Annotation of mandoc/read.c, Revision 1.111
1.111 ! schwarze 1: /* $Id: read.c,v 1.110 2015/01/15 02:29:26 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,
1.111 ! schwarze 83: MANDOCERR_MAX,
1.3 kristaps 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 */
1.108 schwarze 194: NULL,
1.111 ! schwarze 195: "input too large",
1.7 kristaps 196: "input stack limit exceeded, infinite loop?",
197: "skipping bad character",
1.68 schwarze 198: "skipping unknown macro",
1.73 schwarze 199: "skipping item outside list",
1.68 schwarze 200: "skipping column outside column list",
201: "skipping end of block that is not open",
202: "inserting missing end of block",
203: "appending missing end of block",
204:
205: /* related to request and macro arguments */
1.7 kristaps 206: "escaped character not allowed in a name",
207: "argument count wrong",
1.99 schwarze 208: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 209: "missing list type, using -item",
1.70 schwarze 210: "missing manual name, using \"\"",
1.71 schwarze 211: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 212: "unknown standard specifier",
1.71 schwarze 213: "skipping request without numeric argument",
1.109 schwarze 214: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 215: ".so request failed",
1.61 schwarze 216: "skipping all arguments",
217: "skipping excess arguments",
1.92 kristaps 218: "divide by zero",
1.7 kristaps 219: };
220:
221: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
222: "SUCCESS",
223: "RESERVED",
224: "WARNING",
225: "ERROR",
226: "FATAL",
227: "BADARG",
228: "SYSERR"
229: };
230:
1.47 schwarze 231:
1.1 kristaps 232: static void
233: resize_buf(struct buf *buf, size_t initial)
234: {
235:
236: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
237: buf->buf = mandoc_realloc(buf->buf, buf->sz);
238: }
239:
240: static void
1.84 schwarze 241: choose_parser(struct mparse *curp)
1.1 kristaps 242: {
1.83 schwarze 243: char *cp, *ep;
244: int format;
1.1 kristaps 245:
1.83 schwarze 246: /*
247: * If neither command line arguments -mdoc or -man select
248: * a parser nor the roff parser found a .Dd or .TH macro
249: * yet, look ahead in the main input buffer.
250: */
251:
252: if ((format = roff_getformat(curp->roff)) == 0) {
253: cp = curp->primary->buf;
254: ep = cp + curp->primary->sz;
255: while (cp < ep) {
1.85 schwarze 256: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 257: cp++;
258: if (cp[0] == 'D' && cp[1] == 'd') {
259: format = MPARSE_MDOC;
260: break;
261: }
262: if (cp[0] == 'T' && cp[1] == 'H') {
263: format = MPARSE_MAN;
264: break;
265: }
266: }
267: cp = memchr(cp, '\n', ep - cp);
268: if (cp == NULL)
269: break;
270: cp++;
271: }
1.1 kristaps 272: }
273:
1.83 schwarze 274: if (format == MPARSE_MDOC) {
1.47 schwarze 275: if (NULL == curp->pmdoc)
1.44 schwarze 276: curp->pmdoc = mdoc_alloc(
277: curp->roff, curp, curp->defos,
278: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 279: assert(curp->pmdoc);
280: curp->mdoc = curp->pmdoc;
281: return;
1.47 schwarze 282: }
1.1 kristaps 283:
1.83 schwarze 284: /* Fall back to man(7) as a last resort. */
285:
1.47 schwarze 286: if (NULL == curp->pman)
1.106 schwarze 287: curp->pman = man_alloc(
288: curp->roff, curp, curp->defos,
1.44 schwarze 289: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 290: assert(curp->pman);
291: curp->man = curp->pman;
292: }
293:
294: /*
1.95 schwarze 295: * Main parse routine for a buffer.
296: * It assumes encoding and line numbering are already set up.
297: * It can recurse directly (for invocations of user-defined
298: * macros, inline equations, and input line traps)
299: * and indirectly (for .so file inclusion).
