Annotation of mandoc/read.c, Revision 1.99
1.99 ! schwarze 1: /* $Id: read.c,v 1.98 2014/11/26 23:42:14 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.40 schwarze 4: * Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org>
1.42 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 kristaps 6: *
7: * Permission to use, copy, modify, and distribute this software for any
8: * purpose with or without fee is hereby granted, provided that the above
9: * copyright notice and this permission notice appear in all copies.
10: *
11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18: */
1.11 kristaps 19: #include "config.h"
20:
1.80 schwarze 21: #include <sys/types.h>
1.81 schwarze 22: #if HAVE_MMAP
1.82 schwarze 23: #include <sys/mman.h>
1.80 schwarze 24: #include <sys/stat.h>
1.15 kristaps 25: #endif
1.82 schwarze 26: #include <sys/wait.h>
1.1 kristaps 27:
28: #include <assert.h>
29: #include <ctype.h>
1.40 schwarze 30: #include <errno.h>
1.1 kristaps 31: #include <fcntl.h>
1.3 kristaps 32: #include <stdarg.h>
1.28 joerg 33: #include <stdint.h>
1.1 kristaps 34: #include <stdio.h>
35: #include <stdlib.h>
36: #include <string.h>
37: #include <unistd.h>
38:
39: #include "mandoc.h"
1.46 schwarze 40: #include "mandoc_aux.h"
1.3 kristaps 41: #include "libmandoc.h"
1.1 kristaps 42: #include "mdoc.h"
43: #include "man.h"
1.28 joerg 44: #include "main.h"
1.1 kristaps 45:
46: #define REPARSE_LIMIT 1000
47:
48: struct mparse {
49: struct man *pman; /* persistent man parser */
50: struct mdoc *pmdoc; /* persistent mdoc parser */
51: struct man *man; /* man parser */
52: struct mdoc *mdoc; /* mdoc parser */
53: struct roff *roff; /* roff parser (!NULL) */
1.94 schwarze 54: const struct mchars *mchars; /* character table */
1.45 schwarze 55: char *sodest; /* filename pointed to by .so */
1.83 schwarze 56: const char *file; /* filename of current input file */
57: struct buf *primary; /* buffer currently being parsed */
58: struct buf *secondary; /* preprocessed copy of input */
59: const char *defos; /* default operating system */
60: mandocmsg mmsg; /* warning/error message handler */
61: enum mandoclevel file_status; /* status of current parse */
62: enum mandoclevel wlevel; /* ignore messages below this */
63: int options; /* parser options */
1.93 schwarze 64: int filenc; /* encoding of the current file */
1.1 kristaps 65: int reparse_count; /* finite interp. stack */
1.83 schwarze 66: int line; /* line number in the file */
1.97 schwarze 67: pid_t child; /* the gunzip(1) process */
1.1 kristaps 68: };
69:
1.84 schwarze 70: static void choose_parser(struct mparse *);
1.1 kristaps 71: static void resize_buf(struct buf *, size_t);
1.95 schwarze 72: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 73: static int read_whole_file(struct mparse *, const char *, int,
74: struct buf *, int *);
1.1 kristaps 75: static void mparse_end(struct mparse *);
1.37 schwarze 76: static void mparse_parse_buffer(struct mparse *, struct buf,
77: const char *);
1.1 kristaps 78:
1.3 kristaps 79: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
80: MANDOCERR_OK,
81: MANDOCERR_WARNING,
82: MANDOCERR_WARNING,
83: MANDOCERR_ERROR,
84: MANDOCERR_FATAL,
85: MANDOCERR_MAX,
86: MANDOCERR_MAX
87: };
88:
1.7 kristaps 89: static const char * const mandocerrs[MANDOCERR_MAX] = {
90: "ok",
91:
92: "generic warning",
93:
94: /* related to the prologue */
1.79 schwarze 95: "missing manual title, using UNTITLED",
96: "missing manual title, using \"\"",
1.54 schwarze 97: "lower case character in document title",
1.79 schwarze 98: "missing manual section, using \"\"",
1.7 kristaps 99: "unknown manual section",
1.32 schwarze 100: "unknown manual volume or arch",
1.54 schwarze 101: "missing date, using today's date",
1.7 kristaps 102: "cannot parse date, using it verbatim",
1.79 schwarze 103: "missing Os macro, using \"\"",
104: "duplicate prologue macro",
105: "late prologue macro",
106: "skipping late title macro",
1.7 kristaps 107: "prologue macros out of order",
108:
109: /* related to document structure */
110: ".so is fragile, better use ln(1)",
