Annotation of mandoc/read.c, Revision 1.106
1.106 ! schwarze 1: /* $Id: read.c,v 1.105 2014/12/16 23:44:41 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"
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: "cannot dup file descriptor",
224: "cannot exec",
225: "gunzip failed with code",
226: "cannot fork",
1.51 schwarze 227: NULL,
1.82 schwarze 228: "cannot open pipe",
229: "cannot read file",
230: "gunzip died from signal",
1.40 schwarze 231: "cannot stat file",
1.82 schwarze 232: "wait failed",
1.7 kristaps 233: };
234:
235: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
236: "SUCCESS",
237: "RESERVED",
238: "WARNING",
239: "ERROR",
240: "FATAL",
241: "BADARG",
242: "SYSERR"
243: };
244:
1.47 schwarze 245:
1.1 kristaps 246: static void
247: resize_buf(struct buf *buf, size_t initial)
248: {
249:
250: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
251: buf->buf = mandoc_realloc(buf->buf, buf->sz);
252: }
253:
254: static void
1.84 schwarze 255: choose_parser(struct mparse *curp)
1.1 kristaps 256: {
1.83 schwarze 257: char *cp, *ep;
258: int format;
1.1 kristaps 259:
1.83 schwarze 260: /*
261: * If neither command line arguments -mdoc or -man select
262: * a parser nor the roff parser found a .Dd or .TH macro
263: * yet, look ahead in the main input buffer.
264: */
265:
266: if ((format = roff_getformat(curp->roff)) == 0) {
267: cp = curp->primary->buf;
268: ep = cp + curp->primary->sz;
269: while (cp < ep) {
1.85 schwarze 270: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 271: cp++;
272: if (cp[0] == 'D' && cp[1] == 'd') {
273: format = MPARSE_MDOC;
274: break;
275: }
276: if (cp[0] == 'T' && cp[1] == 'H') {
277: format = MPARSE_MAN;
278: break;
279: }
280: }
281: cp = memchr(cp, '\n', ep - cp);
282: if (cp == NULL)
283: break;
284: cp++;
285: }
1.1 kristaps 286: }
287:
1.83 schwarze 288: if (format == MPARSE_MDOC) {
1.47 schwarze 289: if (NULL == curp->pmdoc)
1.44 schwarze 290: curp->pmdoc = mdoc_alloc(
291: curp->roff, curp, curp->defos,
292: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 293: assert(curp->pmdoc);
294: curp->mdoc = curp->pmdoc;
295: return;
1.47 schwarze 296: }
1.1 kristaps 297:
1.83 schwarze 298: /* Fall back to man(7) as a last resort. */
299:
1.47 schwarze 300: if (NULL == curp->pman)
1.106 ! schwarze 301: curp->pman = man_alloc(
! 302: curp->roff, curp, curp->defos,
1.44 schwarze 303: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 304: assert(curp->pman);
305: curp->man = curp->pman;
306: }
307:
308: /*
1.95 schwarze 309: * Main parse routine for a buffer.
310: * It assumes encoding and line numbering are already set up.
311: * It can recurse directly (for invocations of user-defined
312: * macros, inline equations, and input line traps)
313: * and indirectly (for .so file inclusion).
1.1 kristaps 314: */
315: static void
1.95 schwarze 316: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 317: {
318: const struct tbl_span *span;
319: struct buf ln;
1.95 schwarze 320: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 321: enum rofferr rr;
1.100 schwarze 322: int of;
1.1 kristaps 323: int lnn; /* line number in the real file */
324: unsigned char c;
325:
1.95 schwarze 326: memset(&ln, 0, sizeof(ln));
1.1 kristaps 327:
1.47 schwarze 328: lnn = curp->line;
329: pos = 0;
1.1 kristaps 330:
1.95 schwarze 331: while (i < blk.sz) {
1.1 kristaps 332: if (0 == pos && '\0' == blk.buf[i])
333: break;
334:
335: if (start) {
336: curp->line = lnn;
337: curp->reparse_count = 0;
1.93 schwarze 338:
339: if (lnn < 3 &&
340: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 341: curp->filenc & MPARSE_LATIN1)
342: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 343: }
344:
1.95 schwarze 345: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 346:
347: /*
348: * When finding an unescaped newline character,
349: * leave the character loop to process the line.
