]> git.cameronkatri.com Git - mandoc.git/blob - read.c
Clean up warnings related to macros and nesting.
[mandoc.git] / read.c
1 /* $Id: read.c,v 1.56 2014/07/02 11:43:20 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 "missing .TH macro, using \"unknown 1\"",
97 "lower case character in document title",
98 "unknown manual section",
99 "unknown manual volume or arch",
100 "missing date, using today's date",
101 "cannot parse date, using it verbatim",
102 "prologue macros out of order",
103 "duplicate prologue macro",
104 "incomplete prologue, terminated by",
105 "skipping prologue macro in body",
106
107 /* related to document structure */
108 ".so is fragile, better use ln(1)",
109 "no document body",
110 "content before first section header",
111 "first section is not \"NAME\"",
112 "bad NAME section contents",
113 "sections out of conventional order",
114 "duplicate section title",
115 "unexpected section",
116
117 /* related to macros and nesting */
118 "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 "fill mode already enabled, skipping .fi",
126 "fill mode already disabled, skipping .nf",
127 "line scope broken",
128
129 /* related to missing macro arguments */
130 "skipping empty macro",
131 "argument count wrong",
132 "missing display type",
133 "list type must come first",
134 "tag lists require a width argument",
135 "missing font type",
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 "generic error",
157
158 /* related to equations */
159 "unexpected equation scope closure",
160 "equation scope open on exit",
161 "overlapping equation scopes",
162 "unexpected end of equation",
163 "equation syntax error",
164
165 /* related to tables */
166 "bad table syntax",
167 "bad table option",
168 "bad table layout",
169 "no table layout cells specified",
170 "no table data cells specified",
171 "ignore data in cell",
172 "data block still open",
173 "ignoring extra data cells",
174
175 "input stack limit exceeded, infinite loop?",
176 "skipping bad character",
177 "escaped character not allowed in a name",
178 "manual name not yet set",
179 "skipping text before first section header",
180 "skipping unknown macro",
181 "NOT IMPLEMENTED, please use groff: skipping request",
182 "argument count wrong",
183 "skipping column outside column list",
184 "skipping end of block that is not open",
185 "missing end of block",
186 "scope open on exit",
187 "uname(3) system call failed",
188 "macro requires line argument(s)",
189 "macro requires body argument(s)",
190 "macro requires argument(s)",
191 "request requires a numeric argument",
192 "missing list type",
193 "line argument(s) will be lost",
194
195 "generic fatal error",
196
197 "input too large",
198 "not a manual",
199 "column syntax is inconsistent",
200 "NOT IMPLEMENTED: .Bd -file",
201 "argument count wrong, violates syntax",
202 "child violates parent syntax",
203 "argument count wrong, violates syntax",
204 "NOT IMPLEMENTED: .so with absolute path or \"..\"",
205 ".so request failed",
206 "no document prologue",
207 "static buffer exhausted",
208
209 /* system errors */
210 NULL,
211 "cannot stat file",
212 "cannot read file",
213 };
214
215 static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
216 "SUCCESS",
217 "RESERVED",
218 "WARNING",
219 "ERROR",
220 "FATAL",
221 "BADARG",
222 "SYSERR"
223 };
224
225
226 static void
227 resize_buf(struct buf *buf, size_t initial)
228 {
229
230 buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
231 buf->buf = mandoc_realloc(buf->buf, buf->sz);
232 }
233
234 static void
235 pset(const char *buf, int pos, struct mparse *curp)
236 {
237 int i;
238
239 /*
240 * Try to intuit which kind of manual parser should be used. If
241 * passed in by command-line (-man, -mdoc), then use that
242 * explicitly. If passed as -mandoc, then try to guess from the
243 * line: either skip dot-lines, use -mdoc when finding `.Dt', or
244 * default to -man, which is more lenient.
245 *
246 * Separate out pmdoc/pman from mdoc/man: the first persists
247 * through all parsers, while the latter is used per-parse.
