Annotation of mandoc/read.c, Revision 1.37
1.37 ! schwarze 1: /* $Id: read.c,v 1.36 2013/06/01 22:57:35 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.35 schwarze 4: * Copyright (c) 2010, 2011, 2012, 2013 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
9: *
10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17: */
1.11 kristaps 18: #ifdef HAVE_CONFIG_H
19: #include "config.h"
20: #endif
21:
1.15 kristaps 22: #ifdef HAVE_MMAP
23: # include <sys/stat.h>
24: # include <sys/mman.h>
25: #endif
1.1 kristaps 26:
27: #include <assert.h>
28: #include <ctype.h>
29: #include <fcntl.h>
1.3 kristaps 30: #include <stdarg.h>
1.28 joerg 31: #include <stdint.h>
1.1 kristaps 32: #include <stdio.h>
33: #include <stdlib.h>
34: #include <string.h>
35: #include <unistd.h>
36:
37: #include "mandoc.h"
1.3 kristaps 38: #include "libmandoc.h"
1.1 kristaps 39: #include "mdoc.h"
40: #include "man.h"
1.28 joerg 41: #include "main.h"
1.1 kristaps 42:
43: #define REPARSE_LIMIT 1000
44:
45: struct buf {
46: char *buf; /* binary input buffer */
47: size_t sz; /* size of binary buffer */
48: };
49:
50: struct mparse {
51: enum mandoclevel file_status; /* status of current parse */
1.3 kristaps 52: enum mandoclevel wlevel; /* ignore messages below this */
1.1 kristaps 53: int line; /* line number in the file */
54: enum mparset inttype; /* which parser to use */
55: struct man *pman; /* persistent man parser */
56: struct mdoc *pmdoc; /* persistent mdoc parser */
57: struct man *man; /* man parser */
58: struct mdoc *mdoc; /* mdoc parser */
59: struct roff *roff; /* roff parser (!NULL) */
60: int reparse_count; /* finite interp. stack */
61: mandocmsg mmsg; /* warning/error message handler */
62: void *arg; /* argument to mmsg */
1.3 kristaps 63: const char *file;
1.24 kristaps 64: struct buf *secondary;
1.29 schwarze 65: char *defos; /* default operating system */
1.1 kristaps 66: };
67:
68: static void resize_buf(struct buf *, size_t);
69: static void mparse_buf_r(struct mparse *, struct buf, int);
70: static void pset(const char *, int, struct mparse *);
71: static int read_whole_file(const char *, int, struct buf *, int *);
72: static void mparse_end(struct mparse *);
1.37 ! schwarze 73: static void mparse_parse_buffer(struct mparse *, struct buf,
! 74: const char *);
1.1 kristaps 75:
1.3 kristaps 76: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
77: MANDOCERR_OK,
78: MANDOCERR_WARNING,
79: MANDOCERR_WARNING,
80: MANDOCERR_ERROR,
81: MANDOCERR_FATAL,
82: MANDOCERR_MAX,
83: MANDOCERR_MAX
84: };
85:
1.7 kristaps 86: static const char * const mandocerrs[MANDOCERR_MAX] = {
87: "ok",
88:
89: "generic warning",
90:
91: /* related to the prologue */
92: "no title in document",
93: "document title should be all caps",
94: "unknown manual section",
1.32 schwarze 95: "unknown manual volume or arch",
1.7 kristaps 96: "date missing, using today's date",
97: "cannot parse date, using it verbatim",
98: "prologue macros out of order",
99: "duplicate prologue macro",
100: "macro not allowed in prologue",
101: "macro not allowed in body",
102:
103: /* related to document structure */
104: ".so is fragile, better use ln(1)",
105: "NAME section must come first",
106: "bad NAME section contents",
107: "sections out of conventional order",
108: "duplicate section name",
109: "section not in conventional manual section",
110:
111: /* related to macros and nesting */
112: "skipping obsolete macro",
113: "skipping paragraph macro",
1.31 schwarze 114: "moving paragraph macro out of list",
1.7 kristaps 115: "skipping no-space macro",
116: "blocks badly nested",
117: "child violates parent syntax",
118: "nested displays are not portable",
119: "already in literal mode",