350: * Skip a preceding carriage return, if any.
351: */
352:
1.95 schwarze 353: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 354: '\n' == blk.buf[i + 1])
355: ++i;
356: if ('\n' == blk.buf[i]) {
357: ++i;
358: ++lnn;
359: break;
360: }
361:
1.35 schwarze 362: /*
1.93 schwarze 363: * Make sure we have space for the worst
364: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 365: */
366:
1.95 schwarze 367: if (pos + 11 > ln.sz)
1.35 schwarze 368: resize_buf(&ln, 256);
369:
1.47 schwarze 370: /*
1.93 schwarze 371: * Encode 8-bit input.
1.1 kristaps 372: */
373:
1.93 schwarze 374: c = blk.buf[i];
375: if (c & 0x80) {
1.95 schwarze 376: if ( ! (curp->filenc && preconv_encode(
377: &blk, &i, &ln, &pos, &curp->filenc))) {
1.93 schwarze 378: mandoc_vmsg(MANDOCERR_BADCHAR,
379: curp, curp->line, pos,
380: "0x%x", c);
381: ln.buf[pos++] = '?';
382: i++;
383: }
384: continue;
385: }
386:
387: /*
388: * Exclude control characters.
389: */
1.1 kristaps 390:
1.93 schwarze 391: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.78 schwarze 392: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
393: curp->line, pos, "0x%x", c);
1.1 kristaps 394: i++;
1.27 joerg 395: ln.buf[pos++] = '?';
1.1 kristaps 396: continue;
397: }
398:
399: /* Trailing backslash = a plain char. */
400:
1.95 schwarze 401: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 402: ln.buf[pos++] = blk.buf[i++];
403: continue;
404: }
405:
406: /*
407: * Found escape and at least one other character.
408: * When it's a newline character, skip it.
409: * When there is a carriage return in between,
410: * skip that one as well.
411: */
412:
1.95 schwarze 413: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 414: '\n' == blk.buf[i + 2])
415: ++i;
416: if ('\n' == blk.buf[i + 1]) {
417: i += 2;
418: ++lnn;
419: continue;
420: }
421:
1.13 kristaps 422: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 423: i += 2;
424: /* Comment, skip to end of line */
1.95 schwarze 425: for (; i < blk.sz; ++i) {
1.1 kristaps 426: if ('\n' == blk.buf[i]) {
427: ++i;
428: ++lnn;
429: break;
430: }
431: }
432:
433: /* Backout trailing whitespaces */
434: for (; pos > 0; --pos) {
435: if (ln.buf[pos - 1] != ' ')
436: break;
437: if (pos > 2 && ln.buf[pos - 2] == '\\')
438: break;
439: }
440: break;
441: }
442:
1.35 schwarze 443: /* Catch escaped bogus characters. */
444:
445: c = (unsigned char) blk.buf[i+1];
446:
1.47 schwarze 447: if ( ! (isascii(c) &&
448: (isgraph(c) || isblank(c)))) {
1.78 schwarze 449: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
450: curp->line, pos, "0x%x", c);
1.35 schwarze 451: i += 2;
452: ln.buf[pos++] = '?';
453: continue;
454: }
455:
1.1 kristaps 456: /* Some other escape sequence, copy & cont. */
457:
458: ln.buf[pos++] = blk.buf[i++];
459: ln.buf[pos++] = blk.buf[i++];
460: }
461:
1.95 schwarze 462: if (pos >= ln.sz)
1.1 kristaps 463: resize_buf(&ln, 256);
464:
465: ln.buf[pos] = '\0';
466:
467: /*
468: * A significant amount of complexity is contained by
469: * the roff preprocessor. It's line-oriented but can be
470: * expressed on one line, so we need at times to
471: * readjust our starting point and re-run it. The roff
472: * preprocessor can also readjust the buffers with new
473: * data, so we pass them in wholesale.