1.50 schwarze 111: "no document body",
1.54 schwarze 112: "content before first section header",
113: "first section is not \"NAME\"",
1.7 kristaps 114: "bad NAME section contents",
115: "sections out of conventional order",
1.54 schwarze 116: "duplicate section title",
117: "unexpected section",
1.87 schwarze 118: "unusual Xr order",
119: "unusual Xr punctuation",
1.86 schwarze 120: "AUTHORS section without An macro",
1.7 kristaps 121:
122: /* related to macros and nesting */
1.55 schwarze 123: "obsolete macro",
1.7 kristaps 124: "skipping paragraph macro",
1.31 schwarze 125: "moving paragraph macro out of list",
1.7 kristaps 126: "skipping no-space macro",
127: "blocks badly nested",
128: "nested displays are not portable",
1.57 schwarze 129: "moving content out of list",
130: ".Vt block has child macro",
1.78 schwarze 131: "fill mode already enabled, skipping",
132: "fill mode already disabled, skipping",
1.7 kristaps 133: "line scope broken",
134:
135: /* related to missing macro arguments */
1.58 schwarze 136: "skipping empty request",
137: "conditional request controls empty scope",
1.7 kristaps 138: "skipping empty macro",
1.62 schwarze 139: "empty argument, using 0n",
1.7 kristaps 140: "argument count wrong",
1.60 schwarze 141: "missing display type, using -ragged",
142: "list type is not the first argument",
143: "missing -width in -tag list, using 8n",
1.78 schwarze 144: "missing utility name, using \"\"",
1.60 schwarze 145: "empty head in list item",
146: "empty list item",
1.61 schwarze 147: "missing font type, using \\fR",
148: "unknown font type, using \\fR",
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",
191:
1.68 schwarze 192: /* related to document structure and macros */
1.7 kristaps 193: "input stack limit exceeded, infinite loop?",
194: "skipping bad character",
1.68 schwarze 195: "skipping unknown macro",
1.73 schwarze 196: "skipping item outside list",
1.68 schwarze 197: "skipping column outside column list",
198: "skipping end of block that is not open",
199: "inserting missing end of block",
200: "appending missing end of block",
201:
202: /* related to request and macro arguments */
1.7 kristaps 203: "escaped character not allowed in a name",
204: "argument count wrong",
1.99 ! schwarze 205: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 206: "missing list type, using -item",
1.70 schwarze 207: "missing manual name, using \"\"",
1.71 schwarze 208: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 209: "unknown standard specifier",
1.71 schwarze 210: "skipping request without numeric argument",
1.61 schwarze 211: "skipping all arguments",
212: "skipping excess arguments",
1.92 kristaps 213: "divide by zero",
1.7 kristaps 214:
215: "generic fatal error",
216:
1.40 schwarze 217: "input too large",
1.7 kristaps 218: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.52 schwarze 219: ".so request failed",
1.40 schwarze 220:
221: /* system errors */
1.82 schwarze 222: "cannot dup file descriptor",
223: "cannot exec",
224: "gunzip failed with code",
225: "cannot fork",
1.51 schwarze 226: NULL,
1.82 schwarze 227: "cannot open pipe",
228: "cannot read file",
229: "gunzip died from signal",
1.40 schwarze 230: "cannot stat file",
1.82 schwarze 231: "wait failed",
1.7 kristaps 232: };
233:
234: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
235: "SUCCESS",
236: "RESERVED",
237: "WARNING",
238: "ERROR",
239: "FATAL",
240: "BADARG",
241: "SYSERR"
242: };
243:
1.47 schwarze 244:
1.1 kristaps 245: static void
246: resize_buf(struct buf *buf, size_t initial)
247: {
248:
249: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
250: buf->buf = mandoc_realloc(buf->buf, buf->sz);
251: }
252:
253: static void
1.84 schwarze 254: choose_parser(struct mparse *curp)
1.1 kristaps 255: {
1.83 schwarze 256: char *cp, *ep;
257: int format;
1.1 kristaps 258:
1.83 schwarze 259: /*
260: * If neither command line arguments -mdoc or -man select
261: * a parser nor the roff parser found a .Dd or .TH macro
262: * yet, look ahead in the main input buffer.