1.1 kristaps 300: */
301: static void
1.95 schwarze 302: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 303: {
304: const struct tbl_span *span;
305: struct buf ln;
1.110 schwarze 306: char *cp;
1.95 schwarze 307: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 308: enum rofferr rr;
1.100 schwarze 309: int of;
1.1 kristaps 310: int lnn; /* line number in the real file */
1.110 schwarze 311: int fd;
312: pid_t save_child;
1.1 kristaps 313: unsigned char c;
314:
1.95 schwarze 315: memset(&ln, 0, sizeof(ln));
1.1 kristaps 316:
1.47 schwarze 317: lnn = curp->line;
318: pos = 0;
1.1 kristaps 319:
1.95 schwarze 320: while (i < blk.sz) {
1.1 kristaps 321: if (0 == pos && '\0' == blk.buf[i])
322: break;
323:
324: if (start) {
325: curp->line = lnn;
326: curp->reparse_count = 0;
1.93 schwarze 327:
328: if (lnn < 3 &&
329: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 330: curp->filenc & MPARSE_LATIN1)
331: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 332: }
333:
1.95 schwarze 334: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 335:
336: /*
337: * When finding an unescaped newline character,
338: * leave the character loop to process the line.
339: * Skip a preceding carriage return, if any.
340: */
341:
1.95 schwarze 342: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 343: '\n' == blk.buf[i + 1])
344: ++i;
345: if ('\n' == blk.buf[i]) {
346: ++i;
347: ++lnn;
348: break;
349: }
350:
1.35 schwarze 351: /*
1.93 schwarze 352: * Make sure we have space for the worst
353: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 354: */
355:
1.95 schwarze 356: if (pos + 11 > ln.sz)
1.35 schwarze 357: resize_buf(&ln, 256);
358:
1.47 schwarze 359: /*
1.93 schwarze 360: * Encode 8-bit input.
1.1 kristaps 361: */
362:
1.93 schwarze 363: c = blk.buf[i];
364: if (c & 0x80) {
1.95 schwarze 365: if ( ! (curp->filenc && preconv_encode(
366: &blk, &i, &ln, &pos, &curp->filenc))) {
1.93 schwarze 367: mandoc_vmsg(MANDOCERR_BADCHAR,
368: curp, curp->line, pos,
369: "0x%x", c);
370: ln.buf[pos++] = '?';
371: i++;
372: }
373: continue;
374: }
375:
376: /*
377: * Exclude control characters.
378: */
1.1 kristaps 379:
1.93 schwarze 380: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.78 schwarze 381: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
382: curp->line, pos, "0x%x", c);
1.1 kristaps 383: i++;
1.27 joerg 384: ln.buf[pos++] = '?';
1.1 kristaps 385: continue;
386: }
387:
388: /* Trailing backslash = a plain char. */
389:
1.95 schwarze 390: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 391: ln.buf[pos++] = blk.buf[i++];
392: continue;
393: }
394:
395: /*
396: * Found escape and at least one other character.
397: * When it's a newline character, skip it.
398: * When there is a carriage return in between,
399: * skip that one as well.
400: */
401:
1.95 schwarze 402: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 403: '\n' == blk.buf[i + 2])
404: ++i;
405: if ('\n' == blk.buf[i + 1]) {
406: i += 2;
407: ++lnn;
408: continue;
409: }
410:
1.13 kristaps 411: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 412: i += 2;
413: /* Comment, skip to end of line */
1.95 schwarze 414: for (; i < blk.sz; ++i) {
1.1 kristaps 415: if ('\n' == blk.buf[i]) {
416: ++i;
417: ++lnn;
418: break;
419: }
420: }
421:
422: /* Backout trailing whitespaces */
423: for (; pos > 0; --pos) {
424: if (ln.buf[pos - 1] != ' ')
425: break;
426: if (pos > 2 && ln.buf[pos - 2] == '\\')
427: break;
428: }
429: break;
430: }
431:
1.35 schwarze 432: /* Catch escaped bogus characters. */
433:
434: c = (unsigned char) blk.buf[i+1];
435:
1.47 schwarze 436: if ( ! (isascii(c) &&
437: (isgraph(c) || isblank(c)))) {
1.78 schwarze 438: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
439: curp->line, pos, "0x%x", c);
1.35 schwarze 440: i += 2;
441: ln.buf[pos++] = '?';
442: continue;
443: }
444:
1.1 kristaps 445: /* Some other escape sequence, copy & cont. */
446:
447: ln.buf[pos++] = blk.buf[i++];
448: ln.buf[pos++] = blk.buf[i++];
449: }
450:
1.95 schwarze 451: if (pos >= ln.sz)
1.1 kristaps 452: resize_buf(&ln, 256);
453:
454: ln.buf[pos] = '\0';
455:
456: /*
457: * A significant amount of complexity is contained by
458: * the roff preprocessor. It's line-oriented but can be
459: * expressed on one line, so we need at times to
460: * readjust our starting point and re-run it. The roff
461: * preprocessor can also readjust the buffers with new
462: * data, so we pass them in wholesale.
