Annotation of src/usr.bin/mandoc/read.c, Revision 1.149
1.149 ! schwarze 1: /* $OpenBSD: read.c,v 1.148 2017/06/11 19:36:31 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:
1.148 schwarze 87: "Mdocdate found",
88: "Mdocdate missing",
89: "legacy man(7) date format",
1.149 ! schwarze 90: "RCS id missing",
1.138 schwarze 91: "useless macro",
1.139 schwarze 92: "consider using OS macro",
1.144 schwarze 93: "errnos out of order",
94: "duplicate errno",
1.140 schwarze 95: "description line ends with a full stop",
1.146 schwarze 96: "no blank before trailing delimiter",
1.147 schwarze 97: "function name without markup",
1.138 schwarze 98:
1.1 schwarze 99: "generic warning",
100:
101: /* related to the prologue */
1.57 schwarze 102: "missing manual title, using UNTITLED",
103: "missing manual title, using \"\"",
1.32 schwarze 104: "lower case character in document title",
1.57 schwarze 105: "missing manual section, using \"\"",
1.1 schwarze 106: "unknown manual section",
1.32 schwarze 107: "missing date, using today's date",
1.1 schwarze 108: "cannot parse date, using it verbatim",
1.57 schwarze 109: "missing Os macro, using \"\"",
110: "duplicate prologue macro",
111: "late prologue macro",
112: "skipping late title macro",
1.1 schwarze 113: "prologue macros out of order",
114:
115: /* related to document structure */
116: ".so is fragile, better use ln(1)",
1.28 schwarze 117: "no document body",
1.32 schwarze 118: "content before first section header",
119: "first section is not \"NAME\"",
1.129 schwarze 120: "NAME section without Nm before Nd",
1.103 schwarze 121: "NAME section without description",
122: "description not at the end of NAME",
123: "bad NAME section content",
1.129 schwarze 124: "missing comma before name",
1.96 schwarze 125: "missing description line, using \"\"",
1.134 schwarze 126: "description line outside NAME section",
1.1 schwarze 127: "sections out of conventional order",
1.32 schwarze 128: "duplicate section title",
129: "unexpected section",
1.63 schwarze 130: "unusual Xr order",
131: "unusual Xr punctuation",
1.62 schwarze 132: "AUTHORS section without An macro",
1.1 schwarze 133:
134: /* related to macros and nesting */
1.33 schwarze 135: "obsolete macro",
1.78 schwarze 136: "macro neither callable nor escaped",
1.1 schwarze 137: "skipping paragraph macro",
1.10 schwarze 138: "moving paragraph macro out of list",
1.1 schwarze 139: "skipping no-space macro",
140: "blocks badly nested",
141: "nested displays are not portable",
1.35 schwarze 142: "moving content out of list",
1.56 schwarze 143: "fill mode already enabled, skipping",
144: "fill mode already disabled, skipping",
1.1 schwarze 145: "line scope broken",
1.141 schwarze 146: "skipping blank line in line scope",
1.1 schwarze 147:
148: /* related to missing macro arguments */
1.36 schwarze 149: "skipping empty request",
150: "conditional request controls empty scope",
1.1 schwarze 151: "skipping empty macro",
1.99 schwarze 152: "empty block",
1.40 schwarze 153: "empty argument, using 0n",
1.38 schwarze 154: "missing display type, using -ragged",
155: "list type is not the first argument",
1.125 schwarze 156: "missing -width in -tag list, using 6n",
1.56 schwarze 157: "missing utility name, using \"\"",
1.98 schwarze 158: "missing function name, using \"\"",
1.38 schwarze 159: "empty head in list item",
160: "empty list item",
1.39 schwarze 161: "missing font type, using \\fR",
162: "unknown font type, using \\fR",
1.79 schwarze 163: "nothing follows prefix",
1.97 schwarze 164: "empty reference block",
1.128 schwarze 165: "missing section argument",
1.38 schwarze 166: "missing -std argument, adding it",
1.100 schwarze 167: "missing option string, using \"\"",
168: "missing resource identifier, using \"\"",
1.66 schwarze 169: "missing eqn box, using \"\"",
1.1 schwarze 170:
171: /* related to bad macro arguments */
1.42 schwarze 172: "unterminated quoted argument",
1.1 schwarze 173: "duplicate argument",
1.54 schwarze 174: "skipping duplicate argument",
1.41 schwarze 175: "skipping duplicate display type",
176: "skipping duplicate list type",
