]> git.cameronkatri.com Git - mandoc.git/blob - read.c
Start systematic improvements of error reporting.
[mandoc.git] / read.c
1 /* $Id: read.c,v 1.49 2014/06/20 17:24:00 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 "NAME section must come first",
110 "bad NAME section contents",
111 "sections out of conventional order",
112 "duplicate section name",
113 "section header suited to sections 2, 3, and 9 only",
114
115 /* related to macros and nesting */
116 "skipping obsolete macro",
117 "skipping paragraph macro",
118 "moving paragraph macro out of list",
119 "skipping no-space macro",
120 "blocks badly nested",
121 "child violates parent syntax",
122 "nested displays are not portable",
123 "already in literal mode",
124 "line scope broken",
125
126 /* related to missing macro arguments */
127 "skipping empty macro",
128 "argument count wrong",
129 "missing display type",
130 "list type must come first",
131 "tag lists require a width argument",
132 "missing font type",
133 "skipping end of block that is not open",
134
135 /* related to bad macro arguments */
136 "skipping argument",
137 "duplicate argument",
138 "duplicate display type",
139 "duplicate list type",
140 "unknown AT&T UNIX version",
141 "bad Boolean value",
142 "unknown font",
143 "unknown standard specifier",
144 "bad width argument",
145
146 /* related to plain text */
147 "blank line in non-literal context",
148 "tab in non-literal context",
149 "end of line whitespace",
150 "bad comment style",
151 "bad escape sequence",
152 "unterminated quoted string",
153
154 /* related to equations */
155 "unexpected literal in equation",
156
157 "generic error",
158
159 /* related to equations */
160 "unexpected equation scope closure",
161 "equation scope open on exit",
162 "overlapping equation scopes",
163 "unexpected end of equation",
164 "equation syntax error",
165
166 /* related to tables */
167 "bad table syntax",
168 "bad table option",
169 "bad table layout",
170 "no table layout cells specified",
171 "no table data cells specified",
172 "ignore data in cell",
173 "data block still open",
174 "ignoring extra data cells",
175
176 "input stack limit exceeded, infinite loop?",
177 "skipping bad character",
178 "escaped character not allowed in a name",
179 "manual name not yet set",
180 "skipping text before the first section header",
181 "skipping unknown macro",
182 "NOT IMPLEMENTED, please use groff: skipping request",
183 "argument count wrong",
184 "skipping column outside column list",
185 "skipping end of block that is not open",
186 "missing end of block",
187 "scope open on exit",
188 "uname(3) system call failed",
189 "macro requires line argument(s)",
190 "macro requires body argument(s)",
191 "macro requires argument(s)",
192 "request requires a numeric argument",
193 "missing list type",
194 "line argument(s) will be lost",
195 "body argument(s) will be lost",
196
197 "generic fatal error",
198
199 "input too large",
200 "not a manual",
201 "column syntax is inconsistent",
202 "NOT IMPLEMENTED: .Bd -file",
203 "argument count wrong, violates syntax",
204 "child violates parent syntax",
205 "argument count wrong, violates syntax",
206 "NOT IMPLEMENTED: .so with absolute path or \"..\"",
207 "no document body",
208 "no document prologue",
209 "static buffer exhausted",
210
211 /* system errors */
212 "cannot open file",
213 "cannot stat file",
214 "cannot read file",
215 };
216
217 static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
218 "SUCCESS",
219 "RESERVED",
220 "WARNING",
221 "ERROR",
222 "FATAL",
223 "BADARG",
224 "SYSERR"
225 };
226
227
228 static void
229 resize_buf(struct buf *buf, size_t initial)
230 {
231
232 buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
233 buf->buf = mandoc_realloc(buf->buf, buf->sz);
234 }
235
236 static void
237 pset(const char *buf, int pos, struct mparse *curp)
238 {
239 int i;
240
241 /*
242 * Try to intuit which kind of manual parser should be used. If
243 * passed in by command-line (-man, -mdoc), then use that
244 * explicitly. If passed as -mandoc, then try to guess from the
245 * line: either skip dot-lines, use -mdoc when finding `.Dt', or
246 * default to -man, which is more lenient.