474: */
475:
476: of = 0;
477:
1.24 kristaps 478: /*
479: * Maintain a lookaside buffer of all parsed lines. We
480: * only do this if mparse_keep() has been invoked (the
481: * buffer may be accessed with mparse_getkeep()).
482: */
483:
484: if (curp->secondary) {
1.47 schwarze 485: curp->secondary->buf = mandoc_realloc(
486: curp->secondary->buf,
487: curp->secondary->sz + pos + 2);
488: memcpy(curp->secondary->buf +
489: curp->secondary->sz,
490: ln.buf, pos);
1.24 kristaps 491: curp->secondary->sz += pos;
492: curp->secondary->buf
493: [curp->secondary->sz] = '\n';
494: curp->secondary->sz++;
495: curp->secondary->buf
496: [curp->secondary->sz] = '\0';
497: }
1.1 kristaps 498: rerun:
1.96 schwarze 499: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 500:
501: switch (rr) {
1.47 schwarze 502: case ROFF_REPARSE:
1.1 kristaps 503: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 504: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 505: else
1.3 kristaps 506: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 507: curp->line, pos, NULL);
1.1 kristaps 508: pos = 0;
509: continue;
1.47 schwarze 510: case ROFF_APPEND:
1.95 schwarze 511: pos = strlen(ln.buf);
1.1 kristaps 512: continue;
1.47 schwarze 513: case ROFF_RERUN:
1.1 kristaps 514: goto rerun;
1.47 schwarze 515: case ROFF_IGN:
1.1 kristaps 516: pos = 0;
517: continue;
1.47 schwarze 518: case ROFF_ERR:
1.1 kristaps 519: assert(MANDOCLEVEL_FATAL <= curp->file_status);
520: break;
1.47 schwarze 521: case ROFF_SO:
1.95 schwarze 522: if ( ! (curp->options & MPARSE_SO) &&
523: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 524: curp->sodest = mandoc_strdup(ln.buf + of);
525: free(ln.buf);
526: return;
527: }
1.24 kristaps 528: /*
529: * We remove `so' clauses from our lookaside
530: * buffer because we're going to descend into
531: * the file recursively.
532: */
1.47 schwarze 533: if (curp->secondary)
1.25 kristaps 534: curp->secondary->sz -= pos + 1;
1.36 schwarze 535: mparse_readfd(curp, -1, ln.buf + of);
1.52 schwarze 536: if (MANDOCLEVEL_FATAL <= curp->file_status) {
537: mandoc_vmsg(MANDOCERR_SO_FAIL,
538: curp, curp->line, pos,
539: ".so %s", ln.buf + of);
1.1 kristaps 540: break;
1.52 schwarze 541: }
1.1 kristaps 542: pos = 0;
543: continue;
544: default:
545: break;
546: }
547:
548: /*
549: * If we encounter errors in the recursive parse, make
550: * sure we don't continue parsing.
551: */
552:
553: if (MANDOCLEVEL_FATAL <= curp->file_status)
554: break;
555:
556: /*
557: * If input parsers have not been allocated, do so now.
1.14 kristaps 558: * We keep these instanced between parsers, but set them
1.1 kristaps 559: * locally per parse routine since we can use different
560: * parsers with each one.
561: */
562:
563: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 564: choose_parser(curp);
1.1 kristaps 565:
1.47 schwarze 566: /*
1.84 schwarze 567: * Lastly, push down into the parsers themselves.
1.1 kristaps 568: * If libroff returns ROFF_TBL, then add it to the
569: * currently open parse. Since we only get here if
570: * there does exist data (see tbl_data.c), we're
571: * guaranteed that something's been allocated.
572: * Do the same for ROFF_EQN.
573: */
574:
1.100 schwarze 575: if (rr == ROFF_TBL) {
576: while ((span = roff_span(curp->roff)) != NULL)
577: if (curp->man == NULL)
578: mdoc_addspan(curp->mdoc, span);
579: else
580: man_addspan(curp->man, span);
581: } else if (rr == ROFF_EQN) {
582: if (curp->man == NULL)
583: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
584: else
585: man_addeqn(curp->man, roff_eqn(curp->roff));
586: } else if ((curp->man == NULL ?
