3 #include "run-command.h"
9 * convert.c - convert a file when checking it out and checking it in.
11 * This should use the pathname to decide on whether it wants to do some
12 * more interesting conversions (automatic gzip/unzip, general format
13 * conversions etc etc), but by default it just does automatic CRLF<->LF
14 * translation when the "text" attribute or "auto_crlf" option is set.
17 /* Stat bits: When BIN is set, the txt bits are unset */
18 #define CONVERT_STAT_BITS_TXT_LF 0x1
19 #define CONVERT_STAT_BITS_TXT_CRLF 0x2
20 #define CONVERT_STAT_BITS_BIN 0x4
34 /* NUL, CR, LF and CRLF counts */
35 unsigned nul
, lonecr
, lonelf
, crlf
;
37 /* These are just approximations! */
38 unsigned printable
, nonprintable
;
41 static void gather_stats(const char *buf
, unsigned long size
, struct text_stat
*stats
)
45 memset(stats
, 0, sizeof(*stats
));
47 for (i
= 0; i
< size
; i
++) {
48 unsigned char c
= buf
[i
];
50 if (i
+1 < size
&& buf
[i
+1] == '\n') {
63 stats
->nonprintable
++;
66 /* BS, HT, ESC and FF */
67 case '\b': case '\t': case '\033': case '\014':
74 stats
->nonprintable
++;
81 /* If file ends with EOF then don't count this EOF as non-printable. */
82 if (size
>= 1 && buf
[size
-1] == '\032')
83 stats
->nonprintable
--;
87 * The same heuristics as diff.c::mmfile_is_binary()
88 * We treat files with bare CR as binary
90 static int convert_is_binary(unsigned long size
, const struct text_stat
*stats
)
96 if ((stats
->printable
>> 7) < stats
->nonprintable
)
101 static unsigned int gather_convert_stats(const char *data
, unsigned long size
)
103 struct text_stat stats
;
107 gather_stats(data
, size
, &stats
);
108 if (convert_is_binary(size
, &stats
))
109 ret
|= CONVERT_STAT_BITS_BIN
;
111 ret
|= CONVERT_STAT_BITS_TXT_CRLF
;
113 ret
|= CONVERT_STAT_BITS_TXT_LF
;
118 static const char *gather_convert_stats_ascii(const char *data
, unsigned long size
)
120 unsigned int convert_stats
= gather_convert_stats(data
, size
);
122 if (convert_stats
& CONVERT_STAT_BITS_BIN
)
124 switch (convert_stats
) {
125 case CONVERT_STAT_BITS_TXT_LF
:
127 case CONVERT_STAT_BITS_TXT_CRLF
:
129 case CONVERT_STAT_BITS_TXT_LF
| CONVERT_STAT_BITS_TXT_CRLF
:
136 const char *get_cached_convert_stats_ascii(const char *path
)
140 void *data
= read_blob_data_from_cache(path
, &sz
);
141 ret
= gather_convert_stats_ascii(data
, sz
);
146 const char *get_wt_convert_stats_ascii(const char *path
)
148 const char *ret
= "";
149 struct strbuf sb
= STRBUF_INIT
;
150 if (strbuf_read_file(&sb
, path
, 0) >= 0)
151 ret
= gather_convert_stats_ascii(sb
.buf
, sb
.len
);
156 static int text_eol_is_crlf(void)
158 if (auto_crlf
== AUTO_CRLF_TRUE
)
160 else if (auto_crlf
== AUTO_CRLF_INPUT
)
162 if (core_eol
== EOL_CRLF
)
164 if (core_eol
== EOL_UNSET
&& EOL_NATIVE
== EOL_CRLF
)
169 static enum eol
output_eol(enum crlf_action crlf_action
)
171 switch (crlf_action
) {
176 case CRLF_TEXT_INPUT
:
181 case CRLF_AUTO_INPUT
:
186 return text_eol_is_crlf() ? EOL_CRLF
: EOL_LF
;
188 warning("Illegal crlf_action %d\n", (int)crlf_action
);
192 static void check_safe_crlf(const char *path
, enum crlf_action crlf_action
,
193 struct text_stat
*old_stats
, struct text_stat
*new_stats
,
194 enum safe_crlf checksafe
)
196 if (old_stats
->crlf
&& !new_stats
->crlf
) {
198 * CRLFs would not be restored by checkout
200 if (checksafe
== SAFE_CRLF_WARN
)
201 warning("CRLF will be replaced by LF in %s.\nThe file will have its original line endings in your working directory.", path
);
202 else /* i.e. SAFE_CRLF_FAIL */
203 die("CRLF would be replaced by LF in %s.", path
);
204 } else if (old_stats
->lonelf
&& !new_stats
->lonelf
) {
206 * CRLFs would be added by checkout
208 if (checksafe
== SAFE_CRLF_WARN
)
209 warning("LF will be replaced by CRLF in %s.\nThe file will have its original line endings in your working directory.", path
);
210 else /* i.e. SAFE_CRLF_FAIL */
211 die("LF would be replaced by CRLF in %s", path
);
215 static int has_cr_in_index(const char *path
)
221 data
= read_blob_data_from_cache(path
, &sz
);
224 has_cr
= memchr(data
, '\r', sz
) != NULL
;
229 static int will_convert_lf_to_crlf(size_t len
, struct text_stat
*stats
,
230 enum crlf_action crlf_action
)
232 if (output_eol(crlf_action
) != EOL_CRLF
)
234 /* No "naked" LF? Nothing to convert, regardless. */
238 if (crlf_action
== CRLF_AUTO
|| crlf_action
== CRLF_AUTO_INPUT
|| crlf_action
== CRLF_AUTO_CRLF
) {
239 /* If we have any CR or CRLF line endings, we do not touch it */
240 /* This is the new safer autocrlf-handling */
241 if (stats
->lonecr
|| stats
->crlf
)
244 if (convert_is_binary(len
, stats
))
251 static int crlf_to_git(const char *path
, const char *src
, size_t len
,
253 enum crlf_action crlf_action
, enum safe_crlf checksafe
)
255 struct text_stat stats
;
257 int convert_crlf_into_lf
;
259 if (crlf_action
== CRLF_BINARY
||
264 * If we are doing a dry-run and have no source buffer, there is
265 * nothing to analyze; we must assume we would convert.