248 */
249
250 if ('.' == buf[0] || '\'' == buf[0]) {
251 for (i = 1; buf[i]; i++)
252 if (' ' != buf[i] && '\t' != buf[i])
253 break;
254 if ('\0' == buf[i])
255 return;
256 }
257
258 if (MPARSE_MDOC & curp->options) {
259 if (NULL == curp->pmdoc)
260 curp->pmdoc = mdoc_alloc(
261 curp->roff, curp, curp->defos,
262 MPARSE_QUICK & curp->options ? 1 : 0);
263 assert(curp->pmdoc);
264 curp->mdoc = curp->pmdoc;
265 return;
266 } else if (MPARSE_MAN & curp->options) {
267 if (NULL == curp->pman)
268 curp->pman = man_alloc(curp->roff, curp,
269 MPARSE_QUICK & curp->options ? 1 : 0);
270 assert(curp->pman);
271 curp->man = curp->pman;
272 return;
273 }
274
275 if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
276 if (NULL == curp->pmdoc)
277 curp->pmdoc = mdoc_alloc(
278 curp->roff, curp, curp->defos,
279 MPARSE_QUICK & curp->options ? 1 : 0);
280 assert(curp->pmdoc);
281 curp->mdoc = curp->pmdoc;
282 return;
283 }
284
285 if (NULL == curp->pman)
286 curp->pman = man_alloc(curp->roff, curp,
287 MPARSE_QUICK & curp->options ? 1 : 0);
288 assert(curp->pman);
289 curp->man = curp->pman;
290 }
291
292 /*
293 * Main parse routine for an opened file. This is called for each
294 * opened file and simply loops around the full input file, possibly
295 * nesting (i.e., with `so').
296 */
297 static void
298 mparse_buf_r(struct mparse *curp, struct buf blk, int start)
299 {
300 const struct tbl_span *span;
301 struct buf ln;
302 enum rofferr rr;
303 int i, of, rc;
304 int pos; /* byte number in the ln buffer */
305 int lnn; /* line number in the real file */
306 unsigned char c;
307
308 memset(&ln, 0, sizeof(struct buf));
309
310 lnn = curp->line;
311 pos = 0;
312
313 for (i = 0; i < (int)blk.sz; ) {
314 if (0 == pos && '\0' == blk.buf[i])
315 break;
316
317 if (start) {
318 curp->line = lnn;
319 curp->reparse_count = 0;
320 }
321
322 while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
323
324 /*
325 * When finding an unescaped newline character,
326 * leave the character loop to process the line.
327 * Skip a preceding carriage return, if any.
328 */
329
330 if ('\r' == blk.buf[i] && i + 1 < (int)blk.sz &&
331 '\n' == blk.buf[i + 1])
332 ++i;
333 if ('\n' == blk.buf[i]) {
334 ++i;
335 ++lnn;
336 break;
337 }
338
339 /*
340 * Make sure we have space for at least
341 * one backslash and one other character
342 * and the trailing NUL byte.
343 */
344
345 if (pos + 2 >= (int)ln.sz)
346 resize_buf(&ln, 256);
347
348 /*
349 * Warn about bogus characters. If you're using
350 * non-ASCII encoding, you're screwing your
351 * readers. Since I'd rather this not happen,
352 * I'll be helpful and replace these characters
353 * with "?", so we don't display gibberish.
354 * Note to manual writers: use special characters.
355 */
356
357 c = (unsigned char) blk.buf[i];
358
359 if ( ! (isascii(c) &&
360 (isgraph(c) || isblank(c)))) {
361 mandoc_msg(MANDOCERR_BADCHAR, curp,
362 curp->line, pos, NULL);
363 i++;
364 ln.buf[pos++] = '?';
365 continue;
366 }
367
368 /* Trailing backslash = a plain char. */
369
370 if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
371 ln.buf[pos++] = blk.buf[i++];
372 continue;
373 }
374
375 /*
376 * Found escape and at least one other character.
