Annotation of src/usr.bin/mandoc/read.c, Revision 1.146
1.146 ! schwarze 1: /* $OpenBSD: read.c,v 1.145 2017/06/08 00:21:23 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",
1.144 schwarze 89: "errnos out of order",
90: "duplicate errno",
1.140 schwarze 91: "description line ends with a full stop",
1.146 ! schwarze 92: "no blank before trailing delimiter",
1.138 schwarze 93:
1.1 schwarze 94: "generic warning",
95:
96: /* related to the prologue */
1.57 schwarze 97: "missing manual title, using UNTITLED",
98: "missing manual title, using \"\"",
1.32 schwarze 99: "lower case character in document title",
1.57 schwarze 100: "missing manual section, using \"\"",
1.1 schwarze 101: "unknown manual section",
1.32 schwarze 102: "missing date, using today's date",
1.1 schwarze 103: "cannot parse date, using it verbatim",
1.57 schwarze 104: "missing Os macro, using \"\"",
105: "duplicate prologue macro",
106: "late prologue macro",
107: "skipping late title macro",
1.1 schwarze 108: "prologue macros out of order",
109:
110: /* related to document structure */
111: ".so is fragile, better use ln(1)",
1.28 schwarze 112: "no document body",
1.32 schwarze 113: "content before first section header",
114: "first section is not \"NAME\"",
1.129 schwarze 115: "NAME section without Nm before Nd",
1.103 schwarze 116: "NAME section without description",
117: "description not at the end of NAME",
118: "bad NAME section content",
1.129 schwarze 119: "missing comma before name",
1.96 schwarze 120: "missing description line, using \"\"",
1.134 schwarze 121: "description line outside NAME section",
1.1 schwarze 122: "sections out of conventional order",
1.32 schwarze 123: "duplicate section title",
124: "unexpected section",
1.63 schwarze 125: "unusual Xr order",
126: "unusual Xr punctuation",
1.62 schwarze 127: "AUTHORS section without An macro",
1.1 schwarze 128:
129: /* related to macros and nesting */
1.33 schwarze 130: "obsolete macro",
1.78 schwarze 131: "macro neither callable nor escaped",
1.1 schwarze 132: "skipping paragraph macro",
1.10 schwarze 133: "moving paragraph macro out of list",
1.1 schwarze 134: "skipping no-space macro",
135: "blocks badly nested",
136: "nested displays are not portable",
1.35 schwarze 137: "moving content out of list",
1.56 schwarze 138: "fill mode already enabled, skipping",
139: "fill mode already disabled, skipping",
1.1 schwarze 140: "line scope broken",
1.141 schwarze 141: "skipping blank line in line scope",
1.1 schwarze 142:
143: /* related to missing macro arguments */
1.36 schwarze 144: "skipping empty request",
145: "conditional request controls empty scope",
1.1 schwarze 146: "skipping empty macro",
1.99 schwarze 147: "empty block",
1.40 schwarze 148: "empty argument, using 0n",
1.38 schwarze 149: "missing display type, using -ragged",
150: "list type is not the first argument",
1.125 schwarze 151: "missing -width in -tag list, using 6n",
1.56 schwarze 152: "missing utility name, using \"\"",
1.98 schwarze 153: "missing function name, using \"\"",
1.38 schwarze 154: "empty head in list item",
155: "empty list item",
1.39 schwarze 156: "missing font type, using \\fR",
157: "unknown font type, using \\fR",
1.79 schwarze 158: "nothing follows prefix",
1.97 schwarze 159: "empty reference block",
1.128 schwarze 160: "missing section argument",
1.38 schwarze 161: "missing -std argument, adding it",
1.100 schwarze 162: "missing option string, using \"\"",
163: "missing resource identifier, using \"\"",
1.66 schwarze 164: "missing eqn box, using \"\"",
1.1 schwarze 165:
166: /* related to bad macro arguments */
1.42 schwarze 167: "unterminated quoted argument",
1.1 schwarze 168: "duplicate argument",
1.54 schwarze 169: "skipping duplicate argument",
1.41 schwarze 170: "skipping duplicate display type",
171: "skipping duplicate list type",
1.54 schwarze 172: "skipping -width argument",
1.101 schwarze 173: "wrong number of cells",
1.1 schwarze 174: "unknown AT&T UNIX version",