120: "line scope broken",
121:
122: /* related to missing macro arguments */
123: "skipping empty macro",
124: "argument count wrong",
125: "missing display type",
126: "list type must come first",
127: "tag lists require a width argument",
128: "missing font type",
129: "skipping end of block that is not open",
130:
131: /* related to bad macro arguments */
132: "skipping argument",
133: "duplicate argument",
134: "duplicate display type",
135: "duplicate list type",
136: "unknown AT&T UNIX version",
137: "bad Boolean value",
138: "unknown font",
139: "unknown standard specifier",
140: "bad width argument",
141:
142: /* related to plain text */
143: "blank line in non-literal context",
144: "tab in non-literal context",
145: "end of line whitespace",
146: "bad comment style",
1.12 kristaps 147: "bad escape sequence",
1.7 kristaps 148: "unterminated quoted string",
1.16 kristaps 149:
150: /* related to equations */
151: "unexpected literal in equation",
1.7 kristaps 152:
153: "generic error",
1.17 kristaps 154:
155: /* related to equations */
1.20 kristaps 156: "unexpected equation scope closure",
157: "equation scope open on exit",
1.21 kristaps 158: "overlapping equation scopes",
159: "unexpected end of equation",
160: "equation syntax error",
1.7 kristaps 161:
162: /* related to tables */
163: "bad table syntax",
164: "bad table option",
165: "bad table layout",
166: "no table layout cells specified",
167: "no table data cells specified",
168: "ignore data in cell",
169: "data block still open",
170: "ignoring extra data cells",
171:
172: "input stack limit exceeded, infinite loop?",
173: "skipping bad character",
174: "escaped character not allowed in a name",
1.30 schwarze 175: "manual name not yet set",
1.7 kristaps 176: "skipping text before the first section header",
177: "skipping unknown macro",
178: "NOT IMPLEMENTED, please use groff: skipping request",
179: "argument count wrong",
1.34 schwarze 180: "skipping column outside column list",
1.7 kristaps 181: "skipping end of block that is not open",
182: "missing end of block",
183: "scope open on exit",
184: "uname(3) system call failed",
185: "macro requires line argument(s)",
186: "macro requires body argument(s)",
187: "macro requires argument(s)",
188: "missing list type",
189: "line argument(s) will be lost",
190: "body argument(s) will be lost",
191:
192: "generic fatal error",
193:
194: "not a manual",
195: "column syntax is inconsistent",
196: "NOT IMPLEMENTED: .Bd -file",
197: "argument count wrong, violates syntax",
198: "child violates parent syntax",
199: "argument count wrong, violates syntax",
200: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
201: "no document body",
202: "no document prologue",
203: "static buffer exhausted",
204: };
205:
206: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
207: "SUCCESS",
208: "RESERVED",
209: "WARNING",
210: "ERROR",
211: "FATAL",
212: "BADARG",
213: "SYSERR"
214: };
215:
1.1 kristaps 216: static void
217: resize_buf(struct buf *buf, size_t initial)
218: {
219:
220: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
221: buf->buf = mandoc_realloc(buf->buf, buf->sz);
222: }
223:
224: static void
225: pset(const char *buf, int pos, struct mparse *curp)
226: {
227: int i;
228:
229: /*
230: * Try to intuit which kind of manual parser should be used. If
231: * passed in by command-line (-man, -mdoc), then use that
232: * explicitly. If passed as -mandoc, then try to guess from the
233: * line: either skip dot-lines, use -mdoc when finding `.Dt', or
234: * default to -man, which is more lenient.
235: *
236: * Separate out pmdoc/pman from mdoc/man: the first persists
237: * through all parsers, while the latter is used per-parse.
