]> git.cameronkatri.com Git - mandoc.git/blob - read.c
As suggested by jmc@, only include line and column numbers into messages
[mandoc.git] / read.c
1 /* $Id: read.c,v 1.50 2014/06/20 23:02:31 schwarze Exp $ */
2 /*
3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org>
5 * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
6 *
7 * Permission to use, copy, modify, and distribute this software for any
8 * purpose with or without fee is hereby granted, provided that the above
9 * copyright notice and this permission notice appear in all copies.
10 *
11 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
12 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
14 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18 */
19 #ifdef HAVE_CONFIG_H
20 #include "config.h"
21 #endif
22
23 #ifdef HAVE_MMAP
24 # include <sys/stat.h>
25 # include <sys/mman.h>
26 #endif
27
28 #include <assert.h>
29 #include <ctype.h>
30 #include <errno.h>
31 #include <fcntl.h>
32 #include <stdarg.h>
33 #include <stdint.h>
34 #include <stdio.h>
35 #include <stdlib.h>
36 #include <string.h>
37 #include <unistd.h>
38
39 #include "mandoc.h"
40 #include "mandoc_aux.h"
41 #include "libmandoc.h"
42 #include "mdoc.h"
43 #include "man.h"
44 #include "main.h"
45
46 #define REPARSE_LIMIT 1000
47
48 struct buf {
49 char *buf; /* binary input buffer */
50 size_t sz; /* size of binary buffer */
51 };
52
53 struct mparse {
54 enum mandoclevel file_status; /* status of current parse */
55 enum mandoclevel wlevel; /* ignore messages below this */
56 int line; /* line number in the file */
57 int options; /* parser options */
58 struct man *pman; /* persistent man parser */
59 struct mdoc *pmdoc; /* persistent mdoc parser */
60 struct man *man; /* man parser */
61 struct mdoc *mdoc; /* mdoc parser */
62 struct roff *roff; /* roff parser (!NULL) */
63 char *sodest; /* filename pointed to by .so */
64 int reparse_count; /* finite interp. stack */
65 mandocmsg mmsg; /* warning/error message handler */
66 const char *file;
67 struct buf *secondary;
68 char *defos; /* default operating system */
69 };
70
71 static void resize_buf(struct buf *, size_t);
72 static void mparse_buf_r(struct mparse *, struct buf, int);
73 static void pset(const char *, int, struct mparse *);
74 static int read_whole_file(struct mparse *, const char *, int,
75 struct buf *, int *);
76 static void mparse_end(struct mparse *);
77 static void mparse_parse_buffer(struct mparse *, struct buf,
78 const char *);
79
80 static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
81 MANDOCERR_OK,
82 MANDOCERR_WARNING,
83 MANDOCERR_WARNING,
84 MANDOCERR_ERROR,
85 MANDOCERR_FATAL,
86 MANDOCERR_MAX,
87 MANDOCERR_MAX
88 };
89
90 static const char * const mandocerrs[MANDOCERR_MAX] = {
91 "ok",
92
93 "generic warning",
94
95 /* related to the prologue */
96 "no TH macro in document",
97 "document title should be all caps",
98 "unknown manual section",
99 "unknown manual volume or arch",
100 "date missing, using today's date",
101 "cannot parse date, using it verbatim",
102 "prologue macros out of order",
103 "duplicate prologue macro",
104 "macro not allowed in prologue",
105 "macro not allowed in body",
106
107 /* related to document structure */
108 ".so is fragile, better use ln(1)",
109 "no document body",
110 "content before the first section header",
111 "NAME section must come first",
112 "bad NAME section contents",
113 "sections out of conventional order",
114 "duplicate section name",
115 "section header suited to sections 2, 3, and 9 only",
116
117 /* related to macros and nesting */
118 "skipping obsolete macro",
119 "skipping paragraph macro",
120 "moving paragraph macro out of list",
121 "skipping no-space macro",