270 gather_stats(src
, len
, &stats
);
271 /* Optimization: No CRLF? Nothing to convert, regardless. */
272 convert_crlf_into_lf
= !!stats
.crlf
;
274 if (crlf_action
== CRLF_AUTO
|| crlf_action
== CRLF_AUTO_INPUT
|| crlf_action
== CRLF_AUTO_CRLF
) {
275 if (convert_is_binary(len
, &stats
))
278 * If the file in the index has any CR in it, do not convert.
279 * This is the new safer autocrlf handling.
281 if (checksafe
== SAFE_CRLF_RENORMALIZE
)
282 checksafe
= SAFE_CRLF_FALSE
;
283 else if (has_cr_in_index(path
))
284 convert_crlf_into_lf
= 0;
286 if (checksafe
&& len
) {
287 struct text_stat new_stats
;
288 memcpy(&new_stats
, &stats
, sizeof(new_stats
));
289 /* simulate "git add" */
290 if (convert_crlf_into_lf
) {
291 new_stats
.lonelf
+= new_stats
.crlf
;
294 /* simulate "git checkout" */
295 if (will_convert_lf_to_crlf(len
, &new_stats
, crlf_action
)) {
296 new_stats
.crlf
+= new_stats
.lonelf
;
297 new_stats
.lonelf
= 0;
299 check_safe_crlf(path
, crlf_action
, &stats
, &new_stats
, checksafe
);
301 if (!convert_crlf_into_lf
)
305 * At this point all of our source analysis is done, and we are sure we
306 * would convert. If we are in dry-run mode, we can give an answer.
311 /* only grow if not in place */
312 if (strbuf_avail(buf
) + buf
->len
< len
)
313 strbuf_grow(buf
, len
- buf
->len
);
315 if (crlf_action
== CRLF_AUTO
|| crlf_action
== CRLF_AUTO_INPUT
|| crlf_action
== CRLF_AUTO_CRLF
) {
317 * If we guessed, we already know we rejected a file with
318 * lone CR, and we can strip a CR without looking at what
322 unsigned char c
= *src
++;
328 unsigned char c
= *src
++;
329 if (! (c
== '\r' && (1 < len
&& *src
== '\n')))
333 strbuf_setlen(buf
, dst
- buf
->buf
);
337 static int crlf_to_worktree(const char *path
, const char *src
, size_t len
,
338 struct strbuf
*buf
, enum crlf_action crlf_action
)
340 char *to_free
= NULL
;
341 struct text_stat stats
;
343 if (!len
|| output_eol(crlf_action
) != EOL_CRLF
)
346 gather_stats(src
, len
, &stats
);
347 if (!will_convert_lf_to_crlf(len
, &stats
, crlf_action
))
350 /* are we "faking" in place editing ? */
352 to_free
= strbuf_detach(buf
, NULL
);
354 strbuf_grow(buf
, len
+ stats
.lonelf
);
356 const char *nl
= memchr(src
, '\n', len
);
359 if (nl
> src
&& nl
[-1] == '\r') {
360 strbuf_add(buf
, src
, nl
+ 1 - src
);
362 strbuf_add(buf
, src
, nl
- src
);
363 strbuf_addstr(buf
, "\r\n");
368 strbuf_add(buf
, src
, len
);
374 struct filter_params
{
382 static int filter_buffer_or_fd(int in
, int out
, void *data
)
385 * Spawn cmd and feed the buffer contents through its stdin.
387 struct child_process child_process
= CHILD_PROCESS_INIT
;
388 struct filter_params
*params
= (struct filter_params
*)data
;
389 int write_err
, status
;
390 const char *argv
[] = { NULL
, NULL
};
392 /* apply % substitution to cmd */
393 struct strbuf cmd
= STRBUF_INIT
;
394 struct strbuf path
= STRBUF_INIT
;
395 struct strbuf_expand_dict_entry dict
[] = {
400 /* quote the path to preserve spaces, etc. */
401 sq_quote_buf(&path
, params
->path
);
402 dict
[0].value
= path
.buf
;
404 /* expand all %f with the quoted path */
405 strbuf_expand(&cmd
, params
->cmd
, strbuf_expand_dict_cb
, &dict
);
406 strbuf_release(&path
);
410 child_process
.argv
= argv
;
411 child_process
.use_shell
= 1;
412 child_process
.in
= -1;
413 child_process
.out
= out
;
415 if (start_command(&child_process
))
416 return error("cannot fork to run external filter '%s'", params
->cmd
);
418 sigchain_push(SIGPIPE
, SIG_IGN
);
421 write_err
= (write_in_full(child_process
.in
,
422 params
->src
, params
->size
) < 0);
426 write_err
= copy_fd(params
->fd
, child_process
.in
);
427 if (write_err
== COPY_WRITE_ERROR
&& errno
== EPIPE
)
431 if (close(child_process
.in
))
434 error("cannot feed the input to external filter '%s'", params
->cmd
);
436 sigchain_pop(SIGPIPE
);
438 status
= finish_command(&child_process
);
440 error("external filter '%s' failed %d", params
->cmd
, status
);
442 strbuf_release(&cmd
);
443 return (write_err
|| status
);
446 static int apply_single_file_filter(const char *path
, const char *src
, size_t len
, int fd
,
447 struct strbuf
*dst
, const char *cmd
)
450 * Create a pipeline to have the command filter the buffer's
453 * (child --> cmd) --> us
456 struct strbuf nbuf
= STRBUF_INIT
;
458 struct filter_params params
;
460 memset(&async
, 0, sizeof(async
));
461 async
.proc
= filter_buffer_or_fd
;
462 async
.data
= ¶ms
;
471 if (start_async(&async
))
472 return 0; /* error was already reported */
474 if (strbuf_read(&nbuf
, async
.out
, len
) < 0) {
475 err
= error("read from external filter '%s' failed", cmd
);
477 if (close(async
.out
)) {
478 err
= error("read from external filter '%s' failed", cmd
);
480 if (finish_async(&async
)) {
481 err
= error("external filter '%s' failed", cmd
);
485 strbuf_swap(dst
, &nbuf
);
487 strbuf_release(&nbuf
);
491 #define CAP_CLEAN (1u<<0)
492 #define CAP_SMUDGE (1u<<1)
495 struct hashmap_entry ent
; /* must be the first member! */
496 unsigned int supported_capabilities
;
498 struct child_process process
;
501 static int cmd_process_map_initialized
;
502 static struct hashmap cmd_process_map
;
504 static int cmd2process_cmp(const struct cmd2process
*e1
,
505 const struct cmd2process
*e2
,
508 return strcmp(e1
->cmd
, e2
->cmd
);
511 static struct cmd2process
*find_multi_file_filter_entry(struct hashmap
*hashmap
, const char *cmd
)
513 struct cmd2process key
;
514 hashmap_entry_init(&key
, strhash(cmd
));
516 return hashmap_get(hashmap
, &key
, NULL
);
519 static int packet_write_list(int fd
, const char *line
, ...)