238: */
239:
240: if ('.' == buf[0] || '\'' == buf[0]) {
241: for (i = 1; buf[i]; i++)
242: if (' ' != buf[i] && '\t' != buf[i])
243: break;
244: if ('\0' == buf[i])
245: return;
246: }
247:
248: switch (curp->inttype) {
249: case (MPARSE_MDOC):
250: if (NULL == curp->pmdoc)
1.29 schwarze 251: curp->pmdoc = mdoc_alloc(curp->roff, curp,
252: curp->defos);
1.1 kristaps 253: assert(curp->pmdoc);
254: curp->mdoc = curp->pmdoc;
255: return;
256: case (MPARSE_MAN):
257: if (NULL == curp->pman)
1.18 kristaps 258: curp->pman = man_alloc(curp->roff, curp);
1.1 kristaps 259: assert(curp->pman);
260: curp->man = curp->pman;
261: return;
262: default:
263: break;
264: }
265:
266: if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
267: if (NULL == curp->pmdoc)
1.29 schwarze 268: curp->pmdoc = mdoc_alloc(curp->roff, curp,
269: curp->defos);
1.1 kristaps 270: assert(curp->pmdoc);
271: curp->mdoc = curp->pmdoc;
272: return;
273: }
274:
275: if (NULL == curp->pman)
1.18 kristaps 276: curp->pman = man_alloc(curp->roff, curp);
1.1 kristaps 277: assert(curp->pman);
278: curp->man = curp->pman;
279: }
280:
281: /*
282: * Main parse routine for an opened file. This is called for each
283: * opened file and simply loops around the full input file, possibly
284: * nesting (i.e., with `so').
285: */
286: static void
287: mparse_buf_r(struct mparse *curp, struct buf blk, int start)
288: {
289: const struct tbl_span *span;
290: struct buf ln;
291: enum rofferr rr;
292: int i, of, rc;
293: int pos; /* byte number in the ln buffer */
294: int lnn; /* line number in the real file */
295: unsigned char c;
296:
297: memset(&ln, 0, sizeof(struct buf));
298:
299: lnn = curp->line;
300: pos = 0;
301:
302: for (i = 0; i < (int)blk.sz; ) {
303: if (0 == pos && '\0' == blk.buf[i])
304: break;
305:
306: if (start) {
307: curp->line = lnn;
308: curp->reparse_count = 0;
309: }
310:
311: while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
312:
313: /*
314: * When finding an unescaped newline character,
315: * leave the character loop to process the line.
316: * Skip a preceding carriage return, if any.
317: */
318:
319: if ('\r' == blk.buf[i] && i + 1 < (int)blk.sz &&
320: '\n' == blk.buf[i + 1])
321: ++i;
322: if ('\n' == blk.buf[i]) {
323: ++i;
324: ++lnn;
325: break;
326: }
327:
1.35 schwarze 328: /*
329: * Make sure we have space for at least
330: * one backslash and one other character
331: * and the trailing NUL byte.
332: */
333:
334: if (pos + 2 >= (int)ln.sz)
335: resize_buf(&ln, 256);
336:
1.1 kristaps 337: /*
338: * Warn about bogus characters. If you're using
339: * non-ASCII encoding, you're screwing your
340: * readers. Since I'd rather this not happen,
1.27 joerg 341: * I'll be helpful and replace these characters
342: * with "?", so we don't display gibberish.
343: * Note to manual writers: use special characters.
1.1 kristaps 344: */
345:
346: c = (unsigned char) blk.buf[i];
347:
348: if ( ! (isascii(c) &&
349: (isgraph(c) || isblank(c)))) {
1.3 kristaps 350: mandoc_msg(MANDOCERR_BADCHAR, curp,
1.27 joerg 351: curp->line, pos, NULL);
1.1 kristaps 352: i++;
1.27 joerg 353: ln.buf[pos++] = '?';
1.1 kristaps 354: continue;
355: }
356:
357: /* Trailing backslash = a plain char. */
358:
359: if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
360: ln.buf[pos++] = blk.buf[i++];
361: continue;
362: }
363:
364: /*
365: * Found escape and at least one other character.