247 *
248 * Separate out pmdoc/pman from mdoc/man: the first persists
249 * through all parsers, while the latter is used per-parse.
250 */
251
252 if ('.' == buf[0] || '\'' == buf[0]) {
253 for (i = 1; buf[i]; i++)
254 if (' ' != buf[i] && '\t' != buf[i])
255 break;
256 if ('\0' == buf[i])
257 return;
258 }
259
260 if (MPARSE_MDOC & curp->options) {
261 if (NULL == curp->pmdoc)
262 curp->pmdoc = mdoc_alloc(
263 curp->roff, curp, curp->defos,
264 MPARSE_QUICK & curp->options ? 1 : 0);
265 assert(curp->pmdoc);
266 curp->mdoc = curp->pmdoc;
267 return;
268 } else if (MPARSE_MAN & curp->options) {
269 if (NULL == curp->pman)
270 curp->pman = man_alloc(curp->roff, curp,
271 MPARSE_QUICK & curp->options ? 1 : 0);
272 assert(curp->pman);
273 curp->man = curp->pman;
274 return;
275 }
276
277 if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
278 if (NULL == curp->pmdoc)
279 curp->pmdoc = mdoc_alloc(
280 curp->roff, curp, curp->defos,
281 MPARSE_QUICK & curp->options ? 1 : 0);
282 assert(curp->pmdoc);
283 curp->mdoc = curp->pmdoc;
284 return;
285 }
286
287 if (NULL == curp->pman)
288 curp->pman = man_alloc(curp->roff, curp,
289 MPARSE_QUICK & curp->options ? 1 : 0);
290 assert(curp->pman);
291 curp->man = curp->pman;
292 }
293
294 /*
295 * Main parse routine for an opened file. This is called for each
296 * opened file and simply loops around the full input file, possibly
297 * nesting (i.e., with `so').
298 */
299 static void
300 mparse_buf_r(struct mparse *curp, struct buf blk, int start)
301 {
302 const struct tbl_span *span;
303 struct buf ln;
304 enum rofferr rr;
305 int i, of, rc;
306 int pos; /* byte number in the ln buffer */
307 int lnn; /* line number in the real file */
308 unsigned char c;
309
310 memset(&ln, 0, sizeof(struct buf));
311
312 lnn = curp->line;
313 pos = 0;
314
315 for (i = 0; i < (int)blk.sz; ) {
316 if (0 == pos && '\0' == blk.buf[i])
317 break;
318
319 if (start) {
320 curp->line = lnn;
321 curp->reparse_count = 0;
322 }
323
324 while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
325
326 /*
327 * When finding an unescaped newline character,
328 * leave the character loop to process the line.
329 * Skip a preceding carriage return, if any.
330 */
331
332 if ('\r' == blk.buf[i] && i + 1 < (int)blk.sz &&
333 '\n' == blk.buf[i + 1])
334 ++i;
335 if ('\n' == blk.buf[i]) {
336 ++i;
337 ++lnn;
338 break;
339 }
340
341 /*
342 * Make sure we have space for at least
343 * one backslash and one other character
344 * and the trailing NUL byte.
345 */
346
347 if (pos + 2 >= (int)ln.sz)
348 resize_buf(&ln, 256);
349
350 /*
351 * Warn about bogus characters. If you're using
352 * non-ASCII encoding, you're screwing your
353 * readers. Since I'd rather this not happen,
354 * I'll be helpful and replace these characters
355 * with "?", so we don't display gibberish.
356 * Note to manual writers: use special characters.