523 va_start(args
, line
);
527 if (strlen(line
) > LARGE_PACKET_DATA_MAX
)
529 err
= packet_write_fmt_gently(fd
, "%s\n", line
);
532 line
= va_arg(args
, const char*);
535 return packet_flush_gently(fd
);
538 static void read_multi_file_filter_status(int fd
, struct strbuf
*status
)
540 struct strbuf
**pair
;
543 line
= packet_read_line(fd
, NULL
);
546 pair
= strbuf_split_str(line
, '=', 2);
547 if (pair
[0] && pair
[0]->len
&& pair
[1]) {
548 /* the last "status=<foo>" line wins */
549 if (!strcmp(pair
[0]->buf
, "status=")) {
550 strbuf_reset(status
);
551 strbuf_addbuf(status
, pair
[1]);
554 strbuf_list_free(pair
);
558 static void kill_multi_file_filter(struct hashmap
*hashmap
, struct cmd2process
*entry
)
563 entry
->process
.clean_on_exit
= 0;
564 kill(entry
->process
.pid
, SIGTERM
);
565 finish_command(&entry
->process
);
567 hashmap_remove(hashmap
, entry
, NULL
);
571 static void stop_multi_file_filter(struct child_process
*process
)
573 sigchain_push(SIGPIPE
, SIG_IGN
);
574 /* Closing the pipe signals the filter to initiate a shutdown. */
577 sigchain_pop(SIGPIPE
);
578 /* Finish command will wait until the shutdown is complete. */
579 finish_command(process
);
582 static struct cmd2process
*start_multi_file_filter(struct hashmap
*hashmap
, const char *cmd
)
585 struct cmd2process
*entry
;
586 struct child_process
*process
;
587 const char *argv
[] = { cmd
, NULL
};
588 struct string_list cap_list
= STRING_LIST_INIT_NODUP
;
590 const char *cap_name
;
592 entry
= xmalloc(sizeof(*entry
));
594 entry
->supported_capabilities
= 0;
595 process
= &entry
->process
;
597 child_process_init(process
);
598 process
->argv
= argv
;
599 process
->use_shell
= 1;
602 process
->clean_on_exit
= 1;
603 process
->clean_on_exit_handler
= stop_multi_file_filter
;
605 if (start_command(process
)) {
606 error("cannot fork to run external filter '%s'", cmd
);
610 hashmap_entry_init(entry
, strhash(cmd
));
612 sigchain_push(SIGPIPE
, SIG_IGN
);
614 err
= packet_write_list(process
->in
, "git-filter-client", "version=2", NULL
);
618 err
= strcmp(packet_read_line(process
->out
, NULL
), "git-filter-server");
620 error("external filter '%s' does not support filter protocol version 2", cmd
);
623 err
= strcmp(packet_read_line(process
->out
, NULL
), "version=2");
626 err
= packet_read_line(process
->out
, NULL
) != NULL
;
630 err
= packet_write_list(process
->in
, "capability=clean", "capability=smudge", NULL
);
633 cap_buf
= packet_read_line(process
->out
, NULL
);
636 string_list_split_in_place(&cap_list
, cap_buf
, '=', 1);
638 if (cap_list
.nr
!= 2 || strcmp(cap_list
.items
[0].string
, "capability"))
641 cap_name
= cap_list
.items
[1].string
;
642 if (!strcmp(cap_name
, "clean")) {
643 entry
->supported_capabilities
|= CAP_CLEAN
;
644 } else if (!strcmp(cap_name
, "smudge")) {
645 entry
->supported_capabilities
|= CAP_SMUDGE
;
648 "external filter '%s' requested unsupported filter capability '%s'",
653 string_list_clear(&cap_list
, 0);
657 sigchain_pop(SIGPIPE
);
659 if (err
|| errno
== EPIPE
) {
660 error("initialization for external filter '%s' failed", cmd
);
661 kill_multi_file_filter(hashmap
, entry
);
665 hashmap_add(hashmap
, entry
);
669 static int apply_multi_file_filter(const char *path
, const char *src
, size_t len
,
670 int fd
, struct strbuf
*dst
, const char *cmd
,
671 const unsigned int wanted_capability
)
674 struct cmd2process
*entry
;
675 struct child_process
*process
;
676 struct strbuf nbuf
= STRBUF_INIT
;
677 struct strbuf filter_status
= STRBUF_INIT
;
678 const char *filter_type
;
680 if (!cmd_process_map_initialized
) {
681 cmd_process_map_initialized
= 1;
682 hashmap_init(&cmd_process_map
, (hashmap_cmp_fn
) cmd2process_cmp
, 0);
685 entry
= find_multi_file_filter_entry(&cmd_process_map
, cmd
);
691 entry
= start_multi_file_filter(&cmd_process_map
, cmd
);
695 process