377 * When it's a newline character, skip it.
378 * When there is a carriage return in between,
379 * skip that one as well.
380 */
381
382 if ('\r' == blk.buf[i + 1] && i + 2 < (int)blk.sz &&
383 '\n' == blk.buf[i + 2])
384 ++i;
385 if ('\n' == blk.buf[i + 1]) {
386 i += 2;
387 ++lnn;
388 continue;
389 }
390
391 if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
392 i += 2;
393 /* Comment, skip to end of line */
394 for (; i < (int)blk.sz; ++i) {
395 if ('\n' == blk.buf[i]) {
396 ++i;
397 ++lnn;
398 break;
399 }
400 }
401
402 /* Backout trailing whitespaces */
403 for (; pos > 0; --pos) {
404 if (ln.buf[pos - 1] != ' ')
405 break;
406 if (pos > 2 && ln.buf[pos - 2] == '\\')
407 break;
408 }
409 break;
410 }
411
412 /* Catch escaped bogus characters. */
413
414 c = (unsigned char) blk.buf[i+1];
415
416 if ( ! (isascii(c) &&
417 (isgraph(c) || isblank(c)))) {
418 mandoc_msg(MANDOCERR_BADCHAR, curp,
419 curp->line, pos, NULL);
420 i += 2;
421 ln.buf[pos++] = '?';
422 continue;
423 }
424
425 /* Some other escape sequence, copy & cont. */
426
427 ln.buf[pos++] = blk.buf[i++];
428 ln.buf[pos++] = blk.buf[i++];
429 }
430
431 if (pos >= (int)ln.sz)
432 resize_buf(&ln, 256);
433
434 ln.buf[pos] = '\0';
435
436 /*
437 * A significant amount of complexity is contained by
438 * the roff preprocessor. It's line-oriented but can be
439 * expressed on one line, so we need at times to
440 * readjust our starting point and re-run it. The roff
441 * preprocessor can also readjust the buffers with new
442 * data, so we pass them in wholesale.
443 */
444
445 of = 0;
446
447 /*
448 * Maintain a lookaside buffer of all parsed lines. We
449 * only do this if mparse_keep() has been invoked (the
450 * buffer may be accessed with mparse_getkeep()).
451 */
452
453 if (curp->secondary) {
454 curp->secondary->buf = mandoc_realloc(
455 curp->secondary->buf,
456 curp->secondary->sz + pos + 2);
457 memcpy(curp->secondary->buf +
458 curp->secondary->sz,
459 ln.buf, pos);
460 curp->secondary->sz += pos;
461 curp->secondary->buf
462 [curp->secondary->sz] = '\n';
463 curp->secondary->sz++;
464 curp->secondary->buf
465 [curp->secondary->sz] = '\0';
466 }
467 rerun:
468 rr = roff_parseln(curp->roff, curp->line,
469 &ln.buf, &ln.sz, of, &of);
470
471 switch (rr) {
472 case ROFF_REPARSE:
473 if (REPARSE_LIMIT >= ++curp->reparse_count)
474 mparse_buf_r(curp, ln, 0);
475 else
476 mandoc_msg(MANDOCERR_ROFFLOOP, curp,
477 curp->line, pos, NULL);
478 pos = 0;
479 continue;
480 case ROFF_APPEND:
481 pos = (int)strlen(ln.buf);
482 continue;
483 case ROFF_RERUN:
484 goto rerun;
485 case ROFF_IGN:
486 pos = 0;
487 continue;
488 case ROFF_ERR:
489 assert(MANDOCLEVEL_FATAL <= curp->file_status);
490 break;
491 case ROFF_SO:
492 if (0 == (MPARSE_SO & curp->options) &&
493 (i >= (int)blk.sz || '\0' == blk.buf[i])) {
494 curp->sodest = mandoc_strdup(ln.buf + of);
495 free(ln.buf);
496 return;
497 }
498 /*
499 * We remove `so' clauses from our lookaside
500 * buffer because we're going to descend into
501 * the file recursively.