122 "blocks badly nested",
123 "child violates parent syntax",
124 "nested displays are not portable",
125 "already in literal mode",
126 "line scope broken",
127
128 /* related to missing macro arguments */
129 "skipping empty macro",
130 "argument count wrong",
131 "missing display type",
132 "list type must come first",
133 "tag lists require a width argument",
134 "missing font type",
135 "skipping end of block that is not open",
136
137 /* related to bad macro arguments */
138 "skipping argument",
139 "duplicate argument",
140 "duplicate display type",
141 "duplicate list type",
142 "unknown AT&T UNIX version",
143 "bad Boolean value",
144 "unknown font",
145 "unknown standard specifier",
146 "bad width argument",
147
148 /* related to plain text */
149 "blank line in non-literal context",
150 "tab in non-literal context",
151 "end of line whitespace",
152 "bad comment style",
153 "bad escape sequence",
154 "unterminated quoted string",
155
156 /* related to equations */
157 "unexpected literal in equation",
158
159 "generic error",
160
161 /* related to equations */
162 "unexpected equation scope closure",
163 "equation scope open on exit",
164 "overlapping equation scopes",
165 "unexpected end of equation",
166 "equation syntax error",
167
168 /* related to tables */
169 "bad table syntax",
170 "bad table option",
171 "bad table layout",
172 "no table layout cells specified",
173 "no table data cells specified",
174 "ignore data in cell",
175 "data block still open",
176 "ignoring extra data cells",
177
178 "input stack limit exceeded, infinite loop?",
179 "skipping bad character",
180 "escaped character not allowed in a name",
181 "manual name not yet set",
182 "skipping text before the first section header",
183 "skipping unknown macro",
184 "NOT IMPLEMENTED, please use groff: skipping request",
185 "argument count wrong",
186 "skipping column outside column list",
187 "skipping end of block that is not open",
188 "missing end of block",
189 "scope open on exit",
190 "uname(3) system call failed",
191 "macro requires line argument(s)",
192 "macro requires body argument(s)",
193 "macro requires argument(s)",
194 "request requires a numeric argument",
195 "missing list type",
196 "line argument(s) will be lost",
197 "body argument(s) will be lost",
198
199 "generic fatal error",
200
201 "input too large",
202 "not a manual",
203 "column syntax is inconsistent",
204 "NOT IMPLEMENTED: .Bd -file",
205 "argument count wrong, violates syntax",
206 "child violates parent syntax",
207 "argument count wrong, violates syntax",
208 "NOT IMPLEMENTED: .so with absolute path or \"..\"",
209 "no document prologue",
210 "static buffer exhausted",
211
212 /* system errors */
213 "cannot open file",
214 "cannot stat file",
215 "cannot read file",
216 };
217
218 static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
219 "SUCCESS",
220 "RESERVED",
221 "WARNING",
222 "ERROR",
223 "FATAL",
224 "BADARG",
225 "SYSERR"
226 };
227
228
229 static void
230 resize_buf(struct buf *buf, size_t initial)
231 {
232
233 buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
234 buf->buf = mandoc_realloc(buf->buf, buf->sz);
235 }
236
237 static void
238 pset(const char *buf, int pos, struct mparse *curp)
239 {
240 int i;
241
242 /*
243 * Try to intuit which kind of manual parser should be used. If
244 * passed in by command-line (-man, -mdoc), then use that
245 * explicitly. If passed as -mandoc, then try to guess from the
246 * line: either skip dot-lines, use -mdoc when finding `.Dt', or
247 * default to -man, which is more lenient.
248 *
249 * Separate out pmdoc/pman from mdoc/man: the first persists
250 * through all parsers, while the latter is used per-parse.