366: * When it's a newline character, skip it.
367: * When there is a carriage return in between,
368: * skip that one as well.
369: */
370:
371: if ('\r' == blk.buf[i + 1] && i + 2 < (int)blk.sz &&
372: '\n' == blk.buf[i + 2])
373: ++i;
374: if ('\n' == blk.buf[i + 1]) {
375: i += 2;
376: ++lnn;
377: continue;
378: }
379:
1.13 kristaps 380: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 381: i += 2;
382: /* Comment, skip to end of line */
383: for (; i < (int)blk.sz; ++i) {
384: if ('\n' == blk.buf[i]) {
385: ++i;
386: ++lnn;
387: break;
388: }
389: }
390:
391: /* Backout trailing whitespaces */
392: for (; pos > 0; --pos) {
393: if (ln.buf[pos - 1] != ' ')
394: break;
395: if (pos > 2 && ln.buf[pos - 2] == '\\')
396: break;
397: }
398: break;
399: }
400:
1.35 schwarze 401: /* Catch escaped bogus characters. */
402:
403: c = (unsigned char) blk.buf[i+1];
404:
405: if ( ! (isascii(c) &&
406: (isgraph(c) || isblank(c)))) {
407: mandoc_msg(MANDOCERR_BADCHAR, curp,
408: curp->line, pos, NULL);
409: i += 2;
410: ln.buf[pos++] = '?';
411: continue;
412: }
413:
1.1 kristaps 414: /* Some other escape sequence, copy & cont. */
415:
416: ln.buf[pos++] = blk.buf[i++];
417: ln.buf[pos++] = blk.buf[i++];
418: }
419:
420: if (pos >= (int)ln.sz)
421: resize_buf(&ln, 256);
422:
423: ln.buf[pos] = '\0';
424:
425: /*
426: * A significant amount of complexity is contained by
427: * the roff preprocessor. It's line-oriented but can be
428: * expressed on one line, so we need at times to
429: * readjust our starting point and re-run it. The roff
430: * preprocessor can also readjust the buffers with new
431: * data, so we pass them in wholesale.
432: */
433:
434: of = 0;
435:
1.24 kristaps 436: /*
437: * Maintain a lookaside buffer of all parsed lines. We
438: * only do this if mparse_keep() has been invoked (the
439: * buffer may be accessed with mparse_getkeep()).
440: */
441:
442: if (curp->secondary) {
443: curp->secondary->buf =
444: mandoc_realloc
445: (curp->secondary->buf,
446: curp->secondary->sz + pos + 2);
447: memcpy(curp->secondary->buf +
448: curp->secondary->sz,
449: ln.buf, pos);
450: curp->secondary->sz += pos;
451: curp->secondary->buf
452: [curp->secondary->sz] = '\n';
453: curp->secondary->sz++;
454: curp->secondary->buf
455: [curp->secondary->sz] = '\0';
456: }
1.1 kristaps 457: rerun:
458: rr = roff_parseln
459: (curp->roff, curp->line,
460: &ln.buf, &ln.sz, of, &of);
461:
462: switch (rr) {
463: case (ROFF_REPARSE):
464: if (REPARSE_LIMIT >= ++curp->reparse_count)
465: mparse_buf_r(curp, ln, 0);
466: else
1.3 kristaps 467: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.1 kristaps 468: curp->line, pos, NULL);
469: pos = 0;
470: continue;
471: case (ROFF_APPEND):
472: pos = (int)strlen(ln.buf);
473: continue;
474: case (ROFF_RERUN):
475: goto rerun;
476: case (ROFF_IGN):
477: pos = 0;
478: continue;
479: case (ROFF_ERR):
480: assert(MANDOCLEVEL_FATAL <= curp->file_status);
481: break;
482: case (ROFF_SO):
1.24 kristaps 483: /*
484: * We remove `so' clauses from our lookaside
485: * buffer because we're going to descend into
486: * the file recursively.