1.54 schwarze 177: "skipping -width argument",
1.101 schwarze 178: "wrong number of cells",
1.1 schwarze 179: "unknown AT&T UNIX version",
1.64 schwarze 180: "comma in function argument",
1.65 schwarze 181: "parenthesis in function name",
1.145 schwarze 182: "unknown library name",
1.45 schwarze 183: "invalid content in Rs block",
1.41 schwarze 184: "invalid Boolean argument",
185: "unknown font, skipping request",
1.101 schwarze 186: "odd number of characters in request",
1.1 schwarze 187:
188: /* related to plain text */
1.42 schwarze 189: "blank line in fill mode, using .sp",
190: "tab in filled text",
191: "whitespace at end of input line",
1.131 schwarze 192: "new sentence, new line",
1.1 schwarze 193: "bad comment style",
1.42 schwarze 194: "invalid escape sequence",
195: "undefined string, using \"\"",
1.3 schwarze 196:
1.93 schwarze 197: /* related to tables */
198: "tbl line starts with span",
199: "tbl column starts with span",
200: "skipping vertical bar in tbl layout",
201:
1.1 schwarze 202: "generic error",
1.3 schwarze 203:
1.1 schwarze 204: /* related to tables */
1.91 schwarze 205: "non-alphabetic character in tbl options",
206: "skipping unknown tbl option",
207: "missing tbl option argument",
208: "wrong tbl option argument size",
1.93 schwarze 209: "empty tbl layout",
210: "invalid character in tbl layout",
211: "unmatched parenthesis in tbl layout",
1.94 schwarze 212: "tbl without any data cells",
213: "ignoring data in spanned tbl cell",
214: "ignoring extra tbl data cells",
215: "data block open at end of tbl",
1.1 schwarze 216:
1.46 schwarze 217: /* related to document structure and macros */
1.83 schwarze 218: NULL,
1.1 schwarze 219: "input stack limit exceeded, infinite loop?",
220: "skipping bad character",
1.46 schwarze 221: "skipping unknown macro",
1.87 schwarze 222: "skipping insecure request",
1.51 schwarze 223: "skipping item outside list",
1.46 schwarze 224: "skipping column outside column list",
225: "skipping end of block that is not open",
1.90 schwarze 226: "fewer RS blocks open, skipping",
1.46 schwarze 227: "inserting missing end of block",
228: "appending missing end of block",
229:
230: /* related to request and macro arguments */
1.1 schwarze 231: "escaped character not allowed in a name",
1.75 schwarze 232: "NOT IMPLEMENTED: Bd -file",
1.120 schwarze 233: "skipping display without arguments",
1.49 schwarze 234: "missing list type, using -item",
1.143 schwarze 235: "argument is not numeric, using 1",
1.48 schwarze 236: "missing manual name, using \"\"",
1.49 schwarze 237: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 238: "unknown standard specifier",
1.49 schwarze 239: "skipping request without numeric argument",
1.84 schwarze 240: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 241: ".so request failed",
1.39 schwarze 242: "skipping all arguments",
243: "skipping excess arguments",
1.68 schwarze 244: "divide by zero",
1.87 schwarze 245:
246: "unsupported feature",
247: "input too large",
1.89 schwarze 248: "unsupported control character",
1.87 schwarze 249: "unsupported roff request",
1.94 schwarze 250: "eqn delim option in tbl",
1.93 schwarze 251: "unsupported tbl layout modifier",
1.87 schwarze 252: "ignoring macro in table",
1.1 schwarze 253: };
254:
255: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
256: "SUCCESS",
1.137 schwarze 257: "STYLE",
1.1 schwarze 258: "WARNING",
259: "ERROR",
1.87 schwarze 260: "UNSUPP",
1.1 schwarze 261: "BADARG",
262: "SYSERR"
263: };
264:
1.25 schwarze 265:
1.1 schwarze 266: static void
267: resize_buf(struct buf *buf, size_t initial)
268: {
269:
270: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
271: buf->buf = mandoc_realloc(buf->buf, buf->sz);
272: }
273:
274: static void
1.60 schwarze 275: choose_parser(struct mparse *curp)
1.1 schwarze 276: {
1.59 schwarze 277: char *cp, *ep;
278: int format;
1.1 schwarze 279:
1.59 schwarze 280: /*
281: * If neither command line arguments -mdoc or -man select
282: * a parser nor the roff parser found a .Dd or .TH macro
283: * yet, look ahead in the main input buffer.
