Annotation of mandoc/read.c, Revision 1.133
1.133 ! schwarze 1: /* $Id: read.c,v 1.132 2015/03/17 07:33:07 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: *
1.133 ! schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 kristaps 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.133 ! schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 kristaps 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>
1.132 schwarze 32: #include <signal.h>
1.3 kristaps 33: #include <stdarg.h>
1.28 joerg 34: #include <stdint.h>
1.1 kristaps 35: #include <stdio.h>
36: #include <stdlib.h>
37: #include <string.h>
38: #include <unistd.h>
39:
1.133 ! schwarze 40: #include "mandoc_aux.h"
1.1 kristaps 41: #include "mandoc.h"
1.133 ! schwarze 42: #include "roff.h"
1.1 kristaps 43: #include "mdoc.h"
44: #include "man.h"
1.133 ! schwarze 45: #include "libmandoc.h"
1.1 kristaps 46:
47: #define REPARSE_LIMIT 1000
48:
49: struct mparse {
50: struct man *pman; /* persistent man parser */
51: struct mdoc *pmdoc; /* persistent mdoc parser */
52: struct man *man; /* man parser */
53: struct mdoc *mdoc; /* mdoc parser */
54: struct roff *roff; /* roff parser (!NULL) */
1.94 schwarze 55: const struct mchars *mchars; /* character table */
1.45 schwarze 56: char *sodest; /* filename pointed to by .so */
1.83 schwarze 57: const char *file; /* filename of current input file */
58: struct buf *primary; /* buffer currently being parsed */
59: struct buf *secondary; /* preprocessed copy of input */
60: const char *defos; /* default operating system */
61: mandocmsg mmsg; /* warning/error message handler */
62: enum mandoclevel file_status; /* status of current parse */
63: enum mandoclevel wlevel; /* ignore messages below this */
64: int options; /* parser options */
1.93 schwarze 65: int filenc; /* encoding of the current file */
1.1 kristaps 66: int reparse_count; /* finite interp. stack */
1.83 schwarze 67: int line; /* line number in the file */
1.97 schwarze 68: pid_t child; /* the gunzip(1) process */
1.1 kristaps 69: };
70:
1.84 schwarze 71: static void choose_parser(struct mparse *);
1.1 kristaps 72: static void resize_buf(struct buf *, size_t);
1.95 schwarze 73: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 74: static int read_whole_file(struct mparse *, const char *, int,
75: struct buf *, int *);
1.1 kristaps 76: static void mparse_end(struct mparse *);
1.37 schwarze 77: static void mparse_parse_buffer(struct mparse *, struct buf,
78: const char *);
1.1 kristaps 79:
1.3 kristaps 80: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
81: MANDOCERR_OK,
82: MANDOCERR_WARNING,
83: MANDOCERR_WARNING,
84: MANDOCERR_ERROR,
1.112 schwarze 85: MANDOCERR_UNSUPP,
1.3 kristaps 86: MANDOCERR_MAX,
87: MANDOCERR_MAX
88: };
89:
1.7 kristaps 90: static const char * const mandocerrs[MANDOCERR_MAX] = {
91: "ok",
92:
93: "generic warning",
94:
95: /* related to the prologue */
1.79 schwarze 96: "missing manual title, using UNTITLED",
97: "missing manual title, using \"\"",
1.54 schwarze 98: "lower case character in document title",
1.79 schwarze 99: "missing manual section, using \"\"",
1.7 kristaps 100: "unknown manual section",
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\"",
1.128 schwarze 114: "NAME section without name",
115: "NAME section without description",
116: "description not at the end of NAME",
117: "bad NAME section content",
1.121 schwarze 118: "missing description line, using \"\"",
1.7 kristaps 119: "sections out of conventional order",
1.54 schwarze 120: "duplicate section title",
121: "unexpected section",
1.87 schwarze 122: "unusual Xr order",
123: "unusual Xr punctuation",
1.86 schwarze 124: "AUTHORS section without An macro",
1.7 kristaps 125:
126: /* related to macros and nesting */
1.55 schwarze 127: "obsolete macro",
1.102 schwarze 128: "macro neither callable nor escaped",
1.7 kristaps 129: "skipping paragraph macro",
1.31 schwarze 130: "moving paragraph macro out of list",
1.7 kristaps 131: "skipping no-space macro",
132: "blocks badly nested",
133: "nested displays are not portable",