487: */
1.25 kristaps 488: if (curp->secondary)
489: curp->secondary->sz -= pos + 1;
1.36 schwarze 490: mparse_readfd(curp, -1, ln.buf + of);
1.1 kristaps 491: if (MANDOCLEVEL_FATAL <= curp->file_status)
492: break;
493: pos = 0;
494: continue;
495: default:
496: break;
497: }
498:
499: /*
500: * If we encounter errors in the recursive parse, make
501: * sure we don't continue parsing.
502: */
503:
504: if (MANDOCLEVEL_FATAL <= curp->file_status)
505: break;
506:
507: /*
508: * If input parsers have not been allocated, do so now.
1.14 kristaps 509: * We keep these instanced between parsers, but set them
1.1 kristaps 510: * locally per parse routine since we can use different
511: * parsers with each one.
512: */
513:
514: if ( ! (curp->man || curp->mdoc))
515: pset(ln.buf + of, pos - of, curp);
516:
517: /*
518: * Lastly, push down into the parsers themselves. One
519: * of these will have already been set in the pset()
520: * routine.
521: * If libroff returns ROFF_TBL, then add it to the
522: * currently open parse. Since we only get here if
523: * there does exist data (see tbl_data.c), we're
524: * guaranteed that something's been allocated.
525: * Do the same for ROFF_EQN.
526: */
527:
528: rc = -1;
529:
530: if (ROFF_TBL == rr)
531: while (NULL != (span = roff_span(curp->roff))) {
532: rc = curp->man ?
533: man_addspan(curp->man, span) :
534: mdoc_addspan(curp->mdoc, span);
535: if (0 == rc)
536: break;
537: }
538: else if (ROFF_EQN == rr)
539: rc = curp->mdoc ?
540: mdoc_addeqn(curp->mdoc,
541: roff_eqn(curp->roff)) :
542: man_addeqn(curp->man,
543: roff_eqn(curp->roff));
544: else if (curp->man || curp->mdoc)
545: rc = curp->man ?
546: man_parseln(curp->man,
547: curp->line, ln.buf, of) :
548: mdoc_parseln(curp->mdoc,
549: curp->line, ln.buf, of);
550:
551: if (0 == rc) {
552: assert(MANDOCLEVEL_FATAL <= curp->file_status);
553: break;
554: }
555:
556: /* Temporary buffers typically are not full. */
557:
558: if (0 == start && '\0' == blk.buf[i])
559: break;
560:
561: /* Start the next input line. */
562:
563: pos = 0;
564: }
565:
566: free(ln.buf);
567: }
568:
569: static int
570: read_whole_file(const char *file, int fd, struct buf *fb, int *with_mmap)
571: {
572: size_t off;
573: ssize_t ssz;
574:
1.15 kristaps 575: #ifdef HAVE_MMAP
576: struct stat st;
1.1 kristaps 577: if (-1 == fstat(fd, &st)) {
578: perror(file);
579: return(0);
580: }
581:
582: /*
583: * If we're a regular file, try just reading in the whole entry
584: * via mmap(). This is faster than reading it into blocks, and
585: * since each file is only a few bytes to begin with, I'm not
586: * concerned that this is going to tank any machines.
587: */
588:
589: if (S_ISREG(st.st_mode)) {
590: if (st.st_size >= (1U << 31)) {
591: fprintf(stderr, "%s: input too large\n", file);
592: return(0);
593: }
594: *with_mmap = 1;
595: fb->sz = (size_t)st.st_size;
1.37 ! schwarze 596: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 597: if (fb->buf != MAP_FAILED)
598: return(1);
599: }
1.15 kristaps 600: #endif
1.1 kristaps 601:
602: /*
603: * If this isn't a regular file (like, say, stdin), then we must
604: * go the old way and just read things in bit by bit.