284: */
285:
286: if ((format = roff_getformat(curp->roff)) == 0) {
287: cp = curp->primary->buf;
288: ep = cp + curp->primary->sz;
289: while (cp < ep) {
1.61 schwarze 290: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 291: cp++;
292: if (cp[0] == 'D' && cp[1] == 'd') {
293: format = MPARSE_MDOC;
294: break;
295: }
296: if (cp[0] == 'T' && cp[1] == 'H') {
297: format = MPARSE_MAN;
298: break;
299: }
300: }
301: cp = memchr(cp, '\n', ep - cp);
302: if (cp == NULL)
303: break;
304: cp++;
305: }
1.1 schwarze 306: }
307:
1.59 schwarze 308: if (format == MPARSE_MDOC) {
1.112 schwarze 309: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 310: if (curp->man->mdocmac == NULL)
311: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 312: } else {
313: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 314: if (curp->man->manmac == NULL)
315: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 316: }
1.136 schwarze 317: curp->man->first->tok = TOKEN_NONE;
1.1 schwarze 318: }
319:
320: /*
1.71 schwarze 321: * Main parse routine for a buffer.
322: * It assumes encoding and line numbering are already set up.
323: * It can recurse directly (for invocations of user-defined
324: * macros, inline equations, and input line traps)
325: * and indirectly (for .so file inclusion).
1.1 schwarze 326: */
1.135 schwarze 327: static int
1.71 schwarze 328: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 329: {
330: const struct tbl_span *span;
331: struct buf ln;
1.88 schwarze 332: const char *save_file;
1.85 schwarze 333: char *cp;
1.71 schwarze 334: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 335: enum rofferr rr;
1.76 schwarze 336: int of;
1.1 schwarze 337: int lnn; /* line number in the real file */
1.85 schwarze 338: int fd;
1.1 schwarze 339: unsigned char c;
340:
1.71 schwarze 341: memset(&ln, 0, sizeof(ln));
1.1 schwarze 342:
1.25 schwarze 343: lnn = curp->line;
344: pos = 0;
1.1 schwarze 345:
1.71 schwarze 346: while (i < blk.sz) {
1.1 schwarze 347: if (0 == pos && '\0' == blk.buf[i])
348: break;
349:
350: if (start) {
351: curp->line = lnn;
352: curp->reparse_count = 0;
1.70 schwarze 353:
354: if (lnn < 3 &&
355: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 356: curp->filenc & MPARSE_LATIN1)
357: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 358: }
359:
1.71 schwarze 360: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 361:
362: /*
363: * When finding an unescaped newline character,
364: * leave the character loop to process the line.
365: * Skip a preceding carriage return, if any.
366: */
367:
1.71 schwarze 368: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 369: '\n' == blk.buf[i + 1])
370: ++i;
371: if ('\n' == blk.buf[i]) {
372: ++i;
373: ++lnn;
374: break;
375: }
376:
1.13 schwarze 377: /*
1.70 schwarze 378: * Make sure we have space for the worst
379: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 380: */
381:
1.71 schwarze 382: if (pos + 11 > ln.sz)
1.13 schwarze 383: resize_buf(&ln, 256);
384:
1.25 schwarze 385: /*
1.70 schwarze 386: * Encode 8-bit input.