357 */
358
359 c = (unsigned char) blk.buf[i];
360
361 if ( ! (isascii(c) &&
362 (isgraph(c) || isblank(c)))) {
363 mandoc_msg(MANDOCERR_BADCHAR, curp,
364 curp->line, pos, NULL);
365 i++;
366 ln.buf[pos++] = '?';
367 continue;
368 }
369
370 /* Trailing backslash = a plain char. */
371
372 if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
373 ln.buf[pos++] = blk.buf[i++];
374 continue;
375 }
376
377 /*
378 * Found escape and at least one other character.
379 * When it's a newline character, skip it.
380 * When there is a carriage return in between,
381 * skip that one as well.
382 */
383
384 if ('\r' == blk.buf[i + 1] && i + 2 < (int)blk.sz &&
385 '\n' == blk.buf[i + 2])
386 ++i;
387 if ('\n' == blk.buf[i + 1]) {
388 i += 2;
389 ++lnn;
390 continue;
391 }
392
393 if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
394 i += 2;
395 /* Comment, skip to end of line */
396 for (; i < (int)blk.sz; ++i) {
397 if ('\n' == blk.buf[i]) {
398 ++i;
399 ++lnn;
400 break;
401 }
402 }
403
404 /* Backout trailing whitespaces */
405 for (; pos > 0; --pos) {
406 if (ln.buf[pos - 1] != ' ')
407 break;
408 if (pos > 2 && ln.buf[pos - 2] == '\\')
409 break;
410 }
411 break;
412 }
413
414 /* Catch escaped bogus characters. */
415
416 c = (unsigned char) blk.buf[i+1];
417
418 if ( ! (isascii(c) &&
419 (isgraph(c) || isblank(c)))) {
420 mandoc_msg(MANDOCERR_BADCHAR, curp,
421 curp->line, pos, NULL);
422 i += 2;
423 ln.buf[pos++] = '?';
424 continue;
425 }
426
427 /* Some other escape sequence, copy & cont. */
428
429 ln.buf[pos++] = blk.buf[i++];
430 ln.buf[pos++] = blk.buf[i++];
431 }
432
433 if (pos >= (int)ln.sz)
434 resize_buf(&ln, 256);
435
436 ln.buf[pos] = '\0';
437
438 /*
439 * A significant amount of complexity is contained by
440 * the roff preprocessor. It's line-oriented but can be
441 * expressed on one line, so we need at times to
442 * readjust our starting point and re-run it. The roff
443 * preprocessor can also readjust the buffers with new
444 * data, so we pass them in wholesale.
445 */
446
447 of = 0;
448
449 /*
450 * Maintain a lookaside buffer of all parsed lines. We
451 * only do this if mparse_keep() has been invoked (the
452 * buffer may be accessed with mparse_getkeep()).
453 */
454
455 if (curp->secondary) {
456 curp->secondary->buf = mandoc_realloc(
457 curp->secondary->buf,
458 curp->secondary->sz + pos + 2);
459 memcpy(curp->secondary->buf +
460 curp->secondary->sz,
461 ln.buf, pos);
462 curp->secondary->sz += pos;
463 curp->secondary->buf
464 [curp->secondary->sz] = '\n';
465 curp->secondary->sz++;
466 curp->secondary->buf
467 [curp->secondary->sz] = '\0';
468 }
469 rerun:
470 rr = roff_parseln(curp->roff, curp->line,
471 &ln.buf, &ln.sz, of, &of);
472
473 switch (rr) {
474 case ROFF_REPARSE:
475 if (REPARSE_LIMIT >= ++curp->reparse_count)
476 mparse_buf_r(curp, ln, 0);
477 else
478 mandoc_msg(MANDOCERR_ROFFLOOP, curp,
479 curp->line, pos, NULL);
480 pos = 0;
481 continue;
482 case ROFF_APPEND:
483 pos = (int)strlen(ln.buf);
484 continue;
485 case ROFF_RERUN:
486 goto rerun;
487 case ROFF_IGN:
488 pos = 0;
489 continue;
490 case ROFF_ERR:
491 assert(MANDOCLEVEL_FATAL <= curp->file_status);
492 break;
493 case ROFF_SO:
494 if (0 == (MPARSE_SO & curp->options) &&
495 (i >= (int)blk.sz || '\0' == blk.buf[i])) {
496 curp->sodest = mandoc_strdup(ln.buf + of);
497 free(ln.buf);
498 return;
499 }
500 /*
501 * We remove `so' clauses from our lookaside
502 * buffer because we're going to descend into
503 * the file recursively.
