![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to read.c CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [cvsweb.bsd.lv] / mandoc|
Return to read.c CVS log | Up to [cvsweb.bsd.lv] / mandoc |
1.36 ! schwarze 1: /* $Id: read.c,v 1.35 2013/05/30 03:52:59 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: #ifndef MAP_FILE
44: #define MAP_FILE 0
45: #endif
46:
47: #define REPARSE_LIMIT 1000
48:
49: struct buf {
50: char *buf; /* binary input buffer */
51: size_t sz; /* size of binary buffer */
52: };
53:
54: struct mparse {
55: enum mandoclevel file_status; /* status of current parse */
1.3 kristaps 56: enum mandoclevel wlevel; /* ignore messages below this */
1.1 kristaps 57: int line; /* line number in the file */
58: enum mparset inttype; /* which parser to use */
59: struct man *pman; /* persistent man parser */
60: struct mdoc *pmdoc; /* persistent mdoc parser */
61: struct man *man; /* man parser */
62: struct mdoc *mdoc; /* mdoc parser */
63: struct roff *roff; /* roff parser (!NULL) */
64: int reparse_count; /* finite interp. stack */
65: mandocmsg mmsg; /* warning/error message handler */
66: void *arg; /* argument to mmsg */
1.3 kristaps 67: const char *file;
1.24 kristaps 68: struct buf *secondary;
1.29 schwarze 69: char *defos; /* default operating system */
1.1 kristaps 70: };
71:
72: static void resize_buf(struct buf *, size_t);
73: static void mparse_buf_r(struct mparse *, struct buf, int);
74: static void pset(const char *, int, struct mparse *);
75: static int read_whole_file(const char *, int, struct buf *, int *);
76: static void mparse_end(struct mparse *);
77:
1.3 kristaps 78: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
79: MANDOCERR_OK,
80: MANDOCERR_WARNING,
81: MANDOCERR_WARNING,
82: MANDOCERR_ERROR,
83: MANDOCERR_FATAL,
84: MANDOCERR_MAX,
85: MANDOCERR_MAX
86: };
87:
1.7 kristaps 88: static const char * const mandocerrs[MANDOCERR_MAX] = {
89: "ok",
90:
91: "generic warning",
92:
93: /* related to the prologue */
94: "no title in document",
95: "document title should be all caps",
96: "unknown manual section",
1.32 schwarze 97: "unknown manual volume or arch",
1.7 kristaps 98: "date missing, using today's date",
99: "cannot parse date, using it verbatim",
100: "prologue macros out of order",
101: "duplicate prologue macro",
102: "macro not allowed in prologue",
103: "macro not allowed in body",
104:
105: /* related to document structure */
106: ".so is fragile, better use ln(1)",
107: "NAME section must come first",
108: "bad NAME section contents",
109: "sections out of conventional order",
110: "duplicate section name",
111: "section not in conventional manual section",
112:
113: /* related to macros and nesting */
114: "skipping obsolete macro",
115: "skipping paragraph macro",
1.31 schwarze 116: "moving paragraph macro out of list",
1.7 kristaps 117: "skipping no-space macro",
118: "blocks badly nested",
119: "child violates parent syntax",
120: "nested displays are not portable",
121: "already in literal mode",
122: "line scope broken",
123:
124: /* related to missing macro arguments */
125: "skipping empty macro",
126: "argument count wrong",
127: "missing display type",
128: "list type must come first",
129: "tag lists require a width argument",
130: "missing font type",
131: "skipping end of block that is not open",
132:
133: /* related to bad macro arguments */
134: "skipping argument",
135: "duplicate argument",
136: "duplicate display type",
137: "duplicate list type",
138: "unknown AT&T UNIX version",
139: "bad Boolean value",
140: "unknown font",
141: "unknown standard specifier",
142: "bad width argument",
143:
144: /* related to plain text */
145: "blank line in non-literal context",
146: "tab in non-literal context",
147: "end of line whitespace",
148: "bad comment style",
1.12 kristaps 149: "bad escape sequence",
1.7 kristaps 150: "unterminated quoted string",
1.16 kristaps 151:
152: /* related to equations */
153: "unexpected literal in equation",
1.7 kristaps 154:
155: "generic error",
1.17 kristaps 156:
157: /* related to equations */
1.20 kristaps 158: "unexpected equation scope closure",
159: "equation scope open on exit",
1.21 kristaps 160: "overlapping equation scopes",
161: "unexpected end of equation",
162: "equation syntax error",
1.7 kristaps 163:
164: /* related to tables */
165: "bad table syntax",
166: "bad table option",
167: "bad table layout",
168: "no table layout cells specified",
169: "no table data cells specified",
170: "ignore data in cell",
171: "data block still open",
172: "ignoring extra data cells",
173:
174: "input stack limit exceeded, infinite loop?",
175: "skipping bad character",
176: "escaped character not allowed in a name",
1.30 schwarze 177: "manual name not yet set",
1.7 kristaps 178: "skipping text before the first section header",
179: "skipping unknown macro",
180: "NOT IMPLEMENTED, please use groff: skipping request",
181: "argument count wrong",
1.34 schwarze 182: "skipping column outside column list",
1.7 kristaps 183: "skipping end of block that is not open",
184: "missing end of block",
185: "scope open on exit",
186: "uname(3) system call failed",
187: "macro requires line argument(s)",
188: "macro requires body argument(s)",
189: "macro requires argument(s)",
190: "missing list type",
191: "line argument(s) will be lost",
192: "body argument(s) will be lost",
193:
194: "generic fatal error",
195:
196: "not a manual",
197: "column syntax is inconsistent",
198: "NOT IMPLEMENTED: .Bd -file",
199: "argument count wrong, violates syntax",
200: "child violates parent syntax",
201: "argument count wrong, violates syntax",
202: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
203: "no document body",
204: "no document prologue",
205: "static buffer exhausted",
206: };
207:
208: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
209: "SUCCESS",
210: "RESERVED",
211: "WARNING",
212: "ERROR",
213: "FATAL",
214: "BADARG",
215: "SYSERR"
216: };
217:
1.1 kristaps 218: static void
219: resize_buf(struct buf *buf, size_t initial)
220: {
221:
222: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
223: buf->buf = mandoc_realloc(buf->buf, buf->sz);
224: }
225:
226: static void
227: pset(const char *buf, int pos, struct mparse *curp)
228: {
229: int i;
230:
231: /*
232: * Try to intuit which kind of manual parser should be used. If
233: * passed in by command-line (-man, -mdoc), then use that
234: * explicitly. If passed as -mandoc, then try to guess from the
235: * line: either skip dot-lines, use -mdoc when finding `.Dt', or
236: * default to -man, which is more lenient.
237: *
238: * Separate out pmdoc/pman from mdoc/man: the first persists
239: * through all parsers, while the latter is used per-parse.
240: */
241:
242: if ('.' == buf[0] || '\'' == buf[0]) {
243: for (i = 1; buf[i]; i++)
244: if (' ' != buf[i] && '\t' != buf[i])
245: break;
246: if ('\0' == buf[i])
247: return;
248: }
249:
250: switch (curp->inttype) {
251: case (MPARSE_MDOC):
252: if (NULL == curp->pmdoc)
1.29 schwarze 253: curp->pmdoc = mdoc_alloc(curp->roff, curp,
254: curp->defos);
1.1 kristaps 255: assert(curp->pmdoc);
256: curp->mdoc = curp->pmdoc;
257: return;
258: case (MPARSE_MAN):
259: if (NULL == curp->pman)
1.18 kristaps 260: curp->pman = man_alloc(curp->roff, curp);
1.1 kristaps 261: assert(curp->pman);
262: curp->man = curp->pman;
263: return;
264: default:
265: break;
266: }
267:
268: if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
269: if (NULL == curp->pmdoc)
1.29 schwarze 270: curp->pmdoc = mdoc_alloc(curp->roff, curp,
271: curp->defos);
1.1 kristaps 272: assert(curp->pmdoc);
273: curp->mdoc = curp->pmdoc;
274: return;
275: }
276:
277: if (NULL == curp->pman)
1.18 kristaps 278: curp->pman = man_alloc(curp->roff, curp);
1.1 kristaps 279: assert(curp->pman);
280: curp->man = curp->pman;
281: }
282:
283: /*
284: * Main parse routine for an opened file. This is called for each
285: * opened file and simply loops around the full input file, possibly
286: * nesting (i.e., with `so').
