Annotation of src/usr.bin/mandoc/read.c, Revision 1.143
1.143 ! schwarze 1: /* $OpenBSD: read.c,v 1.142 2017/06/04 00:08:56 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.129 schwarze 4: * Copyright (c) 2010-2017 Ingo Schwarze <schwarze@openbsd.org>
1.20 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 schwarze 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: *
1.108 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.108 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 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: */
1.58 schwarze 19: #include <sys/types.h>
20: #include <sys/mman.h>
1.1 schwarze 21: #include <sys/stat.h>
22:
23: #include <assert.h>
24: #include <ctype.h>
1.118 schwarze 25: #include <err.h>
1.18 schwarze 26: #include <errno.h>
1.1 schwarze 27: #include <fcntl.h>
28: #include <stdarg.h>
29: #include <stdio.h>
30: #include <stdlib.h>
31: #include <string.h>
32: #include <unistd.h>
1.115 schwarze 33: #include <zlib.h>
1.1 schwarze 34:
1.108 schwarze 35: #include "mandoc_aux.h"
1.1 schwarze 36: #include "mandoc.h"
1.108 schwarze 37: #include "roff.h"
1.1 schwarze 38: #include "mdoc.h"
39: #include "man.h"
1.108 schwarze 40: #include "libmandoc.h"
1.114 schwarze 41: #include "roff_int.h"
1.1 schwarze 42:
43: #define REPARSE_LIMIT 1000
44:
45: struct mparse {
1.133 schwarze 46: struct roff *roff; /* roff parser (!NULL) */
1.109 schwarze 47: struct roff_man *man; /* man parser */
1.23 schwarze 48: char *sodest; /* filename pointed to by .so */
1.59 schwarze 49: const char *file; /* filename of current input file */
50: struct buf *primary; /* buffer currently being parsed */
51: struct buf *secondary; /* preprocessed copy of input */
52: const char *defos; /* default operating system */
53: mandocmsg mmsg; /* warning/error message handler */
54: enum mandoclevel file_status; /* status of current parse */
55: enum mandoclevel wlevel; /* ignore messages below this */
56: int options; /* parser options */
1.115 schwarze 57: int gzip; /* current input file is gzipped */
1.70 schwarze 58: int filenc; /* encoding of the current file */
1.1 schwarze 59: int reparse_count; /* finite interp. stack */
1.59 schwarze 60: int line; /* line number in the file */
1.1 schwarze 61: };
62:
1.60 schwarze 63: static void choose_parser(struct mparse *);
1.1 schwarze 64: static void resize_buf(struct buf *, size_t);
1.135 schwarze 65: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.18 schwarze 66: static int read_whole_file(struct mparse *, const char *, int,
67: struct buf *, int *);
1.1 schwarze 68: static void mparse_end(struct mparse *);
1.15 schwarze 69: static void mparse_parse_buffer(struct mparse *, struct buf,
70: const char *);
1.1 schwarze 71:
72: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
73: MANDOCERR_OK,
1.137 schwarze 74: MANDOCERR_STYLE,
1.1 schwarze 75: MANDOCERR_WARNING,
76: MANDOCERR_ERROR,
1.87 schwarze 77: MANDOCERR_UNSUPP,
1.1 schwarze 78: MANDOCERR_MAX,
79: MANDOCERR_MAX
80: };
81:
82: static const char * const mandocerrs[MANDOCERR_MAX] = {
83: "ok",
84:
1.137 schwarze 85: "generic style suggestion",
1.138 schwarze 86:
87: "useless macro",
1.139 schwarze 88: "consider using OS macro",
1.140 schwarze 89: "description line ends with a full stop",
1.138 schwarze 90:
1.1 schwarze 91: "generic warning",
92:
93: /* related to the prologue */
1.57 schwarze 94: "missing manual title, using UNTITLED",
95: "missing manual title, using \"\"",
1.32 schwarze 96: "lower case character in document title",
1.57 schwarze 97: "missing manual section, using \"\"",
