Annotation of mandoc/read.c, Revision 1.87
1.87 ! schwarze 1: /* $Id: read.c,v 1.86 2014/09/07 23:25:01 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.40 schwarze 4: * Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org>
1.42 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 kristaps 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: */
1.11 kristaps 19: #include "config.h"
20:
1.80 schwarze 21: #include <sys/types.h>
1.81 schwarze 22: #if HAVE_MMAP
1.82 schwarze 23: #include <sys/mman.h>
1.80 schwarze 24: #include <sys/stat.h>
1.15 kristaps 25: #endif
1.82 schwarze 26: #include <sys/wait.h>
1.1 kristaps 27:
28: #include <assert.h>
29: #include <ctype.h>
1.40 schwarze 30: #include <errno.h>
1.1 kristaps 31: #include <fcntl.h>
1.3 kristaps 32: #include <stdarg.h>
1.28 joerg 33: #include <stdint.h>
1.1 kristaps 34: #include <stdio.h>
35: #include <stdlib.h>
36: #include <string.h>
37: #include <unistd.h>
38:
39: #include "mandoc.h"
1.46 schwarze 40: #include "mandoc_aux.h"
1.3 kristaps 41: #include "libmandoc.h"
1.1 kristaps 42: #include "mdoc.h"
43: #include "man.h"
1.28 joerg 44: #include "main.h"
1.1 kristaps 45:
46: #define REPARSE_LIMIT 1000
47:
48: struct buf {
1.47 schwarze 49: char *buf; /* binary input buffer */
1.1 kristaps 50: size_t sz; /* size of binary buffer */
51: };
52:
53: struct mparse {
54: struct man *pman; /* persistent man parser */
55: struct mdoc *pmdoc; /* persistent mdoc parser */
56: struct man *man; /* man parser */
57: struct mdoc *mdoc; /* mdoc parser */
58: struct roff *roff; /* roff parser (!NULL) */
1.45 schwarze 59: char *sodest; /* filename pointed to by .so */
1.83 schwarze 60: const char *file; /* filename of current input file */
61: struct buf *primary; /* buffer currently being parsed */
62: struct buf *secondary; /* preprocessed copy of input */
63: const char *defos; /* default operating system */
64: mandocmsg mmsg; /* warning/error message handler */
65: enum mandoclevel file_status; /* status of current parse */
66: enum mandoclevel wlevel; /* ignore messages below this */
67: int options; /* parser options */
1.1 kristaps 68: int reparse_count; /* finite interp. stack */
1.83 schwarze 69: int line; /* line number in the file */
1.1 kristaps 70: };
71:
1.84 schwarze 72: static void choose_parser(struct mparse *);
1.1 kristaps 73: static void resize_buf(struct buf *, size_t);
74: static void mparse_buf_r(struct mparse *, struct buf, int);
1.40 schwarze 75: static int read_whole_file(struct mparse *, const char *, int,
76: struct buf *, int *);
1.1 kristaps 77: static void mparse_end(struct mparse *);
1.37 schwarze 78: static void mparse_parse_buffer(struct mparse *, struct buf,
79: const char *);
1.1 kristaps 80:
1.3 kristaps 81: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
82: MANDOCERR_OK,
83: MANDOCERR_WARNING,
84: MANDOCERR_WARNING,
85: MANDOCERR_ERROR,
86: MANDOCERR_FATAL,
87: MANDOCERR_MAX,
88: MANDOCERR_MAX
89: };
90:
1.7 kristaps 91: static const char * const mandocerrs[MANDOCERR_MAX] = {
92: "ok",
93:
94: "generic warning",
95:
96: /* related to the prologue */
1.79 schwarze 97: "missing manual title, using UNTITLED",
98: "missing manual title, using \"\"",
1.54 schwarze 99: "lower case character in document title",
1.79 schwarze 100: "missing manual section, using \"\"",
1.7 kristaps 101: "unknown manual section",
1.32 schwarze 102: "unknown manual volume or arch",
1.54 schwarze 103: "missing date, using today's date",
1.7 kristaps 104: "cannot parse date, using it verbatim",
1.79 schwarze 105: "missing Os macro, using \"\"",
106: "duplicate prologue macro",
107: "late prologue macro",
108: "skipping late title macro",
1.7 kristaps 109: "prologue macros out of order",
110:
111: /* related to document structure */
112: ".so is fragile, better use ln(1)",
1.50 schwarze 113: "no document body",
1.54 schwarze 114: "content before first section header",
115: "first section is not \"NAME\"",
1.7 kristaps 116: "bad NAME section contents",
117: "sections out of conventional order",
1.54 schwarze 118: "duplicate section title",
119: "unexpected section",
1.87 ! schwarze 120: "unusual Xr order",
! 121: "unusual Xr punctuation",
1.86 schwarze 122: "AUTHORS section without An macro",
1.7 kristaps 123:
124: /* related to macros and nesting */
1.55 schwarze 125: "obsolete macro",
1.7 kristaps 126: "skipping paragraph macro",
1.31 schwarze 127: "moving paragraph macro out of list",
1.7 kristaps 128: "skipping no-space macro",
129: "blocks badly nested",
130: "nested displays are not portable",
1.57 schwarze 131: "moving content out of list",
132: ".Vt block has child macro",
1.78 schwarze 133: "fill mode already enabled, skipping",
134: "fill mode already disabled, skipping",
1.7 kristaps 135: "line scope broken",
136:
137: /* related to missing macro arguments */
1.58 schwarze 138: "skipping empty request",
139: "conditional request controls empty scope",
1.7 kristaps 140: "skipping empty macro",
1.62 schwarze 141: "empty argument, using 0n",
1.7 kristaps 142: "argument count wrong",
1.60 schwarze 143: "missing display type, using -ragged",
144: "list type is not the first argument",
145: "missing -width in -tag list, using 8n",
1.78 schwarze 146: "missing utility name, using \"\"",
1.60 schwarze 147: "empty head in list item",
148: "empty list item",
1.61 schwarze 149: "missing font type, using \\fR",
150: "unknown font type, using \\fR",
1.60 schwarze 151: "missing -std argument, adding it",
1.7 kristaps 152:
153: /* related to bad macro arguments */
1.64 schwarze 154: "unterminated quoted argument",
1.7 kristaps 155: "duplicate argument",
1.76 schwarze 156: "skipping duplicate argument",
1.63 schwarze 157: "skipping duplicate display type",
158: "skipping duplicate list type",
1.76 schwarze 159: "skipping -width argument",
1.7 kristaps 160: "unknown AT&T UNIX version",
1.67 schwarze 161: "invalid content in Rs block",
1.63 schwarze 162: "invalid Boolean argument",
163: "unknown font, skipping request",
1.7 kristaps 164:
165: /* related to plain text */
1.64 schwarze 166: "blank line in fill mode, using .sp",
167: "tab in filled text",
168: "whitespace at end of input line",
1.7 kristaps 169: "bad comment style",
1.64 schwarze 170: "invalid escape sequence",
171: "undefined string, using \"\"",
1.16 kristaps 172:
1.7 kristaps 173: "generic error",
1.17 kristaps 174:
175: /* related to equations */
1.20 kristaps 176: "unexpected equation scope closure",
177: "equation scope open on exit",
1.21 kristaps 178: "overlapping equation scopes",
179: "unexpected end of equation",
180: "equation syntax error",
1.7 kristaps 181:
182: /* related to tables */
183: "bad table syntax",
184: "bad table option",
185: "bad table layout",
186: "no table layout cells specified",
187: "no table data cells specified",
188: "ignore data in cell",
189: "data block still open",
190: "ignoring extra data cells",
191:
1.68 schwarze 192: /* related to document structure and macros */
1.7 kristaps 193: "input stack limit exceeded, infinite loop?",
194: "skipping bad character",
1.68 schwarze 195: "skipping unknown macro",
1.73 schwarze 196: "skipping item outside list",
1.68 schwarze 197: "skipping column outside column list",
198: "skipping end of block that is not open",
199: "inserting missing end of block",
200: "appending missing end of block",
201:
202: /* related to request and macro arguments */
1.7 kristaps 203: "escaped character not allowed in a name",
204: "argument count wrong",
1.71 schwarze 205: "missing list type, using -item",
1.70 schwarze 206: "missing manual name, using \"\"",
1.71 schwarze 207: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 208: "unknown standard specifier",
1.71 schwarze 209: "skipping request without numeric argument",
1.61 schwarze 210: "skipping all arguments",
211: "skipping excess arguments",
1.7 kristaps 212:
213: "generic fatal error",
214:
1.40 schwarze 215: "input too large",
1.78 schwarze 216: "NOT IMPLEMENTED: Bd -file",
1.7 kristaps 217: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.52 schwarze 218: ".so request failed",
1.40 schwarze 219:
220: /* system errors */
1.82 schwarze 221: "cannot dup file descriptor",
222: "cannot exec",
223: "gunzip failed with code",
224: "cannot fork",
1.51 schwarze 225: NULL,
1.82 schwarze 226: "cannot open pipe",
227: "cannot read file",
228: "gunzip died from signal",
1.40 schwarze 229: "cannot stat file",
1.82 schwarze 230: "wait failed",
1.7 kristaps 231: };
232:
233: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
234: "SUCCESS",
235: "RESERVED",
236: "WARNING",
237: "ERROR",
238: "FATAL",
239: "BADARG",
240: "SYSERR"
241: };
242:
1.47 schwarze 243:
1.1 kristaps 244: static void
245: resize_buf(struct buf *buf, size_t initial)
246: {
247:
248: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
249: buf->buf = mandoc_realloc(buf->buf, buf->sz);
250: }
251:
252: static void
1.84 schwarze 253: choose_parser(struct mparse *curp)
1.1 kristaps 254: {
1.83 schwarze 255: char *cp, *ep;
256: int format;
1.1 kristaps 257:
1.83 schwarze 258: /*
259: * If neither command line arguments -mdoc or -man select
260: * a parser nor the roff parser found a .Dd or .TH macro
261: * yet, look ahead in the main input buffer.
262: */
263:
264: if ((format = roff_getformat(curp->roff)) == 0) {
265: cp = curp->primary->buf;
266: ep = cp + curp->primary->sz;
267: while (cp < ep) {
1.85 schwarze 268: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 269: cp++;
270: if (cp[0] == 'D' && cp[1] == 'd') {
271: format = MPARSE_MDOC;
272: break;
273: }
274: if (cp[0] == 'T' && cp[1] == 'H') {
275: format = MPARSE_MAN;
276: break;
277: }
278: }
279: cp = memchr(cp, '\n', ep - cp);
280: if (cp == NULL)
281: break;
282: cp++;
283: }
1.1 kristaps 284: }
285:
1.83 schwarze 286: if (format == MPARSE_MDOC) {
1.47 schwarze 287: if (NULL == curp->pmdoc)
1.44 schwarze 288: curp->pmdoc = mdoc_alloc(
289: curp->roff, curp, curp->defos,
290: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 291: assert(curp->pmdoc);
292: curp->mdoc = curp->pmdoc;
293: return;
1.47 schwarze 294: }
1.1 kristaps 295:
1.83 schwarze 296: /* Fall back to man(7) as a last resort. */
297:
1.47 schwarze 298: if (NULL == curp->pman)
1.44 schwarze 299: curp->pman = man_alloc(curp->roff, curp,
300: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 301: assert(curp->pman);
302: curp->man = curp->pman;
303: }
304:
305: /*
306: * Main parse routine for an opened file. This is called for each
307: * opened file and simply loops around the full input file, possibly
308: * nesting (i.e., with `so').
309: */
310: static void
311: mparse_buf_r(struct mparse *curp, struct buf blk, int start)
312: {
313: const struct tbl_span *span;
314: struct buf ln;
315: enum rofferr rr;
316: int i, of, rc;
317: int pos; /* byte number in the ln buffer */
318: int lnn; /* line number in the real file */
319: unsigned char c;
320:
321: memset(&ln, 0, sizeof(struct buf));
322:
1.47 schwarze 323: lnn = curp->line;
324: pos = 0;
1.1 kristaps 325:
326: for (i = 0; i < (int)blk.sz; ) {
327: if (0 == pos && '\0' == blk.buf[i])
328: break;
329:
330: if (start) {
331: curp->line = lnn;
332: curp->reparse_count = 0;
333: }
334:
335: while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
336:
337: /*
338: * When finding an unescaped newline character,
339: * leave the character loop to process the line.
340: * Skip a preceding carriage return, if any.
341: */
342:
343: if ('\r' == blk.buf[i] && i + 1 < (int)blk.sz &&
344: '\n' == blk.buf[i + 1])
345: ++i;
346: if ('\n' == blk.buf[i]) {
347: ++i;
348: ++lnn;
349: break;
350: }
351:
1.35 schwarze 352: /*
353: * Make sure we have space for at least
354: * one backslash and one other character
355: * and the trailing NUL byte.
356: */
357:
358: if (pos + 2 >= (int)ln.sz)
359: resize_buf(&ln, 256);
360:
1.47 schwarze 361: /*
1.1 kristaps 362: * Warn about bogus characters. If you're using
363: * non-ASCII encoding, you're screwing your
364: * readers. Since I'd rather this not happen,
1.27 joerg 365: * I'll be helpful and replace these characters
366: * with "?", so we don't display gibberish.
367: * Note to manual writers: use special characters.
1.1 kristaps 368: */
369:
370: c = (unsigned char) blk.buf[i];
371:
1.47 schwarze 372: if ( ! (isascii(c) &&
373: (isgraph(c) || isblank(c)))) {
1.78 schwarze 374: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
375: curp->line, pos, "0x%x", c);
1.1 kristaps 376: i++;
1.27 joerg 377: ln.buf[pos++] = '?';
1.1 kristaps 378: continue;
379: }
380:
381: /* Trailing backslash = a plain char. */
382:
383: if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
384: ln.buf[pos++] = blk.buf[i++];
385: continue;
386: }
387:
388: /*
389: * Found escape and at least one other character.
390: * When it's a newline character, skip it.
391: * When there is a carriage return in between,
392: * skip that one as well.
393: */
394:
395: if ('\r' == blk.buf[i + 1] && i + 2 < (int)blk.sz &&
396: '\n' == blk.buf[i + 2])
397: ++i;
398: if ('\n' == blk.buf[i + 1]) {
399: i += 2;
400: ++lnn;
401: continue;
402: }
403:
1.13 kristaps 404: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 405: i += 2;
406: /* Comment, skip to end of line */
407: for (; i < (int)blk.sz; ++i) {
408: if ('\n' == blk.buf[i]) {
409: ++i;
410: ++lnn;
411: break;
412: }
413: }
414:
415: /* Backout trailing whitespaces */
416: for (; pos > 0; --pos) {
417: if (ln.buf[pos - 1] != ' ')
418: break;
419: if (pos > 2 && ln.buf[pos - 2] == '\\')
420: break;
421: }
422: break;
423: }
424:
1.35 schwarze 425: /* Catch escaped bogus characters. */
426:
427: c = (unsigned char) blk.buf[i+1];
428:
1.47 schwarze 429: if ( ! (isascii(c) &&
430: (isgraph(c) || isblank(c)))) {
1.78 schwarze 431: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
432: curp->line, pos, "0x%x", c);
1.35 schwarze 433: i += 2;
434: ln.buf[pos++] = '?';
435: continue;
436: }
437:
1.1 kristaps 438: /* Some other escape sequence, copy & cont. */
439:
440: ln.buf[pos++] = blk.buf[i++];
441: ln.buf[pos++] = blk.buf[i++];
442: }
443:
1.47 schwarze 444: if (pos >= (int)ln.sz)
1.1 kristaps 445: resize_buf(&ln, 256);
446:
447: ln.buf[pos] = '\0';
448:
449: /*
450: * A significant amount of complexity is contained by
451: * the roff preprocessor. It's line-oriented but can be
452: * expressed on one line, so we need at times to
453: * readjust our starting point and re-run it. The roff
454: * preprocessor can also readjust the buffers with new
455: * data, so we pass them in wholesale.
456: */
457:
458: of = 0;
459:
1.24 kristaps 460: /*
461: * Maintain a lookaside buffer of all parsed lines. We
462: * only do this if mparse_keep() has been invoked (the
463: * buffer may be accessed with mparse_getkeep()).
464: */
465:
466: if (curp->secondary) {
1.47 schwarze 467: curp->secondary->buf = mandoc_realloc(
468: curp->secondary->buf,
469: curp->secondary->sz + pos + 2);
470: memcpy(curp->secondary->buf +
471: curp->secondary->sz,
472: ln.buf, pos);
1.24 kristaps 473: curp->secondary->sz += pos;
474: curp->secondary->buf
475: [curp->secondary->sz] = '\n';
476: curp->secondary->sz++;
477: curp->secondary->buf
478: [curp->secondary->sz] = '\0';
479: }
1.1 kristaps 480: rerun:
1.47 schwarze 481: rr = roff_parseln(curp->roff, curp->line,
482: &ln.buf, &ln.sz, of, &of);
1.1 kristaps 483:
484: switch (rr) {
1.47 schwarze 485: case ROFF_REPARSE:
1.1 kristaps 486: if (REPARSE_LIMIT >= ++curp->reparse_count)
487: mparse_buf_r(curp, ln, 0);
488: else
1.3 kristaps 489: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 490: curp->line, pos, NULL);
1.1 kristaps 491: pos = 0;
492: continue;
1.47 schwarze 493: case ROFF_APPEND:
1.1 kristaps 494: pos = (int)strlen(ln.buf);
495: continue;
1.47 schwarze 496: case ROFF_RERUN:
1.1 kristaps 497: goto rerun;
1.47 schwarze 498: case ROFF_IGN:
1.1 kristaps 499: pos = 0;
500: continue;
1.47 schwarze 501: case ROFF_ERR:
1.1 kristaps 502: assert(MANDOCLEVEL_FATAL <= curp->file_status);
503: break;
1.47 schwarze 504: case ROFF_SO:
1.45 schwarze 505: if (0 == (MPARSE_SO & curp->options) &&
506: (i >= (int)blk.sz || '\0' == blk.buf[i])) {
507: curp->sodest = mandoc_strdup(ln.buf + of);
508: free(ln.buf);
509: return;
510: }
1.24 kristaps 511: /*
512: * We remove `so' clauses from our lookaside
513: * buffer because we're going to descend into
514: * the file recursively.
515: */
1.47 schwarze 516: if (curp->secondary)
1.25 kristaps 517: curp->secondary->sz -= pos + 1;
1.36 schwarze 518: mparse_readfd(curp, -1, ln.buf + of);
1.52 schwarze 519: if (MANDOCLEVEL_FATAL <= curp->file_status) {
520: mandoc_vmsg(MANDOCERR_SO_FAIL,
521: curp, curp->line, pos,
522: ".so %s", ln.buf + of);
1.1 kristaps 523: break;
1.52 schwarze 524: }
1.1 kristaps 525: pos = 0;
526: continue;
527: default:
528: break;
529: }
530:
531: /*
532: * If we encounter errors in the recursive parse, make
533: * sure we don't continue parsing.
534: */
535:
536: if (MANDOCLEVEL_FATAL <= curp->file_status)
537: break;
538:
539: /*
540: * If input parsers have not been allocated, do so now.
1.14 kristaps 541: * We keep these instanced between parsers, but set them
1.1 kristaps 542: * locally per parse routine since we can use different
543: * parsers with each one.
544: */
545:
546: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 547: choose_parser(curp);
1.1 kristaps 548:
1.47 schwarze 549: /*
1.84 schwarze 550: * Lastly, push down into the parsers themselves.
1.1 kristaps 551: * If libroff returns ROFF_TBL, then add it to the
552: * currently open parse. Since we only get here if
553: * there does exist data (see tbl_data.c), we're
554: * guaranteed that something's been allocated.
555: * Do the same for ROFF_EQN.
556: */
557:
558: rc = -1;
559:
560: if (ROFF_TBL == rr)
561: while (NULL != (span = roff_span(curp->roff))) {
562: rc = curp->man ?
1.47 schwarze 563: man_addspan(curp->man, span) :
564: mdoc_addspan(curp->mdoc, span);
1.1 kristaps 565: if (0 == rc)
566: break;
567: }
568: else if (ROFF_EQN == rr)
1.47 schwarze 569: rc = curp->mdoc ?
570: mdoc_addeqn(curp->mdoc,
571: roff_eqn(curp->roff)) :
572: man_addeqn(curp->man,
573: roff_eqn(curp->roff));
1.1 kristaps 574: else if (curp->man || curp->mdoc)
575: rc = curp->man ?
1.47 schwarze 576: man_parseln(curp->man,
577: curp->line, ln.buf, of) :
578: mdoc_parseln(curp->mdoc,
579: curp->line, ln.buf, of);
1.1 kristaps 580:
581: if (0 == rc) {
582: assert(MANDOCLEVEL_FATAL <= curp->file_status);
583: break;
1.41 schwarze 584: } else if (2 == rc)
585: break;
1.1 kristaps 586:
587: /* Temporary buffers typically are not full. */
588:
589: if (0 == start && '\0' == blk.buf[i])
590: break;
591:
592: /* Start the next input line. */
593:
594: pos = 0;
595: }
596:
597: free(ln.buf);
598: }
599:
600: static int
1.40 schwarze 601: read_whole_file(struct mparse *curp, const char *file, int fd,
602: struct buf *fb, int *with_mmap)
1.1 kristaps 603: {
604: size_t off;
605: ssize_t ssz;
606:
1.81 schwarze 607: #if HAVE_MMAP
1.15 kristaps 608: struct stat st;
1.1 kristaps 609: if (-1 == fstat(fd, &st)) {
1.40 schwarze 610: curp->file_status = MANDOCLEVEL_SYSERR;
611: if (curp->mmsg)
612: (*curp->mmsg)(MANDOCERR_SYSSTAT, curp->file_status,
613: file, 0, 0, strerror(errno));
1.1 kristaps 614: return(0);
615: }
616:
617: /*
618: * If we're a regular file, try just reading in the whole entry
619: * via mmap(). This is faster than reading it into blocks, and
620: * since each file is only a few bytes to begin with, I'm not
621: * concerned that this is going to tank any machines.
622: */
623:
624: if (S_ISREG(st.st_mode)) {
625: if (st.st_size >= (1U << 31)) {
1.40 schwarze 626: curp->file_status = MANDOCLEVEL_FATAL;
627: if (curp->mmsg)
628: (*curp->mmsg)(MANDOCERR_TOOLARGE,
629: curp->file_status, file, 0, 0, NULL);
1.1 kristaps 630: return(0);
631: }
632: *with_mmap = 1;
633: fb->sz = (size_t)st.st_size;
1.37 schwarze 634: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 635: if (fb->buf != MAP_FAILED)
636: return(1);
637: }
1.15 kristaps 638: #endif
1.1 kristaps 639:
640: /*
641: * If this isn't a regular file (like, say, stdin), then we must
642: * go the old way and just read things in bit by bit.
643: */
644:
645: *with_mmap = 0;
646: off = 0;
647: fb->sz = 0;
648: fb->buf = NULL;
649: for (;;) {
650: if (off == fb->sz) {
651: if (fb->sz == (1U << 31)) {
1.40 schwarze 652: curp->file_status = MANDOCLEVEL_FATAL;
653: if (curp->mmsg)
654: (*curp->mmsg)(MANDOCERR_TOOLARGE,
655: curp->file_status,
656: file, 0, 0, NULL);
1.1 kristaps 657: break;
658: }
659: resize_buf(fb, 65536);
660: }
661: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
662: if (ssz == 0) {
663: fb->sz = off;
664: return(1);
665: }
666: if (ssz == -1) {
1.40 schwarze 667: curp->file_status = MANDOCLEVEL_SYSERR;
668: if (curp->mmsg)
669: (*curp->mmsg)(MANDOCERR_SYSREAD,
670: curp->file_status, file, 0, 0,
671: strerror(errno));
1.1 kristaps 672: break;
673: }
674: off += (size_t)ssz;
675: }
676:
677: free(fb->buf);
678: fb->buf = NULL;
679: return(0);
680: }
681:
682: static void
683: mparse_end(struct mparse *curp)
684: {
685:
686: if (MANDOCLEVEL_FATAL <= curp->file_status)
687: return;
688:
1.72 schwarze 689: if (curp->mdoc == NULL &&
690: curp->man == NULL &&
691: curp->sodest == NULL) {
692: if (curp->options & MPARSE_MDOC)
693: curp->mdoc = curp->pmdoc;
694: else {
695: if (curp->pman == NULL)
696: curp->pman = man_alloc(curp->roff, curp,
697: curp->options & MPARSE_QUICK ? 1 : 0);
698: curp->man = curp->pman;
699: }
700: }
701:
1.1 kristaps 702: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
703: assert(MANDOCLEVEL_FATAL <= curp->file_status);
704: return;
705: }
706:
707: if (curp->man && ! man_endparse(curp->man)) {
708: assert(MANDOCLEVEL_FATAL <= curp->file_status);
709: return;
710: }
711:
712: roff_endparse(curp->roff);
713: }
714:
715: static void
1.36 schwarze 716: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 717: {
1.85 schwarze 718: struct buf *svprimary;
1.28 joerg 719: const char *svfile;
1.36 schwarze 720: static int recursion_depth;
721:
722: if (64 < recursion_depth) {
723: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
724: return;
725: }
1.28 joerg 726:
727: /* Line number is per-file. */
728: svfile = curp->file;
729: curp->file = file;
1.85 schwarze 730: svprimary = curp->primary;
1.83 schwarze 731: curp->primary = &blk;
1.28 joerg 732: curp->line = 1;
1.36 schwarze 733: recursion_depth++;
1.28 joerg 734:
735: mparse_buf_r(curp, blk, 1);
736:
1.36 schwarze 737: if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
1.28 joerg 738: mparse_end(curp);
739:
1.85 schwarze 740: curp->primary = svprimary;
1.28 joerg 741: curp->file = svfile;
742: }
743:
744: enum mandoclevel
745: mparse_readmem(struct mparse *curp, const void *buf, size_t len,
746: const char *file)
747: {
748: struct buf blk;
749:
750: blk.buf = UNCONST(buf);
751: blk.sz = len;
752:
1.36 schwarze 753: mparse_parse_buffer(curp, blk, file);
1.28 joerg 754: return(curp->file_status);
755: }
756:
1.36 schwarze 757: enum mandoclevel
758: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 759: {
1.28 joerg 760: struct buf blk;
761: int with_mmap;
1.1 kristaps 762:
1.40 schwarze 763: if (-1 == fd && -1 == (fd = open(file, O_RDONLY, 0))) {
764: curp->file_status = MANDOCLEVEL_SYSERR;
765: if (curp->mmsg)
766: (*curp->mmsg)(MANDOCERR_SYSOPEN,
767: curp->file_status,
768: file, 0, 0, strerror(errno));
769: goto out;
770: }
771:
1.28 joerg 772: /*
773: * Run for each opened file; may be called more than once for
774: * each full parse sequence if the opened file is nested (i.e.,
775: * from `so'). Simply sucks in the whole file and moves into
776: * the parse phase for the file.
777: */
1.1 kristaps 778:
1.40 schwarze 779: if ( ! read_whole_file(curp, file, fd, &blk, &with_mmap))
1.36 schwarze 780: goto out;
1.1 kristaps 781:
1.36 schwarze 782: mparse_parse_buffer(curp, blk, file);
1.1 kristaps 783:
1.81 schwarze 784: #if HAVE_MMAP
1.28 joerg 785: if (with_mmap)
786: munmap(blk.buf, blk.sz);
787: else
788: #endif
789: free(blk.buf);
1.1 kristaps 790:
791: if (STDIN_FILENO != fd && -1 == close(fd))
792: perror(file);
1.36 schwarze 793: out:
1.1 kristaps 794: return(curp->file_status);
1.82 schwarze 795: }
796:
797: enum mandoclevel
798: mparse_open(struct mparse *curp, int *fd, const char *file,
799: pid_t *child_pid)
800: {
801: int pfd[2];
802: char *cp;
803: enum mandocerr err;
804:
805: pfd[1] = -1;
806: curp->file = file;
807: if ((cp = strrchr(file, '.')) == NULL ||
808: strcmp(cp + 1, "gz")) {
809: *child_pid = 0;
810: if ((*fd = open(file, O_RDONLY)) == -1) {
811: err = MANDOCERR_SYSOPEN;
812: goto out;
813: }
814: return(MANDOCLEVEL_OK);
815: }
816:
817: if (pipe(pfd) == -1) {
818: err = MANDOCERR_SYSPIPE;
819: goto out;
820: }
821:
822: switch (*child_pid = fork()) {
823: case -1:
824: err = MANDOCERR_SYSFORK;
825: close(pfd[0]);
826: close(pfd[1]);
827: pfd[1] = -1;
828: break;
829: case 0:
830: close(pfd[0]);
831: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
832: err = MANDOCERR_SYSDUP;
833: break;
834: }
835: execlp("gunzip", "gunzip", "-c", file, NULL);
836: err = MANDOCERR_SYSEXEC;
837: break;
838: default:
839: close(pfd[1]);
840: *fd = pfd[0];
841: return(MANDOCLEVEL_OK);
842: }
843:
844: out:
845: *fd = -1;
846: *child_pid = 0;
847: curp->file_status = MANDOCLEVEL_SYSERR;
848: if (curp->mmsg)
849: (*curp->mmsg)(err, curp->file_status, file,
850: 0, 0, strerror(errno));
851: if (pfd[1] != -1)
852: exit(1);
853: return(curp->file_status);
854: }
855:
856: enum mandoclevel
857: mparse_wait(struct mparse *curp, pid_t child_pid)
858: {
859: int status;
860:
861: if (waitpid(child_pid, &status, 0) == -1) {
862: mandoc_msg(MANDOCERR_SYSWAIT, curp, 0, 0,
863: strerror(errno));
864: curp->file_status = MANDOCLEVEL_SYSERR;
865: return(curp->file_status);
866: }
867: if (WIFSIGNALED(status)) {
868: mandoc_vmsg(MANDOCERR_SYSSIG, curp, 0, 0,
869: "%d", WTERMSIG(status));
870: curp->file_status = MANDOCLEVEL_SYSERR;
871: return(curp->file_status);
872: }
873: if (WEXITSTATUS(status)) {
874: mandoc_vmsg(MANDOCERR_SYSEXIT, curp, 0, 0,
875: "%d", WEXITSTATUS(status));
876: curp->file_status = MANDOCLEVEL_SYSERR;
877: return(curp->file_status);
878: }
879: return(MANDOCLEVEL_OK);
1.1 kristaps 880: }
881:
882: struct mparse *
1.44 schwarze 883: mparse_alloc(int options, enum mandoclevel wlevel,
1.69 schwarze 884: mandocmsg mmsg, const char *defos)
1.1 kristaps 885: {
886: struct mparse *curp;
1.10 kristaps 887:
888: assert(wlevel <= MANDOCLEVEL_FATAL);
1.1 kristaps 889:
890: curp = mandoc_calloc(1, sizeof(struct mparse));
891:
1.44 schwarze 892: curp->options = options;
1.3 kristaps 893: curp->wlevel = wlevel;
1.1 kristaps 894: curp->mmsg = mmsg;
1.29 schwarze 895: curp->defos = defos;
1.1 kristaps 896:
1.44 schwarze 897: curp->roff = roff_alloc(curp, options);
1.72 schwarze 898: if (curp->options & MPARSE_MDOC)
899: curp->pmdoc = mdoc_alloc(
900: curp->roff, curp, curp->defos,
901: curp->options & MPARSE_QUICK ? 1 : 0);
902: if (curp->options & MPARSE_MAN)
903: curp->pman = man_alloc(curp->roff, curp,
904: curp->options & MPARSE_QUICK ? 1 : 0);
905:
1.1 kristaps 906: return(curp);
907: }
908:
909: void
910: mparse_reset(struct mparse *curp)
911: {
912:
913: roff_reset(curp->roff);
914:
915: if (curp->mdoc)
916: mdoc_reset(curp->mdoc);
917: if (curp->man)
918: man_reset(curp->man);
1.24 kristaps 919: if (curp->secondary)
920: curp->secondary->sz = 0;
1.1 kristaps 921:
922: curp->file_status = MANDOCLEVEL_OK;
923: curp->mdoc = NULL;
924: curp->man = NULL;
1.45 schwarze 925:
926: free(curp->sodest);
927: curp->sodest = NULL;
1.1 kristaps 928: }
929:
930: void
931: mparse_free(struct mparse *curp)
932: {
933:
934: if (curp->pmdoc)
935: mdoc_free(curp->pmdoc);
936: if (curp->pman)
937: man_free(curp->pman);
938: if (curp->roff)
939: roff_free(curp->roff);
1.24 kristaps 940: if (curp->secondary)
941: free(curp->secondary->buf);
1.1 kristaps 942:
1.24 kristaps 943: free(curp->secondary);
1.45 schwarze 944: free(curp->sodest);
1.1 kristaps 945: free(curp);
946: }
947:
948: void
1.45 schwarze 949: mparse_result(struct mparse *curp,
950: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 951: {
952:
1.45 schwarze 953: if (sodest && NULL != (*sodest = curp->sodest)) {
954: *mdoc = NULL;
955: *man = NULL;
956: return;
957: }
1.9 kristaps 958: if (mdoc)
959: *mdoc = curp->mdoc;
960: if (man)
961: *man = curp->man;
1.3 kristaps 962: }
963:
964: void
965: mandoc_vmsg(enum mandocerr t, struct mparse *m,
966: int ln, int pos, const char *fmt, ...)
967: {
968: char buf[256];
969: va_list ap;
970:
971: va_start(ap, fmt);
1.48 schwarze 972: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 973: va_end(ap);
974:
975: mandoc_msg(t, m, ln, pos, buf);
976: }
977:
978: void
1.47 schwarze 979: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 980: int ln, int col, const char *msg)
981: {
982: enum mandoclevel level;
983:
984: level = MANDOCLEVEL_FATAL;
985: while (er < mandoclimits[level])
986: level--;
987:
988: if (level < m->wlevel)
989: return;
990:
1.8 kristaps 991: if (m->mmsg)
992: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 993:
994: if (m->file_status < level)
995: m->file_status = level;
1.7 kristaps 996: }
997:
998: const char *
999: mparse_strerror(enum mandocerr er)
1000: {
1001:
1002: return(mandocerrs[er]);
1003: }
1004:
1005: const char *
1006: mparse_strlevel(enum mandoclevel lvl)
1007: {
1008: return(mandoclevels[lvl]);
1.24 kristaps 1009: }
1010:
1011: void
1012: mparse_keep(struct mparse *p)
1013: {
1014:
1015: assert(NULL == p->secondary);
1016: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1017: }
1018:
1019: const char *
1020: mparse_getkeep(const struct mparse *p)
1021: {
1022:
1023: assert(p->secondary);
1024: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1025: }
CVSweb