605: */
606:
607: *with_mmap = 0;
608: off = 0;
609: fb->sz = 0;
610: fb->buf = NULL;
611: for (;;) {
612: if (off == fb->sz) {
613: if (fb->sz == (1U << 31)) {
614: fprintf(stderr, "%s: input too large\n", file);
615: break;
616: }
617: resize_buf(fb, 65536);
618: }
619: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
620: if (ssz == 0) {
621: fb->sz = off;
622: return(1);
623: }
624: if (ssz == -1) {
625: perror(file);
626: break;
627: }
628: off += (size_t)ssz;
629: }
630:
631: free(fb->buf);
632: fb->buf = NULL;
633: return(0);
634: }
635:
636: static void
637: mparse_end(struct mparse *curp)
638: {
639:
640: if (MANDOCLEVEL_FATAL <= curp->file_status)
641: return;
642:
643: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
644: assert(MANDOCLEVEL_FATAL <= curp->file_status);
645: return;
646: }
647:
648: if (curp->man && ! man_endparse(curp->man)) {
649: assert(MANDOCLEVEL_FATAL <= curp->file_status);
650: return;
651: }
652:
653: if ( ! (curp->man || curp->mdoc)) {
1.6 kristaps 654: mandoc_msg(MANDOCERR_NOTMANUAL, curp, 1, 0, NULL);
1.1 kristaps 655: curp->file_status = MANDOCLEVEL_FATAL;
1.6 kristaps 656: return;
1.1 kristaps 657: }
658:
659: roff_endparse(curp->roff);
660: }
661:
662: static void
1.36 schwarze 663: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 664: {
665: const char *svfile;
1.36 schwarze 666: static int recursion_depth;
667:
668: if (64 < recursion_depth) {
669: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
670: return;
671: }
1.28 joerg 672:
673: /* Line number is per-file. */
674: svfile = curp->file;
675: curp->file = file;
676: curp->line = 1;
1.36 schwarze 677: recursion_depth++;
1.28 joerg 678:
679: mparse_buf_r(curp, blk, 1);
680:
1.36 schwarze 681: if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
1.28 joerg 682: mparse_end(curp);
683:
684: curp->file = svfile;
685: }
686:
687: enum mandoclevel
688: mparse_readmem(struct mparse *curp, const void *buf, size_t len,
689: const char *file)
690: {
691: struct buf blk;
692:
693: blk.buf = UNCONST(buf);
694: blk.sz = len;
695:
1.36 schwarze 696: mparse_parse_buffer(curp, blk, file);
1.28 joerg 697: return(curp->file_status);
698: }
699:
1.36 schwarze 700: enum mandoclevel
701: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 702: {
1.28 joerg 703: struct buf blk;
704: int with_mmap;
1.1 kristaps 705:
706: if (-1 == fd)
707: if (-1 == (fd = open(file, O_RDONLY, 0))) {
708: perror(file);
709: curp->file_status = MANDOCLEVEL_SYSERR;
1.36 schwarze 710: goto out;
1.1 kristaps 711: }
1.28 joerg 712: /*
713: * Run for each opened file; may be called more than once for
714: * each full parse sequence if the opened file is nested (i.e.,
715: * from `so'). Simply sucks in the whole file and moves into
716: * the parse phase for the file.