1.1 schwarze 98: "unknown manual section",
1.32 schwarze 99: "missing date, using today's date",
1.1 schwarze 100: "cannot parse date, using it verbatim",
1.57 schwarze 101: "missing Os macro, using \"\"",
102: "duplicate prologue macro",
103: "late prologue macro",
104: "skipping late title macro",
1.1 schwarze 105: "prologue macros out of order",
106:
107: /* related to document structure */
108: ".so is fragile, better use ln(1)",
1.28 schwarze 109: "no document body",
1.32 schwarze 110: "content before first section header",
111: "first section is not \"NAME\"",
1.129 schwarze 112: "NAME section without Nm before Nd",
1.103 schwarze 113: "NAME section without description",
114: "description not at the end of NAME",
115: "bad NAME section content",
1.129 schwarze 116: "missing comma before name",
1.96 schwarze 117: "missing description line, using \"\"",
1.134 schwarze 118: "description line outside NAME section",
1.1 schwarze 119: "sections out of conventional order",
1.32 schwarze 120: "duplicate section title",
121: "unexpected section",
1.63 schwarze 122: "unusual Xr order",
123: "unusual Xr punctuation",
1.62 schwarze 124: "AUTHORS section without An macro",
1.1 schwarze 125:
126: /* related to macros and nesting */
1.33 schwarze 127: "obsolete macro",
1.78 schwarze 128: "macro neither callable nor escaped",
1.1 schwarze 129: "skipping paragraph macro",
1.10 schwarze 130: "moving paragraph macro out of list",
1.1 schwarze 131: "skipping no-space macro",
132: "blocks badly nested",
133: "nested displays are not portable",
1.35 schwarze 134: "moving content out of list",
1.56 schwarze 135: "fill mode already enabled, skipping",
136: "fill mode already disabled, skipping",
1.1 schwarze 137: "line scope broken",
1.141 schwarze 138: "skipping blank line in line scope",
1.1 schwarze 139:
140: /* related to missing macro arguments */
1.36 schwarze 141: "skipping empty request",
142: "conditional request controls empty scope",
1.1 schwarze 143: "skipping empty macro",
1.99 schwarze 144: "empty block",
1.40 schwarze 145: "empty argument, using 0n",
1.38 schwarze 146: "missing display type, using -ragged",
147: "list type is not the first argument",
1.125 schwarze 148: "missing -width in -tag list, using 6n",
1.56 schwarze 149: "missing utility name, using \"\"",
1.98 schwarze 150: "missing function name, using \"\"",
1.38 schwarze 151: "empty head in list item",
152: "empty list item",
1.39 schwarze 153: "missing font type, using \\fR",
154: "unknown font type, using \\fR",
1.79 schwarze 155: "nothing follows prefix",
1.97 schwarze 156: "empty reference block",
1.128 schwarze 157: "missing section argument",
1.38 schwarze 158: "missing -std argument, adding it",
1.100 schwarze 159: "missing option string, using \"\"",
160: "missing resource identifier, using \"\"",
1.66 schwarze 161: "missing eqn box, using \"\"",
1.1 schwarze 162:
163: /* related to bad macro arguments */
1.42 schwarze 164: "unterminated quoted argument",
1.1 schwarze 165: "duplicate argument",
1.54 schwarze 166: "skipping duplicate argument",
1.41 schwarze 167: "skipping duplicate display type",
168: "skipping duplicate list type",
1.54 schwarze 169: "skipping -width argument",
1.101 schwarze 170: "wrong number of cells",
1.1 schwarze 171: "unknown AT&T UNIX version",
1.64 schwarze 172: "comma in function argument",
1.65 schwarze 173: "parenthesis in function name",
1.45 schwarze 174: "invalid content in Rs block",
1.41 schwarze 175: "invalid Boolean argument",
176: "unknown font, skipping request",
1.101 schwarze 177: "odd number of characters in request",
1.1 schwarze 178:
179: /* related to plain text */
1.42 schwarze 180: "blank line in fill mode, using .sp",
181: "tab in filled text",
182: "whitespace at end of input line",
1.131 schwarze 183: "new sentence, new line",
1.1 schwarze 184: "bad comment style",
1.42 schwarze 185: "invalid escape sequence",
186: "undefined string, using \"\"",
1.3 schwarze 187:
1.93 schwarze 188: /* related to tables */
189: "tbl line starts with span",
190: "tbl column starts with span",
191: "skipping vertical bar in tbl layout",
192:
1.1 schwarze 193: "generic error",
1.3 schwarze 194:
1.1 schwarze 195: /* related to tables */
1.91 schwarze 196: "non-alphabetic character in tbl options",
197: "skipping unknown tbl option",
198: "missing tbl option argument",
199: "wrong tbl option argument size",
1.93 schwarze 200: "empty tbl layout",
201: "invalid character in tbl layout",
202: "unmatched parenthesis in tbl layout",
1.94 schwarze 203: "tbl without any data cells",
204: "ignoring data in spanned tbl cell",
205: "ignoring extra tbl data cells",
206: "data block open at end of tbl",
1.1 schwarze 207:
1.46 schwarze 208: /* related to document structure and macros */
1.83 schwarze 209: NULL,
1.1 schwarze 210: "input stack limit exceeded, infinite loop?",
211: "skipping bad character",
1.46 schwarze 212: "skipping unknown macro",
1.87 schwarze 213: "skipping insecure request",
1.51 schwarze 214: "skipping item outside list",
1.46 schwarze 215: "skipping column outside column list",
216: "skipping end of block that is not open",
1.90 schwarze 217: "fewer RS blocks open, skipping",
1.46 schwarze 218: "inserting missing end of block",
219: "appending missing end of block",
220:
221: /* related to request and macro arguments */
1.1 schwarze 222: "escaped character not allowed in a name",
1.75 schwarze 223: "NOT IMPLEMENTED: Bd -file",
1.120 schwarze 224: "skipping display without arguments",
1.49 schwarze 225: "missing list type, using -item",
1.143 ! schwarze 226: "argument is not numeric, using 1",
1.48 schwarze 227: "missing manual name, using \"\"",
1.49 schwarze 228: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 229: "unknown standard specifier",
1.49 schwarze 230: "skipping request without numeric argument",
1.84 schwarze 231: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 232: ".so request failed",
1.39 schwarze 233: "skipping all arguments",
234: "skipping excess arguments",
1.68 schwarze 235: "divide by zero",
1.87 schwarze 236:
237: "unsupported feature",
238: "input too large",
1.89 schwarze 239: "unsupported control character",
1.87 schwarze 240: "unsupported roff request",
1.94 schwarze 241: "eqn delim option in tbl",
1.93 schwarze 242: "unsupported tbl layout modifier",
1.87 schwarze 243: "ignoring macro in table",
1.1 schwarze 244: };
245:
246: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
247: "SUCCESS",
1.137 schwarze 248: "STYLE",
1.1 schwarze 249: "WARNING",
250: "ERROR",
1.87 schwarze 251: "UNSUPP",
1.1 schwarze 252: "BADARG",
253: "SYSERR"
254: };
255:
1.25 schwarze 256:
1.1 schwarze 257: static void
258: resize_buf(struct buf *buf, size_t initial)
259: {
260:
261: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
262: buf->buf = mandoc_realloc(buf->buf, buf->sz);
263: }
264:
265: static void
1.60 schwarze 266: choose_parser(struct mparse *curp)
1.1 schwarze 267: {
1.59 schwarze 268: char *cp, *ep;
269: int format;
1.1 schwarze 270:
1.59 schwarze 271: /*
272: * If neither command line arguments -mdoc or -man select
273: * a parser nor the roff parser found a .Dd or .TH macro
274: * yet, look ahead in the main input buffer.