504 */
505 if (curp->secondary)
506 curp->secondary->sz -= pos + 1;
507 mparse_readfd(curp, -1, ln.buf + of);
508 if (MANDOCLEVEL_FATAL <= curp->file_status)
509 break;
510 pos = 0;
511 continue;
512 default:
513 break;
514 }
515
516 /*
517 * If we encounter errors in the recursive parse, make
518 * sure we don't continue parsing.
519 */
520
521 if (MANDOCLEVEL_FATAL <= curp->file_status)
522 break;
523
524 /*
525 * If input parsers have not been allocated, do so now.
526 * We keep these instanced between parsers, but set them
527 * locally per parse routine since we can use different
528 * parsers with each one.
529 */
530
531 if ( ! (curp->man || curp->mdoc))
532 pset(ln.buf + of, pos - of, curp);
533
534 /*
535 * Lastly, push down into the parsers themselves. One
536 * of these will have already been set in the pset()
537 * routine.
538 * If libroff returns ROFF_TBL, then add it to the
539 * currently open parse. Since we only get here if
540 * there does exist data (see tbl_data.c), we're
541 * guaranteed that something's been allocated.
542 * Do the same for ROFF_EQN.
543 */
544
545 rc = -1;
546
547 if (ROFF_TBL == rr)
548 while (NULL != (span = roff_span(curp->roff))) {
549 rc = curp->man ?
550 man_addspan(curp->man, span) :
551 mdoc_addspan(curp->mdoc, span);
552 if (0 == rc)
553 break;
554 }
555 else if (ROFF_EQN == rr)
556 rc = curp->mdoc ?
557 mdoc_addeqn(curp->mdoc,
558 roff_eqn(curp->roff)) :
559 man_addeqn(curp->man,
560 roff_eqn(curp->roff));
561 else if (curp->man || curp->mdoc)
562 rc = curp->man ?
563 man_parseln(curp->man,
564 curp->line, ln.buf, of) :
565 mdoc_parseln(curp->mdoc,
566 curp->line, ln.buf, of);
567
568 if (0 == rc) {
569 assert(MANDOCLEVEL_FATAL <= curp->file_status);
570 break;
571 } else if (2 == rc)
572 break;
573
574 /* Temporary buffers typically are not full. */
575
576 if (0 == start && '\0' == blk.buf[i])
577 break;
578
579 /* Start the next input line. */
580
581 pos = 0;
582 }
583
584 free(ln.buf);
585 }
586
587 static int
588 read_whole_file(struct mparse *curp, const char *file, int fd,
589 struct buf *fb, int *with_mmap)
590 {
591 size_t off;
592 ssize_t ssz;
593
594 #ifdef HAVE_MMAP
595 struct stat st;
596 if (-1 == fstat(fd, &st)) {
597 curp->file_status = MANDOCLEVEL_SYSERR;
598 if (curp->mmsg)
599 (*curp->mmsg)(MANDOCERR_SYSSTAT, curp->file_status,
600 file, 0, 0, strerror(errno));
601 return(0);
602 }
603
604 /*
605 * If we're a regular file, try just reading in the whole entry
606 * via mmap(). This is faster than reading it into blocks, and
607 * since each file is only a few bytes to begin with, I'm not
608 * concerned that this is going to tank any machines.
609 */
610
611 if (S_ISREG(st.st_mode)) {
612 if (st.st_size >= (1U << 31)) {
613 curp->file_status = MANDOCLEVEL_FATAL;
614 if (curp->mmsg)
615 (*curp->mmsg)(MANDOCERR_TOOLARGE,
616 curp->file_status, file, 0, 0, NULL);
617 return(0);
618 }
619 *with_mmap = 1;
620 fb->sz = (size_t)st.st_size;
621 fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
622 if (fb->buf != MAP_FAILED)
623 return(1);
624 }
625 #endif
626
627 /*
628 * If this isn't a regular file (like, say, stdin), then we must
629 * go the old way and just read things in bit by bit.