= &entry
->process
;
697 if (!(wanted_capability
& entry
->supported_capabilities
))
700 if (CAP_CLEAN
& wanted_capability
)
701 filter_type
= "clean";
702 else if (CAP_SMUDGE
& wanted_capability
)
703 filter_type
= "smudge";
705 die("unexpected filter type");
707 sigchain_push(SIGPIPE
, SIG_IGN
);
709 assert(strlen(filter_type
) < LARGE_PACKET_DATA_MAX
- strlen("command=\n"));
710 err
= packet_write_fmt_gently(process
->in
, "command=%s\n", filter_type
);
714 err
= strlen(path
) > LARGE_PACKET_DATA_MAX
- strlen("pathname=\n");
716 error("path name too long for external filter");
720 err
= packet_write_fmt_gently(process
->in
, "pathname=%s\n", path
);
724 err
= packet_flush_gently(process
->in
);
729 err
= write_packetized_from_fd(fd
, process
->in
);
731 err
= write_packetized_from_buf(src
, len
, process
->in
);
735 read_multi_file_filter_status(process
->out
, &filter_status
);
736 err
= strcmp(filter_status
.buf
, "success");
740 err
= read_packetized_to_strbuf(process
->out
, &nbuf
) < 0;
744 read_multi_file_filter_status(process
->out
, &filter_status
);
745 err
= strcmp(filter_status
.buf
, "success");
748 sigchain_pop(SIGPIPE
);
750 if (err
|| errno
== EPIPE
) {
751 if (!strcmp(filter_status
.buf
, "error")) {
752 /* The filter signaled a problem with the file. */
753 } else if (!strcmp(filter_status
.buf
, "abort")) {
755 * The filter signaled a permanent problem. Don't try to filter
756 * files with the same command for the lifetime of the current
759 entry
->supported_capabilities
&= ~wanted_capability
;
762 * Something went wrong with the protocol filter.
763 * Force shutdown and restart if another blob requires filtering.
765 error("external filter '%s' failed", cmd
);
766 kill_multi_file_filter(&cmd_process_map
, entry
);
769 strbuf_swap(dst
, &nbuf
);
771 strbuf_release(&nbuf
);
775 static struct convert_driver
{
777 struct convert_driver
*next
;
782 } *user_convert
, **user_convert_tail
;
784 static int apply_filter(const char *path
, const char *src
, size_t len
,
785 int fd
, struct strbuf
*dst
, struct convert_driver
*drv
,
786 const unsigned int wanted_capability
)
788 const char *cmd
= NULL
;
796 if ((CAP_CLEAN
& wanted_capability
) && !drv
->process
&& drv
->clean
)
798 else if ((CAP_SMUDGE
& wanted_capability
) && !drv
->process
&& drv
->smudge
)
802 return apply_single_file_filter(path
, src
, len
, fd
, dst
, cmd
);
803 else if (drv
->process
&& *drv
->process
)
804 return apply_multi_file_filter(path
, src
, len
, fd
, dst
, drv
->process
, wanted_capability
);
809 static int read_convert_config(const char *var
, const char *value
, void *cb
)
811 const char *key
, *name
;
813 struct convert_driver
*drv
;
816 * External conversion drivers are configured using
817 * "filter.<name>.variable".
819 if (parse_config_key(var
, "filter", &name
, &namelen
, &key
) < 0 || !name
)
821 for (drv
= user_convert
; drv
; drv
= drv
->next
)
822 if (!strncmp(drv
->name
, name
, namelen
) && !drv
->name
[namelen
])
825 drv
= xcalloc(1, sizeof(struct convert_driver
));
826 drv
->name
= xmemdupz(name
, namelen
);
827 *user_convert_tail
= drv
;
828 user_convert_tail
= &(drv
->next
);
832 * filter.<name>.smudge and filter.<name>.clean specifies
837 * The command-line will not be interpolated in any way.
840 if (!strcmp("smudge", key
))
841 return git_config_string(&drv
->smudge
, var
, value
);
843 if (!strcmp("clean", key
))
844 return git_config_string(&drv
->clean
, var
, value
);
846 if (!strcmp("process", key
))
847 return git_config_string(&drv
->process
, var
, value
);
849 if (!strcmp("required", key
)) {
850 drv
->required
= git_config_bool(var
, value
);
857 static int count_ident(const char *cp
, unsigned long size
)
860 * "$Id: 0000000000000000000000000000000000000000 $" <=> "$Id$"
872 if (memcmp("Id", cp
, 2))
883 * "$Id: ... "; scan up to the closing dollar sign and discard.