587: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
588: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
589: break;
1.1 kristaps 590:
591: /* Temporary buffers typically are not full. */
592:
593: if (0 == start && '\0' == blk.buf[i])
594: break;
595:
596: /* Start the next input line. */
597:
598: pos = 0;
599: }
600:
601: free(ln.buf);
602: }
603:
604: static int
1.40 schwarze 605: read_whole_file(struct mparse *curp, const char *file, int fd,
606: struct buf *fb, int *with_mmap)
1.1 kristaps 607: {
608: size_t off;
609: ssize_t ssz;
610:
1.81 schwarze 611: #if HAVE_MMAP
1.15 kristaps 612: struct stat st;
1.1 kristaps 613: if (-1 == fstat(fd, &st)) {
1.40 schwarze 614: curp->file_status = MANDOCLEVEL_SYSERR;
615: if (curp->mmsg)
616: (*curp->mmsg)(MANDOCERR_SYSSTAT, curp->file_status,
617: file, 0, 0, strerror(errno));
1.1 kristaps 618: return(0);
619: }
620:
621: /*
622: * If we're a regular file, try just reading in the whole entry
623: * via mmap(). This is faster than reading it into blocks, and
624: * since each file is only a few bytes to begin with, I'm not
625: * concerned that this is going to tank any machines.
626: */
627:
628: if (S_ISREG(st.st_mode)) {
629: if (st.st_size >= (1U << 31)) {
1.40 schwarze 630: curp->file_status = MANDOCLEVEL_FATAL;
631: if (curp->mmsg)
632: (*curp->mmsg)(MANDOCERR_TOOLARGE,
633: curp->file_status, file, 0, 0, NULL);
1.1 kristaps 634: return(0);
635: }
636: *with_mmap = 1;
637: fb->sz = (size_t)st.st_size;
1.37 schwarze 638: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 639: if (fb->buf != MAP_FAILED)
640: return(1);
641: }
1.15 kristaps 642: #endif
1.1 kristaps 643:
644: /*
645: * If this isn't a regular file (like, say, stdin), then we must
646: * go the old way and just read things in bit by bit.
647: */
648:
649: *with_mmap = 0;
650: off = 0;
651: fb->sz = 0;
652: fb->buf = NULL;
653: for (;;) {
654: if (off == fb->sz) {
655: if (fb->sz == (1U << 31)) {
1.40 schwarze 656: curp->file_status = MANDOCLEVEL_FATAL;
657: if (curp->mmsg)
658: (*curp->mmsg)(MANDOCERR_TOOLARGE,
659: curp->file_status,
660: file, 0, 0, NULL);
1.1 kristaps 661: break;
662: }
663: resize_buf(fb, 65536);
664: }
665: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
666: if (ssz == 0) {
667: fb->sz = off;
668: return(1);
669: }
670: if (ssz == -1) {
1.40 schwarze 671: curp->file_status = MANDOCLEVEL_SYSERR;
672: if (curp->mmsg)
673: (*curp->mmsg)(MANDOCERR_SYSREAD,
674: curp->file_status, file, 0, 0,
675: strerror(errno));
1.1 kristaps 676: break;
677: }
678: off += (size_t)ssz;
679: }
680:
681: free(fb->buf);
682: fb->buf = NULL;
683: return(0);
684: }
685:
686: static void
687: mparse_end(struct mparse *curp)
688: {
689:
690: if (MANDOCLEVEL_FATAL <= curp->file_status)
691: return;
692:
1.72 schwarze 693: if (curp->mdoc == NULL &&
694: curp->man == NULL &&
695: curp->sodest == NULL) {
696: if (curp->options & MPARSE_MDOC)
697: curp->mdoc = curp->pmdoc;
698: else {
699: if (curp->pman == NULL)
1.106 ! schwarze 700: curp->pman = man_alloc(
! 701: curp->roff, curp, curp->defos,
1.72 schwarze 702: curp->options & MPARSE_QUICK ? 1 : 0);
703: curp->man = curp->pman;
704: }
705: }
706:
1.1 kristaps 707: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
