Annotation of mandoc/read.c, Revision 1.150.2.7
1.150.2.7! schwarze 1: /* $Id: read.c,v 1.150.2.6 2017/01/28 23:35:05 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.150.2.4 schwarze 4: * Copyright (c) 2010-2017 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
1.1 kristaps 26:
27: #include <assert.h>
28: #include <ctype.h>
1.146 schwarze 29: #if HAVE_ERR
1.143 schwarze 30: #include <err.h>
1.146 schwarze 31: #endif
1.40 schwarze 32: #include <errno.h>
1.1 kristaps 33: #include <fcntl.h>
1.3 kristaps 34: #include <stdarg.h>
1.28 joerg 35: #include <stdint.h>
1.1 kristaps 36: #include <stdio.h>
37: #include <stdlib.h>
38: #include <string.h>
39: #include <unistd.h>
1.140 schwarze 40: #include <zlib.h>
1.1 kristaps 41:
1.133 schwarze 42: #include "mandoc_aux.h"
1.1 kristaps 43: #include "mandoc.h"
1.133 schwarze 44: #include "roff.h"
1.1 kristaps 45: #include "mdoc.h"
46: #include "man.h"
1.133 schwarze 47: #include "libmandoc.h"
1.139 schwarze 48: #include "roff_int.h"
1.1 kristaps 49:
50: #define REPARSE_LIMIT 1000
51:
52: struct mparse {
53: struct roff *roff; /* roff parser (!NULL) */
1.150.2.7! schwarze 54: struct roff_man *man; /* man parser */
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 */
1.140 schwarze 64: int gzip; /* current input file is gzipped */
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 */
1.1 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,
1.112 schwarze 84: MANDOCERR_UNSUPP,
1.3 kristaps 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",
1.54 schwarze 100: "missing date, using today's date",
1.7 kristaps 101: "cannot parse date, using it verbatim",
1.79 schwarze 102: "missing Os macro, using \"\"",
103: "duplicate prologue macro",
104: "late prologue macro",
105: "skipping late title macro",
1.7 kristaps 106: "prologue macros out of order",
107:
108: /* related to document structure */
109: ".so is fragile, better use ln(1)",
1.50 schwarze 110: "no document body",
1.54 schwarze 111: "content before first section header",
112: "first section is not \"NAME\"",
1.150.2.4 schwarze 113: "NAME section without Nm before Nd",
1.128 schwarze 114: "NAME section without description",
115: "description not at the end of NAME",
116: "bad NAME section content",
1.150.2.4 schwarze 117: "missing comma before name",
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",
1.78 schwarze 135: "fill mode already enabled, skipping",
136: "fill mode already disabled, skipping",
1.7 kristaps 137: "line scope broken",
138:
139: /* related to missing macro arguments */
1.58 schwarze 140: "skipping empty request",
141: "conditional request controls empty scope",
1.7 kristaps 142: "skipping empty macro",
1.124 schwarze 143: "empty block",
1.62 schwarze 144: "empty argument, using 0n",
1.60 schwarze 145: "missing display type, using -ragged",
146: "list type is not the first argument",
1.150.2.1 schwarze 147: "missing -width in -tag list, using 6n",
1.78 schwarze 148: "missing utility name, using \"\"",
1.123 schwarze 149: "missing function name, using \"\"",
1.60 schwarze 150: "empty head in list item",
151: "empty list item",
1.61 schwarze 152: "missing font type, using \\fR",
153: "unknown font type, using \\fR",
1.103 schwarze 154: "nothing follows prefix",
1.122 schwarze 155: "empty reference block",
1.150.2.4 schwarze 156: "missing section argument",
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",
1.150.2.6 schwarze 182: "new sentence, new line",
1.7 kristaps 183: "bad comment style",
1.64 schwarze 184: "invalid escape sequence",
185: "undefined string, using \"\"",
1.16 kristaps 186:
1.118 schwarze 187: /* related to tables */
188: "tbl line starts with span",
189: "tbl column starts with span",
190: "skipping vertical bar in tbl layout",
191:
1.7 kristaps 192: "generic error",
193:
194: /* related to tables */
1.116 schwarze 195: "non-alphabetic character in tbl options",
196: "skipping unknown tbl option",
197: "missing tbl option argument",
198: "wrong tbl option argument size",