251 */
252
253 if ('.' == buf[0] || '\'' == buf[0]) {
254 for (i = 1; buf[i]; i++)
255 if (' ' != buf[i] && '\t' != buf[i])
256 break;
257 if ('\0' == buf[i])
258 return;
259 }
260
261 if (MPARSE_MDOC & curp->options) {
262 if (NULL == curp->pmdoc)
263 curp->pmdoc = mdoc_alloc(
264 curp->roff, curp, curp->defos,
265 MPARSE_QUICK & curp->options ? 1 : 0);
266 assert(curp->pmdoc);
267 curp->mdoc = curp->pmdoc;
268 return;
269 } else if (MPARSE_MAN & curp->options) {
270 if (NULL == curp->pman)
271 curp->pman = man_alloc(curp->roff, curp,
272 MPARSE_QUICK & curp->options ? 1 : 0);
273 assert(curp->pman);
274 curp->man = curp->pman;
275 return;
276 }
277
278 if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
279 if (NULL == curp->pmdoc)
280 curp->pmdoc = mdoc_alloc(
281 curp->roff, curp, curp->defos,
282 MPARSE_QUICK & curp->options ? 1 : 0);
283 assert(curp->pmdoc);
284 curp->mdoc = curp->pmdoc;
285 return;
286 }
287
288 if (NULL == curp->pman)
289 curp->pman = man_alloc(curp->roff, curp,
290 MPARSE_QUICK & curp->options ? 1 : 0);
291 assert(curp->pman);
292 curp->man = curp->pman;
293 }
294
295 /*
296 * Main parse routine for an opened file. This is called for each
297 * opened file and simply loops around the full input file, possibly
298 * nesting (i.e., with `so').
299 */
300 static void
301 mparse_buf_r(struct mparse *curp, struct buf blk, int start)
302 {
303 const struct tbl_span *span;
304 struct buf ln;
305 enum rofferr rr;
306 int i, of, rc;
307 int pos; /* byte number in the ln buffer */
308 int lnn; /* line number in the real file */
309 unsigned char c;
310
311 memset(&ln, 0, sizeof(struct buf));
312
313 lnn = curp->line;
314 pos = 0;
315
316 for (i = 0; i < (int)blk.sz; ) {
317 if (0 == pos && '\0' == blk.buf[i])
318 break;
319
320 if (start) {
321 curp->line = lnn;
322 curp->reparse_count = 0;
323 }
324
325 while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
326
327 /*
328 * When finding an unescaped newline character,
329 * leave the character loop to process the line.
330 * Skip a preceding carriage return, if any.
331 */
332
333 if ('\r' == blk.buf[i] && i + 1 < (int)blk.sz &&
334 '\n' == blk.buf[i + 1])
335 ++i;
336 if ('\n' == blk.buf[i]) {
337 ++i;
338 ++lnn;
339 break;
340 }
341
342 /*
343 * Make sure we have space for at least
344 * one backslash and one other character
345 * and the trailing NUL byte.
346 */
347
348 if (pos + 2 >= (int)ln.sz)
349 resize_buf(&ln, 256);
350
351 /*
352 * Warn about bogus characters. If you're using
353 * non-ASCII encoding, you're screwing your
354 * readers. Since I'd rather this not happen,
355 * I'll be helpful and replace these characters
356 * with "?", so we don't display gibberish.
357 * Note to manual writers: use special characters.
358 */
359
360 c = (unsigned char) blk.buf[i];
361
362 if ( ! (isascii(c) &&
363 (isgraph(c) || isblank(c)))) {
364 mandoc_msg(MANDOCERR_BADCHAR, curp,
365 curp->line, pos, NULL);
366 i++;
367 ln.buf[pos++] = '?';
368 continue;
369 }
370
371 /* Trailing backslash = a plain char. */
372
373 if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
374 ln.buf[pos++] = blk.buf[i++];
375 continue;
376 }
377
378 /*
379 * Found escape and at least one other character.
380 * When it's a newline character, skip it.
381 * When there is a carriage return in between,
382 * skip that one as well.