1.57 schwarze 134: "moving content out of list",
135: ".Vt block has child macro",
1.78 schwarze 136: "fill mode already enabled, skipping",
137: "fill mode already disabled, skipping",
1.7 kristaps 138: "line scope broken",
139:
140: /* related to missing macro arguments */
1.58 schwarze 141: "skipping empty request",
142: "conditional request controls empty scope",
1.7 kristaps 143: "skipping empty macro",
1.124 schwarze 144: "empty block",
1.62 schwarze 145: "empty argument, using 0n",
1.60 schwarze 146: "missing display type, using -ragged",
147: "list type is not the first argument",
148: "missing -width in -tag list, using 8n",
1.78 schwarze 149: "missing utility name, using \"\"",
1.123 schwarze 150: "missing function name, using \"\"",
1.60 schwarze 151: "empty head in list item",
152: "empty list item",
1.61 schwarze 153: "missing font type, using \\fR",
154: "unknown font type, using \\fR",
1.103 schwarze 155: "nothing follows prefix",
1.122 schwarze 156: "empty reference block",
1.60 schwarze 157: "missing -std argument, adding it",
1.125 schwarze 158: "missing option string, using \"\"",
159: "missing resource identifier, using \"\"",
1.90 schwarze 160: "missing eqn box, using \"\"",
1.7 kristaps 161:
162: /* related to bad macro arguments */
1.64 schwarze 163: "unterminated quoted argument",
1.7 kristaps 164: "duplicate argument",
1.76 schwarze 165: "skipping duplicate argument",
1.63 schwarze 166: "skipping duplicate display type",
167: "skipping duplicate list type",
1.76 schwarze 168: "skipping -width argument",
1.126 schwarze 169: "wrong number of cells",
1.7 kristaps 170: "unknown AT&T UNIX version",
1.88 schwarze 171: "comma in function argument",
1.89 schwarze 172: "parenthesis in function name",
1.67 schwarze 173: "invalid content in Rs block",
1.63 schwarze 174: "invalid Boolean argument",
175: "unknown font, skipping request",
1.126 schwarze 176: "odd number of characters in request",
1.7 kristaps 177:
178: /* related to plain text */
1.64 schwarze 179: "blank line in fill mode, using .sp",
180: "tab in filled text",
181: "whitespace at end of input line",
1.7 kristaps 182: "bad comment style",
1.64 schwarze 183: "invalid escape sequence",
184: "undefined string, using \"\"",
1.16 kristaps 185:
1.118 schwarze 186: /* related to tables */
187: "tbl line starts with span",
188: "tbl column starts with span",
189: "skipping vertical bar in tbl layout",
190:
1.7 kristaps 191: "generic error",
192:
193: /* related to tables */
1.116 schwarze 194: "non-alphabetic character in tbl options",
195: "skipping unknown tbl option",
196: "missing tbl option argument",
197: "wrong tbl option argument size",
1.118 schwarze 198: "empty tbl layout",
199: "invalid character in tbl layout",
200: "unmatched parenthesis in tbl layout",
1.119 schwarze 201: "tbl without any data cells",
202: "ignoring data in spanned tbl cell",
203: "ignoring extra tbl data cells",
204: "data block open at end of tbl",
1.7 kristaps 205:
1.68 schwarze 206: /* related to document structure and macros */
1.108 schwarze 207: NULL,
1.7 kristaps 208: "input stack limit exceeded, infinite loop?",
209: "skipping bad character",
1.68 schwarze 210: "skipping unknown macro",
1.112 schwarze 211: "skipping insecure request",
1.73 schwarze 212: "skipping item outside list",
1.68 schwarze 213: "skipping column outside column list",
214: "skipping end of block that is not open",
1.115 schwarze 215: "fewer RS blocks open, skipping",
1.68 schwarze 216: "inserting missing end of block",
217: "appending missing end of block",
218:
219: /* related to request and macro arguments */
1.7 kristaps 220: "escaped character not allowed in a name",
1.99 schwarze 221: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 222: "missing list type, using -item",
1.70 schwarze 223: "missing manual name, using \"\"",
1.71 schwarze 224: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 225: "unknown standard specifier",
1.71 schwarze 226: "skipping request without numeric argument",