1.64 schwarze 175: "comma in function argument",
1.65 schwarze 176: "parenthesis in function name",
1.145 schwarze 177: "unknown library name",
1.45 schwarze 178: "invalid content in Rs block",
1.41 schwarze 179: "invalid Boolean argument",
180: "unknown font, skipping request",
1.101 schwarze 181: "odd number of characters in request",
1.1 schwarze 182:
183: /* related to plain text */
1.42 schwarze 184: "blank line in fill mode, using .sp",
185: "tab in filled text",
186: "whitespace at end of input line",
1.131 schwarze 187: "new sentence, new line",
1.1 schwarze 188: "bad comment style",
1.42 schwarze 189: "invalid escape sequence",
190: "undefined string, using \"\"",
1.3 schwarze 191:
1.93 schwarze 192: /* related to tables */
193: "tbl line starts with span",
194: "tbl column starts with span",
195: "skipping vertical bar in tbl layout",
196:
1.1 schwarze 197: "generic error",
1.3 schwarze 198:
1.1 schwarze 199: /* related to tables */
1.91 schwarze 200: "non-alphabetic character in tbl options",
201: "skipping unknown tbl option",
202: "missing tbl option argument",
203: "wrong tbl option argument size",
1.93 schwarze 204: "empty tbl layout",
205: "invalid character in tbl layout",
206: "unmatched parenthesis in tbl layout",
1.94 schwarze 207: "tbl without any data cells",
208: "ignoring data in spanned tbl cell",
209: "ignoring extra tbl data cells",
210: "data block open at end of tbl",
1.1 schwarze 211:
1.46 schwarze 212: /* related to document structure and macros */
1.83 schwarze 213: NULL,
1.1 schwarze 214: "input stack limit exceeded, infinite loop?",
215: "skipping bad character",
1.46 schwarze 216: "skipping unknown macro",
1.87 schwarze 217: "skipping insecure request",
1.51 schwarze 218: "skipping item outside list",
1.46 schwarze 219: "skipping column outside column list",
220: "skipping end of block that is not open",
1.90 schwarze 221: "fewer RS blocks open, skipping",
1.46 schwarze 222: "inserting missing end of block",
223: "appending missing end of block",
224:
225: /* related to request and macro arguments */
1.1 schwarze 226: "escaped character not allowed in a name",
1.75 schwarze 227: "NOT IMPLEMENTED: Bd -file",
1.120 schwarze 228: "skipping display without arguments",
1.49 schwarze 229: "missing list type, using -item",
1.143 schwarze 230: "argument is not numeric, using 1",
1.48 schwarze 231: "missing manual name, using \"\"",
1.49 schwarze 232: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 233: "unknown standard specifier",
1.49 schwarze 234: "skipping request without numeric argument",
1.84 schwarze 235: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 236: ".so request failed",
1.39 schwarze 237: "skipping all arguments",
238: "skipping excess arguments",
1.68 schwarze 239: "divide by zero",
1.87 schwarze 240:
241: "unsupported feature",
242: "input too large",
1.89 schwarze 243: "unsupported control character",
1.87 schwarze 244: "unsupported roff request",
1.94 schwarze 245: "eqn delim option in tbl",
1.93 schwarze 246: "unsupported tbl layout modifier",
1.87 schwarze 247: "ignoring macro in table",
1.1 schwarze 248: };
249:
250: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
251: "SUCCESS",
1.137 schwarze 252: "STYLE",
1.1 schwarze 253: "WARNING",
254: "ERROR",
1.87 schwarze 255: "UNSUPP",
1.1 schwarze 256: "BADARG",
257: "SYSERR"
258: };
259:
1.25 schwarze 260:
1.1 schwarze 261: static void
262: resize_buf(struct buf *buf, size_t initial)
263: {
264:
265: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
266: buf->buf = mandoc_realloc(buf->buf, buf->sz);
267: }
268:
269: static void
1.60 schwarze 270: choose_parser(struct mparse *curp)
1.1 schwarze 271: {
1.59 schwarze 272: char *cp, *ep;
273: int format;
1.1 schwarze 274:
1.59 schwarze 275: /*
276: * If neither command line arguments -mdoc or -man select
277: * a parser nor the roff parser found a .Dd or .TH macro
278: * yet, look ahead in the main input buffer.