1.118 schwarze 199: "empty tbl layout",
200: "invalid character in tbl layout",
201: "unmatched parenthesis in tbl layout",
1.119 schwarze 202: "tbl without any data cells",
203: "ignoring data in spanned tbl cell",
204: "ignoring extra tbl data cells",
205: "data block open at end of tbl",
1.7 kristaps 206:
1.68 schwarze 207: /* related to document structure and macros */
1.108 schwarze 208: NULL,
1.7 kristaps 209: "input stack limit exceeded, infinite loop?",
210: "skipping bad character",
1.68 schwarze 211: "skipping unknown macro",
1.112 schwarze 212: "skipping insecure request",
1.73 schwarze 213: "skipping item outside list",
1.68 schwarze 214: "skipping column outside column list",
215: "skipping end of block that is not open",
1.115 schwarze 216: "fewer RS blocks open, skipping",
1.68 schwarze 217: "inserting missing end of block",
218: "appending missing end of block",
219:
220: /* related to request and macro arguments */
1.7 kristaps 221: "escaped character not allowed in a name",
1.99 schwarze 222: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 223: "skipping display without arguments",
1.71 schwarze 224: "missing list type, using -item",
1.70 schwarze 225: "missing manual name, using \"\"",
1.71 schwarze 226: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 227: "unknown standard specifier",
1.71 schwarze 228: "skipping request without numeric argument",
1.109 schwarze 229: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 230: ".so request failed",
1.61 schwarze 231: "skipping all arguments",
232: "skipping excess arguments",
1.92 kristaps 233: "divide by zero",
1.112 schwarze 234:
235: "unsupported feature",
236: "input too large",
1.114 schwarze 237: "unsupported control character",
1.112 schwarze 238: "unsupported roff request",
1.119 schwarze 239: "eqn delim option in tbl",
1.118 schwarze 240: "unsupported tbl layout modifier",
1.112 schwarze 241: "ignoring macro in table",
1.7 kristaps 242: };
243:
244: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
245: "SUCCESS",
246: "RESERVED",
247: "WARNING",
248: "ERROR",
1.112 schwarze 249: "UNSUPP",
1.7 kristaps 250: "BADARG",
251: "SYSERR"
252: };
253:
1.47 schwarze 254:
1.1 kristaps 255: static void
256: resize_buf(struct buf *buf, size_t initial)
257: {
258:
259: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
260: buf->buf = mandoc_realloc(buf->buf, buf->sz);
261: }
262:
263: static void
1.84 schwarze 264: choose_parser(struct mparse *curp)
1.1 kristaps 265: {
1.83 schwarze 266: char *cp, *ep;
267: int format;
1.1 kristaps 268:
1.83 schwarze 269: /*
270: * If neither command line arguments -mdoc or -man select
271: * a parser nor the roff parser found a .Dd or .TH macro
272: * yet, look ahead in the main input buffer.
273: */
274:
275: if ((format = roff_getformat(curp->roff)) == 0) {
276: cp = curp->primary->buf;
277: ep = cp + curp->primary->sz;
278: while (cp < ep) {
1.85 schwarze 279: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 280: cp++;
281: if (cp[0] == 'D' && cp[1] == 'd') {
282: format = MPARSE_MDOC;
283: break;
284: }
285: if (cp[0] == 'T' && cp[1] == 'H') {
286: format = MPARSE_MAN;
287: break;
288: }
289: }
290: cp = memchr(cp, '\n', ep - cp);
291: if (cp == NULL)
292: break;
293: cp++;
294: }
1.1 kristaps 295: }
296:
1.83 schwarze 297: if (format == MPARSE_MDOC) {
1.136 schwarze 298: mdoc_hash_init();
1.137 schwarze 299: curp->man->macroset = MACROSET_MDOC;
1.138 schwarze 300: curp->man->first->tok = TOKEN_NONE;
1.137 schwarze 301: } else {
302: man_hash_init();
303: curp->man->macroset = MACROSET_MAN;
1.138 schwarze 304: curp->man->first->tok = TOKEN_NONE;
1.47 schwarze 305: }
1.1 kristaps 306: }
307:
308: /*
1.95 schwarze 309: * Main parse routine for a buffer.
310: * It assumes encoding and line numbering are already set up.
311: * It can recurse directly (for invocations of user-defined
312: * macros, inline equations, and input line traps)
313: * and indirectly (for .so file inclusion).