287: */
288: static void
289: mparse_buf_r(struct mparse *curp, struct buf blk, int start)
290: {
291: const struct tbl_span *span;
292: struct buf ln;
293: enum rofferr rr;
294: int i, of, rc;
295: int pos; /* byte number in the ln buffer */
296: int lnn; /* line number in the real file */
297: unsigned char c;
298:
299: memset(&ln, 0, sizeof(struct buf));
300:
301: lnn = curp->line;
302: pos = 0;
303:
304: for (i = 0; i < (int)blk.sz; ) {
305: if (0 == pos && '\0' == blk.buf[i])
306: break;
307:
308: if (start) {
309: curp->line = lnn;
310: curp->reparse_count = 0;
311: }
312:
313: while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
314:
315: /*
316: * When finding an unescaped newline character,
317: * leave the character loop to process the line.
318: * Skip a preceding carriage return, if any.
319: */
320:
321: if ('\r' == blk.buf[i] && i + 1 < (int)blk.sz &&
322: '\n' == blk.buf[i + 1])
323: ++i;
324: if ('\n' == blk.buf[i]) {
325: ++i;
326: ++lnn;
327: break;
328: }
329:
1.35 schwarze 330: /*
331: * Make sure we have space for at least
332: * one backslash and one other character
333: * and the trailing NUL byte.
334: */
335:
336: if (pos + 2 >= (int)ln.sz)
337: resize_buf(&ln, 256);
338:
1.1 kristaps 339: /*
340: * Warn about bogus characters. If you're using
341: * non-ASCII encoding, you're screwing your
342: * readers. Since I'd rather this not happen,
1.27 joerg 343: * I'll be helpful and replace these characters
344: * with "?", so we don't display gibberish.
345: * Note to manual writers: use special characters.
1.1 kristaps 346: */
347:
348: c = (unsigned char) blk.buf[i];
349:
350: if ( ! (isascii(c) &&
351: (isgraph(c) || isblank(c)))) {
1.3 kristaps 352: mandoc_msg(MANDOCERR_BADCHAR, curp,
1.27 joerg 353: curp->line, pos, NULL);
1.1 kristaps 354: i++;
1.27 joerg 355: ln.buf[pos++] = '?';
1.1 kristaps 356: continue;
357: }
358:
359: /* Trailing backslash = a plain char. */
360:
361: if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
362: ln.buf[pos++] = blk.buf[i++];
363: continue;
364: }
365:
366: /*
367: * Found escape and at least one other character.
368: * When it's a newline character, skip it.
369: * When there is a carriage return in between,
370: * skip that one as well.
371: */
372:
373: if ('\r' == blk.buf[i + 1] && i + 2 < (int)blk.sz &&
374: '\n' == blk.buf[i + 2])
375: ++i;
376: if ('\n' == blk.buf[i + 1]) {
377: i += 2;
378: ++lnn;
379: continue;
380: }
381:
1.13 kristaps 382: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 383: i += 2;
384: /* Comment, skip to end of line */
385: for (; i < (int)blk.sz; ++i) {
386: if ('\n' == blk.buf[i]) {
387: ++i;
388: ++lnn;
389: break;
390: }
391: }
392:
393: /* Backout trailing whitespaces */
394: for (; pos > 0; --pos) {
395: if (ln.buf[pos - 1] != ' ')
396: break;
397: if (pos > 2 && ln.buf[pos - 2] == '\\')
398: break;
399: }
400: break;
401: }
402:
1.35 schwarze 403: /* Catch escaped bogus characters. */
404:
405: c = (unsigned char) blk.buf[i+1];
406:
407: if ( ! (isascii(c) &&
408: (isgraph(c) || isblank(c)))) {
409: mandoc_msg(MANDOCERR_BADCHAR, curp,
410: curp->line, pos, NULL);
411: i += 2;
412: ln.buf[pos++] = '?';
413: continue;
414: }
415:
1.1 kristaps 416: /* Some other escape sequence, copy & cont. */
417:
418: ln.buf[pos++] = blk.buf[i++];
419: ln.buf[pos++] = blk.buf[i++];
420: }
421:
422: if (pos >= (int)ln.sz)
423: resize_buf(&ln, 256);
424:
425: ln.buf[pos] = '\0';
426:
427: /*
428: * A significant amount of complexity is contained by
429: * the roff preprocessor. It's line-oriented but can be
430: * expressed on one line, so we need at times to
431: * readjust our starting point and re-run it. The roff
432: * preprocessor can also readjust the buffers with new
433: * data, so we pass them in wholesale.
434: */
435:
436: of = 0;
437:
1.24 kristaps 438: /*
439: * Maintain a lookaside buffer of all parsed lines. We
440: * only do this if mparse_keep() has been invoked (the
441: * buffer may be accessed with mparse_getkeep()).