463: */
464:
465: of = 0;
466:
1.24 kristaps 467: /*
468: * Maintain a lookaside buffer of all parsed lines. We
469: * only do this if mparse_keep() has been invoked (the
470: * buffer may be accessed with mparse_getkeep()).
471: */
472:
473: if (curp->secondary) {
1.47 schwarze 474: curp->secondary->buf = mandoc_realloc(
475: curp->secondary->buf,
476: curp->secondary->sz + pos + 2);
477: memcpy(curp->secondary->buf +
478: curp->secondary->sz,
479: ln.buf, pos);
1.24 kristaps 480: curp->secondary->sz += pos;
481: curp->secondary->buf
482: [curp->secondary->sz] = '\n';
483: curp->secondary->sz++;
484: curp->secondary->buf
485: [curp->secondary->sz] = '\0';
486: }
1.1 kristaps 487: rerun:
1.96 schwarze 488: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 489:
490: switch (rr) {
1.47 schwarze 491: case ROFF_REPARSE:
1.1 kristaps 492: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 493: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 494: else
1.3 kristaps 495: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 496: curp->line, pos, NULL);
1.1 kristaps 497: pos = 0;
498: continue;
1.47 schwarze 499: case ROFF_APPEND:
1.95 schwarze 500: pos = strlen(ln.buf);
1.1 kristaps 501: continue;
1.47 schwarze 502: case ROFF_RERUN:
1.1 kristaps 503: goto rerun;
1.47 schwarze 504: case ROFF_IGN:
1.1 kristaps 505: pos = 0;
506: continue;
1.47 schwarze 507: case ROFF_SO:
1.95 schwarze 508: if ( ! (curp->options & MPARSE_SO) &&
509: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 510: curp->sodest = mandoc_strdup(ln.buf + of);
511: free(ln.buf);
512: return;
513: }
1.24 kristaps 514: /*
515: * We remove `so' clauses from our lookaside
516: * buffer because we're going to descend into
517: * the file recursively.
518: */
1.47 schwarze 519: if (curp->secondary)
1.25 kristaps 520: curp->secondary->sz -= pos + 1;
1.110 schwarze 521: save_child = curp->child;
522: if (mparse_open(curp, &fd, ln.buf + of) ==
523: MANDOCLEVEL_OK)
524: mparse_readfd(curp, fd, ln.buf + of);
525: else {
1.52 schwarze 526: mandoc_vmsg(MANDOCERR_SO_FAIL,
527: curp, curp->line, pos,
528: ".so %s", ln.buf + of);
1.110 schwarze 529: ln.sz = mandoc_asprintf(&cp,
530: ".sp\nSee the file %s.\n.sp",
531: ln.buf + of);
532: free(ln.buf);
533: ln.buf = cp;
534: of = 0;
535: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 536: }
1.110 schwarze 537: curp->child = save_child;
1.1 kristaps 538: pos = 0;
539: continue;
540: default:
541: break;
542: }
543:
544: /*
545: * If input parsers have not been allocated, do so now.
1.14 kristaps 546: * We keep these instanced between parsers, but set them
1.1 kristaps 547: * locally per parse routine since we can use different
548: * parsers with each one.
549: */
550:
551: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 552: choose_parser(curp);
1.1 kristaps 553:
1.47 schwarze 554: /*
1.84 schwarze 555: * Lastly, push down into the parsers themselves.
1.1 kristaps 556: * If libroff returns ROFF_TBL, then add it to the
557: * currently open parse. Since we only get here if
558: * there does exist data (see tbl_data.c), we're
559: * guaranteed that something's been allocated.
560: * Do the same for ROFF_EQN.
561: */
562:
1.100 schwarze 563: if (rr == ROFF_TBL) {
564: while ((span = roff_span(curp->roff)) != NULL)
565: if (curp->man == NULL)
566: mdoc_addspan(curp->mdoc, span);
567: else
568: man_addspan(curp->man, span);
569: } else if (rr == ROFF_EQN) {
570: if (curp->man == NULL)
571: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
572: else
573: man_addeqn(curp->man, roff_eqn(curp->roff));
574: } else if ((curp->man == NULL ?
