Annotation of mandoc/read.c, Revision 1.107
1.107 ! schwarze 1: /* $Id: read.c,v 1.106 2014/12/28 14:42:27 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.107 ! schwarze 4: * Copyright (c) 2010-2015 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"
44:
45: #define REPARSE_LIMIT 1000
46:
47: struct mparse {
48: struct man *pman; /* persistent man parser */
49: struct mdoc *pmdoc; /* persistent mdoc parser */
50: struct man *man; /* man parser */
51: struct mdoc *mdoc; /* mdoc parser */
52: struct roff *roff; /* roff parser (!NULL) */
1.94 schwarze 53: const struct mchars *mchars; /* character table */
1.45 schwarze 54: char *sodest; /* filename pointed to by .so */
1.83 schwarze 55: const char *file; /* filename of current input file */
56: struct buf *primary; /* buffer currently being parsed */
57: struct buf *secondary; /* preprocessed copy of input */
58: const char *defos; /* default operating system */
59: mandocmsg mmsg; /* warning/error message handler */
60: enum mandoclevel file_status; /* status of current parse */
61: enum mandoclevel wlevel; /* ignore messages below this */
62: int options; /* parser options */
1.93 schwarze 63: int filenc; /* encoding of the current file */
1.1 kristaps 64: int reparse_count; /* finite interp. stack */
1.83 schwarze 65: int line; /* line number in the file */
1.97 schwarze 66: pid_t child; /* the gunzip(1) process */
1.1 kristaps 67: };
68:
1.84 schwarze 69: static void choose_parser(struct mparse *);
1.1 kristaps 70: static void resize_buf(struct buf *, size_t);
1.95 schwarze 71: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 72: static int read_whole_file(struct mparse *, const char *, int,
73: struct buf *, int *);
1.1 kristaps 74: static void mparse_end(struct mparse *);
1.37 schwarze 75: static void mparse_parse_buffer(struct mparse *, struct buf,
76: const char *);
1.1 kristaps 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 */
1.79 schwarze 94: "missing manual title, using UNTITLED",
95: "missing manual title, using \"\"",
1.54 schwarze 96: "lower case character in document title",
1.79 schwarze 97: "missing manual section, using \"\"",
1.7 kristaps 98: "unknown manual section",
1.54 schwarze 99: "missing date, using today's date",
1.7 kristaps 100: "cannot parse date, using it verbatim",
1.79 schwarze 101: "missing Os macro, using \"\"",
102: "duplicate prologue macro",
103: "late prologue macro",
104: "skipping late title macro",
1.7 kristaps 105: "prologue macros out of order",
106:
107: /* related to document structure */
108: ".so is fragile, better use ln(1)",
1.50 schwarze 109: "no document body",
1.54 schwarze 110: "content before first section header",
111: "first section is not \"NAME\"",
1.7 kristaps 112: "bad NAME section contents",
113: "sections out of conventional order",
1.54 schwarze 114: "duplicate section title",
115: "unexpected section",
1.87 schwarze 116: "unusual Xr order",
117: "unusual Xr punctuation",
1.86 schwarze 118: "AUTHORS section without An macro",
1.7 kristaps 119:
120: /* related to macros and nesting */
1.55 schwarze 121: "obsolete macro",
1.102 schwarze 122: "macro neither callable nor escaped",
1.7 kristaps 123: "skipping paragraph macro",
1.31 schwarze 124: "moving paragraph macro out of list",
1.7 kristaps 125: "skipping no-space macro",
126: "blocks badly nested",
127: "nested displays are not portable",
1.57 schwarze 128: "moving content out of list",
129: ".Vt block has child macro",
1.78 schwarze 130: "fill mode already enabled, skipping",
131: "fill mode already disabled, skipping",
1.7 kristaps 132: "line scope broken",
133:
134: /* related to missing macro arguments */
1.58 schwarze 135: "skipping empty request",
136: "conditional request controls empty scope",
1.7 kristaps 137: "skipping empty macro",
1.62 schwarze 138: "empty argument, using 0n",
1.7 kristaps 139: "argument count wrong",
1.60 schwarze 140: "missing display type, using -ragged",