279: */
280:
281: if ((format = roff_getformat(curp->roff)) == 0) {
282: cp = curp->primary->buf;
283: ep = cp + curp->primary->sz;
284: while (cp < ep) {
1.61 schwarze 285: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 286: cp++;
287: if (cp[0] == 'D' && cp[1] == 'd') {
288: format = MPARSE_MDOC;
289: break;
290: }
291: if (cp[0] == 'T' && cp[1] == 'H') {
292: format = MPARSE_MAN;
293: break;
294: }
295: }
296: cp = memchr(cp, '\n', ep - cp);
297: if (cp == NULL)
298: break;
299: cp++;
300: }
1.1 schwarze 301: }
302:
1.59 schwarze 303: if (format == MPARSE_MDOC) {
1.112 schwarze 304: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 305: if (curp->man->mdocmac == NULL)
306: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 307: } else {
308: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 309: if (curp->man->manmac == NULL)
310: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 311: }
1.136 schwarze 312: curp->man->first->tok = TOKEN_NONE;
1.1 schwarze 313: }
314:
315: /*
1.71 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 schwarze 321: */
1.135 schwarze 322: static int
1.71 schwarze 323: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 324: {
325: const struct tbl_span *span;
326: struct buf ln;
1.88 schwarze 327: const char *save_file;
1.85 schwarze 328: char *cp;
1.71 schwarze 329: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 330: enum rofferr rr;
1.76 schwarze 331: int of;
1.1 schwarze 332: int lnn; /* line number in the real file */
1.85 schwarze 333: int fd;
1.1 schwarze 334: unsigned char c;
335:
1.71 schwarze 336: memset(&ln, 0, sizeof(ln));
1.1 schwarze 337:
1.25 schwarze 338: lnn = curp->line;
339: pos = 0;
1.1 schwarze 340:
1.71 schwarze 341: while (i < blk.sz) {
1.1 schwarze 342: if (0 == pos && '\0' == blk.buf[i])
343: break;
344:
345: if (start) {
346: curp->line = lnn;
347: curp->reparse_count = 0;
1.70 schwarze 348:
349: if (lnn < 3 &&
350: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 351: curp->filenc & MPARSE_LATIN1)
352: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 353: }
354:
1.71 schwarze 355: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 356:
357: /*
358: * When finding an unescaped newline character,
359: * leave the character loop to process the line.
360: * Skip a preceding carriage return, if any.
361: */
362:
1.71 schwarze 363: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 364: '\n' == blk.buf[i + 1])
365: ++i;
366: if ('\n' == blk.buf[i]) {
367: ++i;
368: ++lnn;
369: break;
370: }
371:
1.13 schwarze 372: /*
1.70 schwarze 373: * Make sure we have space for the worst
374: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 375: */
376:
1.71 schwarze 377: if (pos + 11 > ln.sz)
1.13 schwarze 378: resize_buf(&ln, 256);
379:
1.25 schwarze 380: /*
1.70 schwarze 381: * Encode 8-bit input.
1.1 schwarze 382: */
383:
1.70 schwarze 384: c = blk.buf[i];
385: if (c & 0x80) {
1.71 schwarze 386: if ( ! (curp->filenc && preconv_encode(
387: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 388: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
389: curp->line, pos, "0x%x", c);
1.70 schwarze 390: ln.buf[pos++] = '?';
391: i++;
392: }
393: continue;
394: }
395:
396: /*
397: * Exclude control characters.