263: */
264:
265: if ((format = roff_getformat(curp->roff)) == 0) {
266: cp = curp->primary->buf;
267: ep = cp + curp->primary->sz;
268: while (cp < ep) {
1.85 schwarze 269: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 270: cp++;
271: if (cp[0] == 'D' && cp[1] == 'd') {
272: format = MPARSE_MDOC;
273: break;
274: }
275: if (cp[0] == 'T' && cp[1] == 'H') {
276: format = MPARSE_MAN;
277: break;
278: }
279: }
280: cp = memchr(cp, '\n', ep - cp);
281: if (cp == NULL)
282: break;
283: cp++;
284: }
1.1 kristaps 285: }
286:
1.83 schwarze 287: if (format == MPARSE_MDOC) {
1.47 schwarze 288: if (NULL == curp->pmdoc)
1.44 schwarze 289: curp->pmdoc = mdoc_alloc(
290: curp->roff, curp, curp->defos,
291: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 292: assert(curp->pmdoc);
293: curp->mdoc = curp->pmdoc;
294: return;
1.47 schwarze 295: }
1.1 kristaps 296:
1.83 schwarze 297: /* Fall back to man(7) as a last resort. */
298:
1.47 schwarze 299: if (NULL == curp->pman)
1.44 schwarze 300: curp->pman = man_alloc(curp->roff, curp,
301: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 302: assert(curp->pman);
303: curp->man = curp->pman;
304: }
305:
306: /*
1.95 schwarze 307: * Main parse routine for a buffer.
308: * It assumes encoding and line numbering are already set up.
309: * It can recurse directly (for invocations of user-defined
310: * macros, inline equations, and input line traps)
311: * and indirectly (for .so file inclusion).
1.1 kristaps 312: */
313: static void
1.95 schwarze 314: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 315: {
316: const struct tbl_span *span;
317: struct buf ln;
1.95 schwarze 318: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 319: enum rofferr rr;
1.95 schwarze 320: int of, rc;
1.1 kristaps 321: int lnn; /* line number in the real file */
322: unsigned char c;
323:
1.95 schwarze 324: memset(&ln, 0, sizeof(ln));
1.1 kristaps 325:
1.47 schwarze 326: lnn = curp->line;
327: pos = 0;
1.1 kristaps 328:
1.95 schwarze 329: while (i < blk.sz) {
1.1 kristaps 330: if (0 == pos && '\0' == blk.buf[i])
331: break;
332:
333: if (start) {
334: curp->line = lnn;
335: curp->reparse_count = 0;
1.93 schwarze 336:
337: if (lnn < 3 &&
338: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 339: curp->filenc & MPARSE_LATIN1)
340: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 341: }
342:
1.95 schwarze 343: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 344:
345: /*
346: * When finding an unescaped newline character,
347: * leave the character loop to process the line.
348: * Skip a preceding carriage return, if any.
349: */
350:
1.95 schwarze 351: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 352: '\n' == blk.buf[i + 1])
353: ++i;
354: if ('\n' == blk.buf[i]) {
355: ++i;
356: ++lnn;
357: break;
358: }
359:
1.35 schwarze 360: /*
1.93 schwarze 361: * Make sure we have space for the worst
362: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 363: */
364:
1.95 schwarze 365: if (pos + 11 > ln.sz)
1.35 schwarze 366: resize_buf(&ln, 256);
367:
1.47 schwarze 368: /*
1.93 schwarze 369: * Encode 8-bit input.
1.1 kristaps 370: */
371:
1.93 schwarze 372: c = blk.buf[i];
373: if (c & 0x80) {
1.95 schwarze 374: if ( ! (curp->filenc && preconv_encode(
375: &blk, &i, &ln, &pos, &curp->filenc))) {
1.93 schwarze 376: mandoc_vmsg(MANDOCERR_BADCHAR,
377: curp, curp->line, pos,
378: "0x%x", c);
379: ln.buf[pos++] = '?';
380: i++;
381: }
382: continue;
383: }
384:
385: /*
386: * Exclude control characters.
387: */
1.1 kristaps 388:
1.93 schwarze 389: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.78 schwarze 390: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
391: curp->line, pos, "0x%x", c);
1.1 kristaps 392: i++;
1.27 joerg 393: ln.buf[pos++] = '?';
1.1 kristaps 394: continue;
395: }
396:
397: /* Trailing backslash = a plain char. */
398:
1.95 schwarze 399: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 400: ln.buf[pos++] = blk.buf[i++];
401: continue;
402: }
403:
404: /*
405: * Found escape and at least one other character.
406: * When it's a newline character, skip it.
407: * When there is a carriage return in between,
408: * skip that one as well.