1.109 schwarze 227: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 228: ".so request failed",
1.61 schwarze 229: "skipping all arguments",
230: "skipping excess arguments",
1.92 kristaps 231: "divide by zero",
1.112 schwarze 232:
233: "unsupported feature",
234: "input too large",
1.114 schwarze 235: "unsupported control character",
1.112 schwarze 236: "unsupported roff request",
1.119 schwarze 237: "eqn delim option in tbl",
1.118 schwarze 238: "unsupported tbl layout modifier",
1.112 schwarze 239: "ignoring macro in table",
1.7 kristaps 240: };
241:
242: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
243: "SUCCESS",
244: "RESERVED",
245: "WARNING",
246: "ERROR",
1.112 schwarze 247: "UNSUPP",
1.7 kristaps 248: "BADARG",
249: "SYSERR"
250: };
251:
1.47 schwarze 252:
1.1 kristaps 253: static void
254: resize_buf(struct buf *buf, size_t initial)
255: {
256:
257: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
258: buf->buf = mandoc_realloc(buf->buf, buf->sz);
259: }
260:
261: static void
1.84 schwarze 262: choose_parser(struct mparse *curp)
1.1 kristaps 263: {
1.83 schwarze 264: char *cp, *ep;
265: int format;
1.1 kristaps 266:
1.83 schwarze 267: /*
268: * If neither command line arguments -mdoc or -man select
269: * a parser nor the roff parser found a .Dd or .TH macro
270: * yet, look ahead in the main input buffer.
271: */
272:
273: if ((format = roff_getformat(curp->roff)) == 0) {
274: cp = curp->primary->buf;
275: ep = cp + curp->primary->sz;
276: while (cp < ep) {
1.85 schwarze 277: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 278: cp++;
279: if (cp[0] == 'D' && cp[1] == 'd') {
280: format = MPARSE_MDOC;
281: break;
282: }
283: if (cp[0] == 'T' && cp[1] == 'H') {
284: format = MPARSE_MAN;
285: break;
286: }
287: }
288: cp = memchr(cp, '\n', ep - cp);
289: if (cp == NULL)
290: break;
291: cp++;
292: }
1.1 kristaps 293: }
294:
1.83 schwarze 295: if (format == MPARSE_MDOC) {
1.47 schwarze 296: if (NULL == curp->pmdoc)
1.44 schwarze 297: curp->pmdoc = mdoc_alloc(
298: curp->roff, curp, curp->defos,
299: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 300: assert(curp->pmdoc);
301: curp->mdoc = curp->pmdoc;
302: return;
1.47 schwarze 303: }
1.1 kristaps 304:
1.83 schwarze 305: /* Fall back to man(7) as a last resort. */
306:
1.47 schwarze 307: if (NULL == curp->pman)
1.106 schwarze 308: curp->pman = man_alloc(
309: curp->roff, curp, curp->defos,
1.44 schwarze 310: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 311: assert(curp->pman);
312: curp->man = curp->pman;
313: }
314:
315: /*
1.95 schwarze 316: * Main parse routine for a buffer.
317: * It assumes encoding and line numbering are already set up.
318: * It can recurse directly (for invocations of user-defined
319: * macros, inline equations, and input line traps)
320: * and indirectly (for .so file inclusion).
1.1 kristaps 321: */
322: static void
1.95 schwarze 323: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 324: {
325: const struct tbl_span *span;
326: struct buf ln;
1.113 schwarze 327: const char *save_file;
1.110 schwarze 328: char *cp;
1.95 schwarze 329: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 330: enum rofferr rr;
1.100 schwarze 331: int of;
1.1 kristaps 332: int lnn; /* line number in the real file */
1.110 schwarze 333: int fd;
334: pid_t save_child;
1.1 kristaps 335: unsigned char c;
336:
1.95 schwarze 337: memset(&ln, 0, sizeof(ln));
1.1 kristaps 338:
1.47 schwarze 339: lnn = curp->line;
340: pos = 0;
1.1 kristaps 341:
1.95 schwarze 342: while (i < blk.sz) {
1.1 kristaps 343: if (0 == pos && '\0' == blk.buf[i])
344: break;
345:
346: if (start) {
347: curp->line = lnn;
348: curp->reparse_count = 0;
1.93 schwarze 349:
350: if (lnn < 3 &&
351: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 352: curp->filenc & MPARSE_LATIN1)
353: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 354: }
355:
1.95 schwarze 356: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 357:
358: /*
359: * When finding an unescaped newline character,
360: * leave the character loop to process the line.