1.1 schwarze 387: */
388:
1.70 schwarze 389: c = blk.buf[i];
390: if (c & 0x80) {
1.71 schwarze 391: if ( ! (curp->filenc && preconv_encode(
392: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 393: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
394: curp->line, pos, "0x%x", c);
1.70 schwarze 395: ln.buf[pos++] = '?';
396: i++;
397: }
398: continue;
399: }
400:
401: /*
402: * Exclude control characters.
403: */
1.1 schwarze 404:
1.70 schwarze 405: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 406: mandoc_vmsg(c == 0x00 || c == 0x04 ||
407: c > 0x0a ? MANDOCERR_CHAR_BAD :
408: MANDOCERR_CHAR_UNSUPP,
409: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 410: i++;
1.102 schwarze 411: if (c != '\r')
412: ln.buf[pos++] = '?';
1.1 schwarze 413: continue;
414: }
415:
416: ln.buf[pos++] = blk.buf[i++];
417: }
418:
1.142 schwarze 419: if (pos + 1 >= ln.sz)
1.1 schwarze 420: resize_buf(&ln, 256);
421:
1.142 schwarze 422: if (i == blk.sz || blk.buf[i] == '\0')
423: ln.buf[pos++] = '\n';
1.1 schwarze 424: ln.buf[pos] = '\0';
425:
426: /*
427: * A significant amount of complexity is contained by
428: * the roff preprocessor. It's line-oriented but can be
429: * expressed on one line, so we need at times to
430: * readjust our starting point and re-run it. The roff
431: * preprocessor can also readjust the buffers with new
432: * data, so we pass them in wholesale.
433: */
434:
435: of = 0;
436:
1.4 schwarze 437: /*
438: * Maintain a lookaside buffer of all parsed lines. We
439: * only do this if mparse_keep() has been invoked (the
440: * buffer may be accessed with mparse_getkeep()).
441: */
442:
443: if (curp->secondary) {
1.25 schwarze 444: curp->secondary->buf = mandoc_realloc(
445: curp->secondary->buf,
446: curp->secondary->sz + pos + 2);
447: memcpy(curp->secondary->buf +
448: curp->secondary->sz,
449: ln.buf, pos);
1.4 schwarze 450: curp->secondary->sz += pos;
451: curp->secondary->buf
452: [curp->secondary->sz] = '\n';
453: curp->secondary->sz++;
454: curp->secondary->buf
455: [curp->secondary->sz] = '\0';
456: }
1.1 schwarze 457: rerun:
1.72 schwarze 458: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 459:
460: switch (rr) {
1.25 schwarze 461: case ROFF_REPARSE:
1.135 schwarze 462: if (++curp->reparse_count > REPARSE_LIMIT)
1.1 schwarze 463: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 464: curp->line, pos, NULL);
1.135 schwarze 465: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
466: start == 1) {
467: pos = 0;
468: continue;
469: }
470: free(ln.buf);
471: return 0;
1.25 schwarze 472: case ROFF_APPEND:
1.71 schwarze 473: pos = strlen(ln.buf);
1.1 schwarze 474: continue;
1.25 schwarze 475: case ROFF_RERUN:
1.1 schwarze 476: goto rerun;
1.25 schwarze 477: case ROFF_IGN:
1.1 schwarze 478: pos = 0;
479: continue;
1.25 schwarze 480: case ROFF_SO:
1.71 schwarze 481: if ( ! (curp->options & MPARSE_SO) &&
482: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 483: curp->sodest = mandoc_strdup(ln.buf + of);
484: free(ln.buf);
1.135 schwarze 485: return 1;
1.23 schwarze 486: }
1.4 schwarze 487: /*
488: * We remove `so' clauses from our lookaside
489: * buffer because we're going to descend into
490: * the file recursively.