409: */
410:
1.95 schwarze 411: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 412: '\n' == blk.buf[i + 2])
413: ++i;
414: if ('\n' == blk.buf[i + 1]) {
415: i += 2;
416: ++lnn;
417: continue;
418: }
419:
1.13 kristaps 420: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 421: i += 2;
422: /* Comment, skip to end of line */
1.95 schwarze 423: for (; i < blk.sz; ++i) {
1.1 kristaps 424: if ('\n' == blk.buf[i]) {
425: ++i;
426: ++lnn;
427: break;
428: }
429: }
430:
431: /* Backout trailing whitespaces */
432: for (; pos > 0; --pos) {
433: if (ln.buf[pos - 1] != ' ')
434: break;
435: if (pos > 2 && ln.buf[pos - 2] == '\\')
436: break;
437: }
438: break;
439: }
440:
1.35 schwarze 441: /* Catch escaped bogus characters. */
442:
443: c = (unsigned char) blk.buf[i+1];
444:
1.47 schwarze 445: if ( ! (isascii(c) &&
446: (isgraph(c) || isblank(c)))) {
1.78 schwarze 447: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
448: curp->line, pos, "0x%x", c);
1.35 schwarze 449: i += 2;
450: ln.buf[pos++] = '?';
451: continue;
452: }
453:
1.1 kristaps 454: /* Some other escape sequence, copy & cont. */
455:
456: ln.buf[pos++] = blk.buf[i++];
457: ln.buf[pos++] = blk.buf[i++];
458: }
459:
1.95 schwarze 460: if (pos >= ln.sz)
1.1 kristaps 461: resize_buf(&ln, 256);
462:
463: ln.buf[pos] = '\0';
464:
465: /*
466: * A significant amount of complexity is contained by
467: * the roff preprocessor. It's line-oriented but can be
468: * expressed on one line, so we need at times to
469: * readjust our starting point and re-run it. The roff
470: * preprocessor can also readjust the buffers with new
471: * data, so we pass them in wholesale.
472: */
473:
474: of = 0;
475:
1.24 kristaps 476: /*
477: * Maintain a lookaside buffer of all parsed lines. We
478: * only do this if mparse_keep() has been invoked (the
479: * buffer may be accessed with mparse_getkeep()).
480: */
481:
482: if (curp->secondary) {
1.47 schwarze 483: curp->secondary->buf = mandoc_realloc(
484: curp->secondary->buf,
485: curp->secondary->sz + pos + 2);
486: memcpy(curp->secondary->buf +
487: curp->secondary->sz,
488: ln.buf, pos);
1.24 kristaps 489: curp->secondary->sz += pos;
490: curp->secondary->buf
491: [curp->secondary->sz] = '\n';
492: curp->secondary->sz++;
493: curp->secondary->buf
494: [curp->secondary->sz] = '\0';
495: }
1.1 kristaps 496: rerun:
1.96 schwarze 497: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 498:
499: switch (rr) {
1.47 schwarze 500: case ROFF_REPARSE:
1.1 kristaps 501: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 502: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 503: else
1.3 kristaps 504: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 505: curp->line, pos, NULL);
1.1 kristaps 506: pos = 0;
507: continue;
1.47 schwarze 508: case ROFF_APPEND:
1.95 schwarze 509: pos = strlen(ln.buf);
1.1 kristaps 510: continue;
1.47 schwarze 511: case ROFF_RERUN:
1.1 kristaps 512: goto rerun;
1.47 schwarze 513: case ROFF_IGN:
1.1 kristaps 514: pos = 0;
515: continue;
1.47 schwarze 516: case ROFF_ERR:
1.1 kristaps 517: assert(MANDOCLEVEL_FATAL <= curp->file_status);
518: break;
1.47 schwarze 519: case ROFF_SO:
1.95 schwarze 520: if ( ! (curp->options & MPARSE_SO) &&
521: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 522: curp->sodest = mandoc_strdup(ln.buf + of);
523: free(ln.buf);
524: return;
525: }
1.24 kristaps 526: /*
527: * We remove `so' clauses from our lookaside
528: * buffer because we're going to descend into
529: * the file recursively.
530: */
1.47 schwarze 531: if (curp->secondary)
1.25 kristaps 532: curp->secondary->sz -= pos + 1;
1.36 schwarze 533: mparse_readfd(curp, -1, ln.buf + of);
1.52 schwarze 534: if (MANDOCLEVEL_FATAL <= curp->file_status) {
535: mandoc_vmsg(MANDOCERR_SO_FAIL,
536: curp, curp->line, pos,
537: ".so %s", ln.buf + of);
1.1 kristaps 538: break;
1.52 schwarze 539: }
1.1 kristaps 540: pos = 0;
541: continue;
542: default:
543: break;
544: }
545:
546: /*
547: * If we encounter errors in the recursive parse, make
548: * sure we don't continue parsing.