899 static int ident_to_git(const char *path
, const char *src
, size_t len
,
900 struct strbuf
*buf
, int ident
)
904 if (!ident
|| (src
&& !count_ident(src
, len
)))
910 /* only grow if not in place */
911 if (strbuf_avail(buf
) + buf
->len
< len
)
912 strbuf_grow(buf
, len
- buf
->len
);
915 dollar
= memchr(src
, '$', len
);
918 memmove(dst
, src
, dollar
+ 1 - src
);
919 dst
+= dollar
+ 1 - src
;
920 len
-= dollar
+ 1 - src
;
923 if (len
> 3 && !memcmp(src
, "Id:", 3)) {
924 dollar
= memchr(src
+ 3, '$', len
- 3);
927 if (memchr(src
+ 3, '\n', dollar
- src
- 3)) {
928 /* Line break before the next dollar. */
932 memcpy(dst
, "Id$", 3);
934 len
-= dollar
+ 1 - src
;
938 memmove(dst
, src
, len
);
939 strbuf_setlen(buf
, dst
+ len
- buf
->buf
);
943 static int ident_to_worktree(const char *path
, const char *src
, size_t len
,
944 struct strbuf
*buf
, int ident
)
946 unsigned char sha1
[20];
947 char *to_free
= NULL
, *dollar
, *spc
;
953 cnt
= count_ident(src
, len
);
957 /* are we "faking" in place editing ? */
959 to_free
= strbuf_detach(buf
, NULL
);
960 hash_sha1_file(src
, len
, "blob", sha1
);
962 strbuf_grow(buf
, len
+ cnt
* 43);
964 /* step 1: run to the next '$' */
965 dollar
= memchr(src
, '$', len
);
968 strbuf_add(buf
, src
, dollar
+ 1 - src
);
969 len
-= dollar
+ 1 - src
;
972 /* step 2: does it looks like a bit like Id:xxx$ or Id$ ? */
973 if (len
< 3 || memcmp("Id", src
, 2))
976 /* step 3: skip over Id$ or Id:xxxxx$ */
980 } else if (src
[2] == ':') {
982 * It's possible that an expanded Id has crept its way into the
983 * repository, we cope with that by stripping the expansion out.
984 * This is probably not a good idea, since it will cause changes
985 * on checkout, which won't go away by stash, but let's keep it
988 dollar
= memchr(src
+ 3, '$', len
- 3);
990 /* incomplete keyword, no more '$', so just quit the loop */
994 if (memchr(src
+ 3, '\n', dollar
- src
- 3)) {
995 /* Line break before the next dollar. */
999 spc
= memchr(src
+ 4, ' ', dollar
- src
- 4);
1000 if (spc
&& spc
< dollar
-1) {
1001 /* There are spaces in unexpected places.
1002 * This is probably an id from some other
1003 * versioning system. Keep it for now.
1008 len
-= dollar
+ 1 - src
;
1011 /* it wasn't a "Id$" or "Id:xxxx$" */
1015 /* step 4: substitute */
1016 strbuf_addstr(buf
, "Id: ");
1017 strbuf_add(buf
, sha1_to_hex(sha1
), 40);
1018 strbuf_addstr(buf
, " $");
1020 strbuf_add(buf
, src
, len
);
1026 static enum crlf_action
git_path_check_crlf(struct git_attr_check
*check
)
1028 const char *value
= check
->value
;
1030 if (ATTR_TRUE(value
))
1032 else if (ATTR_FALSE(value
))
1034 else if (ATTR_UNSET(value
))
1036 else if (!strcmp(value
, "input"))
1037 return CRLF_TEXT_INPUT
;
1038 else if (!strcmp(value
, "auto"))
1040 return CRLF_UNDEFINED
;
1043 static enum eol
git_path_check_eol(struct git_attr_check
*check
)
1045 const char *value
= check
->value
;
1047 if (ATTR_UNSET(value
))
1049 else if (!strcmp(value
, "lf"))
1051 else if (!strcmp(value
, "crlf"))
1056 static struct convert_driver
*git_path_check_convert(struct git_attr_check
*check
)
1058 const char *value
= check
->value
;
1059 struct convert_driver
*drv
;
1061 if (ATTR_TRUE(value
) || ATTR_FALSE(value
) || ATTR_UNSET(value
))
1063 for (drv
= user_convert
; drv
; drv
= drv
->next
)
1064 if (!strcmp(value
, drv
->name
))
1069 static int git_path_check_ident(struct git_attr_check
*check
)
1071 const char *value
= check
->value
;
1073 return !!ATTR_TRUE(value
);
1077 struct convert_driver
*drv
;
1078 enum crlf_action attr_action
; /* What attr says */
1079 enum crlf_action crlf_action
; /* When no attr is set, use core.autocrlf */
1083 static const char *conv_attr_name
[] = {
1084 "crlf", "ident", "filter", "eol", "text",
1086 #define NUM_CONV_ATTRS ARRAY_SIZE(conv_attr_name)
1088 static void convert_attrs(struct conv_attrs
*ca
, const char *path
)
1091 static struct git_attr_check ccheck
[NUM_CONV_ATTRS
];
1093 if (!ccheck
[0].attr
) {
1094 for (i
= 0; i
< NUM_CONV_ATTRS
; i
++)
1095 ccheck
[i
].attr
= git_attr(conv_attr_name
[i
]);
1096 user_convert_tail
= &user_convert
;
1097 git_config(read_convert_config
, NULL
);
1100 if (!git_check_attr(path
, NUM_CONV_ATTRS
, ccheck
)) {
1101 ca
->crlf_action
= git_path_check_crlf(ccheck
+ 4);
1102 if (ca
->crlf_action
== CRLF_UNDEFINED
)
1103 ca
->crlf_action
= git_path_check_crlf(ccheck
+ 0);
1104 ca
->attr_action
= ca
->crlf_action
;
1105 ca
->ident
= git_path_check_ident(ccheck
+ 1);
1106 ca
->drv
= git_path_check_convert(ccheck
+ 2);
1107 if (ca
->crlf_action
!= CRLF_BINARY
) {
1108 enum eol eol_attr
= git_path_check_eol(ccheck
+ 3);
1109 if (ca
->crlf_action
== CRLF_AUTO
&& eol_attr
== EOL_LF
)
1110 ca
->crlf_action
= CRLF_AUTO_INPUT
;
1111 else if (ca
->crlf_action
== CRLF_AUTO
&& eol_attr
== EOL_CRLF
)
1112 ca
->crlf_action
= CRLF_AUTO_CRLF
;
1113 else if (eol_attr
== EOL_LF
)
1114 ca
->crlf_action
= CRLF_TEXT_INPUT
;
1115 else if (eol_attr
== EOL_CRLF
)
1116 ca
->crlf_action
= CRLF_TEXT_CRLF
;
1118 ca
->attr_action
= ca
->crlf_action
;
1121 ca
->crlf_action
= CRLF_UNDEFINED
;
1124 if (ca
->crlf_action
== CRLF_TEXT
)
1125 ca
->crlf_action
= text_eol_is_crlf() ? CRLF_TEXT_CRLF
: CRLF_TEXT_INPUT
;
1126 if (ca
->crlf_action
== CRLF_UNDEFINED
&& auto_crlf
== AUTO_CRLF_FALSE
)
1127 ca
->crlf_action
= CRLF_BINARY
;
1128 if (ca
->crlf_action
== CRLF_UNDEFINED
&& auto_crlf
== AUTO_CRLF_TRUE
)
1129 ca
->crlf_action
= CRLF_AUTO_CRLF
;
1130 if (ca
->crlf_action
== CRLF_UNDEFINED
&& auto_crlf
== AUTO_CRLF_INPUT
)
1131 ca
->crlf_action
= CRLF_AUTO_INPUT
;
1134 int would_convert_to_git_filter_fd(const char *path
)
1136 struct conv_attrs ca
;
1138 convert_attrs(&ca
, path
);
1143 * Apply a filter to an fd only if the filter is required to succeed.