575: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
576: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
577: break;
1.1 kristaps 578:
579: /* Temporary buffers typically are not full. */
580:
581: if (0 == start && '\0' == blk.buf[i])
582: break;
583:
584: /* Start the next input line. */
585:
586: pos = 0;
587: }
588:
589: free(ln.buf);
590: }
591:
592: static int
1.40 schwarze 593: read_whole_file(struct mparse *curp, const char *file, int fd,
594: struct buf *fb, int *with_mmap)
1.1 kristaps 595: {
596: size_t off;
597: ssize_t ssz;
598:
1.81 schwarze 599: #if HAVE_MMAP
1.15 kristaps 600: struct stat st;
1.1 kristaps 601: if (-1 == fstat(fd, &st)) {
1.107 schwarze 602: perror(file);
603: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 604: }
605:
606: /*
607: * If we're a regular file, try just reading in the whole entry
608: * via mmap(). This is faster than reading it into blocks, and
609: * since each file is only a few bytes to begin with, I'm not
610: * concerned that this is going to tank any machines.
611: */
612:
613: if (S_ISREG(st.st_mode)) {
614: if (st.st_size >= (1U << 31)) {
1.111 ! schwarze 615: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 616: return(0);
617: }
618: *with_mmap = 1;
619: fb->sz = (size_t)st.st_size;
1.37 schwarze 620: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 621: if (fb->buf != MAP_FAILED)
622: return(1);
623: }
1.15 kristaps 624: #endif
1.1 kristaps 625:
626: /*
627: * If this isn't a regular file (like, say, stdin), then we must
628: * go the old way and just read things in bit by bit.
629: */
630:
631: *with_mmap = 0;
632: off = 0;
633: fb->sz = 0;
634: fb->buf = NULL;
635: for (;;) {
636: if (off == fb->sz) {
637: if (fb->sz == (1U << 31)) {
1.111 ! schwarze 638: mandoc_msg(MANDOCERR_TOOLARGE, curp,
! 639: 0, 0, NULL);
1.1 kristaps 640: break;
641: }
642: resize_buf(fb, 65536);
643: }
644: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
645: if (ssz == 0) {
646: fb->sz = off;
647: return(1);
648: }
649: if (ssz == -1) {
1.107 schwarze 650: perror(file);
651: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 652: }
653: off += (size_t)ssz;
654: }
655:
656: free(fb->buf);
657: fb->buf = NULL;
658: return(0);
659: }
660:
661: static void
662: mparse_end(struct mparse *curp)
663: {
664:
1.72 schwarze 665: if (curp->mdoc == NULL &&
666: curp->man == NULL &&
667: curp->sodest == NULL) {
668: if (curp->options & MPARSE_MDOC)
669: curp->mdoc = curp->pmdoc;
670: else {
671: if (curp->pman == NULL)
1.106 schwarze 672: curp->pman = man_alloc(
673: curp->roff, curp, curp->defos,
1.72 schwarze 674: curp->options & MPARSE_QUICK ? 1 : 0);
675: curp->man = curp->pman;
676: }
677: }
1.111 ! schwarze 678: if (curp->mdoc)
! 679: mdoc_endparse(curp->mdoc);
! 680: if (curp->man)
! 681: man_endparse(curp->man);
1.1 kristaps 682: roff_endparse(curp->roff);
683: }
684:
685: static void
1.36 schwarze 686: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 687: {
1.85 schwarze 688: struct buf *svprimary;
1.28 joerg 689: const char *svfile;
1.95 schwarze 690: size_t offset;
1.36 schwarze 691: static int recursion_depth;
692:
693: if (64 < recursion_depth) {
694: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
695: return;
696: }
1.28 joerg 697:
698: /* Line number is per-file. */
699: svfile = curp->file;
700: curp->file = file;
1.85 schwarze 701: svprimary = curp->primary;
1.83 schwarze 702: curp->primary = &blk;
1.28 joerg 703: curp->line = 1;
1.36 schwarze 704: recursion_depth++;
1.28 joerg 705:
1.93 schwarze 706: /* Skip an UTF-8 byte order mark. */
707: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
708: (unsigned char)blk.buf[0] == 0xef &&
709: (unsigned char)blk.buf[1] == 0xbb &&
710: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 711: offset = 3;
1.93 schwarze 712: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 713: } else
714: offset = 0;
1.93 schwarze 715:
1.95 schwarze 716: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 717:
1.111 ! schwarze 718: if (--recursion_depth == 0)
1.28 joerg 719: mparse_end(curp);
720:
1.85 schwarze 721: curp->primary = svprimary;
1.28 joerg 722: curp->file = svfile;
723: }
724:
725: enum mandoclevel
1.104 schwarze 726: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 727: const char *file)
728: {
729: struct buf blk;
730:
1.104 schwarze 731: blk.buf = buf;
1.28 joerg 732: blk.sz = len;
733:
1.36 schwarze 734: mparse_parse_buffer(curp, blk, file);
1.28 joerg 735: return(curp->file_status);
736: }
737:
1.98 schwarze 738: /*
739: * Read the whole file into memory and call the parsers.