549: */
550:
551: if (MANDOCLEVEL_FATAL <= curp->file_status)
552: break;
553:
554: /*
555: * If input parsers have not been allocated, do so now.
1.14 kristaps 556: * We keep these instanced between parsers, but set them
1.1 kristaps 557: * locally per parse routine since we can use different
558: * parsers with each one.
559: */
560:
561: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 562: choose_parser(curp);
1.1 kristaps 563:
1.47 schwarze 564: /*
1.84 schwarze 565: * Lastly, push down into the parsers themselves.
1.1 kristaps 566: * If libroff returns ROFF_TBL, then add it to the
567: * currently open parse. Since we only get here if
568: * there does exist data (see tbl_data.c), we're
569: * guaranteed that something's been allocated.
570: * Do the same for ROFF_EQN.
571: */
572:
573: rc = -1;
574:
575: if (ROFF_TBL == rr)
576: while (NULL != (span = roff_span(curp->roff))) {
577: rc = curp->man ?
1.47 schwarze 578: man_addspan(curp->man, span) :
579: mdoc_addspan(curp->mdoc, span);
1.1 kristaps 580: if (0 == rc)
581: break;
582: }
583: else if (ROFF_EQN == rr)
1.47 schwarze 584: rc = curp->mdoc ?
585: mdoc_addeqn(curp->mdoc,
586: roff_eqn(curp->roff)) :
587: man_addeqn(curp->man,
588: roff_eqn(curp->roff));
1.1 kristaps 589: else if (curp->man || curp->mdoc)
590: rc = curp->man ?
1.47 schwarze 591: man_parseln(curp->man,
592: curp->line, ln.buf, of) :
593: mdoc_parseln(curp->mdoc,
594: curp->line, ln.buf, of);
1.1 kristaps 595:
596: if (0 == rc) {
597: assert(MANDOCLEVEL_FATAL <= curp->file_status);
598: break;
1.41 schwarze 599: } else if (2 == rc)
600: break;
1.1 kristaps 601:
602: /* Temporary buffers typically are not full. */
603:
604: if (0 == start && '\0' == blk.buf[i])
605: break;
606:
607: /* Start the next input line. */
608:
609: pos = 0;
610: }
611:
612: free(ln.buf);
613: }
614:
615: static int
1.40 schwarze 616: read_whole_file(struct mparse *curp, const char *file, int fd,
617: struct buf *fb, int *with_mmap)
1.1 kristaps 618: {
619: size_t off;
620: ssize_t ssz;
621:
1.81 schwarze 622: #if HAVE_MMAP
1.15 kristaps 623: struct stat st;
1.1 kristaps 624: if (-1 == fstat(fd, &st)) {
1.40 schwarze 625: curp->file_status = MANDOCLEVEL_SYSERR;
626: if (curp->mmsg)
627: (*curp->mmsg)(MANDOCERR_SYSSTAT, curp->file_status,
628: file, 0, 0, strerror(errno));
1.1 kristaps 629: return(0);
630: }
631:
632: /*
633: * If we're a regular file, try just reading in the whole entry
634: * via mmap(). This is faster than reading it into blocks, and
635: * since each file is only a few bytes to begin with, I'm not
636: * concerned that this is going to tank any machines.
637: */
638:
639: if (S_ISREG(st.st_mode)) {
640: if (st.st_size >= (1U << 31)) {
1.40 schwarze 641: curp->file_status = MANDOCLEVEL_FATAL;
642: if (curp->mmsg)
643: (*curp->mmsg)(MANDOCERR_TOOLARGE,
644: curp->file_status, file, 0, 0, NULL);
1.1 kristaps 645: return(0);
646: }
647: *with_mmap = 1;
648: fb->sz = (size_t)st.st_size;
1.37 schwarze 649: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 650: if (fb->buf != MAP_FAILED)
651: return(1);
652: }
1.15 kristaps 653: #endif
1.1 kristaps 654:
655: /*
656: * If this isn't a regular file (like, say, stdin), then we must
657: * go the old way and just read things in bit by bit.