1144 * We must die if the filter fails, because the original data before
1145 * filtering is not available.
1147 if (!ca
.drv
->required
)
1150 return apply_filter(path
, NULL
, 0, -1, NULL
, ca
.drv
, CAP_CLEAN
);
1153 const char *get_convert_attr_ascii(const char *path
)
1155 struct conv_attrs ca
;
1157 convert_attrs(&ca
, path
);
1158 switch (ca
.attr_action
) {
1159 case CRLF_UNDEFINED
:
1165 case CRLF_TEXT_INPUT
:
1166 return "text eol=lf";
1167 case CRLF_TEXT_CRLF
:
1168 return "text eol=crlf";
1171 case CRLF_AUTO_CRLF
:
1172 return "text=auto eol=crlf";
1173 case CRLF_AUTO_INPUT
:
1174 return "text=auto eol=lf";
1179 int convert_to_git(const char *path
, const char *src
, size_t len
,
1180 struct strbuf
*dst
, enum safe_crlf checksafe
)
1183 struct conv_attrs ca
;
1185 convert_attrs(&ca
, path
);
1187 ret
|= apply_filter(path
, src
, len
, -1, dst
, ca
.drv
, CAP_CLEAN
);
1188 if (!ret
&& ca
.drv
&& ca
.drv
->required
)
1189 die("%s: clean filter '%s' failed", path
, ca
.drv
->name
);
1195 ret
|= crlf_to_git(path
, src
, len
, dst
, ca
.crlf_action
, checksafe
);
1200 return ret
| ident_to_git(path
, src
, len
, dst
, ca
.ident
);
1203 void convert_to_git_filter_fd(const char *path
, int fd
, struct strbuf
*dst
,
1204 enum safe_crlf checksafe
)
1206 struct conv_attrs ca
;
1207 convert_attrs(&ca
, path
);
1210 assert(ca
.drv
->clean
|| ca
.drv
->process
);
1212 if (!apply_filter(path
, NULL
, 0, fd
, dst
, ca
.drv
, CAP_CLEAN
))
1213 die("%s: clean filter '%s' failed", path
, ca
.drv
->name
);
1215 crlf_to_git(path
, dst
->buf
, dst
->len
, dst
, ca
.crlf_action
, checksafe
);
1216 ident_to_git(path
, dst
->buf
, dst
->len
, dst
, ca
.ident
);
1219 static int convert_to_working_tree_internal(const char *path
, const char *src
,
1220 size_t len
, struct strbuf
*dst
,
1223 int ret
= 0, ret_filter
= 0;
1224 struct conv_attrs ca
;
1226 convert_attrs(&ca
, path
);
1228 ret
|= ident_to_worktree(path
, src
, len
, dst
, ca
.ident
);
1234 * CRLF conversion can be skipped if normalizing, unless there
1235 * is a smudge or process filter (even if the process filter doesn't
1236 * support smudge). The filters might expect CRLFs.