442: */
443:
444: if (curp->secondary) {
445: curp->secondary->buf =
446: mandoc_realloc
447: (curp->secondary->buf,
448: curp->secondary->sz + pos + 2);
449: memcpy(curp->secondary->buf +
450: curp->secondary->sz,
451: ln.buf, pos);
452: curp->secondary->sz += pos;
453: curp->secondary->buf
454: [curp->secondary->sz] = '\n';
455: curp->secondary->sz++;
456: curp->secondary->buf
457: [curp->secondary->sz] = '\0';
458: }
1.1 kristaps 459: rerun:
460: rr = roff_parseln
461: (curp->roff, curp->line,
462: &ln.buf, &ln.sz, of, &of);
463:
464: switch (rr) {
465: case (ROFF_REPARSE):
466: if (REPARSE_LIMIT >= ++curp->reparse_count)
467: mparse_buf_r(curp, ln, 0);
468: else
1.3 kristaps 469: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.1 kristaps 470: curp->line, pos, NULL);
471: pos = 0;
472: continue;
473: case (ROFF_APPEND):
474: pos = (int)strlen(ln.buf);
475: continue;
476: case (ROFF_RERUN):
477: goto rerun;
478: case (ROFF_IGN):
479: pos = 0;
480: continue;
481: case (ROFF_ERR):
482: assert(MANDOCLEVEL_FATAL <= curp->file_status);
483: break;
484: case (ROFF_SO):
1.24 kristaps 485: /*
486: * We remove `so' clauses from our lookaside
487: * buffer because we're going to descend into
488: * the file recursively.
489: */
1.25 kristaps 490: if (curp->secondary)
491: curp->secondary->sz -= pos + 1;
1.36 ! schwarze 492: mparse_readfd(curp, -1, ln.buf + of);
1.1 kristaps 493: if (MANDOCLEVEL_FATAL <= curp->file_status)
494: break;
495: pos = 0;
496: continue;
497: default:
498: break;
499: }
500:
501: /*
502: * If we encounter errors in the recursive parse, make
503: * sure we don't continue parsing.
504: */
505:
506: if (MANDOCLEVEL_FATAL <= curp->file_status)
507: break;
508:
509: /*
510: * If input parsers have not been allocated, do so now.
1.14 kristaps 511: * We keep these instanced between parsers, but set them
1.1 kristaps 512: * locally per parse routine since we can use different
513: * parsers with each one.
514: */
515:
516: if ( ! (curp->man || curp->mdoc))
517: pset(ln.buf + of, pos - of, curp);
518:
519: /*
520: * Lastly, push down into the parsers themselves. One
521: * of these will have already been set in the pset()
522: * routine.
523: * If libroff returns ROFF_TBL, then add it to the
524: * currently open parse. Since we only get here if
525: * there does exist data (see tbl_data.c), we're
526: * guaranteed that something's been allocated.
527: * Do the same for ROFF_EQN.
528: */
529:
530: rc = -1;
531:
532: if (ROFF_TBL == rr)
533: while (NULL != (span = roff_span(curp->roff))) {
534: rc = curp->man ?
535: man_addspan(curp->man, span) :
536: mdoc_addspan(curp->mdoc, span);
537: if (0 == rc)
538: break;
539: }
540: else if (ROFF_EQN == rr)
541: rc = curp->mdoc ?
542: mdoc_addeqn(curp->mdoc,
543: roff_eqn(curp->roff)) :
544: man_addeqn(curp->man,
545: roff_eqn(curp->roff));
546: else if (curp->man || curp->mdoc)
547: rc = curp->man ?
548: man_parseln(curp->man,
549: curp->line, ln.buf, of) :
550: mdoc_parseln(curp->mdoc,
551: curp->line, ln.buf, of);
552:
553: if (0 == rc) {
554: assert(MANDOCLEVEL_FATAL <= curp->file_status);
555: break;
556: }
557:
558: /* Temporary buffers typically are not full. */
559:
560: if (0 == start && '\0' == blk.buf[i])
561: break;
562:
563: /* Start the next input line. */
564:
565: pos = 0;
566: }
567:
568: free(ln.buf);
569: }
570:
571: static int
572: read_whole_file(const char *file, int fd, struct buf *fb, int *with_mmap)
573: {
574: size_t off;
575: ssize_t ssz;
576:
1.15 kristaps 577: #ifdef HAVE_MMAP
578: struct stat st;
1.1 kristaps 579: if (-1 == fstat(fd, &st)) {
580: perror(file);
581: return(0);
582: }
583:
584: /*
585: * If we're a regular file, try just reading in the whole entry
586: * via mmap(). This is faster than reading it into blocks, and
587: * since each file is only a few bytes to begin with, I'm not
588: * concerned that this is going to tank any machines.