658: */
659:
660: *with_mmap = 0;
661: off = 0;
662: fb->sz = 0;
663: fb->buf = NULL;
664: for (;;) {
665: if (off == fb->sz) {
666: if (fb->sz == (1U << 31)) {
1.40 schwarze 667: curp->file_status = MANDOCLEVEL_FATAL;
668: if (curp->mmsg)
669: (*curp->mmsg)(MANDOCERR_TOOLARGE,
670: curp->file_status,
671: file, 0, 0, NULL);
1.1 kristaps 672: break;
673: }
674: resize_buf(fb, 65536);
675: }
676: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
677: if (ssz == 0) {
678: fb->sz = off;
679: return(1);
680: }
681: if (ssz == -1) {
1.40 schwarze 682: curp->file_status = MANDOCLEVEL_SYSERR;
683: if (curp->mmsg)
684: (*curp->mmsg)(MANDOCERR_SYSREAD,
685: curp->file_status, file, 0, 0,
686: strerror(errno));
1.1 kristaps 687: break;
688: }
689: off += (size_t)ssz;
690: }
691:
692: free(fb->buf);
693: fb->buf = NULL;
694: return(0);
695: }
696:
697: static void
698: mparse_end(struct mparse *curp)
699: {
700:
701: if (MANDOCLEVEL_FATAL <= curp->file_status)
702: return;
703:
1.72 schwarze 704: if (curp->mdoc == NULL &&
705: curp->man == NULL &&
706: curp->sodest == NULL) {
707: if (curp->options & MPARSE_MDOC)
708: curp->mdoc = curp->pmdoc;
709: else {
710: if (curp->pman == NULL)
711: curp->pman = man_alloc(curp->roff, curp,
712: curp->options & MPARSE_QUICK ? 1 : 0);
713: curp->man = curp->pman;
714: }
715: }
716:
1.1 kristaps 717: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
718: assert(MANDOCLEVEL_FATAL <= curp->file_status);
719: return;
720: }
721:
722: if (curp->man && ! man_endparse(curp->man)) {
723: assert(MANDOCLEVEL_FATAL <= curp->file_status);
724: return;
725: }
726:
727: roff_endparse(curp->roff);
728: }
729:
730: static void
1.36 schwarze 731: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 732: {
1.85 schwarze 733: struct buf *svprimary;
1.28 joerg 734: const char *svfile;
1.95 schwarze 735: size_t offset;
1.36 schwarze 736: static int recursion_depth;
737:
738: if (64 < recursion_depth) {
739: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
740: return;
741: }
1.28 joerg 742:
743: /* Line number is per-file. */
744: svfile = curp->file;
745: curp->file = file;
1.85 schwarze 746: svprimary = curp->primary;
1.83 schwarze 747: curp->primary = &blk;
1.28 joerg 748: curp->line = 1;
1.36 schwarze 749: recursion_depth++;
1.28 joerg 750:
1.93 schwarze 751: /* Skip an UTF-8 byte order mark. */
752: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
753: (unsigned char)blk.buf[0] == 0xef &&
754: (unsigned char)blk.buf[1] == 0xbb &&
755: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 756: offset = 3;
1.93 schwarze 757: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 758: } else
759: offset = 0;
1.93 schwarze 760:
1.95 schwarze 761: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 762:
1.36 schwarze 763: if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
1.28 joerg 764: mparse_end(curp);
765:
1.85 schwarze 766: curp->primary = svprimary;
1.28 joerg 767: curp->file = svfile;
768: }
769:
770: enum mandoclevel
771: mparse_readmem(struct mparse *curp, const void *buf, size_t len,
772: const char *file)
773: {
774: struct buf blk;
775:
776: blk.buf = UNCONST(buf);
777: blk.sz = len;
778:
1.36 schwarze 779: mparse_parse_buffer(curp, blk, file);
1.28 joerg 780: return(curp->file_status);
781: }
782:
1.98 schwarze 783: /*
784: * If a file descriptor is given, use it and assume it points
785: * to the named file. Otherwise, open the named file.
786: * Read the whole file into memory and call the parsers.
787: * Called recursively when an .so request is encountered.