1238 if ((ca
.drv
&& (ca
.drv
->smudge
|| ca
.drv
->process
)) || !normalizing
) {
1239 ret
|= crlf_to_worktree(path
, src
, len
, dst
, ca
.crlf_action
);
1246 ret_filter
= apply_filter(path
, src
, len
, -1, dst
, ca
.drv
, CAP_SMUDGE
);
1247 if (!ret_filter
&& ca
.drv
&& ca
.drv
->required
)
1248 die("%s: smudge filter %s failed", path
, ca
.drv
->name
);
1250 return ret
| ret_filter
;
1253 int convert_to_working_tree(const char *path
, const char *src
, size_t len
, struct strbuf
*dst
)
1255 return convert_to_working_tree_internal(path
, src
, len
, dst
, 0);
1258 int renormalize_buffer(const char *path
, const char *src
, size_t len
, struct strbuf
*dst
)
1260 int ret
= convert_to_working_tree_internal(path
, src
, len
, dst
, 1);
1265 return ret
| convert_to_git(path
, src
, len
, dst
, SAFE_CRLF_RENORMALIZE
);
1268 /*****************************************************************
1270 * Streaming conversion support
1272 *****************************************************************/
1274 typedef int (*filter_fn
)(struct stream_filter
*,
1275 const char *input
, size_t *isize_p
,
1276 char *output
, size_t *osize_p
);
1277 typedef void (*free_fn
)(struct stream_filter
*);
1279 struct stream_filter_vtbl
{
1284 struct stream_filter
{
1285 struct stream_filter_vtbl
*vtbl
;
1288 static int null_filter_fn(struct stream_filter
*filter
,
1289 const char *input
, size_t *isize_p
,
1290 char *output
, size_t *osize_p
)
1295 return 0; /* we do not keep any states */
1297 if (*osize_p
< count
)
1300 memmove(output
, input
, count
);
1307 static void null_free_fn(struct stream_filter
*filter
)
1309 ; /* nothing -- null instances are shared */
1312 static struct stream_filter_vtbl null_vtbl
= {
1317 static struct stream_filter null_filter_singleton
= {
1321 int is_null_stream_filter(struct stream_filter
*filter
)
1323 return filter
== &null_filter_singleton
;
1331 struct lf_to_crlf_filter
{
1332 struct stream_filter filter
;
1333 unsigned has_held
:1;
1337 static int lf_to_crlf_filter_fn(struct stream_filter
*filter
,
1338 const char *input
, size_t *isize_p
,
1339 char *output
, size_t *osize_p
)
1341 size_t count
, o
= 0;
1342 struct lf_to_crlf_filter
*lf_to_crlf
= (struct lf_to_crlf_filter
*)filter
;
1345 * We may be holding onto the CR to see if it is followed by a
1346 * LF, in which case we would need to go to the main loop.
1347 * Otherwise, just emit it to the output stream.
1349 if (lf_to_crlf
->has_held
&& (lf_to_crlf
->held
!= '\r' || !input
)) {
1350 output
[o
++] = lf_to_crlf
->held
;
1351 lf_to_crlf
->has_held
= 0;
1354 /* We are told to drain */
1361 if (count
|| lf_to_crlf
->has_held
) {
1365 if (lf_to_crlf
->has_held
) {
1367 lf_to_crlf
->has_held
= 0;
1370 for (i
= 0; o
< *osize_p
&& i
< count
; i
++) {
1375 } else if (was_cr
) {
1377 * Previous round saw CR and it is not followed
1378 * by a LF; emit the CR before processing the
1379 * current character.
1385 * We may have consumed the last output slot,
1386 * in which case we need to break out of this
1387 * loop; hold the current character before
1390 if (*osize_p
<= o
) {
1391 lf_to_crlf
->has_held
= 1;
1392 lf_to_crlf
->held
= ch
;
1393 continue; /* break but increment i */
1408 if (!lf_to_crlf
->has_held
&& was_cr
) {
1409 lf_to_crlf
->has_held
= 1;
1410 lf_to_crlf
->held
= '\r';
1416 static void lf_to_crlf_free_fn(struct stream_filter
*filter
)
1421 static struct stream_filter_vtbl lf_to_crlf_vtbl
= {
1422 lf_to_crlf_filter_fn
,
1426 static struct stream_filter
*lf_to_crlf_filter(void)
1428 struct lf_to_crlf_filter
*lf_to_crlf
= xcalloc(1, sizeof(*lf_to_crlf
));
1430 lf_to_crlf
->filter
.vtbl
= &lf_to_crlf_vtbl
;
1431 return (struct stream_filter
*)lf_to_crlf
;
1437 #define FILTER_BUFFER 1024
1438 struct cascade_filter
{
1439 struct stream_filter filter
;
1440 struct stream_filter
*one
;
1441 struct stream_filter
*two
;
1442 char buf
[FILTER_BUFFER
];
1446 static int cascade_filter_fn(struct stream_filter
*filter
,
1447 const char *input
, size_t *isize_p
,
1448 char *output
, size_t *osize_p
)
1450 struct cascade_filter
*cas
= (struct cascade_filter
*) filter
;
1452 size_t sz
= *osize_p
;
1453 size_t to_feed
, remaining
;
1456 * input -- (one) --> buf -- (two) --> output
1458 while (filled
< sz
) {
1459 remaining
= sz
- filled
;
1461 /* do we already have something to feed two with? */
1462 if (cas
->ptr
< cas
->end
) {
1463 to_feed
= cas
->end
- cas
->ptr
;
1464 if (stream_filter(cas
->two
,
1465 cas
->buf
+ cas
->ptr
, &to_feed
,
1466 output
+ filled
, &remaining
))
1468 cas
->ptr
+= (cas
->end
- cas
->ptr
) - to_feed
;
1469 filled
= sz
- remaining
;
1473 /* feed one from upstream and have it emit into our buffer */
1474 to_feed
= input
? *isize_p
: 0;
1475 if (input
&& !to_feed
)
1477 remaining
= sizeof(cas
->buf
);
1478 if (stream_filter(cas
->one
,
1480 cas
->buf
, &remaining
))
1482 cas
->end
= sizeof(cas
->buf
) - remaining
;
1485 size_t fed
= *isize_p
- to_feed
;
1490 /* do we know that we drained one completely? */