502 */
503 if (curp->secondary)
504 curp->secondary->sz -= pos + 1;
505 mparse_readfd(curp, -1, ln.buf + of);
506 if (MANDOCLEVEL_FATAL <= curp->file_status) {
507 mandoc_vmsg(MANDOCERR_SO_FAIL,
508 curp, curp->line, pos,
509 ".so %s", ln.buf + of);
510 break;
511 }
512 pos = 0;
513 continue;
514 default:
515 break;
516 }
517
518 /*
519 * If we encounter errors in the recursive parse, make
520 * sure we don't continue parsing.
521 */
522
523 if (MANDOCLEVEL_FATAL <= curp->file_status)
524 break;
525
526 /*
527 * If input parsers have not been allocated, do so now.
528 * We keep these instanced between parsers, but set them
529 * locally per parse routine since we can use different
530 * parsers with each one.
531 */
532
533 if ( ! (curp->man || curp->mdoc))
534 pset(ln.buf + of, pos - of, curp);
535
536 /*
537 * Lastly, push down into the parsers themselves. One
538 * of these will have already been set in the pset()
539 * routine.
540 * If libroff returns ROFF_TBL, then add it to the
541 * currently open parse. Since we only get here if
542 * there does exist data (see tbl_data.c), we're
543 * guaranteed that something's been allocated.
544 * Do the same for ROFF_EQN.
545 */
546
547 rc = -1;
548
549 if (ROFF_TBL == rr)
550 while (NULL != (span = roff_span(curp->roff))) {
551 rc = curp->man ?
552 man_addspan(curp->man, span) :
553 mdoc_addspan(curp->mdoc, span);
554 if (0 == rc)
555 break;
556 }
557 else if (ROFF_EQN == rr)
558 rc = curp->mdoc ?
559 mdoc_addeqn(curp->mdoc,
560 roff_eqn(curp->roff)) :
561 man_addeqn(curp->man,
562 roff_eqn(curp->roff));
563 else if (curp->man || curp->mdoc)
564 rc = curp->man ?
565 man_parseln(curp->man,
566 curp->line, ln.buf, of) :
567 mdoc_parseln(curp->mdoc,
568 curp->line, ln.buf, of);
569
570 if (0 == rc) {
571 assert(MANDOCLEVEL_FATAL <= curp->file_status);
572 break;
573 } else if (2 == rc)
574 break;
575
576 /* Temporary buffers typically are not full. */
577
578 if (0 == start && '\0' == blk.buf[i])
579 break;
580
581 /* Start the next input line. */
582
583 pos = 0;
584 }
585
586 free(ln.buf);
587 }
588
589 static int
590 read_whole_file(struct mparse *curp, const char *file, int fd,
591 struct buf *fb, int *with_mmap)
592 {
593 size_t off;
594 ssize_t ssz;
595
596 #ifdef HAVE_MMAP
597 struct stat st;
598 if (-1 == fstat(fd, &st)) {
599 curp->file_status = MANDOCLEVEL_SYSERR;
600 if (curp->mmsg)
601 (*curp->mmsg)(MANDOCERR_SYSSTAT, curp->file_status,
602 file, 0, 0, strerror(errno));
603 return(0);
604 }
605
606 /*
607 * If we're a regular file, try just reading in the whole entry
608 * via mmap(). This is faster than reading it into blocks, and
609 * since each file is only a few bytes to begin with, I'm not
610 * concerned that this is going to tank any machines.
611 */
612
613 if (S_ISREG(st.st_mode)) {
614 if (st.st_size >= (1U << 31)) {
615 curp->file_status = MANDOCLEVEL_FATAL;
616 if (curp->mmsg)
617 (*curp->mmsg)(MANDOCERR_TOOLARGE,
618 curp->file_status, file, 0, 0, NULL);
619 return(0);
620 }
621 *with_mmap = 1;
622 fb->sz = (size_t)st.st_size;
623 fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
624 if (fb->buf != MAP_FAILED)
625 return(1);
626 }
627 #endif
628
629 /*
630 * If this isn't a regular file (like, say, stdin), then we must
631 * go the old way and just read things in bit by bit.