491: */
1.25 schwarze 492: if (curp->secondary)
1.4 schwarze 493: curp->secondary->sz -= pos + 1;
1.88 schwarze 494: save_file = curp->file;
1.122 schwarze 495: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 496: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 497: close(fd);
1.88 schwarze 498: curp->file = save_file;
499: } else {
500: curp->file = save_file;
1.30 schwarze 501: mandoc_vmsg(MANDOCERR_SO_FAIL,
502: curp, curp->line, pos,
503: ".so %s", ln.buf + of);
1.85 schwarze 504: ln.sz = mandoc_asprintf(&cp,
505: ".sp\nSee the file %s.\n.sp",
506: ln.buf + of);
507: free(ln.buf);
508: ln.buf = cp;
509: of = 0;
510: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 511: }
1.1 schwarze 512: pos = 0;
513: continue;
514: default:
515: break;
516: }
517:
1.124 schwarze 518: if (curp->man->macroset == MACROSET_NONE)
1.60 schwarze 519: choose_parser(curp);
1.1 schwarze 520:
1.25 schwarze 521: /*
1.60 schwarze 522: * Lastly, push down into the parsers themselves.
1.1 schwarze 523: * If libroff returns ROFF_TBL, then add it to the
524: * currently open parse. Since we only get here if
525: * there does exist data (see tbl_data.c), we're
526: * guaranteed that something's been allocated.
527: * Do the same for ROFF_EQN.
528: */
529:
1.114 schwarze 530: if (rr == ROFF_TBL)
1.76 schwarze 531: while ((span = roff_span(curp->roff)) != NULL)
1.114 schwarze 532: roff_addtbl(curp->man, span);
533: else if (rr == ROFF_EQN)
534: roff_addeqn(curp->man, roff_eqn(curp->roff));
535: else if ((curp->man->macroset == MACROSET_MDOC ?
1.110 schwarze 536: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 537: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
538: break;
1.1 schwarze 539:
540: /* Temporary buffers typically are not full. */
541:
542: if (0 == start && '\0' == blk.buf[i])
543: break;
544:
545: /* Start the next input line. */
546:
547: pos = 0;
548: }
549:
550: free(ln.buf);
1.135 schwarze 551: return 1;
1.1 schwarze 552: }
553:
554: static int
1.18 schwarze 555: read_whole_file(struct mparse *curp, const char *file, int fd,
556: struct buf *fb, int *with_mmap)
1.1 schwarze 557: {
558: struct stat st;
1.115 schwarze 559: gzFile gz;
1.1 schwarze 560: size_t off;
561: ssize_t ssz;
562:
1.118 schwarze 563: if (fstat(fd, &st) == -1)
564: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 565:
566: /*
567: * If we're a regular file, try just reading in the whole entry
568: * via mmap(). This is faster than reading it into blocks, and
569: * since each file is only a few bytes to begin with, I'm not
570: * concerned that this is going to tank any machines.
571: */
572:
1.115 schwarze 573: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 574: if (st.st_size > 0x7fffffff) {
1.86 schwarze 575: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.117 schwarze 576: return 0;
1.1 schwarze 577: }
578: *with_mmap = 1;
579: fb->sz = (size_t)st.st_size;
1.15 schwarze 580: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 581: if (fb->buf != MAP_FAILED)
1.117 schwarze 582: return 1;
1.1 schwarze 583: }
584:
1.115 schwarze 585: if (curp->gzip) {
1.118 schwarze 586: if ((gz = gzdopen(fd, "rb")) == NULL)
587: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.115 schwarze 588: } else
589: gz = NULL;
590:
1.1 schwarze 591: /*
592: * If this isn't a regular file (like, say, stdin), then we must
593: * go the old way and just read things in bit by bit.
594: */
595:
596: *with_mmap = 0;
597: off = 0;
598: fb->sz = 0;
599: fb->buf = NULL;
600: for (;;) {
601: if (off == fb->sz) {
602: if (fb->sz == (1U << 31)) {
1.86 schwarze 603: mandoc_msg(MANDOCERR_TOOLARGE, curp,
604: 0, 0, NULL);
1.1 schwarze 605: break;
606: }
607: resize_buf(fb, 65536);
608: }
1.115 schwarze 609: ssz = curp->gzip ?