275: */
276:
277: if ((format = roff_getformat(curp->roff)) == 0) {
278: cp = curp->primary->buf;
279: ep = cp + curp->primary->sz;
280: while (cp < ep) {
1.61 schwarze 281: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 282: cp++;
283: if (cp[0] == 'D' && cp[1] == 'd') {
284: format = MPARSE_MDOC;
285: break;
286: }
287: if (cp[0] == 'T' && cp[1] == 'H') {
288: format = MPARSE_MAN;
289: break;
290: }
291: }
292: cp = memchr(cp, '\n', ep - cp);
293: if (cp == NULL)
294: break;
295: cp++;
296: }
1.1 schwarze 297: }
298:
1.59 schwarze 299: if (format == MPARSE_MDOC) {
1.112 schwarze 300: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 301: if (curp->man->mdocmac == NULL)
302: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 303: } else {
304: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 305: if (curp->man->manmac == NULL)
306: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 307: }
1.136 schwarze 308: curp->man->first->tok = TOKEN_NONE;
1.1 schwarze 309: }
310:
311: /*
1.71 schwarze 312: * Main parse routine for a buffer.
313: * It assumes encoding and line numbering are already set up.
314: * It can recurse directly (for invocations of user-defined
315: * macros, inline equations, and input line traps)
316: * and indirectly (for .so file inclusion).
1.1 schwarze 317: */
1.135 schwarze 318: static int
1.71 schwarze 319: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 320: {
321: const struct tbl_span *span;
322: struct buf ln;
1.88 schwarze 323: const char *save_file;
1.85 schwarze 324: char *cp;
1.71 schwarze 325: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 326: enum rofferr rr;
1.76 schwarze 327: int of;
1.1 schwarze 328: int lnn; /* line number in the real file */
1.85 schwarze 329: int fd;
1.1 schwarze 330: unsigned char c;
331:
1.71 schwarze 332: memset(&ln, 0, sizeof(ln));
1.1 schwarze 333:
1.25 schwarze 334: lnn = curp->line;
335: pos = 0;
1.1 schwarze 336:
1.71 schwarze 337: while (i < blk.sz) {
1.1 schwarze 338: if (0 == pos && '\0' == blk.buf[i])
339: break;
340:
341: if (start) {
342: curp->line = lnn;
343: curp->reparse_count = 0;
1.70 schwarze 344:
345: if (lnn < 3 &&
346: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 347: curp->filenc & MPARSE_LATIN1)
348: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 349: }
350:
1.71 schwarze 351: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 352:
353: /*
354: * When finding an unescaped newline character,
355: * leave the character loop to process the line.
356: * Skip a preceding carriage return, if any.
357: */
358:
1.71 schwarze 359: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 360: '\n' == blk.buf[i + 1])
361: ++i;
362: if ('\n' == blk.buf[i]) {
363: ++i;
364: ++lnn;
365: break;
366: }
367:
1.13 schwarze 368: /*
1.70 schwarze 369: * Make sure we have space for the worst
370: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 371: */
372:
1.71 schwarze 373: if (pos + 11 > ln.sz)
1.13 schwarze 374: resize_buf(&ln, 256);
375:
1.25 schwarze 376: /*
1.70 schwarze 377: * Encode 8-bit input.
1.1 schwarze 378: */
379:
1.70 schwarze 380: c = blk.buf[i];
381: if (c & 0x80) {
1.71 schwarze 382: if ( ! (curp->filenc && preconv_encode(
383: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 384: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
385: curp->line, pos, "0x%x", c);
1.70 schwarze 386: ln.buf[pos++] = '?';
387: i++;
388: }
389: continue;
390: }
391:
392: /*
393: * Exclude control characters.