1.1 kristaps 314: */
315: static void
1.95 schwarze 316: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 317: {
318: const struct tbl_span *span;
319: struct buf ln;
1.113 schwarze 320: const char *save_file;
1.110 schwarze 321: char *cp;
1.95 schwarze 322: size_t pos; /* byte number in the ln buffer */
1.150.2.3 schwarze 323: size_t j; /* auxiliary byte number in the blk buffer */
1.1 kristaps 324: enum rofferr rr;
1.100 schwarze 325: int of;
1.1 kristaps 326: int lnn; /* line number in the real file */
1.110 schwarze 327: int fd;
1.1 kristaps 328: unsigned char c;
329:
1.95 schwarze 330: memset(&ln, 0, sizeof(ln));
1.1 kristaps 331:
1.47 schwarze 332: lnn = curp->line;
333: pos = 0;
1.1 kristaps 334:
1.95 schwarze 335: while (i < blk.sz) {
1.1 kristaps 336: if (0 == pos && '\0' == blk.buf[i])
337: break;
338:
339: if (start) {
340: curp->line = lnn;
341: curp->reparse_count = 0;
1.93 schwarze 342:
343: if (lnn < 3 &&
344: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 345: curp->filenc & MPARSE_LATIN1)
346: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 347: }
348:
1.95 schwarze 349: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 350:
351: /*
352: * When finding an unescaped newline character,
353: * leave the character loop to process the line.
354: * Skip a preceding carriage return, if any.
355: */
356:
1.95 schwarze 357: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 358: '\n' == blk.buf[i + 1])
359: ++i;
360: if ('\n' == blk.buf[i]) {
361: ++i;
362: ++lnn;
363: break;
364: }
365:
1.35 schwarze 366: /*
1.93 schwarze 367: * Make sure we have space for the worst
368: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 369: */
370:
1.95 schwarze 371: if (pos + 11 > ln.sz)
1.35 schwarze 372: resize_buf(&ln, 256);
373:
1.47 schwarze 374: /*
1.93 schwarze 375: * Encode 8-bit input.
1.1 kristaps 376: */
377:
1.93 schwarze 378: c = blk.buf[i];
379: if (c & 0x80) {
1.95 schwarze 380: if ( ! (curp->filenc && preconv_encode(
381: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 382: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
383: curp->line, pos, "0x%x", c);
1.93 schwarze 384: ln.buf[pos++] = '?';
385: i++;
386: }
387: continue;
388: }
389:
390: /*
391: * Exclude control characters.
392: */
1.1 kristaps 393:
1.93 schwarze 394: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 395: mandoc_vmsg(c == 0x00 || c == 0x04 ||
396: c > 0x0a ? MANDOCERR_CHAR_BAD :
397: MANDOCERR_CHAR_UNSUPP,
398: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 399: i++;
1.127 schwarze 400: if (c != '\r')
401: ln.buf[pos++] = '?';
1.1 kristaps 402: continue;
403: }
404:
405: /* Trailing backslash = a plain char. */
406:
1.95 schwarze 407: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 408: ln.buf[pos++] = blk.buf[i++];
409: continue;
410: }
411:
412: /*
413: * Found escape and at least one other character.
414: * When it's a newline character, skip it.
415: * When there is a carriage return in between,
416: * skip that one as well.
417: */
418:
1.95 schwarze 419: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 420: '\n' == blk.buf[i + 2])
421: ++i;
422: if ('\n' == blk.buf[i + 1]) {
423: i += 2;
424: ++lnn;
425: continue;
426: }
427:
1.13 kristaps 428: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.150.2.3 schwarze 429: j = i;
1.1 kristaps 430: i += 2;
431: /* Comment, skip to end of line */
1.95 schwarze 432: for (; i < blk.sz; ++i) {
1.150.2.2 schwarze 433: if (blk.buf[i] != '\n')
434: continue;
435: if (blk.buf[i - 1] == ' ' ||
436: blk.buf[i - 1] == '\t')
437: mandoc_msg(
438: MANDOCERR_SPACE_EOL,
439: curp, curp->line,
1.150.2.3 schwarze 440: pos + i-1 - j, NULL);
1.150.2.2 schwarze 441: ++i;
442: ++lnn;
443: break;
1.1 kristaps 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;
1.148 schwarze 546: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 547: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 548: close(fd);
1.113 schwarze 549: curp->file = save_file;
550: } else {
551: curp->file = save_file;
1.52 schwarze 552: mandoc_vmsg(MANDOCERR_SO_FAIL,
553: curp, curp->line, pos,
554: ".so %s", ln.buf + of);
1.110 schwarze 555: ln.sz = mandoc_asprintf(&cp,
556: ".sp\nSee the file %s.\n.sp",
557: ln.buf + of);
558: free(ln.buf);
559: ln.buf = cp;
560: of = 0;
561: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 562: }
1.1 kristaps 563: pos = 0;
564: continue;
565: default:
566: break;
567: }
568:
1.150 schwarze 569: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 570: choose_parser(curp);
1.1 kristaps 571:
1.47 schwarze 572: /*
1.84 schwarze 573: * Lastly, push down into the parsers themselves.