788: */
1.36 schwarze 789: enum mandoclevel
790: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 791: {
1.28 joerg 792: struct buf blk;
793: int with_mmap;
1.93 schwarze 794: int save_filenc;
1.98 schwarze 795: pid_t save_child;
1.1 kristaps 796:
1.98 schwarze 797: save_child = curp->child;
798: if (fd != -1)
799: curp->child = 0;
800: else if (mparse_open(curp, &fd, file) >= MANDOCLEVEL_SYSERR)
801: goto out;
1.1 kristaps 802:
1.91 schwarze 803: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 804: save_filenc = curp->filenc;
805: curp->filenc = curp->options &
806: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 807: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 808: curp->filenc = save_filenc;
1.81 schwarze 809: #if HAVE_MMAP
1.91 schwarze 810: if (with_mmap)
811: munmap(blk.buf, blk.sz);
812: else
1.28 joerg 813: #endif
1.91 schwarze 814: free(blk.buf);
815: }
1.1 kristaps 816:
1.98 schwarze 817: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 818: perror(file);
1.91 schwarze 819:
1.98 schwarze 820: mparse_wait(curp);
821: out:
822: curp->child = save_child;
1.1 kristaps 823: return(curp->file_status);
1.82 schwarze 824: }
825:
826: enum mandoclevel
1.97 schwarze 827: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 828: {
829: int pfd[2];
1.98 schwarze 830: int save_errno;
1.82 schwarze 831: char *cp;
832: enum mandocerr err;
833:
834: pfd[1] = -1;
835: curp->file = file;
1.98 schwarze 836:
837: /* Unless zipped, try to just open the file. */
838:
1.82 schwarze 839: if ((cp = strrchr(file, '.')) == NULL ||
840: strcmp(cp + 1, "gz")) {
1.97 schwarze 841: curp->child = 0;
1.98 schwarze 842: if ((*fd = open(file, O_RDONLY)) != -1)
843: return(MANDOCLEVEL_OK);
844:
845: /* Open failed; try to append ".gz". */
846:
847: mandoc_asprintf(&cp, "%s.gz", file);
848: file = cp;
849: } else
850: cp = NULL;
851:
852: /* Before forking, make sure the file can be read. */
853:
854: save_errno = errno;
855: if (access(file, R_OK) == -1) {
856: if (cp != NULL)
857: errno = save_errno;
858: err = MANDOCERR_SYSOPEN;
859: goto out;
1.82 schwarze 860: }
861:
1.98 schwarze 862: /* Run gunzip(1). */
863:
1.82 schwarze 864: if (pipe(pfd) == -1) {
865: err = MANDOCERR_SYSPIPE;
866: goto out;
867: }
868:
1.97 schwarze 869: switch (curp->child = fork()) {
1.82 schwarze 870: case -1:
871: err = MANDOCERR_SYSFORK;
872: close(pfd[0]);
873: close(pfd[1]);
874: pfd[1] = -1;
875: break;
876: case 0:
877: close(pfd[0]);
878: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
879: err = MANDOCERR_SYSDUP;
880: break;
881: }
882: execlp("gunzip", "gunzip", "-c", file, NULL);
883: err = MANDOCERR_SYSEXEC;
884: break;
885: default:
886: close(pfd[1]);
887: *fd = pfd[0];
888: return(MANDOCLEVEL_OK);
889: }
890:
891: out:
1.98 schwarze 892: free(cp);
1.82 schwarze 893: *fd = -1;
1.97 schwarze 894: curp->child = 0;
1.82 schwarze 895: curp->file_status = MANDOCLEVEL_SYSERR;
896: if (curp->mmsg)
1.98 schwarze 897: (*curp->mmsg)(err, curp->file_status, curp->file,
1.82 schwarze 898: 0, 0, strerror(errno));
899: if (pfd[1] != -1)
900: exit(1);
901: return(curp->file_status);
902: }
903:
904: enum mandoclevel
1.97 schwarze 905: mparse_wait(struct mparse *curp)
1.82 schwarze 906: {
907: int status;
908:
1.97 schwarze 909: if (curp->child == 0)
910: return(MANDOCLEVEL_OK);
911:
912: if (waitpid(curp->child, &status, 0) == -1) {
1.82 schwarze 913: mandoc_msg(MANDOCERR_SYSWAIT, curp, 0, 0,
914: strerror(errno));
915: curp->file_status = MANDOCLEVEL_SYSERR;
916: return(curp->file_status);
917: }
918: if (WIFSIGNALED(status)) {
919: mandoc_vmsg(MANDOCERR_SYSSIG, curp, 0, 0,
920: "%d", WTERMSIG(status));
921: curp->file_status = MANDOCLEVEL_SYSERR;
922: return(curp->file_status);
923: }
924: if (WEXITSTATUS(status)) {
925: mandoc_vmsg(MANDOCERR_SYSEXIT, curp, 0, 0,