717: */
1.1 kristaps 718:
1.28 joerg 719: if ( ! read_whole_file(file, fd, &blk, &with_mmap)) {
720: curp->file_status = MANDOCLEVEL_SYSERR;
1.36 schwarze 721: goto out;
1.28 joerg 722: }
1.1 kristaps 723:
1.36 schwarze 724: mparse_parse_buffer(curp, blk, file);
1.1 kristaps 725:
1.28 joerg 726: #ifdef HAVE_MMAP
727: if (with_mmap)
728: munmap(blk.buf, blk.sz);
729: else
730: #endif
731: free(blk.buf);
1.1 kristaps 732:
733: if (STDIN_FILENO != fd && -1 == close(fd))
734: perror(file);
1.36 schwarze 735: out:
1.1 kristaps 736: return(curp->file_status);
737: }
738:
739: struct mparse *
1.29 schwarze 740: mparse_alloc(enum mparset inttype, enum mandoclevel wlevel,
741: mandocmsg mmsg, void *arg, char *defos)
1.1 kristaps 742: {
743: struct mparse *curp;
1.10 kristaps 744:
745: assert(wlevel <= MANDOCLEVEL_FATAL);
1.1 kristaps 746:
747: curp = mandoc_calloc(1, sizeof(struct mparse));
748:
1.3 kristaps 749: curp->wlevel = wlevel;
1.1 kristaps 750: curp->mmsg = mmsg;
751: curp->arg = arg;
752: curp->inttype = inttype;
1.29 schwarze 753: curp->defos = defos;
1.1 kristaps 754:
1.33 schwarze 755: curp->roff = roff_alloc(inttype, curp);
1.1 kristaps 756: return(curp);
757: }
758:
759: void
760: mparse_reset(struct mparse *curp)
761: {
762:
763: roff_reset(curp->roff);
764:
765: if (curp->mdoc)
766: mdoc_reset(curp->mdoc);
767: if (curp->man)
768: man_reset(curp->man);
1.24 kristaps 769: if (curp->secondary)
770: curp->secondary->sz = 0;
1.1 kristaps 771:
772: curp->file_status = MANDOCLEVEL_OK;
773: curp->mdoc = NULL;
774: curp->man = NULL;
775: }
776:
777: void
778: mparse_free(struct mparse *curp)
779: {
780:
781: if (curp->pmdoc)
782: mdoc_free(curp->pmdoc);
783: if (curp->pman)
784: man_free(curp->pman);
785: if (curp->roff)
786: roff_free(curp->roff);
1.24 kristaps 787: if (curp->secondary)
788: free(curp->secondary->buf);
1.1 kristaps 789:
1.24 kristaps 790: free(curp->secondary);
1.1 kristaps 791: free(curp);
792: }
793:
794: void
795: mparse_result(struct mparse *curp, struct mdoc **mdoc, struct man **man)
796: {
797:
1.9 kristaps 798: if (mdoc)
799: *mdoc = curp->mdoc;
800: if (man)
801: *man = curp->man;
1.3 kristaps 802: }
803:
804: void
805: mandoc_vmsg(enum mandocerr t, struct mparse *m,
806: int ln, int pos, const char *fmt, ...)
807: {
808: char buf[256];
809: va_list ap;
810:
811: va_start(ap, fmt);
812: vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
813: va_end(ap);
814:
815: mandoc_msg(t, m, ln, pos, buf);
816: }
817:
818: void
819: mandoc_msg(enum mandocerr er, struct mparse *m,
820: int ln, int col, const char *msg)
821: {
822: enum mandoclevel level;
823:
824: level = MANDOCLEVEL_FATAL;
825: while (er < mandoclimits[level])
826: level--;
827:
828: if (level < m->wlevel)
829: return;
830:
1.8 kristaps 831: if (m->mmsg)
832: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 833:
834: if (m->file_status < level)
835: m->file_status = level;
1.7 kristaps 836: }
837:
838: const char *
839: mparse_strerror(enum mandocerr er)
840: {
841:
842: return(mandocerrs[er]);
843: }
844:
845: const char *
846: mparse_strlevel(enum mandoclevel lvl)
847: {
848: return(mandoclevels[lvl]);
1.24 kristaps 849: }
850:
851: void
852: mparse_keep(struct mparse *p)
853: {
854:
855: assert(NULL == p->secondary);
856: p->secondary = mandoc_calloc(1, sizeof(struct buf));
857: }
858:
859: const char *
860: mparse_getkeep(const struct mparse *p)
861: {
862:
863: assert(p->secondary);
864: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 865: }
CVSweb