630 */
631
632 *with_mmap = 0;
633 off = 0;
634 fb->sz = 0;
635 fb->buf = NULL;
636 for (;;) {
637 if (off == fb->sz) {
638 if (fb->sz == (1U << 31)) {
639 curp->file_status = MANDOCLEVEL_FATAL;
640 if (curp->mmsg)
641 (*curp->mmsg)(MANDOCERR_TOOLARGE,
642 curp->file_status,
643 file, 0, 0, NULL);
644 break;
645 }
646 resize_buf(fb, 65536);
647 }
648 ssz = read(fd, fb->buf + (int)off, fb->sz - off);
649 if (ssz == 0) {
650 fb->sz = off;
651 return(1);
652 }
653 if (ssz == -1) {
654 curp->file_status = MANDOCLEVEL_SYSERR;
655 if (curp->mmsg)
656 (*curp->mmsg)(MANDOCERR_SYSREAD,
657 curp->file_status, file, 0, 0,
658 strerror(errno));
659 break;
660 }
661 off += (size_t)ssz;
662 }
663
664 free(fb->buf);
665 fb->buf = NULL;
666 return(0);
667 }
668
669 static void
670 mparse_end(struct mparse *curp)
671 {
672
673 if (MANDOCLEVEL_FATAL <= curp->file_status)
674 return;
675
676 if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
677 assert(MANDOCLEVEL_FATAL <= curp->file_status);
678 return;
679 }
680
681 if (curp->man && ! man_endparse(curp->man)) {
682 assert(MANDOCLEVEL_FATAL <= curp->file_status);
683 return;
684 }
685
686 if ( ! (curp->mdoc || curp->man || curp->sodest)) {
687 mandoc_msg(MANDOCERR_NOTMANUAL, curp, 1, 0, NULL);
688 curp->file_status = MANDOCLEVEL_FATAL;
689 return;
690 }
691
692 roff_endparse(curp->roff);
693 }
694
695 static void
696 mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
697 {
698 const char *svfile;
699 static int recursion_depth;
700
701 if (64 < recursion_depth) {
702 mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
703 return;
704 }
705
706 /* Line number is per-file. */
707 svfile = curp->file;
708 curp->file = file;
709 curp->line = 1;
710 recursion_depth++;
711
712 mparse_buf_r(curp, blk, 1);
713
714 if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
715 mparse_end(curp);
716
717 curp->file = svfile;
718 }
719
720 enum mandoclevel
721 mparse_readmem(struct mparse *curp, const void *buf, size_t len,
722 const char *file)
723 {
724 struct buf blk;
725
726 blk.buf = UNCONST(buf);
727 blk.sz = len;
728
729 mparse_parse_buffer(curp, blk, file);
730 return(curp->file_status);
731 }
732
733 enum mandoclevel
734 mparse_readfd(struct mparse *curp, int fd, const char *file)
735 {
736 struct buf blk;
737 int with_mmap;
738
739 if (-1 == fd && -1 == (fd = open(file, O_RDONLY, 0))) {
740 curp->file_status = MANDOCLEVEL_SYSERR;
741 if (curp->mmsg)
742 (*curp->mmsg)(MANDOCERR_SYSOPEN,
743 curp->file_status,
744 file, 0, 0, strerror(errno));
745 goto out;
746 }
747
748 /*
749 * Run for each opened file; may be called more than once for
750 * each full parse sequence if the opened file is nested (i.e.,
751 * from `so'). Simply sucks in the whole file and moves into
752 * the parse phase for the file.