708: assert(MANDOCLEVEL_FATAL <= curp->file_status);
709: return;
710: }
711:
712: if (curp->man && ! man_endparse(curp->man)) {
713: assert(MANDOCLEVEL_FATAL <= curp->file_status);
714: return;
715: }
716:
717: roff_endparse(curp->roff);
718: }
719:
720: static void
1.36 schwarze 721: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 722: {
1.85 schwarze 723: struct buf *svprimary;
1.28 joerg 724: const char *svfile;
1.95 schwarze 725: size_t offset;
1.36 schwarze 726: static int recursion_depth;
727:
728: if (64 < recursion_depth) {
729: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
730: return;
731: }
1.28 joerg 732:
733: /* Line number is per-file. */
734: svfile = curp->file;
735: curp->file = file;
1.85 schwarze 736: svprimary = curp->primary;
1.83 schwarze 737: curp->primary = &blk;
1.28 joerg 738: curp->line = 1;
1.36 schwarze 739: recursion_depth++;
1.28 joerg 740:
1.93 schwarze 741: /* Skip an UTF-8 byte order mark. */
742: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
743: (unsigned char)blk.buf[0] == 0xef &&
744: (unsigned char)blk.buf[1] == 0xbb &&
745: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 746: offset = 3;
1.93 schwarze 747: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 748: } else
749: offset = 0;
1.93 schwarze 750:
1.95 schwarze 751: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 752:
1.36 schwarze 753: if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
1.28 joerg 754: mparse_end(curp);
755:
1.85 schwarze 756: curp->primary = svprimary;
1.28 joerg 757: curp->file = svfile;
758: }
759:
760: enum mandoclevel
1.104 schwarze 761: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 762: const char *file)
763: {
764: struct buf blk;
765:
1.104 schwarze 766: blk.buf = buf;
1.28 joerg 767: blk.sz = len;
768:
1.36 schwarze 769: mparse_parse_buffer(curp, blk, file);
1.28 joerg 770: return(curp->file_status);
771: }
772:
1.98 schwarze 773: /*
774: * If a file descriptor is given, use it and assume it points
775: * to the named file. Otherwise, open the named file.
776: * Read the whole file into memory and call the parsers.
777: * Called recursively when an .so request is encountered.
778: */
1.36 schwarze 779: enum mandoclevel
780: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 781: {
1.28 joerg 782: struct buf blk;
783: int with_mmap;
1.93 schwarze 784: int save_filenc;
1.98 schwarze 785: pid_t save_child;
1.1 kristaps 786:
1.98 schwarze 787: save_child = curp->child;
788: if (fd != -1)
789: curp->child = 0;
790: else if (mparse_open(curp, &fd, file) >= MANDOCLEVEL_SYSERR)
791: goto out;
1.1 kristaps 792:
1.91 schwarze 793: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 794: save_filenc = curp->filenc;
795: curp->filenc = curp->options &
796: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 797: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 798: curp->filenc = save_filenc;
1.81 schwarze 799: #if HAVE_MMAP
1.91 schwarze 800: if (with_mmap)
801: munmap(blk.buf, blk.sz);
802: else
1.28 joerg 803: #endif
1.91 schwarze 804: free(blk.buf);
805: }
1.1 kristaps 806:
1.98 schwarze 807: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 808: perror(file);
1.91 schwarze 809:
1.98 schwarze 810: mparse_wait(curp);
811: out:
812: curp->child = save_child;