610: gzread(gz, fb->buf + (int)off, fb->sz - off) :
611: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 612: if (ssz == 0) {
613: fb->sz = off;
1.117 schwarze 614: return 1;
1.1 schwarze 615: }
1.118 schwarze 616: if (ssz == -1)
617: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 618: off += (size_t)ssz;
619: }
620:
621: free(fb->buf);
622: fb->buf = NULL;
1.117 schwarze 623: return 0;
1.1 schwarze 624: }
625:
626: static void
627: mparse_end(struct mparse *curp)
628: {
1.110 schwarze 629: if (curp->man->macroset == MACROSET_NONE)
630: curp->man->macroset = MACROSET_MAN;
631: if (curp->man->macroset == MACROSET_MDOC)
632: mdoc_endparse(curp->man);
633: else
1.86 schwarze 634: man_endparse(curp->man);
1.1 schwarze 635: roff_endparse(curp->roff);
636: }
637:
1.15 schwarze 638: static void
639: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 640: {
1.61 schwarze 641: struct buf *svprimary;
1.1 schwarze 642: const char *svfile;
1.71 schwarze 643: size_t offset;
1.14 schwarze 644: static int recursion_depth;
645:
646: if (64 < recursion_depth) {
647: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 648: return;
1.14 schwarze 649: }
1.1 schwarze 650:
1.15 schwarze 651: /* Line number is per-file. */
652: svfile = curp->file;
653: curp->file = file;
1.61 schwarze 654: svprimary = curp->primary;
1.59 schwarze 655: curp->primary = &blk;
1.15 schwarze 656: curp->line = 1;
657: recursion_depth++;
658:
1.70 schwarze 659: /* Skip an UTF-8 byte order mark. */
660: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
661: (unsigned char)blk.buf[0] == 0xef &&
662: (unsigned char)blk.buf[1] == 0xbb &&
663: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 664: offset = 3;
1.70 schwarze 665: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 666: } else
667: offset = 0;
1.70 schwarze 668:
1.71 schwarze 669: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 670:
1.86 schwarze 671: if (--recursion_depth == 0)
1.15 schwarze 672: mparse_end(curp);
673:
1.61 schwarze 674: curp->primary = svprimary;
1.15 schwarze 675: curp->file = svfile;
676: }
677:
1.74 schwarze 678: /*
679: * Read the whole file into memory and call the parsers.
680: * Called recursively when an .so request is encountered.
681: */
1.15 schwarze 682: enum mandoclevel
683: mparse_readfd(struct mparse *curp, int fd, const char *file)
684: {
685: struct buf blk;
686: int with_mmap;
1.70 schwarze 687: int save_filenc;
1.1 schwarze 688:
1.67 schwarze 689: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 690: save_filenc = curp->filenc;
691: curp->filenc = curp->options &
692: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 693: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 694: curp->filenc = save_filenc;
1.67 schwarze 695: if (with_mmap)
696: munmap(blk.buf, blk.sz);
697: else
698: free(blk.buf);
699: }
1.117 schwarze 700: return curp->file_status;
1.58 schwarze 701: }
702:
1.122 schwarze 703: int
704: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 705: {
706: char *cp;
1.122 schwarze 707: int fd;
1.58 schwarze 708:
709: curp->file = file;
1.115 schwarze 710: cp = strrchr(file, '.');
711: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 712:
1.115 schwarze 713: /* First try to use the filename as it is. */
1.74 schwarze 714:
1.122 schwarze 715: if ((fd = open(file, O_RDONLY)) != -1)
716: return fd;
1.74 schwarze 717:
1.115 schwarze 718: /*
719: * If that doesn't work and the filename doesn't
720: * already end in .gz, try appending .gz.