1491 if (input
|| cas
->end
)
1494 /* tell two to drain; we have nothing more to give it */
1496 remaining
= sz
- filled
;
1497 if (stream_filter(cas
->two
,
1499 output
+ filled
, &remaining
))
1501 if (remaining
== (sz
- filled
))
1502 break; /* completely drained two */
1503 filled
= sz
- remaining
;
1509 static void cascade_free_fn(struct stream_filter
*filter
)
1511 struct cascade_filter
*cas
= (struct cascade_filter
*)filter
;
1512 free_stream_filter(cas
->one
);
1513 free_stream_filter(cas
->two
);
1517 static struct stream_filter_vtbl cascade_vtbl
= {
1522 static struct stream_filter
*cascade_filter(struct stream_filter
*one
,
1523 struct stream_filter
*two
)
1525 struct cascade_filter
*cascade
;
1527 if (!one
|| is_null_stream_filter(one
))
1529 if (!two
|| is_null_stream_filter(two
))
1532 cascade
= xmalloc(sizeof(*cascade
));
1535 cascade
->end
= cascade
->ptr
= 0;
1536 cascade
->filter
.vtbl
= &cascade_vtbl
;
1537 return (struct stream_filter
*)cascade
;
1543 #define IDENT_DRAINING (-1)
1544 #define IDENT_SKIPPING (-2)
1545 struct ident_filter
{
1546 struct stream_filter filter
;
1549 char ident
[45]; /* ": x40 $" */
1552 static int is_foreign_ident(const char *str
)
1556 if (!skip_prefix(str
, "$Id: ", &str
))
1558 for (i
= 0; str
[i
]; i
++) {
1559 if (isspace(str
[i
]) && str
[i
+1] != '$')
1565 static void ident_drain(struct ident_filter
*ident
, char **output_p
, size_t *osize_p
)
1567 size_t to_drain
= ident
->left
.len
;
1569 if (*osize_p
< to_drain
)
1570 to_drain
= *osize_p
;
1572 memcpy(*output_p
, ident
->left
.buf
, to_drain
);
1573 strbuf_remove(&ident
->left
, 0, to_drain
);
1574 *output_p
+= to_drain
;
1575 *osize_p
-= to_drain
;
1577 if (!ident
->left
.len
)
1581 static int ident_filter_fn(struct stream_filter
*filter
,
1582 const char *input
, size_t *isize_p
,
1583 char *output
, size_t *osize_p
)
1585 struct ident_filter
*ident
= (struct ident_filter
*)filter
;
1586 static const char head
[] = "$Id";
1589 /* drain upon eof */
1590 switch (ident
->state
) {
1592 strbuf_add(&ident
->left
, head
, ident
->state
);
1593 case IDENT_SKIPPING
:
1595 case IDENT_DRAINING
:
1596 ident_drain(ident
, &output
, osize_p
);
1601 while (*isize_p
|| (ident
->state
== IDENT_DRAINING
)) {
1604 if (ident
->state
== IDENT_DRAINING
) {
1605 ident_drain(ident
, &output
, osize_p
);
1614 if (ident
->state
== IDENT_SKIPPING
) {
1616 * Skipping until '$' or LF, but keeping them
1617 * in case it is a foreign ident.
1619 strbuf_addch(&ident
->left
, ch
);
1620 if (ch
!= '\n' && ch
!= '$')
1622 if (ch
== '$' && !is_foreign_ident(ident
->left
.buf
)) {
1623 strbuf_setlen(&ident
->left
, sizeof(head
) - 1);
1624 strbuf_addstr(&ident
->left
, ident
->ident
);
1626 ident
->state
= IDENT_DRAINING
;
1630 if (ident
->state
< sizeof(head
) &&
1631 head
[ident
->state
] == ch
) {
1637 strbuf_add(&ident
->left
, head
, ident
->state
);
1638 if (ident
->state
== sizeof(head
) - 1) {
1639 if (ch
!= ':' && ch
!= '$') {
1640 strbuf_addch(&ident
->left
, ch
);
1646 strbuf_addch(&ident
->left
, ch
);
1647 ident
->state
= IDENT_SKIPPING
;
1649 strbuf_addstr(&ident
->left
, ident
->ident
);
1650 ident
->state
= IDENT_DRAINING
;
1655 strbuf_addch(&ident
->left
, ch
);
1656 ident
->state
= IDENT_DRAINING
;
1661 static void ident_free_fn(struct stream_filter
*filter
)
1663 struct ident_filter
*ident
= (struct ident_filter
*)filter
;
1664 strbuf_release(&ident
->left
);
1668 static struct stream_filter_vtbl ident_vtbl
= {
1673 static struct stream_filter
*ident_filter(const unsigned char *sha1
)
1675 struct ident_filter
*ident
= xmalloc(sizeof(*ident
));
1677 xsnprintf(ident
->ident
, sizeof(ident
->ident
),
1678 ": %s $", sha1_to_hex(sha1
));
1679 strbuf_init(&ident
->left
, 0);
1680 ident
->filter
.vtbl
= &ident_vtbl
;
1682 return (struct stream_filter
*)ident
;
1686 * Return an appropriately constructed filter for the path, or NULL if
1687 * the contents cannot be filtered without reading the whole thing
1690 * Note that you would be crazy to set CRLF, smuge/clean or ident to a
1691 * large binary blob you would want us not to slurp into the memory!
1693 struct stream_filter
*get_stream_filter(const char *path
, const unsigned char *sha1
)
1695 struct conv_attrs ca
;
1696 struct stream_filter
*filter
= NULL
;
1698 convert_attrs(&ca
, path
);
1699 if (ca
.drv
&& (ca
.drv
->process
|| ca
.drv
->smudge
|| ca
.drv
->clean
))
1702 if (ca
.crlf_action
== CRLF_AUTO
|| ca
.crlf_action
== CRLF_AUTO_CRLF
)
1706 filter
= ident_filter(sha1
);
1708 if (output_eol(ca
.crlf_action
) == EOL_CRLF
)
1709 filter
= cascade_filter(filter
, lf_to_crlf_filter());
1711 filter
= cascade_filter(filter
, &null_filter_singleton
);
1716 void free_stream_filter(struct stream_filter
*filter
)
1718 filter
->vtbl
->free(filter
);
1721 int stream_filter(struct stream_filter
*filter
,
1722 const char *input
, size_t *isize_p
,
1723 char *output
, size_t *osize_p
)
1725 return filter
->vtbl
->filter(filter
, input
, isize_p
, output
, osize_p
);