398: */
1.1 schwarze 399:
1.70 schwarze 400: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 401: mandoc_vmsg(c == 0x00 || c == 0x04 ||
402: c > 0x0a ? MANDOCERR_CHAR_BAD :
403: MANDOCERR_CHAR_UNSUPP,
404: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 405: i++;
1.102 schwarze 406: if (c != '\r')
407: ln.buf[pos++] = '?';
1.1 schwarze 408: continue;
409: }
410:
411: ln.buf[pos++] = blk.buf[i++];
412: }
413:
1.142 schwarze 414: if (pos + 1 >= ln.sz)
1.1 schwarze 415: resize_buf(&ln, 256);
416:
1.142 schwarze 417: if (i == blk.sz || blk.buf[i] == '\0')
418: ln.buf[pos++] = '\n';
1.1 schwarze 419: ln.buf[pos] = '\0';
420:
421: /*
422: * A significant amount of complexity is contained by
423: * the roff preprocessor. It's line-oriented but can be
424: * expressed on one line, so we need at times to
425: * readjust our starting point and re-run it. The roff
426: * preprocessor can also readjust the buffers with new
427: * data, so we pass them in wholesale.
428: */
429:
430: of = 0;
431:
1.4 schwarze 432: /*
433: * Maintain a lookaside buffer of all parsed lines. We
434: * only do this if mparse_keep() has been invoked (the
435: * buffer may be accessed with mparse_getkeep()).
436: */
437:
438: if (curp->secondary) {
1.25 schwarze 439: curp->secondary->buf = mandoc_realloc(
440: curp->secondary->buf,
441: curp->secondary->sz + pos + 2);
442: memcpy(curp->secondary->buf +
443: curp->secondary->sz,
444: ln.buf, pos);
1.4 schwarze 445: curp->secondary->sz += pos;
446: curp->secondary->buf
447: [curp->secondary->sz] = '\n';
448: curp->secondary->sz++;
449: curp->secondary->buf
450: [curp->secondary->sz] = '\0';
451: }
1.1 schwarze 452: rerun:
1.72 schwarze 453: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 454:
455: switch (rr) {
1.25 schwarze 456: case ROFF_REPARSE:
1.135 schwarze 457: if (++curp->reparse_count > REPARSE_LIMIT)
1.1 schwarze 458: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 459: curp->line, pos, NULL);
1.135 schwarze 460: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
461: start == 1) {
462: pos = 0;
463: continue;
464: }
465: free(ln.buf);
466: return 0;
1.25 schwarze 467: case ROFF_APPEND:
1.71 schwarze 468: pos = strlen(ln.buf);
1.1 schwarze 469: continue;
1.25 schwarze 470: case ROFF_RERUN:
1.1 schwarze 471: goto rerun;
1.25 schwarze 472: case ROFF_IGN:
1.1 schwarze 473: pos = 0;
474: continue;
1.25 schwarze 475: case ROFF_SO:
1.71 schwarze 476: if ( ! (curp->options & MPARSE_SO) &&
477: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 478: curp->sodest = mandoc_strdup(ln.buf + of);
479: free(ln.buf);
1.135 schwarze 480: return 1;
1.23 schwarze 481: }
1.4 schwarze 482: /*
483: * We remove `so' clauses from our lookaside
484: * buffer because we're going to descend into
485: * the file recursively.
486: */
1.25 schwarze 487: if (curp->secondary)
1.4 schwarze 488: curp->secondary->sz -= pos + 1;
1.88 schwarze 489: save_file = curp->file;
1.122 schwarze 490: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 491: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 492: close(fd);
1.88 schwarze 493: curp->file = save_file;
494: } else {
495: curp->file = save_file;
1.30 schwarze 496: mandoc_vmsg(MANDOCERR_SO_FAIL,
497: curp, curp->line, pos,
498: ".so %s", ln.buf + of);
1.85 schwarze 499: ln.sz = mandoc_asprintf(&cp,
500: ".sp\nSee the file %s.\n.sp",
501: ln.buf + of);
502: free(ln.buf);
503: ln.buf = cp;
504: of = 0;
505: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 506: }
1.1 schwarze 507: pos = 0;
508: continue;
509: default:
510: break;
511: }
512:
1.124 schwarze 513: if (curp->man->macroset == MACROSET_NONE)
1.60 schwarze 514: choose_parser(curp);
1.1 schwarze 515:
1.25 schwarze 516: /*
1.60 schwarze 517: * Lastly, push down into the parsers themselves.