394: */
1.1 schwarze 395:
1.70 schwarze 396: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 397: mandoc_vmsg(c == 0x00 || c == 0x04 ||
398: c > 0x0a ? MANDOCERR_CHAR_BAD :
399: MANDOCERR_CHAR_UNSUPP,
400: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 401: i++;
1.102 schwarze 402: if (c != '\r')
403: ln.buf[pos++] = '?';
1.1 schwarze 404: continue;
405: }
406:
407: ln.buf[pos++] = blk.buf[i++];
408: }
409:
1.142 schwarze 410: if (pos + 1 >= ln.sz)
1.1 schwarze 411: resize_buf(&ln, 256);
412:
1.142 schwarze 413: if (i == blk.sz || blk.buf[i] == '\0')
414: ln.buf[pos++] = '\n';
1.1 schwarze 415: ln.buf[pos] = '\0';
416:
417: /*
418: * A significant amount of complexity is contained by
419: * the roff preprocessor. It's line-oriented but can be
420: * expressed on one line, so we need at times to
421: * readjust our starting point and re-run it. The roff
422: * preprocessor can also readjust the buffers with new
423: * data, so we pass them in wholesale.
424: */
425:
426: of = 0;
427:
1.4 schwarze 428: /*
429: * Maintain a lookaside buffer of all parsed lines. We
430: * only do this if mparse_keep() has been invoked (the
431: * buffer may be accessed with mparse_getkeep()).
432: */
433:
434: if (curp->secondary) {
1.25 schwarze 435: curp->secondary->buf = mandoc_realloc(
436: curp->secondary->buf,
437: curp->secondary->sz + pos + 2);
438: memcpy(curp->secondary->buf +
439: curp->secondary->sz,
440: ln.buf, pos);
1.4 schwarze 441: curp->secondary->sz += pos;
442: curp->secondary->buf
443: [curp->secondary->sz] = '\n';
444: curp->secondary->sz++;
445: curp->secondary->buf
446: [curp->secondary->sz] = '\0';
447: }
1.1 schwarze 448: rerun:
1.72 schwarze 449: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 450:
451: switch (rr) {
1.25 schwarze 452: case ROFF_REPARSE:
1.135 schwarze 453: if (++curp->reparse_count > REPARSE_LIMIT)
1.1 schwarze 454: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 455: curp->line, pos, NULL);
1.135 schwarze 456: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
457: start == 1) {
458: pos = 0;
459: continue;
460: }
461: free(ln.buf);
462: return 0;
1.25 schwarze 463: case ROFF_APPEND:
1.71 schwarze 464: pos = strlen(ln.buf);
1.1 schwarze 465: continue;
1.25 schwarze 466: case ROFF_RERUN:
1.1 schwarze 467: goto rerun;
1.25 schwarze 468: case ROFF_IGN:
1.1 schwarze 469: pos = 0;
470: continue;
1.25 schwarze 471: case ROFF_SO:
1.71 schwarze 472: if ( ! (curp->options & MPARSE_SO) &&
473: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 474: curp->sodest = mandoc_strdup(ln.buf + of);
475: free(ln.buf);
1.135 schwarze 476: return 1;
1.23 schwarze 477: }
1.4 schwarze 478: /*
479: * We remove `so' clauses from our lookaside
480: * buffer because we're going to descend into
481: * the file recursively.
482: */
1.25 schwarze 483: if (curp->secondary)
1.4 schwarze 484: curp->secondary->sz -= pos + 1;
1.88 schwarze 485: save_file = curp->file;
1.122 schwarze 486: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 487: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 488: close(fd);
1.88 schwarze 489: curp->file = save_file;
490: } else {
491: curp->file = save_file;
1.30 schwarze 492: mandoc_vmsg(MANDOCERR_SO_FAIL,
493: curp, curp->line, pos,
494: ".so %s", ln.buf + of);
1.85 schwarze 495: ln.sz = mandoc_asprintf(&cp,
496: ".sp\nSee the file %s.\n.sp",
497: ln.buf + of);
498: free(ln.buf);
499: ln.buf = cp;
500: of = 0;
501: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 502: }
1.1 schwarze 503: pos = 0;
504: continue;
505: default:
506: break;
507: }
508:
1.124 schwarze 509: if (curp->man->macroset == MACROSET_NONE)
1.60 schwarze 510: choose_parser(curp);
1.1 schwarze 511:
1.25 schwarze 512: /*
1.60 schwarze 513: * Lastly, push down into the parsers themselves.