383 */
384
385 if ('\r' == blk.buf[i + 1] && i + 2 < (int)blk.sz &&
386 '\n' == blk.buf[i + 2])
387 ++i;
388 if ('\n' == blk.buf[i + 1]) {
389 i += 2;
390 ++lnn;
391 continue;
392 }
393
394 if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
395 i += 2;
396 /* Comment, skip to end of line */
397 for (; i < (int)blk.sz; ++i) {
398 if ('\n' == blk.buf[i]) {
399 ++i;
400 ++lnn;
401 break;
402 }
403 }
404
405 /* Backout trailing whitespaces */
406 for (; pos > 0; --pos) {
407 if (ln.buf[pos - 1] != ' ')
408 break;
409 if (pos > 2 && ln.buf[pos - 2] == '\\')
410 break;
411 }
412 break;
413 }
414
415 /* Catch escaped bogus characters. */
416
417 c = (unsigned char) blk.buf[i+1];
418
419 if ( ! (isascii(c) &&
420 (isgraph(c) || isblank(c)))) {
421 mandoc_msg(MANDOCERR_BADCHAR, curp,
422 curp->line, pos, NULL);
423 i += 2;
424 ln.buf[pos++] = '?';
425 continue;
426 }
427
428 /* Some other escape sequence, copy & cont. */
429
430 ln.buf[pos++] = blk.buf[i++];
431 ln.buf[pos++] = blk.buf[i++];
432 }
433
434 if (pos >= (int)ln.sz)
435 resize_buf(&ln, 256);
436
437 ln.buf[pos] = '\0';
438
439 /*
440 * A significant amount of complexity is contained by
441 * the roff preprocessor. It's line-oriented but can be
442 * expressed on one line, so we need at times to
443 * readjust our starting point and re-run it. The roff
444 * preprocessor can also readjust the buffers with new
445 * data, so we pass them in wholesale.
446 */
447
448 of = 0;
449
450 /*
451 * Maintain a lookaside buffer of all parsed lines. We
452 * only do this if mparse_keep() has been invoked (the
453 * buffer may be accessed with mparse_getkeep()).
454 */
455
456 if (curp->secondary) {
457 curp->secondary->buf = mandoc_realloc(
458 curp->secondary->buf,
459 curp->secondary->sz + pos + 2);
460 memcpy(curp->secondary->buf +
461 curp->secondary->sz,
462 ln.buf, pos);
463 curp->secondary->sz += pos;
464 curp->secondary->buf
465 [curp->secondary->sz] = '\n';
466 curp->secondary->sz++;
467 curp->secondary->buf
468 [curp->secondary->sz] = '\0';
469 }
470 rerun:
471 rr = roff_parseln(curp->roff, curp->line,
472 &ln.buf, &ln.sz, of, &of);
473
474 switch (rr) {
475 case ROFF_REPARSE:
476 if (REPARSE_LIMIT >= ++curp->reparse_count)
477 mparse_buf_r(curp, ln, 0);
478 else
479 mandoc_msg(MANDOCERR_ROFFLOOP, curp,
480 curp->line, pos, NULL);
481 pos = 0;
482 continue;
483 case ROFF_APPEND:
484 pos = (int)strlen(ln.buf);
485 continue;
486 case ROFF_RERUN:
487 goto rerun;
488 case ROFF_IGN:
489 pos = 0;
490 continue;
491 case ROFF_ERR:
492 assert(MANDOCLEVEL_FATAL <= curp->file_status);
493 break;
494 case ROFF_SO:
495 if (0 == (MPARSE_SO & curp->options) &&
496 (i >= (int)blk.sz || '\0' == blk.buf[i])) {
497 curp->sodest = mandoc_strdup(ln.buf + of);
498 free(ln.buf);
499 return;
500 }
501 /*
502 * We remove `so' clauses from our lookaside
503 * buffer because we're going to descend into
504 * the file recursively.