361: * Skip a preceding carriage return, if any.
362: */
363:
1.95 schwarze 364: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 365: '\n' == blk.buf[i + 1])
366: ++i;
367: if ('\n' == blk.buf[i]) {
368: ++i;
369: ++lnn;
370: break;
371: }
372:
1.35 schwarze 373: /*
1.93 schwarze 374: * Make sure we have space for the worst
375: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 376: */
377:
1.95 schwarze 378: if (pos + 11 > ln.sz)
1.35 schwarze 379: resize_buf(&ln, 256);
380:
1.47 schwarze 381: /*
1.93 schwarze 382: * Encode 8-bit input.
1.1 kristaps 383: */
384:
1.93 schwarze 385: c = blk.buf[i];
386: if (c & 0x80) {
1.95 schwarze 387: if ( ! (curp->filenc && preconv_encode(
388: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 389: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
390: curp->line, pos, "0x%x", c);
1.93 schwarze 391: ln.buf[pos++] = '?';
392: i++;
393: }
394: continue;
395: }
396:
397: /*
398: * Exclude control characters.
399: */
1.1 kristaps 400:
1.93 schwarze 401: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 402: mandoc_vmsg(c == 0x00 || c == 0x04 ||
403: c > 0x0a ? MANDOCERR_CHAR_BAD :
404: MANDOCERR_CHAR_UNSUPP,
405: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 406: i++;
1.127 schwarze 407: if (c != '\r')
408: ln.buf[pos++] = '?';
1.1 kristaps 409: continue;
410: }
411:
412: /* Trailing backslash = a plain char. */
413:
1.95 schwarze 414: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 415: ln.buf[pos++] = blk.buf[i++];
416: continue;
417: }
418:
419: /*
420: * Found escape and at least one other character.
421: * When it's a newline character, skip it.
422: * When there is a carriage return in between,
423: * skip that one as well.
424: */
425:
1.95 schwarze 426: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 427: '\n' == blk.buf[i + 2])
428: ++i;
429: if ('\n' == blk.buf[i + 1]) {
430: i += 2;
431: ++lnn;
432: continue;
433: }
434:
1.13 kristaps 435: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 436: i += 2;
437: /* Comment, skip to end of line */
1.95 schwarze 438: for (; i < blk.sz; ++i) {
1.1 kristaps 439: if ('\n' == blk.buf[i]) {
440: ++i;
441: ++lnn;
442: break;
443: }
444: }
445:
446: /* Backout trailing whitespaces */
447: for (; pos > 0; --pos) {
448: if (ln.buf[pos - 1] != ' ')
449: break;
450: if (pos > 2 && ln.buf[pos - 2] == '\\')
451: break;
452: }
453: break;
454: }
455:
1.35 schwarze 456: /* Catch escaped bogus characters. */
457:
458: c = (unsigned char) blk.buf[i+1];
459:
1.47 schwarze 460: if ( ! (isascii(c) &&
461: (isgraph(c) || isblank(c)))) {
1.114 schwarze 462: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 463: curp->line, pos, "0x%x", c);
1.35 schwarze 464: i += 2;
465: ln.buf[pos++] = '?';
466: continue;
467: }
468:
1.1 kristaps 469: /* Some other escape sequence, copy & cont. */
470:
471: ln.buf[pos++] = blk.buf[i++];
472: ln.buf[pos++] = blk.buf[i++];
473: }
474:
1.95 schwarze 475: if (pos >= ln.sz)
1.1 kristaps 476: resize_buf(&ln, 256);
477:
478: ln.buf[pos] = '\0';
479:
480: /*
481: * A significant amount of complexity is contained by
482: * the roff preprocessor. It's line-oriented but can be
483: * expressed on one line, so we need at times to
484: * readjust our starting point and re-run it. The roff
485: * preprocessor can also readjust the buffers with new
486: * data, so we pass them in wholesale.
487: */
488:
489: of = 0;
490:
1.24 kristaps 491: /*
492: * Maintain a lookaside buffer of all parsed lines. We
493: * only do this if mparse_keep() has been invoked (the
494: * buffer may be accessed with mparse_getkeep()).