740: * Called recursively when an .so request is encountered.
741: */
1.36 schwarze 742: enum mandoclevel
743: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 744: {
1.28 joerg 745: struct buf blk;
746: int with_mmap;
1.93 schwarze 747: int save_filenc;
1.1 kristaps 748:
1.91 schwarze 749: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 750: save_filenc = curp->filenc;
751: curp->filenc = curp->options &
752: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 753: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 754: curp->filenc = save_filenc;
1.81 schwarze 755: #if HAVE_MMAP
1.91 schwarze 756: if (with_mmap)
757: munmap(blk.buf, blk.sz);
758: else
1.28 joerg 759: #endif
1.91 schwarze 760: free(blk.buf);
761: }
1.1 kristaps 762:
1.98 schwarze 763: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 764: perror(file);
1.91 schwarze 765:
1.98 schwarze 766: mparse_wait(curp);
1.1 kristaps 767: return(curp->file_status);
1.82 schwarze 768: }
769:
770: enum mandoclevel
1.97 schwarze 771: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 772: {
773: int pfd[2];
1.98 schwarze 774: int save_errno;
1.82 schwarze 775: char *cp;
776:
777: curp->file = file;
1.98 schwarze 778:
779: /* Unless zipped, try to just open the file. */
780:
1.82 schwarze 781: if ((cp = strrchr(file, '.')) == NULL ||
782: strcmp(cp + 1, "gz")) {
1.97 schwarze 783: curp->child = 0;
1.98 schwarze 784: if ((*fd = open(file, O_RDONLY)) != -1)
785: return(MANDOCLEVEL_OK);
786:
787: /* Open failed; try to append ".gz". */
788:
789: mandoc_asprintf(&cp, "%s.gz", file);
790: file = cp;
791: } else
792: cp = NULL;
793:
794: /* Before forking, make sure the file can be read. */
795:
796: save_errno = errno;
797: if (access(file, R_OK) == -1) {
798: if (cp != NULL)
799: errno = save_errno;
1.108 schwarze 800: free(cp);
801: *fd = -1;
802: curp->child = 0;
803: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
804: return(MANDOCLEVEL_ERROR);
1.82 schwarze 805: }
806:
1.98 schwarze 807: /* Run gunzip(1). */
808:
1.82 schwarze 809: if (pipe(pfd) == -1) {
1.107 schwarze 810: perror("pipe");
811: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 812: }
813:
1.97 schwarze 814: switch (curp->child = fork()) {
1.82 schwarze 815: case -1:
1.107 schwarze 816: perror("fork");
817: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 818: case 0:
819: close(pfd[0]);
820: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 821: perror("dup");
822: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 823: }
824: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 825: perror("exec");
826: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 827: default:
828: close(pfd[1]);
829: *fd = pfd[0];
830: return(MANDOCLEVEL_OK);
831: }
832: }
833:
834: enum mandoclevel
1.97 schwarze 835: mparse_wait(struct mparse *curp)
1.82 schwarze 836: {
837: int status;
838:
1.97 schwarze 839: if (curp->child == 0)
840: return(MANDOCLEVEL_OK);
841:
842: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 843: perror("wait");
844: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 845: }
846: if (WIFSIGNALED(status)) {
1.108 schwarze 847: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
848: "gunzip died from signal %d", WTERMSIG(status));
849: return(MANDOCLEVEL_ERROR);
1.82 schwarze 850: }
851: if (WEXITSTATUS(status)) {
1.108 schwarze 852: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
853: "gunzip failed with code %d", WEXITSTATUS(status));
854: return(MANDOCLEVEL_ERROR);
1.82 schwarze 855: }
856: return(MANDOCLEVEL_OK);
1.1 kristaps 857: }
858:
859: struct mparse *