505 */
506 if (curp->secondary)
507 curp->secondary->sz -= pos + 1;
508 mparse_readfd(curp, -1, ln.buf + of);
509 if (MANDOCLEVEL_FATAL <= curp->file_status)
510 break;
511 pos = 0;
512 continue;
513 default:
514 break;
515 }
516
517 /*
518 * If we encounter errors in the recursive parse, make
519 * sure we don't continue parsing.
520 */
521
522 if (MANDOCLEVEL_FATAL <= curp->file_status)
523 break;
524
525 /*
526 * If input parsers have not been allocated, do so now.
527 * We keep these instanced between parsers, but set them
528 * locally per parse routine since we can use different
529 * parsers with each one.
530 */
531
532 if ( ! (curp->man || curp->mdoc))
533 pset(ln.buf + of, pos - of, curp);
534
535 /*
536 * Lastly, push down into the parsers themselves. One
537 * of these will have already been set in the pset()
538 * routine.
539 * If libroff returns ROFF_TBL, then add it to the
540 * currently open parse. Since we only get here if
541 * there does exist data (see tbl_data.c), we're
542 * guaranteed that something's been allocated.
543 * Do the same for ROFF_EQN.
544 */
545
546 rc = -1;
547
548 if (ROFF_TBL == rr)
549 while (NULL != (span = roff_span(curp->roff))) {
550 rc = curp->man ?
551 man_addspan(curp->man, span) :
552 mdoc_addspan(curp->mdoc, span);
553 if (0 == rc)
554 break;
555 }
556 else if (ROFF_EQN == rr)
557 rc = curp->mdoc ?
558 mdoc_addeqn(curp->mdoc,
559 roff_eqn(curp->roff)) :
560 man_addeqn(curp->man,
561 roff_eqn(curp->roff));
562 else if (curp->man || curp->mdoc)
563 rc = curp->man ?
564 man_parseln(curp->man,
565 curp->line, ln.buf, of) :
566 mdoc_parseln(curp->mdoc,
567 curp->line, ln.buf, of);
568
569 if (0 == rc) {
570 assert(MANDOCLEVEL_FATAL <= curp->file_status);
571 break;
572 } else if (2 == rc)
573 break;
574
575 /* Temporary buffers typically are not full. */
576
577 if (0 == start && '\0' == blk.buf[i])
578 break;
579
580 /* Start the next input line. */
581
582 pos = 0;
583 }
584
585 free(ln.buf);
586 }
587
588 static int
589 read_whole_file(struct mparse *curp, const char *file, int fd,
590 struct buf *fb, int *with_mmap)
591 {
592 size_t off;
593 ssize_t ssz;
594
595 #ifdef HAVE_MMAP
596 struct stat st;
597 if (-1 == fstat(fd, &st)) {
598 curp->file_status = MANDOCLEVEL_SYSERR;
599 if (curp->mmsg)
600 (*curp->mmsg)(MANDOCERR_SYSSTAT, curp->file_status,
601 file, 0, 0, strerror(errno));
602 return(0);
603 }
604
605 /*
606 * If we're a regular file, try just reading in the whole entry
607 * via mmap(). This is faster than reading it into blocks, and
608 * since each file is only a few bytes to begin with, I'm not
609 * concerned that this is going to tank any machines.
610 */
611
612 if (S_ISREG(st.st_mode)) {
613 if (st.st_size >= (1U << 31)) {
614 curp->file_status = MANDOCLEVEL_FATAL;
615 if (curp->mmsg)
616 (*curp->mmsg)(MANDOCERR_TOOLARGE,
617 curp->file_status, file, 0, 0, NULL);
618 return(0);
619 }
620 *with_mmap = 1;
621 fb->sz = (size_t)st.st_size;
622 fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
623 if (fb->buf != MAP_FAILED)
624 return(1);
625 }
626 #endif
627
628 /*
629 * If this isn't a regular file (like, say, stdin), then we must
630 * go the old way and just read things in bit by bit.