721: */
1.74 schwarze 722:
1.115 schwarze 723: if ( ! curp->gzip) {
1.74 schwarze 724: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 725: fd = open(cp, O_RDONLY);
1.83 schwarze 726: free(cp);
1.122 schwarze 727: if (fd != -1) {
1.115 schwarze 728: curp->gzip = 1;
1.122 schwarze 729: return fd;
1.58 schwarze 730: }
731: }
732:
1.115 schwarze 733: /* Neither worked, give up. */
1.73 schwarze 734:
1.115 schwarze 735: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.122 schwarze 736: return -1;
1.1 schwarze 737: }
738:
739: struct mparse *
1.69 schwarze 740: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.119 schwarze 741: const char *defos)
1.1 schwarze 742: {
743: struct mparse *curp;
744:
745: curp = mandoc_calloc(1, sizeof(struct mparse));
746:
1.22 schwarze 747: curp->options = options;
1.1 schwarze 748: curp->wlevel = wlevel;
749: curp->mmsg = mmsg;
1.7 schwarze 750: curp->defos = defos;
1.1 schwarze 751:
1.119 schwarze 752: curp->roff = roff_alloc(curp, options);
1.112 schwarze 753: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
754: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 755: if (curp->options & MPARSE_MDOC) {
1.112 schwarze 756: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 757: if (curp->man->mdocmac == NULL)
758: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 759: } else if (curp->options & MPARSE_MAN) {
760: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 761: if (curp->man->manmac == NULL)
762: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 763: }
1.113 schwarze 764: curp->man->first->tok = TOKEN_NONE;
1.117 schwarze 765: return curp;
1.1 schwarze 766: }
767:
768: void
769: mparse_reset(struct mparse *curp)
770: {
771: roff_reset(curp->roff);
1.124 schwarze 772: roff_man_reset(curp->man);
1.133 schwarze 773:
774: free(curp->sodest);
775: curp->sodest = NULL;
776:
1.4 schwarze 777: if (curp->secondary)
778: curp->secondary->sz = 0;
1.1 schwarze 779:
780: curp->file_status = MANDOCLEVEL_OK;
1.132 schwarze 781: curp->gzip = 0;
1.1 schwarze 782: }
783:
784: void
785: mparse_free(struct mparse *curp)
786: {
787:
1.136 schwarze 788: roffhash_free(curp->man->mdocmac);
789: roffhash_free(curp->man->manmac);
1.112 schwarze 790: roff_man_free(curp->man);
1.133 schwarze 791: roff_free(curp->roff);
1.4 schwarze 792: if (curp->secondary)
793: free(curp->secondary->buf);
1.1 schwarze 794:
1.4 schwarze 795: free(curp->secondary);
1.23 schwarze 796: free(curp->sodest);
1.1 schwarze 797: free(curp);
798: }
799:
800: void
1.110 schwarze 801: mparse_result(struct mparse *curp, struct roff_man **man,
802: char **sodest)
1.1 schwarze 803: {
804:
1.23 schwarze 805: if (sodest && NULL != (*sodest = curp->sodest)) {
806: *man = NULL;
807: return;
808: }
1.1 schwarze 809: if (man)
810: *man = curp->man;
1.130 schwarze 811: }
812:
813: void
814: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
815: {
816: if (curp->file_status > *rc)
817: *rc = curp->file_status;
1.1 schwarze 818: }
819:
820: void
821: mandoc_vmsg(enum mandocerr t, struct mparse *m,
822: int ln, int pos, const char *fmt, ...)
823: {
824: char buf[256];
825: va_list ap;
826:
827: va_start(ap, fmt);
1.26 schwarze 828: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 829: va_end(ap);
830:
831: mandoc_msg(t, m, ln, pos, buf);
832: }
833:
834: void
1.25 schwarze 835: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 836: int ln, int col, const char *msg)
837: {
838: enum mandoclevel level;
839:
1.87 schwarze 840: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 841: while (er < mandoclimits[level])
842: level--;
843:
1.83 schwarze 844: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 845: return;
846:
847: if (m->mmsg)
848: (*m->mmsg)(er, level, m->file, ln, col, msg);
849:
850: if (m->file_status < level)
851: m->file_status = level;
852: }
853:
854: const char *
855: mparse_strerror(enum mandocerr er)
856: {
857:
1.117 schwarze 858: return mandocerrs[er];
1.1 schwarze 859: }
860:
861: const char *
862: mparse_strlevel(enum mandoclevel lvl)
863: {
1.117 schwarze 864: return mandoclevels[lvl];
1.4 schwarze 865: }
866:
867: void
868: mparse_keep(struct mparse *p)
869: {
870:
871: assert(NULL == p->secondary);
872: p->secondary = mandoc_calloc(1, sizeof(struct buf));
873: }
874:
875: const char *
876: mparse_getkeep(const struct mparse *p)
877: {
878:
879: assert(p->secondary);
1.117 schwarze 880: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 schwarze 881: }