1.1 schwarze 514: * If libroff returns ROFF_TBL, then add it to the
515: * currently open parse. Since we only get here if
516: * there does exist data (see tbl_data.c), we're
517: * guaranteed that something's been allocated.
518: * Do the same for ROFF_EQN.
519: */
520:
1.114 schwarze 521: if (rr == ROFF_TBL)
1.76 schwarze 522: while ((span = roff_span(curp->roff)) != NULL)
1.114 schwarze 523: roff_addtbl(curp->man, span);
524: else if (rr == ROFF_EQN)
525: roff_addeqn(curp->man, roff_eqn(curp->roff));
526: else if ((curp->man->macroset == MACROSET_MDOC ?
1.110 schwarze 527: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 528: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
529: break;
1.1 schwarze 530:
531: /* Temporary buffers typically are not full. */
532:
533: if (0 == start && '\0' == blk.buf[i])
534: break;
535:
536: /* Start the next input line. */
537:
538: pos = 0;
539: }
540:
541: free(ln.buf);
1.135 schwarze 542: return 1;
1.1 schwarze 543: }
544:
545: static int
1.18 schwarze 546: read_whole_file(struct mparse *curp, const char *file, int fd,
547: struct buf *fb, int *with_mmap)
1.1 schwarze 548: {
549: struct stat st;
1.115 schwarze 550: gzFile gz;
1.1 schwarze 551: size_t off;
552: ssize_t ssz;
553:
1.118 schwarze 554: if (fstat(fd, &st) == -1)
555: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 556:
557: /*
558: * If we're a regular file, try just reading in the whole entry
559: * via mmap(). This is faster than reading it into blocks, and
560: * since each file is only a few bytes to begin with, I'm not
561: * concerned that this is going to tank any machines.
562: */
563:
1.115 schwarze 564: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 565: if (st.st_size > 0x7fffffff) {
1.86 schwarze 566: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.117 schwarze 567: return 0;
1.1 schwarze 568: }
569: *with_mmap = 1;
570: fb->sz = (size_t)st.st_size;
1.15 schwarze 571: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 572: if (fb->buf != MAP_FAILED)
1.117 schwarze 573: return 1;
1.1 schwarze 574: }
575:
1.115 schwarze 576: if (curp->gzip) {
1.118 schwarze 577: if ((gz = gzdopen(fd, "rb")) == NULL)
578: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.115 schwarze 579: } else
580: gz = NULL;
581:
1.1 schwarze 582: /*
583: * If this isn't a regular file (like, say, stdin), then we must
584: * go the old way and just read things in bit by bit.
585: */
586:
587: *with_mmap = 0;
588: off = 0;
589: fb->sz = 0;
590: fb->buf = NULL;
591: for (;;) {
592: if (off == fb->sz) {
593: if (fb->sz == (1U << 31)) {
1.86 schwarze 594: mandoc_msg(MANDOCERR_TOOLARGE, curp,
595: 0, 0, NULL);
1.1 schwarze 596: break;
597: }
598: resize_buf(fb, 65536);
599: }
1.115 schwarze 600: ssz = curp->gzip ?