631 */
632
633 *with_mmap = 0;
634 off = 0;
635 fb->sz = 0;
636 fb->buf = NULL;
637 for (;;) {
638 if (off == fb->sz) {
639 if (fb->sz == (1U << 31)) {
640 curp->file_status = MANDOCLEVEL_FATAL;
641 if (curp->mmsg)
642 (*curp->mmsg)(MANDOCERR_TOOLARGE,
643 curp->file_status,
644 file, 0, 0, NULL);
645 break;
646 }
647 resize_buf(fb, 65536);
648 }
649 ssz = read(fd, fb->buf + (int)off, fb->sz - off);
650 if (ssz == 0) {
651 fb->sz = off;
652 return(1);
653 }
654 if (ssz == -1) {
655 curp->file_status = MANDOCLEVEL_SYSERR;
656 if (curp->mmsg)
657 (*curp->mmsg)(MANDOCERR_SYSREAD,
658 curp->file_status, file, 0, 0,
659 strerror(errno));
660 break;
661 }
662 off += (size_t)ssz;
663 }
664
665 free(fb->buf);
666 fb->buf = NULL;
667 return(0);
668 }
669
670 static void
671 mparse_end(struct mparse *curp)
672 {
673
674 if (MANDOCLEVEL_FATAL <= curp->file_status)
675 return;
676
677 if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
678 assert(MANDOCLEVEL_FATAL <= curp->file_status);
679 return;
680 }
681
682 if (curp->man && ! man_endparse(curp->man)) {
683 assert(MANDOCLEVEL_FATAL <= curp->file_status);
684 return;
685 }
686
687 if ( ! (curp->mdoc || curp->man || curp->sodest)) {
688 mandoc_msg(MANDOCERR_NOTMANUAL, curp, 0, 0, NULL);
689 curp->file_status = MANDOCLEVEL_FATAL;
690 return;
691 }
692
693 roff_endparse(curp->roff);
694 }
695
696 static void
697 mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
698 {
699 const char *svfile;
700 static int recursion_depth;
701
702 if (64 < recursion_depth) {
703 mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
704 return;
705 }
706
707 /* Line number is per-file. */
708 svfile = curp->file;
709 curp->file = file;
710 curp->line = 1;
711 recursion_depth++;
712
713 mparse_buf_r(curp, blk, 1);
714
715 if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
716 mparse_end(curp);
717
718 curp->file = svfile;
719 }
720
721 enum mandoclevel
722 mparse_readmem(struct mparse *curp, const void *buf, size_t len,
723 const char *file)
724 {
725 struct buf blk;
726
727 blk.buf = UNCONST(buf);
728 blk.sz = len;
729
730 mparse_parse_buffer(curp, blk, file);
731 return(curp->file_status);
732 }
733
734 enum mandoclevel
735 mparse_readfd(struct mparse *curp, int fd, const char *file)
736 {
737 struct buf blk;
738 int with_mmap;
739
740 if (-1 == fd && -1 == (fd = open(file, O_RDONLY, 0))) {
741 curp->file_status = MANDOCLEVEL_SYSERR;
742 if (curp->mmsg)
743 (*curp->mmsg)(MANDOCERR_SYSOPEN,
744 curp->file_status,
745 file, 0, 0, strerror(errno));
746 goto out;
747 }
748
749 /*
750 * Run for each opened file; may be called more than once for
751 * each full parse sequence if the opened file is nested (i.e.,
752 * from `so'). Simply sucks in the whole file and moves into
753 * the parse phase for the file.