589: */
590:
591: if (S_ISREG(st.st_mode)) {
592: if (st.st_size >= (1U << 31)) {
593: fprintf(stderr, "%s: input too large\n", file);
594: return(0);
595: }
596: *with_mmap = 1;
597: fb->sz = (size_t)st.st_size;
598: fb->buf = mmap(NULL, fb->sz, PROT_READ,
599: MAP_FILE|MAP_SHARED, fd, 0);
600: if (fb->buf != MAP_FAILED)
601: return(1);
602: }
1.15 kristaps 603: #endif
1.1 kristaps 604:
605: /*
606: * If this isn't a regular file (like, say, stdin), then we must
607: * go the old way and just read things in bit by bit.
608: */
609:
610: *with_mmap = 0;
611: off = 0;
612: fb->sz = 0;
613: fb->buf = NULL;
614: for (;;) {
615: if (off == fb->sz) {
616: if (fb->sz == (1U << 31)) {
617: fprintf(stderr, "%s: input too large\n", file);
618: break;
619: }
620: resize_buf(fb, 65536);
621: }
622: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
623: if (ssz == 0) {
624: fb->sz = off;
625: return(1);
626: }
627: if (ssz == -1) {
628: perror(file);
629: break;
630: }
631: off += (size_t)ssz;
632: }
633:
634: free(fb->buf);
635: fb->buf = NULL;
636: return(0);
637: }
638:
639: static void
640: mparse_end(struct mparse *curp)
641: {
642:
643: if (MANDOCLEVEL_FATAL <= curp->file_status)
644: return;
645:
646: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
647: assert(MANDOCLEVEL_FATAL <= curp->file_status);
648: return;
649: }
650:
651: if (curp->man && ! man_endparse(curp->man)) {
652: assert(MANDOCLEVEL_FATAL <= curp->file_status);
653: return;
654: }
655:
656: if ( ! (curp->man || curp->mdoc)) {
1.6 kristaps 657: mandoc_msg(MANDOCERR_NOTMANUAL, curp, 1, 0, NULL);
1.1 kristaps 658: curp->file_status = MANDOCLEVEL_FATAL;
1.6 kristaps 659: return;
1.1 kristaps 660: }
661:
662: roff_endparse(curp->roff);
663: }
664:
665: static void
1.36 ! schwarze 666: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 667: {
668: const char *svfile;
1.36 ! schwarze 669: static int recursion_depth;
! 670:
! 671: if (64 < recursion_depth) {
! 672: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
! 673: return;
! 674: }
1.28 joerg 675:
676: /* Line number is per-file. */
677: svfile = curp->file;
678: curp->file = file;
679: curp->line = 1;
1.36 ! schwarze 680: recursion_depth++;
1.28 joerg 681:
682: mparse_buf_r(curp, blk, 1);
683:
1.36 ! schwarze 684: if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
1.28 joerg 685: mparse_end(curp);
686:
687: curp->file = svfile;
688: }
689:
690: enum mandoclevel
691: mparse_readmem(struct mparse *curp, const void *buf, size_t len,
692: const char *file)
693: {
694: struct buf blk;
695:
696: blk.buf = UNCONST(buf);
697: blk.sz = len;
698:
1.36 ! schwarze 699: mparse_parse_buffer(curp, blk, file);
1.28 joerg 700: return(curp->file_status);
701: }
702:
1.36 ! schwarze 703: enum mandoclevel
! 704: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 705: {
1.28 joerg 706: struct buf blk;
707: int with_mmap;
1.1 kristaps 708:
709: if (-1 == fd)
710: if (-1 == (fd = open(file, O_RDONLY, 0))) {
711: perror(file);
712: curp->file_status = MANDOCLEVEL_SYSERR;
1.36 ! schwarze 713: goto out;
1.1 kristaps 714: }
1.28 joerg 715: /*
716: * Run for each opened file; may be called more than once for
717: * each full parse sequence if the opened file is nested (i.e.,
718: * from `so'). Simply sucks in the whole file and moves into
719: * the parse phase for the file.
720: */
1.1 kristaps 721:
1.28 joerg 722: if ( ! read_whole_file(file, fd, &blk, &with_mmap)) {
723: curp->file_status = MANDOCLEVEL_SYSERR;
1.36 ! schwarze 724: goto out;
1.28 joerg 725: }
1.1 kristaps 726:
1.36 ! schwarze 727: mparse_parse_buffer(curp, blk, file);
1.1 kristaps 728:
1.28 joerg 729: #ifdef HAVE_MMAP
730: if (with_mmap)
731: munmap(blk.buf, blk.sz);
732: else
733: #endif
734: free(blk.buf);
1.1 kristaps 735:
736: if (STDIN_FILENO != fd && -1 == close(fd))
737: perror(file);
1.36 ! schwarze 738: out:
1.1 kristaps 739: return(curp->file_status);
740: }
741:
742: struct mparse *
1.29 schwarze 743: mparse_alloc(enum mparset inttype, enum mandoclevel wlevel,
744: mandocmsg mmsg, void *arg, char *defos)
1.1 kristaps 745: {
746: struct mparse *curp;
1.10 kristaps 747:
748: assert(wlevel <= MANDOCLEVEL_FATAL);
1.1 kristaps 749:
750: curp = mandoc_calloc(1, sizeof(struct mparse));
751:
1.3 kristaps 752: curp->wlevel = wlevel;
1.1 kristaps 753: curp->mmsg = mmsg;
754: curp->arg = arg;
755: curp->inttype = inttype;
1.29 schwarze 756: curp->defos = defos;
1.1 kristaps 757:
1.33 schwarze 758: curp->roff = roff_alloc(inttype, curp);
1.1 kristaps 759: return(curp);
760: }
761:
762: void
763: mparse_reset(struct mparse *curp)
764: {
765:
766: roff_reset(curp->roff);
767:
768: if (curp->mdoc)
769: mdoc_reset(curp->mdoc);
770: if (curp->man)
771: man_reset(curp->man);
1.24 kristaps 772: if (curp->secondary)
773: curp->secondary->sz = 0;
1.1 kristaps 774:
775: curp->file_status = MANDOCLEVEL_OK;
776: curp->mdoc = NULL;
777: curp->man = NULL;
778: }
779:
780: void
781: mparse_free(struct mparse *curp)
782: {
783:
784: if (curp->pmdoc)
785: mdoc_free(curp->pmdoc);
786: if (curp->pman)
787: man_free(curp->pman);
788: if (curp->roff)
789: roff_free(curp->roff);
1.24 kristaps 790: if (curp->secondary)
791: free(curp->secondary->buf);
1.1 kristaps 792:
1.24 kristaps 793: free(curp->secondary);
1.1 kristaps 794: free(curp);
795: }
796:
797: void
798: mparse_result(struct mparse *curp, struct mdoc **mdoc, struct man **man)
799: {
800:
1.9 kristaps 801: if (mdoc)
802: *mdoc = curp->mdoc;
803: if (man)
804: *man = curp->man;
1.3 kristaps 805: }
806:
807: void
808: mandoc_vmsg(enum mandocerr t, struct mparse *m,
809: int ln, int pos, const char *fmt, ...)
810: {
811: char buf[256];
812: va_list ap;
813:
814: va_start(ap, fmt);
815: vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
816: va_end(ap);
817:
818: mandoc_msg(t, m, ln, pos, buf);
819: }
820:
821: void
822: mandoc_msg(enum mandocerr er, struct mparse *m,
823: int ln, int col, const char *msg)
824: {
825: enum mandoclevel level;
826:
827: level = MANDOCLEVEL_FATAL;
828: while (er < mandoclimits[level])
829: level--;
830:
831: if (level < m->wlevel)
832: return;
833:
1.8 kristaps 834: if (m->mmsg)
835: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 836:
837: if (m->file_status < level)
838: m->file_status = level;
1.7 kristaps 839: }
840:
841: const char *
842: mparse_strerror(enum mandocerr er)
843: {
844:
845: return(mandocerrs[er]);
846: }
847:
848: const char *
849: mparse_strlevel(enum mandoclevel lvl)
850: {
851: return(mandoclevels[lvl]);
1.24 kristaps 852: }
853:
854: void
855: mparse_keep(struct mparse *p)
856: {
857:
858: assert(NULL == p->secondary);
859: p->secondary = mandoc_calloc(1, sizeof(struct buf));
860: }
861:
862: const char *
863: mparse_getkeep(const struct mparse *p)
864: {
865:
866: assert(p->secondary);
867: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 868: }