601: gzread(gz, fb->buf + (int)off, fb->sz - off) :
602: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 603: if (ssz == 0) {
604: fb->sz = off;
1.117 schwarze 605: return 1;
1.1 schwarze 606: }
1.118 schwarze 607: if (ssz == -1)
608: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 609: off += (size_t)ssz;
610: }
611:
612: free(fb->buf);
613: fb->buf = NULL;
1.117 schwarze 614: return 0;
1.1 schwarze 615: }
616:
617: static void
618: mparse_end(struct mparse *curp)
619: {
1.110 schwarze 620: if (curp->man->macroset == MACROSET_NONE)
621: curp->man->macroset = MACROSET_MAN;
622: if (curp->man->macroset == MACROSET_MDOC)
623: mdoc_endparse(curp->man);
624: else
1.86 schwarze 625: man_endparse(curp->man);
1.1 schwarze 626: roff_endparse(curp->roff);
627: }
628:
1.15 schwarze 629: static void
630: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 631: {
1.61 schwarze 632: struct buf *svprimary;
1.1 schwarze 633: const char *svfile;
1.71 schwarze 634: size_t offset;
1.14 schwarze 635: static int recursion_depth;
636:
637: if (64 < recursion_depth) {
638: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 639: return;
1.14 schwarze 640: }
1.1 schwarze 641:
1.15 schwarze 642: /* Line number is per-file. */
643: svfile = curp->file;
644: curp->file = file;
1.61 schwarze 645: svprimary = curp->primary;
1.59 schwarze 646: curp->primary = &blk;
1.15 schwarze 647: curp->line = 1;
648: recursion_depth++;
649:
1.70 schwarze 650: /* Skip an UTF-8 byte order mark. */
651: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
652: (unsigned char)blk.buf[0] == 0xef &&
653: (unsigned char)blk.buf[1] == 0xbb &&
654: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 655: offset = 3;
1.70 schwarze 656: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 657: } else
658: offset = 0;
1.70 schwarze 659:
1.71 schwarze 660: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 661:
1.86 schwarze 662: if (--recursion_depth == 0)
1.15 schwarze 663: mparse_end(curp);
664:
1.61 schwarze 665: curp->primary = svprimary;
1.15 schwarze 666: curp->file = svfile;
667: }
668:
1.74 schwarze 669: /*
670: * Read the whole file into memory and call the parsers.
671: * Called recursively when an .so request is encountered.
672: */
1.15 schwarze 673: enum mandoclevel
674: mparse_readfd(struct mparse *curp, int fd, const char *file)
675: {
676: struct buf blk;
677: int with_mmap;
1.70 schwarze 678: int save_filenc;
1.1 schwarze 679:
1.67 schwarze 680: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 681: save_filenc = curp->filenc;
682: curp->filenc = curp->options &
683: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 684: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 685: curp->filenc = save_filenc;
1.67 schwarze 686: if (with_mmap)
687: munmap(blk.buf, blk.sz);
688: else
689: free(blk.buf);
690: }
1.117 schwarze 691: return curp->file_status;
1.58 schwarze 692: }
693:
1.122 schwarze 694: int
695: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 696: {
697: char *cp;
1.122 schwarze 698: int fd;
1.58 schwarze 699:
700: curp->file = file;
1.115 schwarze 701: cp = strrchr(file, '.');
702: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 703:
1.115 schwarze 704: /* First try to use the filename as it is. */
1.74 schwarze 705:
1.122 schwarze 706: if ((fd = open(file, O_RDONLY)) != -1)
707: return fd;
1.74 schwarze 708:
1.115 schwarze 709: /*
710: * If that doesn't work and the filename doesn't
711: * already end in .gz, try appending .gz.