141: "list type is not the first argument",
142: "missing -width in -tag list, using 8n",
1.78 schwarze 143: "missing utility name, using \"\"",
1.60 schwarze 144: "empty head in list item",
145: "empty list item",
1.61 schwarze 146: "missing font type, using \\fR",
147: "unknown font type, using \\fR",
1.103 schwarze 148: "nothing follows prefix",
1.60 schwarze 149: "missing -std argument, adding it",
1.90 schwarze 150: "missing eqn box, using \"\"",
1.7 kristaps 151:
152: /* related to bad macro arguments */
1.64 schwarze 153: "unterminated quoted argument",
1.7 kristaps 154: "duplicate argument",
1.76 schwarze 155: "skipping duplicate argument",
1.63 schwarze 156: "skipping duplicate display type",
157: "skipping duplicate list type",
1.76 schwarze 158: "skipping -width argument",
1.7 kristaps 159: "unknown AT&T UNIX version",
1.88 schwarze 160: "comma in function argument",
1.89 schwarze 161: "parenthesis in function name",
1.67 schwarze 162: "invalid content in Rs block",
1.63 schwarze 163: "invalid Boolean argument",
164: "unknown font, skipping request",
1.7 kristaps 165:
166: /* related to plain text */
1.64 schwarze 167: "blank line in fill mode, using .sp",
168: "tab in filled text",
169: "whitespace at end of input line",
1.7 kristaps 170: "bad comment style",
1.64 schwarze 171: "invalid escape sequence",
172: "undefined string, using \"\"",
1.16 kristaps 173:
1.7 kristaps 174: "generic error",
1.17 kristaps 175:
176: /* related to equations */
1.20 kristaps 177: "unexpected equation scope closure",
178: "equation scope open on exit",
1.21 kristaps 179: "overlapping equation scopes",
180: "unexpected end of equation",
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",
1.105 schwarze 191: "ignoring macro in table",
1.7 kristaps 192:
1.68 schwarze 193: /* related to document structure and macros */
1.7 kristaps 194: "input stack limit exceeded, infinite loop?",
195: "skipping bad character",
1.68 schwarze 196: "skipping unknown macro",
1.73 schwarze 197: "skipping item outside list",
1.68 schwarze 198: "skipping column outside column list",
199: "skipping end of block that is not open",
200: "inserting missing end of block",
201: "appending missing end of block",
202:
203: /* related to request and macro arguments */
1.7 kristaps 204: "escaped character not allowed in a name",
205: "argument count wrong",
1.99 schwarze 206: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 207: "missing list type, using -item",
1.70 schwarze 208: "missing manual name, using \"\"",
1.71 schwarze 209: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 210: "unknown standard specifier",
1.71 schwarze 211: "skipping request without numeric argument",
1.61 schwarze 212: "skipping all arguments",
213: "skipping excess arguments",
1.92 kristaps 214: "divide by zero",
1.7 kristaps 215:
216: "generic fatal error",
217:
1.40 schwarze 218: "input too large",
1.7 kristaps 219: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.52 schwarze 220: ".so request failed",
1.40 schwarze 221:
222: /* system errors */
1.82 schwarze 223: "gunzip failed with code",
1.51 schwarze 224: NULL,
1.82 schwarze 225: "gunzip died from signal",
1.7 kristaps 226: };
227:
228: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
229: "SUCCESS",
230: "RESERVED",
231: "WARNING",
232: "ERROR",
233: "FATAL",
234: "BADARG",
235: "SYSERR"
236: };
237:
1.47 schwarze 238:
1.1 kristaps 239: static void
240: resize_buf(struct buf *buf, size_t initial)
241: {
242:
243: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
244: buf->buf = mandoc_realloc(buf->buf, buf->sz);
245: }
246:
247: static void
1.84 schwarze 248: choose_parser(struct mparse *curp)
1.1 kristaps 249: {
1.83 schwarze 250: char *cp, *ep;
251: int format;
1.1 kristaps 252:
1.83 schwarze 253: /*
254: * If neither command line arguments -mdoc or -man select
255: * a parser nor the roff parser found a .Dd or .TH macro
256: * yet, look ahead in the main input buffer.