1.1 kristaps 813: return(curp->file_status);
1.82 schwarze 814: }
815:
816: enum mandoclevel
1.97 schwarze 817: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 818: {
819: int pfd[2];
1.98 schwarze 820: int save_errno;
1.82 schwarze 821: char *cp;
822: enum mandocerr err;
823:
824: pfd[1] = -1;
825: curp->file = file;
1.98 schwarze 826:
827: /* Unless zipped, try to just open the file. */
828:
1.82 schwarze 829: if ((cp = strrchr(file, '.')) == NULL ||
830: strcmp(cp + 1, "gz")) {
1.97 schwarze 831: curp->child = 0;
1.98 schwarze 832: if ((*fd = open(file, O_RDONLY)) != -1)
833: return(MANDOCLEVEL_OK);
834:
835: /* Open failed; try to append ".gz". */
836:
837: mandoc_asprintf(&cp, "%s.gz", file);
838: file = cp;
839: } else
840: cp = NULL;
841:
842: /* Before forking, make sure the file can be read. */
843:
844: save_errno = errno;
845: if (access(file, R_OK) == -1) {
846: if (cp != NULL)
847: errno = save_errno;
848: err = MANDOCERR_SYSOPEN;
849: goto out;
1.82 schwarze 850: }
851:
1.98 schwarze 852: /* Run gunzip(1). */
853:
1.82 schwarze 854: if (pipe(pfd) == -1) {
855: err = MANDOCERR_SYSPIPE;
856: goto out;
857: }
858:
1.97 schwarze 859: switch (curp->child = fork()) {
1.82 schwarze 860: case -1:
861: err = MANDOCERR_SYSFORK;
862: close(pfd[0]);
863: close(pfd[1]);
864: pfd[1] = -1;
865: break;
866: case 0:
867: close(pfd[0]);
868: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
869: err = MANDOCERR_SYSDUP;
870: break;
871: }
872: execlp("gunzip", "gunzip", "-c", file, NULL);
873: err = MANDOCERR_SYSEXEC;
874: break;
875: default:
876: close(pfd[1]);
877: *fd = pfd[0];
878: return(MANDOCLEVEL_OK);
879: }
880:
881: out:
1.98 schwarze 882: free(cp);
1.82 schwarze 883: *fd = -1;
1.97 schwarze 884: curp->child = 0;
1.82 schwarze 885: curp->file_status = MANDOCLEVEL_SYSERR;
886: if (curp->mmsg)
1.98 schwarze 887: (*curp->mmsg)(err, curp->file_status, curp->file,
1.82 schwarze 888: 0, 0, strerror(errno));
889: if (pfd[1] != -1)
890: exit(1);
891: return(curp->file_status);
892: }
893:
894: enum mandoclevel
1.97 schwarze 895: mparse_wait(struct mparse *curp)
1.82 schwarze 896: {
897: int status;
898:
1.97 schwarze 899: if (curp->child == 0)
900: return(MANDOCLEVEL_OK);
901:
902: if (waitpid(curp->child, &status, 0) == -1) {
1.82 schwarze 903: mandoc_msg(MANDOCERR_SYSWAIT, curp, 0, 0,
904: strerror(errno));
905: curp->file_status = MANDOCLEVEL_SYSERR;
906: return(curp->file_status);
907: }
908: if (WIFSIGNALED(status)) {
909: mandoc_vmsg(MANDOCERR_SYSSIG, curp, 0, 0,
910: "%d", WTERMSIG(status));
911: curp->file_status = MANDOCLEVEL_SYSERR;
912: return(curp->file_status);
913: }
914: if (WEXITSTATUS(status)) {
915: mandoc_vmsg(MANDOCERR_SYSEXIT, curp, 0, 0,
916: "%d", WEXITSTATUS(status));
917: curp->file_status = MANDOCLEVEL_SYSERR;
918: return(curp->file_status);
919: }
920: return(MANDOCLEVEL_OK);
1.1 kristaps 921: }
922:
923: struct mparse *
1.94 schwarze 924: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
925: const struct mchars *mchars, const char *defos)
1.1 kristaps 926: {
927: struct mparse *curp;
1.10 kristaps 928:
929: assert(wlevel <= MANDOCLEVEL_FATAL);
1.1 kristaps 930:
931: curp = mandoc_calloc(1, sizeof(struct mparse));
932:
1.44 schwarze 933: curp->options = options;
1.3 kristaps 934: curp->wlevel = wlevel;