712: */
1.74 schwarze 713:
1.115 schwarze 714: if ( ! curp->gzip) {
1.74 schwarze 715: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 716: fd = open(cp, O_RDONLY);
1.83 schwarze 717: free(cp);
1.122 schwarze 718: if (fd != -1) {
1.115 schwarze 719: curp->gzip = 1;
1.122 schwarze 720: return fd;
1.58 schwarze 721: }
722: }
723:
1.115 schwarze 724: /* Neither worked, give up. */
1.73 schwarze 725:
1.115 schwarze 726: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.122 schwarze 727: return -1;
1.1 schwarze 728: }
729:
730: struct mparse *
1.69 schwarze 731: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.119 schwarze 732: const char *defos)
1.1 schwarze 733: {
734: struct mparse *curp;
735:
736: curp = mandoc_calloc(1, sizeof(struct mparse));
737:
1.22 schwarze 738: curp->options = options;
1.1 schwarze 739: curp->wlevel = wlevel;
740: curp->mmsg = mmsg;
1.7 schwarze 741: curp->defos = defos;
1.1 schwarze 742:
1.119 schwarze 743: curp->roff = roff_alloc(curp, options);
1.112 schwarze 744: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
745: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 746: if (curp->options & MPARSE_MDOC) {
1.112 schwarze 747: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 748: if (curp->man->mdocmac == NULL)
749: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 750: } else if (curp->options & MPARSE_MAN) {
751: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 752: if (curp->man->manmac == NULL)
753: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 754: }
1.113 schwarze 755: curp->man->first->tok = TOKEN_NONE;
1.117 schwarze 756: return curp;
1.1 schwarze 757: }
758:
759: void
760: mparse_reset(struct mparse *curp)
761: {
762: roff_reset(curp->roff);
1.124 schwarze 763: roff_man_reset(curp->man);
1.133 schwarze 764:
765: free(curp->sodest);
766: curp->sodest = NULL;
767:
1.4 schwarze 768: if (curp->secondary)
769: curp->secondary->sz = 0;
1.1 schwarze 770:
771: curp->file_status = MANDOCLEVEL_OK;
1.132 schwarze 772: curp->gzip = 0;
1.1 schwarze 773: }
774:
775: void
776: mparse_free(struct mparse *curp)
777: {
778:
1.136 schwarze 779: roffhash_free(curp->man->mdocmac);
780: roffhash_free(curp->man->manmac);
1.112 schwarze 781: roff_man_free(curp->man);
1.133 schwarze 782: roff_free(curp->roff);
1.4 schwarze 783: if (curp->secondary)
784: free(curp->secondary->buf);
1.1 schwarze 785:
1.4 schwarze 786: free(curp->secondary);
1.23 schwarze 787: free(curp->sodest);
1.1 schwarze 788: free(curp);
789: }
790:
791: void
1.110 schwarze 792: mparse_result(struct mparse *curp, struct roff_man **man,
793: char **sodest)
1.1 schwarze 794: {
795:
1.23 schwarze 796: if (sodest && NULL != (*sodest = curp->sodest)) {
797: *man = NULL;
798: return;
799: }
1.1 schwarze 800: if (man)
801: *man = curp->man;
1.130 schwarze 802: }
803:
804: void
805: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
806: {
807: if (curp->file_status > *rc)
808: *rc = curp->file_status;
1.1 schwarze 809: }
810:
811: void
812: mandoc_vmsg(enum mandocerr t, struct mparse *m,
813: int ln, int pos, const char *fmt, ...)
814: {
815: char buf[256];
816: va_list ap;
817:
818: va_start(ap, fmt);
1.26 schwarze 819: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 820: va_end(ap);
821:
822: mandoc_msg(t, m, ln, pos, buf);
823: }
824:
825: void
1.25 schwarze 826: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 827: int ln, int col, const char *msg)
828: {
829: enum mandoclevel level;
830:
1.87 schwarze 831: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 832: while (er < mandoclimits[level])
833: level--;
834:
1.83 schwarze 835: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 836: return;
837:
838: if (m->mmsg)
839: (*m->mmsg)(er, level, m->file, ln, col, msg);
840:
841: if (m->file_status < level)
842: m->file_status = level;
843: }
844:
845: const char *
846: mparse_strerror(enum mandocerr er)
847: {
848:
1.117 schwarze 849: return mandocerrs[er];
1.1 schwarze 850: }
851:
852: const char *
853: mparse_strlevel(enum mandoclevel lvl)
854: {
1.117 schwarze 855: return mandoclevels[lvl];
1.4 schwarze 856: }
857:
858: void
859: mparse_keep(struct mparse *p)
860: {
861:
862: assert(NULL == p->secondary);
863: p->secondary = mandoc_calloc(1, sizeof(struct buf));
864: }
865:
866: const char *
867: mparse_getkeep(const struct mparse *p)
868: {
869:
870: assert(p->secondary);
1.117 schwarze 871: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 schwarze 872: }