632 */
633
634 *with_mmap = 0;
635 off = 0;
636 fb->sz = 0;
637 fb->buf = NULL;
638 for (;;) {
639 if (off == fb->sz) {
640 if (fb->sz == (1U << 31)) {
641 curp->file_status = MANDOCLEVEL_FATAL;
642 if (curp->mmsg)
643 (*curp->mmsg)(MANDOCERR_TOOLARGE,
644 curp->file_status,
645 file, 0, 0, NULL);
646 break;
647 }
648 resize_buf(fb, 65536);
649 }
650 ssz = read(fd, fb->buf + (int)off, fb->sz - off);
651 if (ssz == 0) {
652 fb->sz = off;
653 return(1);
654 }
655 if (ssz == -1) {
656 curp->file_status = MANDOCLEVEL_SYSERR;
657 if (curp->mmsg)
658 (*curp->mmsg)(MANDOCERR_SYSREAD,
659 curp->file_status, file, 0, 0,
660 strerror(errno));
661 break;
662 }
663 off += (size_t)ssz;
664 }
665
666 free(fb->buf);
667 fb->buf = NULL;
668 return(0);
669 }
670
671 static void
672 mparse_end(struct mparse *curp)
673 {
674
675 if (MANDOCLEVEL_FATAL <= curp->file_status)
676 return;
677
678 if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
679 assert(MANDOCLEVEL_FATAL <= curp->file_status);
680 return;
681 }
682
683 if (curp->man && ! man_endparse(curp->man)) {
684 assert(MANDOCLEVEL_FATAL <= curp->file_status);
685 return;
686 }
687
688 if ( ! (curp->mdoc || curp->man || curp->sodest)) {
689 mandoc_msg(MANDOCERR_NOTMANUAL, curp, 0, 0, NULL);
690 curp->file_status = MANDOCLEVEL_FATAL;
691 return;
692 }
693
694 roff_endparse(curp->roff);
695 }
696
697 static void
698 mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
699 {
700 const char *svfile;
701 static int recursion_depth;
702
703 if (64 < recursion_depth) {
704 mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
705 return;
706 }
707
708 /* Line number is per-file. */
709 svfile = curp->file;
710 curp->file = file;
711 curp->line = 1;
712 recursion_depth++;
713
714 mparse_buf_r(curp, blk, 1);
715
716 if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
717 mparse_end(curp);
718
719 curp->file = svfile;
720 }
721
722 enum mandoclevel
723 mparse_readmem(struct mparse *curp, const void *buf, size_t len,
724 const char *file)
725 {
726 struct buf blk;
727
728 blk.buf = UNCONST(buf);
729 blk.sz = len;
730
731 mparse_parse_buffer(curp, blk, file);
732 return(curp->file_status);
733 }
734
735 enum mandoclevel
736 mparse_readfd(struct mparse *curp, int fd, const char *file)
737 {
738 struct buf blk;
739 int with_mmap;
740
741 if (-1 == fd && -1 == (fd = open(file, O_RDONLY, 0))) {
742 curp->file_status = MANDOCLEVEL_SYSERR;
743 if (curp->mmsg)
744 (*curp->mmsg)(MANDOCERR_SYSOPEN,
745 curp->file_status,
746 file, 0, 0, strerror(errno));
747 goto out;
748 }
749
750 /*
751 * Run for each opened file; may be called more than once for
752 * each full parse sequence if the opened file is nested (i.e.,
753 * from `so'). Simply sucks in the whole file and moves into
754 * the parse phase for the file.