495: */
496:
497: if (curp->secondary) {
1.47 schwarze 498: curp->secondary->buf = mandoc_realloc(
499: curp->secondary->buf,
500: curp->secondary->sz + pos + 2);
501: memcpy(curp->secondary->buf +
502: curp->secondary->sz,
503: ln.buf, pos);
1.24 kristaps 504: curp->secondary->sz += pos;
505: curp->secondary->buf
506: [curp->secondary->sz] = '\n';
507: curp->secondary->sz++;
508: curp->secondary->buf
509: [curp->secondary->sz] = '\0';
510: }
1.1 kristaps 511: rerun:
1.96 schwarze 512: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 513:
514: switch (rr) {
1.47 schwarze 515: case ROFF_REPARSE:
1.1 kristaps 516: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 517: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 518: else
1.3 kristaps 519: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 520: curp->line, pos, NULL);
1.1 kristaps 521: pos = 0;
522: continue;
1.47 schwarze 523: case ROFF_APPEND:
1.95 schwarze 524: pos = strlen(ln.buf);
1.1 kristaps 525: continue;
1.47 schwarze 526: case ROFF_RERUN:
1.1 kristaps 527: goto rerun;
1.47 schwarze 528: case ROFF_IGN:
1.1 kristaps 529: pos = 0;
530: continue;
1.47 schwarze 531: case ROFF_SO:
1.95 schwarze 532: if ( ! (curp->options & MPARSE_SO) &&
533: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 534: curp->sodest = mandoc_strdup(ln.buf + of);
535: free(ln.buf);
536: return;
537: }
1.24 kristaps 538: /*
539: * We remove `so' clauses from our lookaside
540: * buffer because we're going to descend into
541: * the file recursively.
542: */
1.47 schwarze 543: if (curp->secondary)
1.25 kristaps 544: curp->secondary->sz -= pos + 1;
1.113 schwarze 545: save_file = curp->file;
1.110 schwarze 546: save_child = curp->child;
547: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 schwarze 548: MANDOCLEVEL_OK) {
1.110 schwarze 549: mparse_readfd(curp, fd, ln.buf + of);
1.113 schwarze 550: curp->file = save_file;
551: } else {
552: curp->file = save_file;
1.52 schwarze 553: mandoc_vmsg(MANDOCERR_SO_FAIL,
554: curp, curp->line, pos,
555: ".so %s", ln.buf + of);
1.110 schwarze 556: ln.sz = mandoc_asprintf(&cp,
557: ".sp\nSee the file %s.\n.sp",
558: ln.buf + of);
559: free(ln.buf);
560: ln.buf = cp;
561: of = 0;
562: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 563: }
1.110 schwarze 564: curp->child = save_child;
1.1 kristaps 565: pos = 0;
566: continue;
567: default:
568: break;
569: }
570:
571: /*
572: * If input parsers have not been allocated, do so now.
1.14 kristaps 573: * We keep these instanced between parsers, but set them
1.1 kristaps 574: * locally per parse routine since we can use different
575: * parsers with each one.
576: */
577:
578: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 579: choose_parser(curp);
1.1 kristaps 580:
1.47 schwarze 581: /*
1.84 schwarze 582: * Lastly, push down into the parsers themselves.
1.1 kristaps 583: * If libroff returns ROFF_TBL, then add it to the
584: * currently open parse. Since we only get here if
585: * there does exist data (see tbl_data.c), we're
586: * guaranteed that something's been allocated.
587: * Do the same for ROFF_EQN.
588: */
589:
1.100 schwarze 590: if (rr == ROFF_TBL) {
591: while ((span = roff_span(curp->roff)) != NULL)
592: if (curp->man == NULL)
593: mdoc_addspan(curp->mdoc, span);
594: else
595: man_addspan(curp->man, span);
596: } else if (rr == ROFF_EQN) {
597: if (curp->man == NULL)
598: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
599: else
600: man_addeqn(curp->man, roff_eqn(curp->roff));
601: } else if ((curp->man == NULL ?
602: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
603: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
604: break;
1.1 kristaps 605:
606: /* Temporary buffers typically are not full. */
607:
608: if (0 == start && '\0' == blk.buf[i])
609: break;
610:
611: /* Start the next input line. */
612:
613: pos = 0;
614: }
615:
616: free(ln.buf);
617: }
618:
619: static int
1.40 schwarze 620: read_whole_file(struct mparse *curp, const char *file, int fd,
621: struct buf *fb, int *with_mmap)
1.1 kristaps 622: {
623: size_t off;
624: ssize_t ssz;
625:
1.81 schwarze 626: #if HAVE_MMAP
1.15 kristaps 627: struct stat st;
1.1 kristaps 628: if (-1 == fstat(fd, &st)) {
1.107 schwarze 629: perror(file);
630: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 631: }
632:
633: /*
634: * If we're a regular file, try just reading in the whole entry
635: * via mmap(). This is faster than reading it into blocks, and
636: * since each file is only a few bytes to begin with, I'm not
637: * concerned that this is going to tank any machines.
638: */
639:
640: if (S_ISREG(st.st_mode)) {
1.131 schwarze 641: if (st.st_size > 0x7fffffff) {
1.111 schwarze 642: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 643: return(0);
644: }
645: *with_mmap = 1;
646: fb->sz = (size_t)st.st_size;
1.37 schwarze 647: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 648: if (fb->buf != MAP_FAILED)
649: return(1);
650: }
1.15 kristaps 651: #endif
1.1 kristaps 652:
653: /*
654: * If this isn't a regular file (like, say, stdin), then we must
655: * go the old way and just read things in bit by bit.
656: */
657:
658: *with_mmap = 0;
659: off = 0;
660: fb->sz = 0;
661: fb->buf = NULL;
662: for (;;) {
663: if (off == fb->sz) {
664: if (fb->sz == (1U << 31)) {
1.111 schwarze 665: mandoc_msg(MANDOCERR_TOOLARGE, curp,
666: 0, 0, NULL);
1.1 kristaps 667: break;
668: }
669: resize_buf(fb, 65536);
670: }
671: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
672: if (ssz == 0) {
673: fb->sz = off;
674: return(1);
675: }
676: if (ssz == -1) {
1.107 schwarze 677: perror(file);
678: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 679: }
680: off += (size_t)ssz;
681: }
682:
683: free(fb->buf);
684: fb->buf = NULL;
685: return(0);
686: }
687:
688: static void
689: mparse_end(struct mparse *curp)
690: {
691:
1.72 schwarze 692: if (curp->mdoc == NULL &&
693: curp->man == NULL &&
694: curp->sodest == NULL) {
695: if (curp->options & MPARSE_MDOC)
696: curp->mdoc = curp->pmdoc;
697: else {
698: if (curp->pman == NULL)
1.106 schwarze 699: curp->pman = man_alloc(
700: curp->roff, curp, curp->defos,
1.72 schwarze 701: curp->options & MPARSE_QUICK ? 1 : 0);
702: curp->man = curp->pman;
703: }
704: }
1.111 schwarze 705: if (curp->mdoc)
706: mdoc_endparse(curp->mdoc);
707: if (curp->man)
708: man_endparse(curp->man);
1.1 kristaps 709: roff_endparse(curp->roff);
710: }
711:
712: static void
1.36 schwarze 713: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 714: {
1.85 schwarze 715: struct buf *svprimary;
1.28 joerg 716: const char *svfile;
1.95 schwarze 717: size_t offset;
1.36 schwarze 718: static int recursion_depth;
719:
720: if (64 < recursion_depth) {
721: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
722: return;
723: }
1.28 joerg 724:
725: /* Line number is per-file. */
726: svfile = curp->file;
727: curp->file = file;
1.85 schwarze 728: svprimary = curp->primary;
1.83 schwarze 729: curp->primary = &blk;
1.28 joerg 730: curp->line = 1;
1.36 schwarze 731: recursion_depth++;
1.28 joerg 732:
1.93 schwarze 733: /* Skip an UTF-8 byte order mark. */
734: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
735: (unsigned char)blk.buf[0] == 0xef &&
736: (unsigned char)blk.buf[1] == 0xbb &&
737: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 738: offset = 3;
1.93 schwarze 739: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 740: } else
741: offset = 0;
1.93 schwarze 742:
1.95 schwarze 743: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 744:
1.111 schwarze 745: if (--recursion_depth == 0)
1.28 joerg 746: mparse_end(curp);
747:
1.85 schwarze 748: curp->primary = svprimary;
1.28 joerg 749: curp->file = svfile;
750: }
751:
752: enum mandoclevel
1.104 schwarze 753: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 754: const char *file)
755: {
756: struct buf blk;
757:
1.104 schwarze 758: blk.buf = buf;
1.28 joerg 759: blk.sz = len;
760:
1.36 schwarze 761: mparse_parse_buffer(curp, blk, file);
1.28 joerg 762: return(curp->file_status);
763: }
764:
1.98 schwarze 765: /*
766: * Read the whole file into memory and call the parsers.
767: * Called recursively when an .so request is encountered.
768: */
1.36 schwarze 769: enum mandoclevel
770: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 771: {
1.28 joerg 772: struct buf blk;
773: int with_mmap;
1.93 schwarze 774: int save_filenc;
1.1 kristaps 775:
1.91 schwarze 776: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 777: save_filenc = curp->filenc;
778: curp->filenc = curp->options &
779: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 780: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 781: curp->filenc = save_filenc;
1.81 schwarze 782: #if HAVE_MMAP
1.91 schwarze 783: if (with_mmap)
784: munmap(blk.buf, blk.sz);
785: else
1.28 joerg 786: #endif
1.91 schwarze 787: free(blk.buf);
788: }
1.1 kristaps 789:
1.98 schwarze 790: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 791: perror(file);
1.91 schwarze 792:
1.98 schwarze 793: mparse_wait(curp);
1.1 kristaps 794: return(curp->file_status);
1.82 schwarze 795: }
796:
797: enum mandoclevel
1.97 schwarze 798: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 799: {
800: int pfd[2];
1.98 schwarze 801: int save_errno;
1.82 schwarze 802: char *cp;
803:
804: curp->file = file;
1.98 schwarze 805:
806: /* Unless zipped, try to just open the file. */
807:
1.82 schwarze 808: if ((cp = strrchr(file, '.')) == NULL ||
809: strcmp(cp + 1, "gz")) {
1.97 schwarze 810: curp->child = 0;
1.98 schwarze 811: if ((*fd = open(file, O_RDONLY)) != -1)
812: return(MANDOCLEVEL_OK);
813:
814: /* Open failed; try to append ".gz". */
815:
816: mandoc_asprintf(&cp, "%s.gz", file);
817: file = cp;
818: } else
819: cp = NULL;
820:
821: /* Before forking, make sure the file can be read. */
822:
823: save_errno = errno;
824: if (access(file, R_OK) == -1) {
825: if (cp != NULL)
826: errno = save_errno;
1.108 schwarze 827: free(cp);
828: *fd = -1;
829: curp->child = 0;
830: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
831: return(MANDOCLEVEL_ERROR);
1.82 schwarze 832: }
833:
1.98 schwarze 834: /* Run gunzip(1). */
835:
1.82 schwarze 836: if (pipe(pfd) == -1) {
1.107 schwarze 837: perror("pipe");
838: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 839: }
840:
1.97 schwarze 841: switch (curp->child = fork()) {
1.82 schwarze 842: case -1:
1.107 schwarze 843: perror("fork");
844: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 845: case 0:
846: close(pfd[0]);
847: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 848: perror("dup");
849: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 850: }
1.132 schwarze 851: signal(SIGPIPE, SIG_DFL);
1.82 schwarze 852: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 853: perror("exec");
854: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 855: default:
856: close(pfd[1]);
857: *fd = pfd[0];
858: return(MANDOCLEVEL_OK);
859: }
860: }
861:
862: enum mandoclevel
1.97 schwarze 863: mparse_wait(struct mparse *curp)
1.82 schwarze 864: {
865: int status;
866:
1.97 schwarze 867: if (curp->child == 0)
868: return(MANDOCLEVEL_OK);
869:
870: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 871: perror("wait");
872: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 873: }
1.129 schwarze 874: curp->child = 0;
1.82 schwarze 875: if (WIFSIGNALED(status)) {
1.108 schwarze 876: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
877: "gunzip died from signal %d", WTERMSIG(status));
878: return(MANDOCLEVEL_ERROR);
1.82 schwarze 879: }
880: if (WEXITSTATUS(status)) {
1.108 schwarze 881: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
882: "gunzip failed with code %d", WEXITSTATUS(status));
883: return(MANDOCLEVEL_ERROR);
1.82 schwarze 884: }
885: return(MANDOCLEVEL_OK);
1.1 kristaps 886: }
887:
888: struct mparse *
1.94 schwarze 889: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
890: const struct mchars *mchars, const char *defos)
1.1 kristaps 891: {
892: struct mparse *curp;
1.10 kristaps 893:
1.1 kristaps 894: curp = mandoc_calloc(1, sizeof(struct mparse));
895:
1.44 schwarze 896: curp->options = options;
1.3 kristaps 897: curp->wlevel = wlevel;
1.1 kristaps 898: curp->mmsg = mmsg;
1.29 schwarze 899: curp->defos = defos;
1.1 kristaps 900:
1.94 schwarze 901: curp->mchars = mchars;
902: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 903: if (curp->options & MPARSE_MDOC)
904: curp->pmdoc = mdoc_alloc(
905: curp->roff, curp, curp->defos,
906: curp->options & MPARSE_QUICK ? 1 : 0);
907: if (curp->options & MPARSE_MAN)
1.106 schwarze 908: curp->pman = man_alloc(
909: curp->roff, curp, curp->defos,
1.72 schwarze 910: curp->options & MPARSE_QUICK ? 1 : 0);
911:
1.1 kristaps 912: return(curp);
913: }
914:
915: void
916: mparse_reset(struct mparse *curp)
917: {
918:
919: roff_reset(curp->roff);
920:
921: if (curp->mdoc)
922: mdoc_reset(curp->mdoc);
923: if (curp->man)
924: man_reset(curp->man);
1.24 kristaps 925: if (curp->secondary)
926: curp->secondary->sz = 0;
1.1 kristaps 927:
928: curp->file_status = MANDOCLEVEL_OK;
929: curp->mdoc = NULL;
930: curp->man = NULL;
1.45 schwarze 931:
932: free(curp->sodest);
933: curp->sodest = NULL;
1.1 kristaps 934: }
935:
936: void
937: mparse_free(struct mparse *curp)
938: {
939:
940: if (curp->pmdoc)
941: mdoc_free(curp->pmdoc);
942: if (curp->pman)
943: man_free(curp->pman);
944: if (curp->roff)
945: roff_free(curp->roff);
1.24 kristaps 946: if (curp->secondary)
947: free(curp->secondary->buf);
1.1 kristaps 948:
1.24 kristaps 949: free(curp->secondary);
1.45 schwarze 950: free(curp->sodest);
1.1 kristaps 951: free(curp);
952: }
953:
954: void
1.45 schwarze 955: mparse_result(struct mparse *curp,
956: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 957: {
958:
1.45 schwarze 959: if (sodest && NULL != (*sodest = curp->sodest)) {
960: *mdoc = NULL;
961: *man = NULL;
962: return;
963: }
1.9 kristaps 964: if (mdoc)
965: *mdoc = curp->mdoc;
966: if (man)
967: *man = curp->man;
1.3 kristaps 968: }
969:
970: void
971: mandoc_vmsg(enum mandocerr t, struct mparse *m,
972: int ln, int pos, const char *fmt, ...)
973: {
974: char buf[256];
975: va_list ap;
976:
977: va_start(ap, fmt);
1.48 schwarze 978: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 979: va_end(ap);
980:
981: mandoc_msg(t, m, ln, pos, buf);
982: }
983:
984: void
1.47 schwarze 985: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 986: int ln, int col, const char *msg)
987: {
988: enum mandoclevel level;
989:
1.112 schwarze 990: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 991: while (er < mandoclimits[level])
992: level--;
993:
1.108 schwarze 994: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 995: return;
996:
1.8 kristaps 997: if (m->mmsg)
998: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 999:
1000: if (m->file_status < level)
1001: m->file_status = level;
1.7 kristaps 1002: }
1003:
1004: const char *
1005: mparse_strerror(enum mandocerr er)
1006: {
1007:
1008: return(mandocerrs[er]);
1009: }
1010:
1011: const char *
1012: mparse_strlevel(enum mandoclevel lvl)
1013: {
1014: return(mandoclevels[lvl]);
1.24 kristaps 1015: }
1016:
1017: void
1018: mparse_keep(struct mparse *p)
1019: {
1020:
1021: assert(NULL == p->secondary);
1022: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1023: }
1024:
1025: const char *
1026: mparse_getkeep(const struct mparse *p)
1027: {
1028:
1029: assert(p->secondary);
1030: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1031: }
CVSweb