1.94 schwarze 860: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
861: const struct mchars *mchars, const char *defos)
1.1 kristaps 862: {
863: struct mparse *curp;
1.10 kristaps 864:
1.1 kristaps 865: curp = mandoc_calloc(1, sizeof(struct mparse));
866:
1.44 schwarze 867: curp->options = options;
1.3 kristaps 868: curp->wlevel = wlevel;
1.1 kristaps 869: curp->mmsg = mmsg;
1.29 schwarze 870: curp->defos = defos;
1.1 kristaps 871:
1.94 schwarze 872: curp->mchars = mchars;
873: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 874: if (curp->options & MPARSE_MDOC)
875: curp->pmdoc = mdoc_alloc(
876: curp->roff, curp, curp->defos,
877: curp->options & MPARSE_QUICK ? 1 : 0);
878: if (curp->options & MPARSE_MAN)
1.106 schwarze 879: curp->pman = man_alloc(
880: curp->roff, curp, curp->defos,
1.72 schwarze 881: curp->options & MPARSE_QUICK ? 1 : 0);
882:
1.1 kristaps 883: return(curp);
884: }
885:
886: void
887: mparse_reset(struct mparse *curp)
888: {
889:
890: roff_reset(curp->roff);
891:
892: if (curp->mdoc)
893: mdoc_reset(curp->mdoc);
894: if (curp->man)
895: man_reset(curp->man);
1.24 kristaps 896: if (curp->secondary)
897: curp->secondary->sz = 0;
1.1 kristaps 898:
899: curp->file_status = MANDOCLEVEL_OK;
900: curp->mdoc = NULL;
901: curp->man = NULL;
1.45 schwarze 902:
903: free(curp->sodest);
904: curp->sodest = NULL;
1.1 kristaps 905: }
906:
907: void
908: mparse_free(struct mparse *curp)
909: {
910:
911: if (curp->pmdoc)
912: mdoc_free(curp->pmdoc);
913: if (curp->pman)
914: man_free(curp->pman);
915: if (curp->roff)
916: roff_free(curp->roff);
1.24 kristaps 917: if (curp->secondary)
918: free(curp->secondary->buf);
1.1 kristaps 919:
1.24 kristaps 920: free(curp->secondary);
1.45 schwarze 921: free(curp->sodest);
1.1 kristaps 922: free(curp);
923: }
924:
925: void
1.45 schwarze 926: mparse_result(struct mparse *curp,
927: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 928: {
929:
1.45 schwarze 930: if (sodest && NULL != (*sodest = curp->sodest)) {
931: *mdoc = NULL;
932: *man = NULL;
933: return;
934: }
1.9 kristaps 935: if (mdoc)
936: *mdoc = curp->mdoc;
937: if (man)
938: *man = curp->man;
1.3 kristaps 939: }
940:
941: void
942: mandoc_vmsg(enum mandocerr t, struct mparse *m,
943: int ln, int pos, const char *fmt, ...)
944: {
945: char buf[256];
946: va_list ap;
947:
948: va_start(ap, fmt);
1.48 schwarze 949: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 950: va_end(ap);
951:
952: mandoc_msg(t, m, ln, pos, buf);
953: }
954:
955: void
1.47 schwarze 956: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 957: int ln, int col, const char *msg)
958: {
959: enum mandoclevel level;
960:
1.111 ! schwarze 961: level = MANDOCLEVEL_ERROR;
1.3 kristaps 962: while (er < mandoclimits[level])
963: level--;
964:
1.108 schwarze 965: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 966: return;
967:
1.8 kristaps 968: if (m->mmsg)
969: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 970:
971: if (m->file_status < level)
972: m->file_status = level;
1.7 kristaps 973: }
974:
975: const char *
976: mparse_strerror(enum mandocerr er)
977: {
978:
979: return(mandocerrs[er]);
980: }
981:
982: const char *
983: mparse_strlevel(enum mandoclevel lvl)
984: {
985: return(mandoclevels[lvl]);
1.24 kristaps 986: }
987:
988: void
989: mparse_keep(struct mparse *p)
990: {
991:
992: assert(NULL == p->secondary);
993: p->secondary = mandoc_calloc(1, sizeof(struct buf));
994: }
995:
996: const char *
997: mparse_getkeep(const struct mparse *p)
998: {
999:
1000: assert(p->secondary);
1001: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1002: }
CVSweb