;
1476 }
1477 return 0;
1478}
1479
1480static void ident_drain(struct ident_filter *ident, char **output_p, size_t *osize_p)
1481{
1482 size_t to_drain = ident->left.len;
1483
1484 if (*osize_p < to_drain)
1485 to_drain = *osize_p;
1486 if (to_drain) {
1487 memcpy(*output_p, ident->left.buf, to_drain);
1488 strbuf_remove(&ident->left, 0, to_drain);
1489 *output_p += to_drain;
1490 *osize_p -= to_drain;
1491 }
1492 if (!ident->left.len)
1493 ident->state = 0;
1494}
1495
1496static int ident_filter_fn(struct stream_filter *filter,
1497 const char *input, size_t *isize_p,
1498 char *output, size_t *osize_p)
1499{
1500 struct ident_filter *ident = (struct ident_filter *)filter;
1501 static const char head[] = "$Id";
1502
1503 if (!input) {
1504 /* drain upon eof */
1505 switch (ident->state) {
1506 default:
1507 strbuf_add(&ident->left, head, ident->state);
1508 case IDENT_SKIPPING:
1509 /* fallthru */
1510 case IDENT_DRAINING:
1511 ident_drain(ident, &output, osize_p);
1512 }
1513 return 0;
1514 }
1515
1516 while (*isize_p || (ident->state == IDENT_DRAINING)) {
1517 int ch;
1518
1519 if (ident->state == IDENT_DRAINING) {
1520 ident_drain(ident, &output, osize_p);
1521 if (!*osize_p)
1522 break;
1523 continue;
1524 }
1525
1526 ch = *(input++);
1527 (*isize_p)--;
1528
1529 if (ident->state == IDENT_SKIPPING) {
1530 /*
1531 * Skipping until '$' or LF, but keeping them
1532 * in case it is a foreign ident.
1533 */
1534 strbuf_addch(&ident->left, ch);
1535 if (ch != '\n' && ch != '$')
1536 continue;
1537 if (ch == '$' && !is_foreign_ident(ident->left.buf)) {
1538 strbuf_setlen(&ident->left, sizeof(head) - 1);
1539 strbuf_addstr(&ident->left, ident->ident);
1540 }
1541 ident->state = IDENT_DRAINING;
1542 continue;
1543 }
1544
1545 if (ident->state < sizeof(head) &&
1546 head[ident->state] == ch) {
1547 ident->state++;
1548 continue;
1549 }
1550
1551 if (ident->state)
1552 strbuf_add(&ident->left, head, ident->state);
1553 if (ident->state == sizeof(head) - 1) {
1554 if (ch != ':' && ch != '$') {
1555 strbuf_addch(&ident->left, ch);
1556 ident->state = 0;
1557 continue;
1558 }
1559
1560 if (ch == ':') {
1561 strbuf_addch(&ident->left, ch);
1562 ident->state = IDENT_SKIPPING;
1563 } else {
1564 strbuf_addstr(&ident->left, ident->ident);
1565 ident->state = IDENT_DRAINING;
1566 }
1567 continue;
1568 }
1569
1570 strbuf_addch(&ident->left, ch);
1571 ident->state = IDENT_DRAINING;
1572 }
1573 return 0;
1574}
1575
1576static void ident_free_fn(struct stream_filter *filter)
1577{
1578 struct ident_filter *ident = (struct ident_filter *)filter;
1579 strbuf_release(&ident->left);
1580 free(filter);
1581}
1582
1583static struct stream_filter_vtbl ident_vtbl = {
1584 ident_filter_fn,
1585 ident_free_fn,
1586};
1587
1588static struct stream_filter *ident_filter(const unsigned char *sha1)
1589{
1590 struct ident_filter *ident = xmalloc(sizeof(*ident));
1591
5096d490
JK
1592 xsnprintf(ident->ident, sizeof(ident->ident),
1593 ": %s $", sha1_to_hex(sha1));
b84c7839
JH
1594 strbuf_init(&ident->left, 0);
1595 ident->filter.vtbl = &ident_vtbl;
1596 ident->state = 0;
1597 return (struct stream_filter *)ident;
1598}
1599
dd8e9121 1600/*
b6691092
JH
1601 * Return an appropriately constructed filter for the path, or NULL if
1602 * the contents cannot be filtered without reading the whole thing
1603 * in-core.
1604 *
1605 * Note that you would be crazy to set CRLF, smuge/clean or ident to a
1606 * large binary blob you would want us not to slurp into the memory!
dd8e9121 1607 */
b6691092 1608struct stream_filter *get_stream_filter(const char *path, const unsigned char *sha1)
dd8e9121
JH
1609{
1610 struct conv_attrs ca;
b84c7839 1611 struct stream_filter *filter = NULL;
dd8e9121
JH
1612
1613 convert_attrs(&ca, path);
edcc8581 1614 if (ca.drv && (ca.drv->process || ca.drv->smudge || ca.drv->clean))
caa47adc
TB
1615 return NULL;
1616
1617 if (ca.crlf_action == CRLF_AUTO || ca.crlf_action == CRLF_AUTO_CRLF)
1618 return NULL;
b84c7839
JH
1619
1620 if (ca.ident)
1621 filter = ident_filter(sha1);
dd8e9121 1622
caa47adc 1623 if (output_eol(ca.crlf_action) == EOL_CRLF)
284e3d28 1624 filter = cascade_filter(filter, lf_to_crlf_filter());
caa47adc
TB
1625 else
1626 filter = cascade_filter(filter, &null_filter_singleton);
e322ee38 1627
b84c7839 1628 return filter;
b6691092
JH
1629}
1630
1631void free_stream_filter(struct stream_filter *filter)
1632{
1633 filter->vtbl->free(filter);
1634}
1635
1636int stream_filter(struct stream_filter *filter,
1637 const char *input, size_t *isize_p,
1638 char *output, size_t *osize_p)
1639{
1640 return filter->vtbl->filter(filter, input, isize_p, output, osize_p);
dd8e9121 1641}