1.1 kristaps 935: curp->mmsg = mmsg;
1.29 schwarze 936: curp->defos = defos;
1.1 kristaps 937:
1.94 schwarze 938: curp->mchars = mchars;
939: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 940: if (curp->options & MPARSE_MDOC)
941: curp->pmdoc = mdoc_alloc(
942: curp->roff, curp, curp->defos,
943: curp->options & MPARSE_QUICK ? 1 : 0);
944: if (curp->options & MPARSE_MAN)
1.106 ! schwarze 945: curp->pman = man_alloc(
! 946: curp->roff, curp, curp->defos,
1.72 schwarze 947: curp->options & MPARSE_QUICK ? 1 : 0);
948:
1.1 kristaps 949: return(curp);
950: }
951:
952: void
953: mparse_reset(struct mparse *curp)
954: {
955:
956: roff_reset(curp->roff);
957:
958: if (curp->mdoc)
959: mdoc_reset(curp->mdoc);
960: if (curp->man)
961: man_reset(curp->man);
1.24 kristaps 962: if (curp->secondary)
963: curp->secondary->sz = 0;
1.1 kristaps 964:
965: curp->file_status = MANDOCLEVEL_OK;
966: curp->mdoc = NULL;
967: curp->man = NULL;
1.45 schwarze 968:
969: free(curp->sodest);
970: curp->sodest = NULL;
1.1 kristaps 971: }
972:
973: void
974: mparse_free(struct mparse *curp)
975: {
976:
977: if (curp->pmdoc)
978: mdoc_free(curp->pmdoc);
979: if (curp->pman)
980: man_free(curp->pman);
981: if (curp->roff)
982: roff_free(curp->roff);
1.24 kristaps 983: if (curp->secondary)
984: free(curp->secondary->buf);
1.1 kristaps 985:
1.24 kristaps 986: free(curp->secondary);
1.45 schwarze 987: free(curp->sodest);
1.1 kristaps 988: free(curp);
989: }
990:
991: void
1.45 schwarze 992: mparse_result(struct mparse *curp,
993: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 994: {
995:
1.45 schwarze 996: if (sodest && NULL != (*sodest = curp->sodest)) {
997: *mdoc = NULL;
998: *man = NULL;
999: return;
1000: }
1.9 kristaps 1001: if (mdoc)
1002: *mdoc = curp->mdoc;
1003: if (man)
1004: *man = curp->man;
1.3 kristaps 1005: }
1006:
1007: void
1008: mandoc_vmsg(enum mandocerr t, struct mparse *m,
1009: int ln, int pos, const char *fmt, ...)
1010: {
1011: char buf[256];
1012: va_list ap;
1013:
1014: va_start(ap, fmt);
1.48 schwarze 1015: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 1016: va_end(ap);
1017:
1018: mandoc_msg(t, m, ln, pos, buf);
1019: }
1020:
1021: void
1.47 schwarze 1022: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 1023: int ln, int col, const char *msg)
1024: {
1025: enum mandoclevel level;
1026:
1027: level = MANDOCLEVEL_FATAL;
1028: while (er < mandoclimits[level])
1029: level--;
1030:
1031: if (level < m->wlevel)
1032: return;
1033:
1.8 kristaps 1034: if (m->mmsg)
1035: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 1036:
1037: if (m->file_status < level)
1038: m->file_status = level;
1.7 kristaps 1039: }
1040:
1041: const char *
1042: mparse_strerror(enum mandocerr er)
1043: {
1044:
1045: return(mandocerrs[er]);
1046: }
1047:
1048: const char *
1049: mparse_strlevel(enum mandoclevel lvl)
1050: {
1051: return(mandoclevels[lvl]);
1.24 kristaps 1052: }
1053:
1054: void
1055: mparse_keep(struct mparse *p)
1056: {
1057:
1058: assert(NULL == p->secondary);
1059: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1060: }
1061:
1062: const char *
1063: mparse_getkeep(const struct mparse *p)
1064: {
1065:
1066: assert(p->secondary);
1067: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1068: }
CVSweb