753 */
754
755 if ( ! read_whole_file(curp, file, fd, &blk, &with_mmap))
756 goto out;
757
758 mparse_parse_buffer(curp, blk, file);
759
760 #ifdef HAVE_MMAP
761 if (with_mmap)
762 munmap(blk.buf, blk.sz);
763 else
764 #endif
765 free(blk.buf);
766
767 if (STDIN_FILENO != fd && -1 == close(fd))
768 perror(file);
769 out:
770 return(curp->file_status);
771 }
772
773 struct mparse *
774 mparse_alloc(int options, enum mandoclevel wlevel,
775 mandocmsg mmsg, char *defos)
776 {
777 struct mparse *curp;
778
779 assert(wlevel <= MANDOCLEVEL_FATAL);
780
781 curp = mandoc_calloc(1, sizeof(struct mparse));
782
783 curp->options = options;
784 curp->wlevel = wlevel;
785 curp->mmsg = mmsg;
786 curp->defos = defos;
787
788 curp->roff = roff_alloc(curp, options);
789 return(curp);
790 }
791
792 void
793 mparse_reset(struct mparse *curp)
794 {
795
796 roff_reset(curp->roff);
797
798 if (curp->mdoc)
799 mdoc_reset(curp->mdoc);
800 if (curp->man)
801 man_reset(curp->man);
802 if (curp->secondary)
803 curp->secondary->sz = 0;
804
805 curp->file_status = MANDOCLEVEL_OK;
806 curp->mdoc = NULL;
807 curp->man = NULL;
808
809 free(curp->sodest);
810 curp->sodest = NULL;
811 }
812
813 void
814 mparse_free(struct mparse *curp)
815 {
816
817 if (curp->pmdoc)
818 mdoc_free(curp->pmdoc);
819 if (curp->pman)
820 man_free(curp->pman);
821 if (curp->roff)
822 roff_free(curp->roff);
823 if (curp->secondary)
824 free(curp->secondary->buf);
825
826 free(curp->secondary);
827 free(curp->sodest);
828 free(curp);
829 }
830
831 void
832 mparse_result(struct mparse *curp,
833 struct mdoc **mdoc, struct man **man, char **sodest)
834 {
835
836 if (sodest && NULL != (*sodest = curp->sodest)) {
837 *mdoc = NULL;
838 *man = NULL;
839 return;
840 }
841 if (mdoc)
842 *mdoc = curp->mdoc;
843 if (man)
844 *man = curp->man;
845 }
846
847 void
848 mandoc_vmsg(enum mandocerr t, struct mparse *m,
849 int ln, int pos, const char *fmt, ...)
850 {
851 char buf[256];
852 va_list ap;
853
854 va_start(ap, fmt);
855 (void)vsnprintf(buf, sizeof(buf), fmt, ap);
856 va_end(ap);
857
858 mandoc_msg(t, m, ln, pos, buf);
859 }
860
861 void
862 mandoc_msg(enum mandocerr er, struct mparse *m,
863 int ln, int col, const char *msg)
864 {
865 enum mandoclevel level;
866
867 level = MANDOCLEVEL_FATAL;
868 while (er < mandoclimits[level])
869 level--;
870
871 if (level < m->wlevel)
872 return;
873
874 if (m->mmsg)
875 (*m->mmsg)(er, level, m->file, ln, col, msg);
876
877 if (m->file_status < level)
878 m->file_status = level;
879 }
880
881 const char *
882 mparse_strerror(enum mandocerr er)
883 {
884
885 return(mandocerrs[er]);
886 }
887
888 const char *
889 mparse_strlevel(enum mandoclevel lvl)
890 {
891 return(mandoclevels[lvl]);
892 }
893
894 void
895 mparse_keep(struct mparse *p)
896 {
897
898 assert(NULL == p->secondary);
899 p->secondary = mandoc_calloc(1, sizeof(struct buf));
900 }
901
902 const char *
903 mparse_getkeep(const struct mparse *p)
904 {
905
906 assert(p->secondary);
907 return(p->secondary->sz ? p->secondary->buf : NULL);
908 }