1.1 schwarze 518: * If libroff returns ROFF_TBL, then add it to the
519: * currently open parse. Since we only get here if
520: * there does exist data (see tbl_data.c), we're
521: * guaranteed that something's been allocated.
522: * Do the same for ROFF_EQN.
523: */
524:
1.114 schwarze 525: if (rr == ROFF_TBL)
1.76 schwarze 526: while ((span = roff_span(curp->roff)) != NULL)
1.114 schwarze 527: roff_addtbl(curp->man, span);
528: else if (rr == ROFF_EQN)
529: roff_addeqn(curp->man, roff_eqn(curp->roff));
530: else if ((curp->man->macroset == MACROSET_MDOC ?
1.110 schwarze 531: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 532: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
533: break;
1.1 schwarze 534:
535: /* Temporary buffers typically are not full. */
536:
537: if (0 == start && '\0' == blk.buf[i])
538: break;
539:
540: /* Start the next input line. */
541:
542: pos = 0;
543: }
544:
545: free(ln.buf);
1.135 schwarze 546: return 1;
1.1 schwarze 547: }
548:
549: static int
1.18 schwarze 550: read_whole_file(struct mparse *curp, const char *file, int fd,
551: struct buf *fb, int *with_mmap)
1.1 schwarze 552: {
553: struct stat st;
1.115 schwarze 554: gzFile gz;
1.1 schwarze 555: size_t off;
556: ssize_t ssz;
557:
1.118 schwarze 558: if (fstat(fd, &st) == -1)
559: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 560:
561: /*
562: * If we're a regular file, try just reading in the whole entry
563: * via mmap(). This is faster than reading it into blocks, and
564: * since each file is only a few bytes to begin with, I'm not
565: * concerned that this is going to tank any machines.
566: */
567:
1.115 schwarze 568: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 569: if (st.st_size > 0x7fffffff) {
1.86 schwarze 570: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.117 schwarze 571: return 0;
1.1 schwarze 572: }
573: *with_mmap = 1;
574: fb->sz = (size_t)st.st_size;
1.15 schwarze 575: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 576: if (fb->buf != MAP_FAILED)
1.117 schwarze 577: return 1;
1.1 schwarze 578: }
579:
1.115 schwarze 580: if (curp->gzip) {
1.118 schwarze 581: if ((gz = gzdopen(fd, "rb")) == NULL)
582: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.115 schwarze 583: } else
584: gz = NULL;
585:
1.1 schwarze 586: /*
587: * If this isn't a regular file (like, say, stdin), then we must
588: * go the old way and just read things in bit by bit.
589: */
590:
591: *with_mmap = 0;
592: off = 0;
593: fb->sz = 0;
594: fb->buf = NULL;
595: for (;;) {
596: if (off == fb->sz) {
597: if (fb->sz == (1U << 31)) {
1.86 schwarze 598: mandoc_msg(MANDOCERR_TOOLARGE, curp,
599: 0, 0, NULL);
1.1 schwarze 600: break;
601: }
602: resize_buf(fb, 65536);
603: }
1.115 schwarze 604: ssz = curp->gzip ?