257: */
258:
259: if ((format = roff_getformat(curp->roff)) == 0) {
260: cp = curp->primary->buf;
261: ep = cp + curp->primary->sz;
262: while (cp < ep) {
1.85 schwarze 263: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 264: cp++;
265: if (cp[0] == 'D' && cp[1] == 'd') {
266: format = MPARSE_MDOC;
267: break;
268: }
269: if (cp[0] == 'T' && cp[1] == 'H') {
270: format = MPARSE_MAN;
271: break;
272: }
273: }
274: cp = memchr(cp, '\n', ep - cp);
275: if (cp == NULL)
276: break;
277: cp++;
278: }
1.1 kristaps 279: }
280:
1.83 schwarze 281: if (format == MPARSE_MDOC) {
1.47 schwarze 282: if (NULL == curp->pmdoc)
1.44 schwarze 283: curp->pmdoc = mdoc_alloc(
284: curp->roff, curp, curp->defos,
285: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 286: assert(curp->pmdoc);
287: curp->mdoc = curp->pmdoc;
288: return;
1.47 schwarze 289: }
1.1 kristaps 290:
1.83 schwarze 291: /* Fall back to man(7) as a last resort. */
292:
1.47 schwarze 293: if (NULL == curp->pman)
1.106 schwarze 294: curp->pman = man_alloc(
295: curp->roff, curp, curp->defos,
1.44 schwarze 296: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 297: assert(curp->pman);
298: curp->man = curp->pman;
299: }
300:
301: /*
1.95 schwarze 302: * Main parse routine for a buffer.
303: * It assumes encoding and line numbering are already set up.
304: * It can recurse directly (for invocations of user-defined
305: * macros, inline equations, and input line traps)
306: * and indirectly (for .so file inclusion).
1.1 kristaps 307: */
308: static void
1.95 schwarze 309: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 310: {
311: const struct tbl_span *span;
312: struct buf ln;
1.95 schwarze 313: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 314: enum rofferr rr;
1.100 schwarze 315: int of;
1.1 kristaps 316: int lnn; /* line number in the real file */
317: unsigned char c;
318:
1.95 schwarze 319: memset(&ln, 0, sizeof(ln));
1.1 kristaps 320:
1.47 schwarze 321: lnn = curp->line;
322: pos = 0;
1.1 kristaps 323:
1.95 schwarze 324: while (i < blk.sz) {
1.1 kristaps 325: if (0 == pos && '\0' == blk.buf[i])
326: break;
327:
328: if (start) {
329: curp->line = lnn;
330: curp->reparse_count = 0;
1.93 schwarze 331:
332: if (lnn < 3 &&
333: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 334: curp->filenc & MPARSE_LATIN1)
335: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 336: }
337:
1.95 schwarze 338: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 339:
340: /*
341: * When finding an unescaped newline character,
342: * leave the character loop to process the line.
343: * Skip a preceding carriage return, if any.
344: */
345:
1.95 schwarze 346: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 347: '\n' == blk.buf[i + 1])
348: ++i;
349: if ('\n' == blk.buf[i]) {
350: ++i;
351: ++lnn;
352: break;
353: }
354:
1.35 schwarze 355: /*
1.93 schwarze 356: * Make sure we have space for the worst
357: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 358: */
359:
1.95 schwarze 360: if (pos + 11 > ln.sz)
1.35 schwarze 361: resize_buf(&ln, 256);
362:
1.47 schwarze 363: /*
1.93 schwarze 364: * Encode 8-bit input.
1.1 kristaps 365: */
366:
1.93 schwarze 367: c = blk.buf[i];
368: if (c & 0x80) {
1.95 schwarze 369: if ( ! (curp->filenc && preconv_encode(
370: &blk, &i, &ln, &pos, &curp->filenc))) {
1.93 schwarze 371: mandoc_vmsg(MANDOCERR_BADCHAR,
372: curp, curp->line, pos,
373: "0x%x", c);
374: ln.buf[pos++] = '?';
375: i++;
376: }
377: continue;
378: }
379:
380: /*
381: * Exclude control characters.
382: */
1.1 kristaps 383:
1.93 schwarze 384: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.78 schwarze 385: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
386: curp->line, pos, "0x%x", c);
1.1 kristaps 387: i++;
1.27 joerg 388: ln.buf[pos++] = '?';
1.1 kristaps 389: continue;
390: }
391:
392: /* Trailing backslash = a plain char. */
393:
1.95 schwarze 394: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 395: ln.buf[pos++] = blk.buf[i++];
396: continue;
397: }
398:
399: /*
400: * Found escape and at least one other character.
401: * When it's a newline character, skip it.
402: * When there is a carriage return in between,
403: * skip that one as well.
404: */
405:
1.95 schwarze 406: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 407: '\n' == blk.buf[i + 2])
408: ++i;
409: if ('\n' == blk.buf[i + 1]) {
410: i += 2;
411: ++lnn;
412: continue;
413: }
414:
1.13 kristaps 415: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 416: i += 2;
417: /* Comment, skip to end of line */
1.95 schwarze 418: for (; i < blk.sz; ++i) {
1.1 kristaps 419: if ('\n' == blk.buf[i]) {
420: ++i;
421: ++lnn;
422: break;
423: }
424: }
425:
426: /* Backout trailing whitespaces */
427: for (; pos > 0; --pos) {
428: if (ln.buf[pos - 1] != ' ')
429: break;
430: if (pos > 2 && ln.buf[pos - 2] == '\\')
431: break;
432: }
433: break;
434: }
435:
1.35 schwarze 436: /* Catch escaped bogus characters. */
437:
438: c = (unsigned char) blk.buf[i+1];
439:
1.47 schwarze 440: if ( ! (isascii(c) &&
441: (isgraph(c) || isblank(c)))) {
1.78 schwarze 442: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
443: curp->line, pos, "0x%x", c);
1.35 schwarze 444: i += 2;
445: ln.buf[pos++] = '?';
446: continue;
447: }
448:
1.1 kristaps 449: /* Some other escape sequence, copy & cont. */
450:
451: ln.buf[pos++] = blk.buf[i++];
452: ln.buf[pos++] = blk.buf[i++];
453: }
454:
1.95 schwarze 455: if (pos >= ln.sz)
1.1 kristaps 456: resize_buf(&ln, 256);
457:
458: ln.buf[pos] = '\0';
459:
460: /*
461: * A significant amount of complexity is contained by
462: * the roff preprocessor. It's line-oriented but can be
463: * expressed on one line, so we need at times to
464: * readjust our starting point and re-run it. The roff
465: * preprocessor can also readjust the buffers with new
466: * data, so we pass them in wholesale.
467: */
468:
469: of = 0;
470:
1.24 kristaps 471: /*
472: * Maintain a lookaside buffer of all parsed lines. We
473: * only do this if mparse_keep() has been invoked (the
474: * buffer may be accessed with mparse_getkeep()).
475: */
476:
477: if (curp->secondary) {
1.47 schwarze 478: curp->secondary->buf = mandoc_realloc(
479: curp->secondary->buf,
480: curp->secondary->sz + pos + 2);
481: memcpy(curp->secondary->buf +
482: curp->secondary->sz,
483: ln.buf, pos);
1.24 kristaps 484: curp->secondary->sz += pos;
485: curp->secondary->buf
486: [curp->secondary->sz] = '\n';
487: curp->secondary->sz++;
488: curp->secondary->buf
489: [curp->secondary->sz] = '\0';
490: }
1.1 kristaps 491: rerun:
1.96 schwarze 492: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 493:
494: switch (rr) {
1.47 schwarze 495: case ROFF_REPARSE:
1.1 kristaps 496: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 497: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 498: else
1.3 kristaps 499: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 500: curp->line, pos, NULL);
1.1 kristaps 501: pos = 0;
502: continue;
1.47 schwarze 503: case ROFF_APPEND:
1.95 schwarze 504: pos = strlen(ln.buf);
1.1 kristaps 505: continue;
1.47 schwarze 506: case ROFF_RERUN:
1.1 kristaps 507: goto rerun;
1.47 schwarze 508: case ROFF_IGN:
1.1 kristaps 509: pos = 0;
510: continue;
1.47 schwarze 511: case ROFF_ERR:
1.1 kristaps 512: assert(MANDOCLEVEL_FATAL <= curp->file_status);
513: break;
1.47 schwarze 514: case ROFF_SO:
1.95 schwarze 515: if ( ! (curp->options & MPARSE_SO) &&
516: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 517: curp->sodest = mandoc_strdup(ln.buf + of);
518: free(ln.buf);
519: return;
520: }
1.24 kristaps 521: /*
522: * We remove `so' clauses from our lookaside
523: * buffer because we're going to descend into
524: * the file recursively.
525: */
1.47 schwarze 526: if (curp->secondary)
1.25 kristaps 527: curp->secondary->sz -= pos + 1;
1.36 schwarze 528: mparse_readfd(curp, -1, ln.buf + of);
1.52 schwarze 529: if (MANDOCLEVEL_FATAL <= curp->file_status) {
530: mandoc_vmsg(MANDOCERR_SO_FAIL,
531: curp, curp->line, pos,
532: ".so %s", ln.buf + of);
1.1 kristaps 533: break;
1.52 schwarze 534: }
1.1 kristaps 535: pos = 0;
536: continue;
537: default:
538: break;
539: }
540:
541: /*
542: * If we encounter errors in the recursive parse, make
543: * sure we don't continue parsing.
544: */
545:
546: if (MANDOCLEVEL_FATAL <= curp->file_status)
547: break;
548:
549: /*
550: * If input parsers have not been allocated, do so now.
1.14 kristaps 551: * We keep these instanced between parsers, but set them
1.1 kristaps 552: * locally per parse routine since we can use different
553: * parsers with each one.
554: */
555:
556: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 557: choose_parser(curp);
1.1 kristaps 558:
1.47 schwarze 559: /*
1.84 schwarze 560: * Lastly, push down into the parsers themselves.
1.1 kristaps 561: * If libroff returns ROFF_TBL, then add it to the
562: * currently open parse. Since we only get here if
563: * there does exist data (see tbl_data.c), we're
564: * guaranteed that something's been allocated.
565: * Do the same for ROFF_EQN.
566: */
567:
1.100 schwarze 568: if (rr == ROFF_TBL) {
569: while ((span = roff_span(curp->roff)) != NULL)
570: if (curp->man == NULL)
571: mdoc_addspan(curp->mdoc, span);
572: else
573: man_addspan(curp->man, span);
574: } else if (rr == ROFF_EQN) {
575: if (curp->man == NULL)
576: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
577: else
578: man_addeqn(curp->man, roff_eqn(curp->roff));
579: } else if ((curp->man == NULL ?
580: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
581: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
582: break;
1.1 kristaps 583:
584: /* Temporary buffers typically are not full. */
585:
586: if (0 == start && '\0' == blk.buf[i])
587: break;
588:
589: /* Start the next input line. */
590:
591: pos = 0;
592: }
593:
594: free(ln.buf);
595: }
596:
597: static int
1.40 schwarze 598: read_whole_file(struct mparse *curp, const char *file, int fd,
599: struct buf *fb, int *with_mmap)
1.1 kristaps 600: {
601: size_t off;
602: ssize_t ssz;
603:
1.81 schwarze 604: #if HAVE_MMAP
1.15 kristaps 605: struct stat st;
1.1 kristaps 606: if (-1 == fstat(fd, &st)) {
1.107 ! schwarze 607: perror(file);
! 608: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 609: }
610:
611: /*
612: * If we're a regular file, try just reading in the whole entry
613: * via mmap(). This is faster than reading it into blocks, and
614: * since each file is only a few bytes to begin with, I'm not
615: * concerned that this is going to tank any machines.
616: */
617:
618: if (S_ISREG(st.st_mode)) {
619: if (st.st_size >= (1U << 31)) {
1.40 schwarze 620: curp->file_status = MANDOCLEVEL_FATAL;
621: if (curp->mmsg)
622: (*curp->mmsg)(MANDOCERR_TOOLARGE,
623: curp->file_status, file, 0, 0, NULL);
1.1 kristaps 624: return(0);
625: }
626: *with_mmap = 1;
627: fb->sz = (size_t)st.st_size;
1.37 schwarze 628: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 629: if (fb->buf != MAP_FAILED)
630: return(1);
631: }
1.15 kristaps 632: #endif
1.1 kristaps 633:
634: /*
635: * If this isn't a regular file (like, say, stdin), then we must
636: * go the old way and just read things in bit by bit.
637: */
638:
639: *with_mmap = 0;
640: off = 0;
641: fb->sz = 0;
642: fb->buf = NULL;
643: for (;;) {
644: if (off == fb->sz) {
645: if (fb->sz == (1U << 31)) {
1.40 schwarze 646: curp->file_status = MANDOCLEVEL_FATAL;
647: if (curp->mmsg)
648: (*curp->mmsg)(MANDOCERR_TOOLARGE,
649: curp->file_status,
650: file, 0, 0, NULL);
1.1 kristaps 651: break;
652: }
653: resize_buf(fb, 65536);
654: }
655: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
656: if (ssz == 0) {
657: fb->sz = off;
658: return(1);
659: }
660: if (ssz == -1) {
1.107 ! schwarze 661: perror(file);
! 662: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 663: }
664: off += (size_t)ssz;
665: }
666:
667: free(fb->buf);
668: fb->buf = NULL;
669: return(0);
670: }
671:
672: static void
673: mparse_end(struct mparse *curp)
674: {
675:
676: if (MANDOCLEVEL_FATAL <= curp->file_status)
677: return;
678:
1.72 schwarze 679: if (curp->mdoc == NULL &&
680: curp->man == NULL &&
681: curp->sodest == NULL) {
682: if (curp->options & MPARSE_MDOC)
683: curp->mdoc = curp->pmdoc;
684: else {
685: if (curp->pman == NULL)
1.106 schwarze 686: curp->pman = man_alloc(
687: curp->roff, curp, curp->defos,
1.72 schwarze 688: curp->options & MPARSE_QUICK ? 1 : 0);
689: curp->man = curp->pman;
690: }
691: }
692:
1.1 kristaps 693: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
694: assert(MANDOCLEVEL_FATAL <= curp->file_status);
695: return;
696: }
697:
698: if (curp->man && ! man_endparse(curp->man)) {
699: assert(MANDOCLEVEL_FATAL <= curp->file_status);
700: return;
701: }
702:
703: roff_endparse(curp->roff);
704: }
705:
706: static void
1.36 schwarze 707: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 708: {
1.85 schwarze 709: struct buf *svprimary;
1.28 joerg 710: const char *svfile;
1.95 schwarze 711: size_t offset;
1.36 schwarze 712: static int recursion_depth;
713:
714: if (64 < recursion_depth) {
715: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
716: return;
717: }
1.28 joerg 718:
719: /* Line number is per-file. */
720: svfile = curp->file;
721: curp->file = file;
1.85 schwarze 722: svprimary = curp->primary;
1.83 schwarze 723: curp->primary = &blk;
1.28 joerg 724: curp->line = 1;
1.36 schwarze 725: recursion_depth++;
1.28 joerg 726:
1.93 schwarze 727: /* Skip an UTF-8 byte order mark. */
728: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
729: (unsigned char)blk.buf[0] == 0xef &&
730: (unsigned char)blk.buf[1] == 0xbb &&
731: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 732: offset = 3;
1.93 schwarze 733: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 734: } else
735: offset = 0;
1.93 schwarze 736:
1.95 schwarze 737: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 738:
1.36 schwarze 739: if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
1.28 joerg 740: mparse_end(curp);
741:
1.85 schwarze 742: curp->primary = svprimary;
1.28 joerg 743: curp->file = svfile;
744: }
745:
746: enum mandoclevel
1.104 schwarze 747: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 748: const char *file)
749: {
750: struct buf blk;
751:
1.104 schwarze 752: blk.buf = buf;
1.28 joerg 753: blk.sz = len;
754:
1.36 schwarze 755: mparse_parse_buffer(curp, blk, file);
1.28 joerg 756: return(curp->file_status);
757: }
758:
1.98 schwarze 759: /*
760: * If a file descriptor is given, use it and assume it points
761: * to the named file. Otherwise, open the named file.
762: * Read the whole file into memory and call the parsers.
763: * Called recursively when an .so request is encountered.
764: */
1.36 schwarze 765: enum mandoclevel
766: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 767: {
1.28 joerg 768: struct buf blk;
769: int with_mmap;
1.93 schwarze 770: int save_filenc;
1.98 schwarze 771: pid_t save_child;
1.1 kristaps 772:
1.98 schwarze 773: save_child = curp->child;
774: if (fd != -1)
775: curp->child = 0;
776: else if (mparse_open(curp, &fd, file) >= MANDOCLEVEL_SYSERR)
777: goto out;
1.1 kristaps 778:
1.91 schwarze 779: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 780: save_filenc = curp->filenc;
781: curp->filenc = curp->options &
782: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 783: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 784: curp->filenc = save_filenc;
1.81 schwarze 785: #if HAVE_MMAP
1.91 schwarze 786: if (with_mmap)
787: munmap(blk.buf, blk.sz);
788: else
1.28 joerg 789: #endif
1.91 schwarze 790: free(blk.buf);
791: }
1.1 kristaps 792:
1.98 schwarze 793: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 794: perror(file);
1.91 schwarze 795:
1.98 schwarze 796: mparse_wait(curp);
797: out:
798: curp->child = save_child;
1.1 kristaps 799: return(curp->file_status);
1.82 schwarze 800: }
801:
802: enum mandoclevel
1.97 schwarze 803: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 804: {
805: int pfd[2];
1.98 schwarze 806: int save_errno;
1.82 schwarze 807: char *cp;
808: enum mandocerr err;
809:
810: pfd[1] = -1;
811: curp->file = file;
1.98 schwarze 812:
813: /* Unless zipped, try to just open the file. */
814:
1.82 schwarze 815: if ((cp = strrchr(file, '.')) == NULL ||
816: strcmp(cp + 1, "gz")) {
1.97 schwarze 817: curp->child = 0;
1.98 schwarze 818: if ((*fd = open(file, O_RDONLY)) != -1)
819: return(MANDOCLEVEL_OK);
820:
821: /* Open failed; try to append ".gz". */
822:
823: mandoc_asprintf(&cp, "%s.gz", file);
824: file = cp;
825: } else
826: cp = NULL;
827:
828: /* Before forking, make sure the file can be read. */
829:
830: save_errno = errno;
831: if (access(file, R_OK) == -1) {
832: if (cp != NULL)
833: errno = save_errno;
834: err = MANDOCERR_SYSOPEN;
835: goto out;
1.82 schwarze 836: }
837:
1.98 schwarze 838: /* Run gunzip(1). */
839:
1.82 schwarze 840: if (pipe(pfd) == -1) {
1.107 ! schwarze 841: perror("pipe");
! 842: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 843: }
844:
1.97 schwarze 845: switch (curp->child = fork()) {
1.82 schwarze 846: case -1:
1.107 ! schwarze 847: perror("fork");
! 848: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 849: case 0:
850: close(pfd[0]);
851: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 ! schwarze 852: perror("dup");
! 853: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 854: }
855: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 ! schwarze 856: perror("exec");
! 857: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 858: default:
859: close(pfd[1]);
860: *fd = pfd[0];
861: return(MANDOCLEVEL_OK);
862: }
863:
864: out:
1.98 schwarze 865: free(cp);
1.82 schwarze 866: *fd = -1;
1.97 schwarze 867: curp->child = 0;
1.82 schwarze 868: curp->file_status = MANDOCLEVEL_SYSERR;
869: if (curp->mmsg)
1.98 schwarze 870: (*curp->mmsg)(err, curp->file_status, curp->file,
1.82 schwarze 871: 0, 0, strerror(errno));
872: if (pfd[1] != -1)
873: exit(1);
874: return(curp->file_status);
875: }
876:
877: enum mandoclevel
1.97 schwarze 878: mparse_wait(struct mparse *curp)
1.82 schwarze 879: {
880: int status;
881:
1.97 schwarze 882: if (curp->child == 0)
883: return(MANDOCLEVEL_OK);
884:
885: if (waitpid(curp->child, &status, 0) == -1) {
1.107 ! schwarze 886: perror("wait");
! 887: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 888: }
889: if (WIFSIGNALED(status)) {
890: mandoc_vmsg(MANDOCERR_SYSSIG, curp, 0, 0,
891: "%d", WTERMSIG(status));
892: curp->file_status = MANDOCLEVEL_SYSERR;
893: return(curp->file_status);
894: }
895: if (WEXITSTATUS(status)) {
896: mandoc_vmsg(MANDOCERR_SYSEXIT, curp, 0, 0,
897: "%d", WEXITSTATUS(status));
898: curp->file_status = MANDOCLEVEL_SYSERR;
899: return(curp->file_status);
900: }
901: return(MANDOCLEVEL_OK);
1.1 kristaps 902: }
903:
904: struct mparse *
1.94 schwarze 905: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
906: const struct mchars *mchars, const char *defos)
1.1 kristaps 907: {
908: struct mparse *curp;
1.10 kristaps 909:
910: assert(wlevel <= MANDOCLEVEL_FATAL);
1.1 kristaps 911:
912: curp = mandoc_calloc(1, sizeof(struct mparse));
913:
1.44 schwarze 914: curp->options = options;
1.3 kristaps 915: curp->wlevel = wlevel;
1.1 kristaps 916: curp->mmsg = mmsg;
1.29 schwarze 917: curp->defos = defos;
1.1 kristaps 918:
1.94 schwarze 919: curp->mchars = mchars;
920: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 921: if (curp->options & MPARSE_MDOC)
922: curp->pmdoc = mdoc_alloc(
923: curp->roff, curp, curp->defos,
924: curp->options & MPARSE_QUICK ? 1 : 0);
925: if (curp->options & MPARSE_MAN)
1.106 schwarze 926: curp->pman = man_alloc(
927: curp->roff, curp, curp->defos,
1.72 schwarze 928: curp->options & MPARSE_QUICK ? 1 : 0);
929:
1.1 kristaps 930: return(curp);
931: }
932:
933: void
934: mparse_reset(struct mparse *curp)
935: {
936:
937: roff_reset(curp->roff);
938:
939: if (curp->mdoc)
940: mdoc_reset(curp->mdoc);
941: if (curp->man)
942: man_reset(curp->man);
1.24 kristaps 943: if (curp->secondary)
944: curp->secondary->sz = 0;
1.1 kristaps 945:
946: curp->file_status = MANDOCLEVEL_OK;
947: curp->mdoc = NULL;
948: curp->man = NULL;
1.45 schwarze 949:
950: free(curp->sodest);
951: curp->sodest = NULL;
1.1 kristaps 952: }
953:
954: void
955: mparse_free(struct mparse *curp)
956: {
957:
958: if (curp->pmdoc)
959: mdoc_free(curp->pmdoc);
960: if (curp->pman)
961: man_free(curp->pman);
962: if (curp->roff)
963: roff_free(curp->roff);
1.24 kristaps 964: if (curp->secondary)
965: free(curp->secondary->buf);
1.1 kristaps 966:
1.24 kristaps 967: free(curp->secondary);
1.45 schwarze 968: free(curp->sodest);
1.1 kristaps 969: free(curp);
970: }
971:
972: void
1.45 schwarze 973: mparse_result(struct mparse *curp,
974: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 975: {
976:
1.45 schwarze 977: if (sodest && NULL != (*sodest = curp->sodest)) {
978: *mdoc = NULL;
979: *man = NULL;
980: return;
981: }
1.9 kristaps 982: if (mdoc)
983: *mdoc = curp->mdoc;
984: if (man)
985: *man = curp->man;
1.3 kristaps 986: }
987:
988: void
989: mandoc_vmsg(enum mandocerr t, struct mparse *m,
990: int ln, int pos, const char *fmt, ...)
991: {
992: char buf[256];
993: va_list ap;
994:
995: va_start(ap, fmt);
1.48 schwarze 996: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 997: va_end(ap);
998:
999: mandoc_msg(t, m, ln, pos, buf);
1000: }
1001:
1002: void
1.47 schwarze 1003: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 1004: int ln, int col, const char *msg)
1005: {
1006: enum mandoclevel level;
1007:
1008: level = MANDOCLEVEL_FATAL;
1009: while (er < mandoclimits[level])
1010: level--;
1011:
1012: if (level < m->wlevel)
1013: return;
1014:
1.8 kristaps 1015: if (m->mmsg)
1016: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 1017:
1018: if (m->file_status < level)
1019: m->file_status = level;
1.7 kristaps 1020: }
1021:
1022: const char *
1023: mparse_strerror(enum mandocerr er)
1024: {
1025:
1026: return(mandocerrs[er]);
1027: }
1028:
1029: const char *
1030: mparse_strlevel(enum mandoclevel lvl)
1031: {
1032: return(mandoclevels[lvl]);
1.24 kristaps 1033: }
1034:
1035: void
1036: mparse_keep(struct mparse *p)
1037: {
1038:
1039: assert(NULL == p->secondary);
1040: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1041: }
1042:
1043: const char *
1044: mparse_getkeep(const struct mparse *p)
1045: {
1046:
1047: assert(p->secondary);
1048: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1049: }
CVSweb