926: "%d", WEXITSTATUS(status));
927: curp->file_status = MANDOCLEVEL_SYSERR;
928: return(curp->file_status);
929: }
930: return(MANDOCLEVEL_OK);
1.1 kristaps 931: }
932:
933: struct mparse *
1.94 schwarze 934: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
935: const struct mchars *mchars, const char *defos)
1.1 kristaps 936: {
937: struct mparse *curp;
1.10 kristaps 938:
939: assert(wlevel <= MANDOCLEVEL_FATAL);
1.1 kristaps 940:
941: curp = mandoc_calloc(1, sizeof(struct mparse));
942:
1.44 schwarze 943: curp->options = options;
1.3 kristaps 944: curp->wlevel = wlevel;
1.1 kristaps 945: curp->mmsg = mmsg;
1.29 schwarze 946: curp->defos = defos;
1.1 kristaps 947:
1.94 schwarze 948: curp->mchars = mchars;
949: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 950: if (curp->options & MPARSE_MDOC)
951: curp->pmdoc = mdoc_alloc(
952: curp->roff, curp, curp->defos,
953: curp->options & MPARSE_QUICK ? 1 : 0);
954: if (curp->options & MPARSE_MAN)
955: curp->pman = man_alloc(curp->roff, curp,
956: curp->options & MPARSE_QUICK ? 1 : 0);
957:
1.1 kristaps 958: return(curp);
959: }
960:
961: void
962: mparse_reset(struct mparse *curp)
963: {
964:
965: roff_reset(curp->roff);
966:
967: if (curp->mdoc)
968: mdoc_reset(curp->mdoc);
969: if (curp->man)
970: man_reset(curp->man);
1.24 kristaps 971: if (curp->secondary)
972: curp->secondary->sz = 0;
1.1 kristaps 973:
974: curp->file_status = MANDOCLEVEL_OK;
975: curp->mdoc = NULL;
976: curp->man = NULL;
1.45 schwarze 977:
978: free(curp->sodest);
979: curp->sodest = NULL;
1.1 kristaps 980: }
981:
982: void
983: mparse_free(struct mparse *curp)
984: {
985:
986: if (curp->pmdoc)
987: mdoc_free(curp->pmdoc);
988: if (curp->pman)
989: man_free(curp->pman);
990: if (curp->roff)
991: roff_free(curp->roff);
1.24 kristaps 992: if (curp->secondary)
993: free(curp->secondary->buf);
1.1 kristaps 994:
1.24 kristaps 995: free(curp->secondary);
1.45 schwarze 996: free(curp->sodest);
1.1 kristaps 997: free(curp);
998: }
999:
1000: void
1.45 schwarze 1001: mparse_result(struct mparse *curp,
1002: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 1003: {
1004:
1.45 schwarze 1005: if (sodest && NULL != (*sodest = curp->sodest)) {
1006: *mdoc = NULL;
1007: *man = NULL;
1008: return;
1009: }
1.9 kristaps 1010: if (mdoc)
1011: *mdoc = curp->mdoc;
1012: if (man)
1013: *man = curp->man;
1.3 kristaps 1014: }
1015:
1016: void
1017: mandoc_vmsg(enum mandocerr t, struct mparse *m,
1018: int ln, int pos, const char *fmt, ...)
1019: {
1020: char buf[256];
1021: va_list ap;
1022:
1023: va_start(ap, fmt);
1.48 schwarze 1024: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 1025: va_end(ap);
1026:
1027: mandoc_msg(t, m, ln, pos, buf);
1028: }
1029:
1030: void
1.47 schwarze 1031: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 1032: int ln, int col, const char *msg)
1033: {
1034: enum mandoclevel level;
1035:
1036: level = MANDOCLEVEL_FATAL;
1037: while (er < mandoclimits[level])
1038: level--;
1039:
1040: if (level < m->wlevel)
1041: return;
1042:
1.8 kristaps 1043: if (m->mmsg)
1044: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 1045:
1046: if (m->file_status < level)
1047: m->file_status = level;
1.7 kristaps 1048: }
1049:
1050: const char *
1051: mparse_strerror(enum mandocerr er)
1052: {
1053:
1054: return(mandocerrs[er]);
1055: }
1056:
1057: const char *
1058: mparse_strlevel(enum mandoclevel lvl)
1059: {
1060: return(mandoclevels[lvl]);
1.24 kristaps 1061: }
1062:
1063: void
1064: mparse_keep(struct mparse *p)
1065: {
1066:
1067: assert(NULL == p->secondary);
1068: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1069: }
1070:
1071: const char *
1072: mparse_getkeep(const struct mparse *p)
1073: {
1074:
1075: assert(p->secondary);
1076: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1077: }
CVSweb