1.1 kristaps 574: * If libroff returns ROFF_TBL, then add it to the
575: * currently open parse. Since we only get here if
576: * there does exist data (see tbl_data.c), we're
577: * guaranteed that something's been allocated.
578: * Do the same for ROFF_EQN.
579: */
580:
1.139 schwarze 581: if (rr == ROFF_TBL)
1.100 schwarze 582: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 583: roff_addtbl(curp->man, span);
584: else if (rr == ROFF_EQN)
585: roff_addeqn(curp->man, roff_eqn(curp->roff));
586: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 587: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 588: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
589: break;
1.1 kristaps 590:
591: /* Temporary buffers typically are not full. */
592:
593: if (0 == start && '\0' == blk.buf[i])
594: break;
595:
596: /* Start the next input line. */
597:
598: pos = 0;
599: }
600:
601: free(ln.buf);
602: }
603:
604: static int
1.40 schwarze 605: read_whole_file(struct mparse *curp, const char *file, int fd,
606: struct buf *fb, int *with_mmap)
1.1 kristaps 607: {
1.140 schwarze 608: gzFile gz;
1.1 kristaps 609: size_t off;
610: ssize_t ssz;
611:
1.81 schwarze 612: #if HAVE_MMAP
1.15 kristaps 613: struct stat st;
1.143 schwarze 614:
615: if (fstat(fd, &st) == -1)
616: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 617:
618: /*
619: * If we're a regular file, try just reading in the whole entry
620: * via mmap(). This is faster than reading it into blocks, and
621: * since each file is only a few bytes to begin with, I'm not
622: * concerned that this is going to tank any machines.
623: */
624:
1.140 schwarze 625: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 626: if (st.st_size > 0x7fffffff) {
1.111 schwarze 627: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 628: return 0;
1.1 kristaps 629: }
630: *with_mmap = 1;
631: fb->sz = (size_t)st.st_size;
1.37 schwarze 632: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 633: if (fb->buf != MAP_FAILED)
1.142 schwarze 634: return 1;
1.1 kristaps 635: }
1.15 kristaps 636: #endif
1.1 kristaps 637:
1.140 schwarze 638: if (curp->gzip) {
1.143 schwarze 639: if ((gz = gzdopen(fd, "rb")) == NULL)
640: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 641: } else
642: gz = NULL;
643:
1.1 kristaps 644: /*
645: * If this isn't a regular file (like, say, stdin), then we must
646: * go the old way and just read things in bit by bit.
647: */
648:
649: *with_mmap = 0;
650: off = 0;
651: fb->sz = 0;
652: fb->buf = NULL;
653: for (;;) {
654: if (off == fb->sz) {
655: if (fb->sz == (1U << 31)) {
1.111 schwarze 656: mandoc_msg(MANDOCERR_TOOLARGE, curp,
657: 0, 0, NULL);
1.1 kristaps 658: break;
659: }
660: resize_buf(fb, 65536);
661: }
1.140 schwarze 662: ssz = curp->gzip ?