755 */
756
757 if ( ! read_whole_file(curp, file, fd, &blk, &with_mmap))
758 goto out;
759
760 mparse_parse_buffer(curp, blk, file);
761
762 #ifdef HAVE_MMAP
763 if (with_mmap)
764 munmap(blk.buf, blk.sz);
765 else
766 #endif
767 free(blk.buf);
768
769 if (STDIN_FILENO != fd && -1 == close(fd))
770 perror(file);
771 out:
772 return(curp->file_status);
773 }
774
775 struct mparse *
776 mparse_alloc(int options, enum mandoclevel wlevel,
777 mandocmsg mmsg, char *defos)
778 {
779 struct mparse *curp;
780
781 assert(wlevel <= MANDOCLEVEL_FATAL);
782
783 curp = mandoc_calloc(1, sizeof(struct mparse));
784
785 curp->options = options;
786 curp->wlevel = wlevel;
787 curp->mmsg = mmsg;
788 curp->defos = defos;
789
790 curp->roff = roff_alloc(curp, options);
791 return(curp);
792 }
793
794 void
795 mparse_reset(struct mparse *curp)
796 {
797
798 roff_reset(curp->roff);
799
800 if (curp->mdoc)
801 mdoc_reset(curp->mdoc);
802 if (curp->man)
803 man_reset(curp->man);
804 if (curp->secondary)
805 curp->secondary->sz = 0;
806
807 curp->file_status = MANDOCLEVEL_OK;
808 curp->mdoc = NULL;
809 curp->man = NULL;
810
811 free(curp->sodest);
812 curp->sodest = NULL;
813 }
814
815 void
816 mparse_free(struct mparse *curp)
817 {
818
819 if (curp->pmdoc)
820 mdoc_free(curp->pmdoc);
821 if (curp->pman)
822 man_free(curp->pman);
823 if (curp->roff)
824 roff_free(curp->roff);
825 if (curp->secondary)
826 free(curp->secondary->buf);
827
828 free(curp->secondary);
829 free(curp->sodest);
830 free(curp);
831 }
832
833 void
834 mparse_result(struct mparse *curp,
835 struct mdoc **mdoc, struct man **man, char **sodest)
836 {
837
838 if (sodest && NULL != (*sodest = curp->sodest)) {
839 *mdoc = NULL;
840 *man = NULL;
841 return;
842 }
843 if (mdoc)
844 *mdoc = curp->mdoc;
845 if (man)
846 *man = curp->man;
847 }
848
849 void
850 mandoc_vmsg(enum mandocerr t, struct mparse *m,
851 int ln, int pos, const char *fmt, ...)
852 {
853 char buf[256];
854 va_list ap;
855
856 va_start(ap, fmt);
857 (void)vsnprintf(buf, sizeof(buf), fmt, ap);
858 va_end(ap);
859
860 mandoc_msg(t, m, ln, pos, buf);
861 }
862
863 void
864 mandoc_msg(enum mandocerr er, struct mparse *m,
865 int ln, int col, const char *msg)
866 {
867 enum mandoclevel level;
868
869 level = MANDOCLEVEL_FATAL;
870 while (er < mandoclimits[level])
871 level--;
872
873 if (level < m->wlevel)
874 return;
875
876 if (m->mmsg)
877 (*m->mmsg)(er, level, m->file, ln, col, msg);
878
879 if (m->file_status < level)
880 m->file_status = level;
881 }
882
883 const char *
884 mparse_strerror(enum mandocerr er)
885 {
886
887 return(mandocerrs[er]);
888 }
889
890 const char *
891 mparse_strlevel(enum mandoclevel lvl)
892 {
893 return(mandoclevels[lvl]);
894 }
895
896 void
897 mparse_keep(struct mparse *p)
898 {
899
900 assert(NULL == p->secondary);
901 p->secondary = mandoc_calloc(1, sizeof(struct buf));
902 }
903
904 const char *
905 mparse_getkeep(const struct mparse *p)
906 {
907
908 assert(p->secondary);
909 return(p->secondary->sz ? p->secondary->buf : NULL);
910 }