605: gzread(gz, fb->buf + (int)off, fb->sz - off) :
606: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 607: if (ssz == 0) {
608: fb->sz = off;
1.117 schwarze 609: return 1;
1.1 schwarze 610: }
1.118 schwarze 611: if (ssz == -1)
612: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 613: off += (size_t)ssz;
614: }
615:
616: free(fb->buf);
617: fb->buf = NULL;
1.117 schwarze 618: return 0;
1.1 schwarze 619: }
620:
621: static void
622: mparse_end(struct mparse *curp)
623: {
1.110 schwarze 624: if (curp->man->macroset == MACROSET_NONE)
625: curp->man->macroset = MACROSET_MAN;
626: if (curp->man->macroset == MACROSET_MDOC)
627: mdoc_endparse(curp->man);
628: else
1.86 schwarze 629: man_endparse(curp->man);
1.1 schwarze 630: roff_endparse(curp->roff);
631: }
632:
1.15 schwarze 633: static void
634: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 635: {
1.61 schwarze 636: struct buf *svprimary;
1.1 schwarze 637: const char *svfile;
1.71 schwarze 638: size_t offset;
1.14 schwarze 639: static int recursion_depth;
640:
641: if (64 < recursion_depth) {
642: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 643: return;
1.14 schwarze 644: }
1.1 schwarze 645:
1.15 schwarze 646: /* Line number is per-file. */
647: svfile = curp->file;
648: curp->file = file;
1.61 schwarze 649: svprimary = curp->primary;
1.59 schwarze 650: curp->primary = &blk;
1.15 schwarze 651: curp->line = 1;
652: recursion_depth++;
653:
1.70 schwarze 654: /* Skip an UTF-8 byte order mark. */
655: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
656: (unsigned char)blk.buf[0] == 0xef &&
657: (unsigned char)blk.buf[1] == 0xbb &&
658: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 659: offset = 3;
1.70 schwarze 660: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 661: } else
662: offset = 0;
1.70 schwarze 663:
1.71 schwarze 664: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 665:
1.86 schwarze 666: if (--recursion_depth == 0)
1.15 schwarze 667: mparse_end(curp);
668:
1.61 schwarze 669: curp->primary = svprimary;
1.15 schwarze 670: curp->file = svfile;
671: }
672:
1.74 schwarze 673: /*
674: * Read the whole file into memory and call the parsers.
675: * Called recursively when an .so request is encountered.
676: */
1.15 schwarze 677: enum mandoclevel
678: mparse_readfd(struct mparse *curp, int fd, const char *file)
679: {
680: struct buf blk;
681: int with_mmap;
1.70 schwarze 682: int save_filenc;
1.1 schwarze 683:
1.67 schwarze 684: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 685: save_filenc = curp->filenc;
686: curp->filenc = curp->options &
687: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 688: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 689: curp->filenc = save_filenc;
1.67 schwarze 690: if (with_mmap)
691: munmap(blk.buf, blk.sz);
692: else
693: free(blk.buf);
694: }
1.117 schwarze 695: return curp->file_status;
1.58 schwarze 696: }
697:
1.122 schwarze 698: int
699: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 700: {
701: char *cp;
1.122 schwarze 702: int fd;
1.58 schwarze 703:
704: curp->file = file;
1.115 schwarze 705: cp = strrchr(file, '.');
706: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 707:
1.115 schwarze 708: /* First try to use the filename as it is. */
1.74 schwarze 709:
1.122 schwarze 710: if ((fd = open(file, O_RDONLY)) != -1)
711: return fd;
1.74 schwarze 712:
1.115 schwarze 713: /*
714: * If that doesn't work and the filename doesn't
715: * already end in .gz, try appending .gz.