663: gzread(gz, fb->buf + (int)off, fb->sz - off) :
664: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 665: if (ssz == 0) {
666: fb->sz = off;
1.142 schwarze 667: return 1;
1.1 kristaps 668: }
1.143 schwarze 669: if (ssz == -1)
670: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 671: off += (size_t)ssz;
672: }
673:
674: free(fb->buf);
675: fb->buf = NULL;
1.142 schwarze 676: return 0;
1.1 kristaps 677: }
678:
679: static void
680: mparse_end(struct mparse *curp)
681: {
1.135 schwarze 682: if (curp->man->macroset == MACROSET_NONE)
683: curp->man->macroset = MACROSET_MAN;
684: if (curp->man->macroset == MACROSET_MDOC)
685: mdoc_endparse(curp->man);
686: else
1.111 schwarze 687: man_endparse(curp->man);
1.1 kristaps 688: roff_endparse(curp->roff);
689: }
690:
691: static void
1.36 schwarze 692: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 693: {
1.85 schwarze 694: struct buf *svprimary;
1.28 joerg 695: const char *svfile;
1.95 schwarze 696: size_t offset;
1.36 schwarze 697: static int recursion_depth;
698:
699: if (64 < recursion_depth) {
700: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
701: return;
702: }
1.28 joerg 703:
704: /* Line number is per-file. */
705: svfile = curp->file;
706: curp->file = file;
1.85 schwarze 707: svprimary = curp->primary;
1.83 schwarze 708: curp->primary = &blk;
1.28 joerg 709: curp->line = 1;
1.36 schwarze 710: recursion_depth++;
1.28 joerg 711:
1.93 schwarze 712: /* Skip an UTF-8 byte order mark. */
713: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
714: (unsigned char)blk.buf[0] == 0xef &&
715: (unsigned char)blk.buf[1] == 0xbb &&
716: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 717: offset = 3;
1.93 schwarze 718: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 719: } else
720: offset = 0;
1.93 schwarze 721:
1.95 schwarze 722: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 723:
1.111 schwarze 724: if (--recursion_depth == 0)
1.28 joerg 725: mparse_end(curp);
726:
1.85 schwarze 727: curp->primary = svprimary;
1.28 joerg 728: curp->file = svfile;
729: }
730:
731: enum mandoclevel
1.104 schwarze 732: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 733: const char *file)
734: {
735: struct buf blk;
736:
1.104 schwarze 737: blk.buf = buf;
1.28 joerg 738: blk.sz = len;
739:
1.36 schwarze 740: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 741: return curp->file_status;
1.28 joerg 742: }
743:
1.98 schwarze 744: /*
745: * Read the whole file into memory and call the parsers.
746: * Called recursively when an .so request is encountered.
747: */
1.36 schwarze 748: enum mandoclevel
749: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 750: {
1.28 joerg 751: struct buf blk;
752: int with_mmap;
1.93 schwarze 753: int save_filenc;
1.1 kristaps 754:
1.91 schwarze 755: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 756: save_filenc = curp->filenc;
757: curp->filenc = curp->options &
758: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 759: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 760: curp->filenc = save_filenc;
1.81 schwarze 761: #if HAVE_MMAP
1.91 schwarze 762: if (with_mmap)
763: munmap(blk.buf, blk.sz);
764: else
1.28 joerg 765: #endif
1.91 schwarze 766: free(blk.buf);
767: }
1.142 schwarze 768: return curp->file_status;
1.82 schwarze 769: }
770:
1.148 schwarze 771: int
772: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 773: {
774: char *cp;
1.148 schwarze 775: int fd;
1.82 schwarze 776:
777: curp->file = file;
1.140 schwarze 778: cp = strrchr(file, '.');
779: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 780:
1.140 schwarze 781: /* First try to use the filename as it is. */
1.98 schwarze 782:
1.148 schwarze 783: if ((fd = open(file, O_RDONLY)) != -1)
784: return fd;
1.98 schwarze 785:
1.140 schwarze 786: /*
787: * If that doesn't work and the filename doesn't
788: * already end in .gz, try appending .gz.
789: */
1.98 schwarze 790:
1.140 schwarze 791: if ( ! curp->gzip) {
1.98 schwarze 792: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 793: fd = open(cp, O_RDONLY);
1.108 schwarze 794: free(cp);
1.148 schwarze 795: if (fd != -1) {
1.140 schwarze 796: curp->gzip = 1;
1.148 schwarze 797: return fd;
1.82 schwarze 798: }
799: }
800:
1.140 schwarze 801: /* Neither worked, give up. */
1.97 schwarze 802:
1.140 schwarze 803: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 804: return -1;
1.1 kristaps 805: }
806:
807: struct mparse *
1.94 schwarze 808: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.144 schwarze 809: const char *defos)
1.1 kristaps 810: {
811: struct mparse *curp;
1.10 kristaps 812:
1.1 kristaps 813: curp = mandoc_calloc(1, sizeof(struct mparse));
814:
1.44 schwarze 815: curp->options = options;
1.3 kristaps 816: curp->wlevel = wlevel;
1.1 kristaps 817: curp->mmsg = mmsg;
1.29 schwarze 818: curp->defos = defos;
1.1 kristaps 819:
1.144 schwarze 820: curp->roff = roff_alloc(curp, options);
1.137 schwarze 821: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
822: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 823: if (curp->options & MPARSE_MDOC) {
824: mdoc_hash_init();
1.137 schwarze 825: curp->man->macroset = MACROSET_MDOC;
826: } else if (curp->options & MPARSE_MAN) {
1.136 schwarze 827: man_hash_init();
1.137 schwarze 828: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 829: }
1.138 schwarze 830: curp->man->first->tok = TOKEN_NONE;
1.142 schwarze 831: return curp;
1.1 kristaps 832: }
833:
834: void
835: mparse_reset(struct mparse *curp)
836: {
837: roff_reset(curp->roff);
1.150 schwarze 838: roff_man_reset(curp->man);
1.150.2.7! schwarze 839:
! 840: free(curp->sodest);
! 841: curp->sodest = NULL;
! 842:
1.24 kristaps 843: if (curp->secondary)
844: curp->secondary->sz = 0;
1.1 kristaps 845:
846: curp->file_status = MANDOCLEVEL_OK;
1.150.2.7! schwarze 847: curp->gzip = 0;
1.1 kristaps 848: }
849:
850: void
851: mparse_free(struct mparse *curp)
852: {
853:
1.137 schwarze 854: roff_man_free(curp->man);
1.150.2.7! schwarze 855: roff_free(curp->roff);
1.24 kristaps 856: if (curp->secondary)
857: free(curp->secondary->buf);
1.1 kristaps 858:
1.24 kristaps 859: free(curp->secondary);
1.45 schwarze 860: free(curp->sodest);
1.1 kristaps 861: free(curp);
862: }
863:
864: void
1.135 schwarze 865: mparse_result(struct mparse *curp, struct roff_man **man,
866: char **sodest)
1.1 kristaps 867: {
868:
1.45 schwarze 869: if (sodest && NULL != (*sodest = curp->sodest)) {
870: *man = NULL;
871: return;
872: }
1.9 kristaps 873: if (man)
874: *man = curp->man;
1.150.2.5 schwarze 875: }
876:
877: void
878: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
879: {
880: if (curp->file_status > *rc)
881: *rc = curp->file_status;
1.3 kristaps 882: }
883:
884: void
885: mandoc_vmsg(enum mandocerr t, struct mparse *m,
886: int ln, int pos, const char *fmt, ...)
887: {
888: char buf[256];
889: va_list ap;
890:
891: va_start(ap, fmt);
1.48 schwarze 892: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 893: va_end(ap);
894:
895: mandoc_msg(t, m, ln, pos, buf);
896: }
897:
898: void
1.47 schwarze 899: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 900: int ln, int col, const char *msg)
901: {
902: enum mandoclevel level;
903:
1.112 schwarze 904: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 905: while (er < mandoclimits[level])
906: level--;
907:
1.108 schwarze 908: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 909: return;
910:
1.8 kristaps 911: if (m->mmsg)
912: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 913:
914: if (m->file_status < level)
915: m->file_status = level;
1.7 kristaps 916: }
917:
918: const char *
919: mparse_strerror(enum mandocerr er)
920: {
921:
1.142 schwarze 922: return mandocerrs[er];
1.7 kristaps 923: }
924:
925: const char *
926: mparse_strlevel(enum mandoclevel lvl)
927: {
1.142 schwarze 928: return mandoclevels[lvl];
1.24 kristaps 929: }
930:
931: void
932: mparse_keep(struct mparse *p)
933: {
934:
935: assert(NULL == p->secondary);
936: p->secondary = mandoc_calloc(1, sizeof(struct buf));
937: }
938:
939: const char *
940: mparse_getkeep(const struct mparse *p)
941: {
942:
943: assert(p->secondary);
1.142 schwarze 944: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 945: }
CVSweb