754 */
755
756 if ( ! read_whole_file(curp, file, fd, &blk, &with_mmap))
757 goto out;
758
759 mparse_parse_buffer(curp, blk, file);
760
761 #ifdef HAVE_MMAP
762 if (with_mmap)
763 munmap(blk.buf, blk.sz);
764 else
765 #endif
766 free(blk.buf);
767
768 if (STDIN_FILENO != fd && -1 == close(fd))
769 perror(file);
770 out:
771 return(curp->file_status);
772 }
773
774 struct mparse *
775 mparse_alloc(int options, enum mandoclevel wlevel,
776 mandocmsg mmsg, char *defos)
777 {
778 struct mparse *curp;
779
780 assert(wlevel <= MANDOCLEVEL_FATAL);
781
782 curp = mandoc_calloc(1, sizeof(struct mparse));
783
784 curp->options = options;
785 curp->wlevel = wlevel;
786 curp->mmsg = mmsg;
787 curp->defos = defos;
788
789 curp->roff = roff_alloc(curp, options);
790 return(curp);
791 }
792
793 void
794 mparse_reset(struct mparse *curp)
795 {
796
797 roff_reset(curp->roff);
798
799 if (curp->mdoc)
800 mdoc_reset(curp->mdoc);
801 if (curp->man)
802 man_reset(curp->man);
803 if (curp->secondary)
804 curp->secondary->sz = 0;
805
806 curp->file_status = MANDOCLEVEL_OK;
807 curp->mdoc = NULL;
808 curp->man = NULL;
809
810 free(curp->sodest);
811 curp->sodest = NULL;
812 }
813
814 void
815 mparse_free(struct mparse *curp)
816 {
817
818 if (curp->pmdoc)
819 mdoc_free(curp->pmdoc);
820 if (curp->pman)
821 man_free(curp->pman);
822 if (curp->roff)
823 roff_free(curp->roff);
824 if (curp->secondary)
825 free(curp->secondary->buf);
826
827 free(curp->secondary);
828 free(curp->sodest);
829 free(curp);
830 }
831
832 void
833 mparse_result(struct mparse *curp,
834 struct mdoc **mdoc, struct man **man, char **sodest)
835 {
836
837 if (sodest && NULL != (*sodest = curp->sodest)) {
838 *mdoc = NULL;
839 *man = NULL;
840 return;
841 }
842 if (mdoc)
843 *mdoc = curp->mdoc;
844 if (man)
845 *man = curp->man;
846 }
847
848 void
849 mandoc_vmsg(enum mandocerr t, struct mparse *m,
850 int ln, int pos, const char *fmt, ...)
851 {
852 char buf[256];
853 va_list ap;
854
855 va_start(ap, fmt);
856 (void)vsnprintf(buf, sizeof(buf), fmt, ap);
857 va_end(ap);
858
859 mandoc_msg(t, m, ln, pos, buf);
860 }
861
862 void
863 mandoc_msg(enum mandocerr er, struct mparse *m,
864 int ln, int col, const char *msg)
865 {
866 enum mandoclevel level;
867
868 level = MANDOCLEVEL_FATAL;
869 while (er < mandoclimits[level])
870 level--;
871
872 if (level < m->wlevel)
873 return;
874
875 if (m->mmsg)
876 (*m->mmsg)(er, level, m->file, ln, col, msg);
877
878 if (m->file_status < level)
879 m->file_status = level;
880 }
881
882 const char *
883 mparse_strerror(enum mandocerr er)
884 {
885
886 return(mandocerrs[er]);
887 }
888
889 const char *
890 mparse_strlevel(enum mandoclevel lvl)
891 {
892 return(mandoclevels[lvl]);
893 }
894
895 void
896 mparse_keep(struct mparse *p)
897 {
898
899 assert(NULL == p->secondary);
900 p->secondary = mandoc_calloc(1, sizeof(struct buf));
901 }
902
903 const char *
904 mparse_getkeep(const struct mparse *p)
905 {
906
907 assert(p->secondary);
908 return(p->secondary->sz ? p->secondary->buf : NULL);
909 }