716: */
1.74 schwarze 717:
1.115 schwarze 718: if ( ! curp->gzip) {
1.74 schwarze 719: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 720: fd = open(cp, O_RDONLY);
1.83 schwarze 721: free(cp);
1.122 schwarze 722: if (fd != -1) {
1.115 schwarze 723: curp->gzip = 1;
1.122 schwarze 724: return fd;
1.58 schwarze 725: }
726: }
727:
1.115 schwarze 728: /* Neither worked, give up. */
1.73 schwarze 729:
1.115 schwarze 730: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.122 schwarze 731: return -1;
1.1 schwarze 732: }
733:
734: struct mparse *
1.69 schwarze 735: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.119 schwarze 736: const char *defos)
1.1 schwarze 737: {
738: struct mparse *curp;
739:
740: curp = mandoc_calloc(1, sizeof(struct mparse));
741:
1.22 schwarze 742: curp->options = options;
1.1 schwarze 743: curp->wlevel = wlevel;
744: curp->mmsg = mmsg;
1.7 schwarze 745: curp->defos = defos;
1.1 schwarze 746:
1.119 schwarze 747: curp->roff = roff_alloc(curp, options);
1.112 schwarze 748: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
749: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 750: if (curp->options & MPARSE_MDOC) {
1.112 schwarze 751: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 752: if (curp->man->mdocmac == NULL)
753: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 754: } else if (curp->options & MPARSE_MAN) {
755: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 756: if (curp->man->manmac == NULL)
757: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 758: }
1.113 schwarze 759: curp->man->first->tok = TOKEN_NONE;
1.117 schwarze 760: return curp;
1.1 schwarze 761: }
762:
763: void
764: mparse_reset(struct mparse *curp)
765: {
766: roff_reset(curp->roff);
1.124 schwarze 767: roff_man_reset(curp->man);
1.133 schwarze 768:
769: free(curp->sodest);
770: curp->sodest = NULL;
771:
1.4 schwarze 772: if (curp->secondary)
773: curp->secondary->sz = 0;
1.1 schwarze 774:
775: curp->file_status = MANDOCLEVEL_OK;
1.132 schwarze 776: curp->gzip = 0;
1.1 schwarze 777: }
778:
779: void
780: mparse_free(struct mparse *curp)
781: {
782:
1.136 schwarze 783: roffhash_free(curp->man->mdocmac);
784: roffhash_free(curp->man->manmac);
1.112 schwarze 785: roff_man_free(curp->man);
1.133 schwarze 786: roff_free(curp->roff);
1.4 schwarze 787: if (curp->secondary)
788: free(curp->secondary->buf);
1.1 schwarze 789:
1.4 schwarze 790: free(curp->secondary);
1.23 schwarze 791: free(curp->sodest);
1.1 schwarze 792: free(curp);
793: }
794:
795: void
1.110 schwarze 796: mparse_result(struct mparse *curp, struct roff_man **man,
797: char **sodest)
1.1 schwarze 798: {
799:
1.23 schwarze 800: if (sodest && NULL != (*sodest = curp->sodest)) {
801: *man = NULL;
802: return;
803: }
1.1 schwarze 804: if (man)
805: *man = curp->man;
1.130 schwarze 806: }
807:
808: void
809: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
810: {
811: if (curp->file_status > *rc)
812: *rc = curp->file_status;
1.1 schwarze 813: }
814:
815: void
816: mandoc_vmsg(enum mandocerr t, struct mparse *m,
817: int ln, int pos, const char *fmt, ...)
818: {
819: char buf[256];
820: va_list ap;
821:
822: va_start(ap, fmt);
1.26 schwarze 823: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 824: va_end(ap);
825:
826: mandoc_msg(t, m, ln, pos, buf);
827: }
828:
829: void
1.25 schwarze 830: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 831: int ln, int col, const char *msg)
832: {
833: enum mandoclevel level;
834:
1.87 schwarze 835: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 836: while (er < mandoclimits[level])
837: level--;
838:
1.83 schwarze 839: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 840: return;
841:
842: if (m->mmsg)
843: (*m->mmsg)(er, level, m->file, ln, col, msg);
844:
845: if (m->file_status < level)
846: m->file_status = level;
847: }
848:
849: const char *
850: mparse_strerror(enum mandocerr er)
851: {
852:
1.117 schwarze 853: return mandocerrs[er];
1.1 schwarze 854: }
855:
856: const char *
857: mparse_strlevel(enum mandoclevel lvl)
858: {
1.117 schwarze 859: return mandoclevels[lvl];
1.4 schwarze 860: }
861:
862: void
863: mparse_keep(struct mparse *p)
864: {
865:
866: assert(NULL == p->secondary);
867: p->secondary = mandoc_calloc(1, sizeof(struct buf));
868: }
869:
870: const char *
871: mparse_getkeep(const struct mparse *p)
872: {
873:
874: assert(p->secondary);
1.117 schwarze 875: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 schwarze 876: }
![[BACK]](/icons/back.gif){kind=icon}
Return to read.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc