Annotation of src/usr.bin/mandoc/read.c, Revision 1.139
1.139 ! schwarze 1: /* $OpenBSD: read.c,v 1.138 2017/05/30 19:29:31 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:
87: "useless macro",
1.139 ! schwarze 88: "consider using OS macro",
1.138 schwarze 89:
1.1 schwarze 90: "generic warning",
91:
92: /* related to the prologue */
1.57 schwarze 93: "missing manual title, using UNTITLED",
94: "missing manual title, using \"\"",
1.32 schwarze 95: "lower case character in document title",
1.57 schwarze 96: "missing manual section, using \"\"",
1.1 schwarze 97: "unknown manual section",
1.32 schwarze 98: "missing date, using today's date",
1.1 schwarze 99: "cannot parse date, using it verbatim",
1.57 schwarze 100: "missing Os macro, using \"\"",
101: "duplicate prologue macro",
102: "late prologue macro",
103: "skipping late title macro",
1.1 schwarze 104: "prologue macros out of order",
105:
106: /* related to document structure */
107: ".so is fragile, better use ln(1)",
1.28 schwarze 108: "no document body",
1.32 schwarze 109: "content before first section header",
110: "first section is not \"NAME\"",
1.129 schwarze 111: "NAME section without Nm before Nd",
1.103 schwarze 112: "NAME section without description",
113: "description not at the end of NAME",
114: "bad NAME section content",
1.129 schwarze 115: "missing comma before name",
1.96 schwarze 116: "missing description line, using \"\"",
1.134 schwarze 117: "description line outside NAME section",
1.1 schwarze 118: "sections out of conventional order",
1.32 schwarze 119: "duplicate section title",
120: "unexpected section",
1.63 schwarze 121: "unusual Xr order",
122: "unusual Xr punctuation",
1.62 schwarze 123: "AUTHORS section without An macro",
1.1 schwarze 124:
125: /* related to macros and nesting */
1.33 schwarze 126: "obsolete macro",
1.78 schwarze 127: "macro neither callable nor escaped",
1.1 schwarze 128: "skipping paragraph macro",
1.10 schwarze 129: "moving paragraph macro out of list",
1.1 schwarze 130: "skipping no-space macro",
131: "blocks badly nested",
132: "nested displays are not portable",
1.35 schwarze 133: "moving content out of list",
1.56 schwarze 134: "fill mode already enabled, skipping",
135: "fill mode already disabled, skipping",
1.1 schwarze 136: "line scope broken",
137:
138: /* related to missing macro arguments */
1.36 schwarze 139: "skipping empty request",
140: "conditional request controls empty scope",
1.1 schwarze 141: "skipping empty macro",
1.99 schwarze 142: "empty block",
1.40 schwarze 143: "empty argument, using 0n",
1.38 schwarze 144: "missing display type, using -ragged",
145: "list type is not the first argument",
1.125 schwarze 146: "missing -width in -tag list, using 6n",
1.56 schwarze 147: "missing utility name, using \"\"",
1.98 schwarze 148: "missing function name, using \"\"",
1.38 schwarze 149: "empty head in list item",
150: "empty list item",
1.39 schwarze 151: "missing font type, using \\fR",
152: "unknown font type, using \\fR",
1.79 schwarze 153: "nothing follows prefix",
1.97 schwarze 154: "empty reference block",
1.128 schwarze 155: "missing section argument",
1.38 schwarze 156: "missing -std argument, adding it",
1.100 schwarze 157: "missing option string, using \"\"",
158: "missing resource identifier, using \"\"",
1.66 schwarze 159: "missing eqn box, using \"\"",
1.1 schwarze 160:
161: /* related to bad macro arguments */
1.42 schwarze 162: "unterminated quoted argument",
1.1 schwarze 163: "duplicate argument",
1.54 schwarze 164: "skipping duplicate argument",
1.41 schwarze 165: "skipping duplicate display type",
166: "skipping duplicate list type",
1.54 schwarze 167: "skipping -width argument",
1.101 schwarze 168: "wrong number of cells",
1.1 schwarze 169: "unknown AT&T UNIX version",
1.64 schwarze 170: "comma in function argument",
1.65 schwarze 171: "parenthesis in function name",
1.45 schwarze 172: "invalid content in Rs block",
1.41 schwarze 173: "invalid Boolean argument",
174: "unknown font, skipping request",
1.101 schwarze 175: "odd number of characters in request",
1.1 schwarze 176:
177: /* related to plain text */
1.42 schwarze 178: "blank line in fill mode, using .sp",
179: "tab in filled text",
180: "whitespace at end of input line",
1.131 schwarze 181: "new sentence, new line",
1.1 schwarze 182: "bad comment style",
1.42 schwarze 183: "invalid escape sequence",
184: "undefined string, using \"\"",
1.3 schwarze 185:
1.93 schwarze 186: /* related to tables */
187: "tbl line starts with span",
188: "tbl column starts with span",
189: "skipping vertical bar in tbl layout",
190:
1.1 schwarze 191: "generic error",
1.3 schwarze 192:
1.1 schwarze 193: /* related to tables */
1.91 schwarze 194: "non-alphabetic character in tbl options",
195: "skipping unknown tbl option",
196: "missing tbl option argument",
197: "wrong tbl option argument size",
1.93 schwarze 198: "empty tbl layout",
199: "invalid character in tbl layout",
200: "unmatched parenthesis in tbl layout",
1.94 schwarze 201: "tbl without any data cells",
202: "ignoring data in spanned tbl cell",
203: "ignoring extra tbl data cells",
204: "data block open at end of tbl",
1.1 schwarze 205:
1.46 schwarze 206: /* related to document structure and macros */
1.83 schwarze 207: NULL,
1.1 schwarze 208: "input stack limit exceeded, infinite loop?",
209: "skipping bad character",
1.46 schwarze 210: "skipping unknown macro",
1.87 schwarze 211: "skipping insecure request",
1.51 schwarze 212: "skipping item outside list",
1.46 schwarze 213: "skipping column outside column list",
214: "skipping end of block that is not open",
1.90 schwarze 215: "fewer RS blocks open, skipping",
1.46 schwarze 216: "inserting missing end of block",
217: "appending missing end of block",
218:
219: /* related to request and macro arguments */
1.1 schwarze 220: "escaped character not allowed in a name",
1.75 schwarze 221: "NOT IMPLEMENTED: Bd -file",
1.120 schwarze 222: "skipping display without arguments",
1.49 schwarze 223: "missing list type, using -item",
1.48 schwarze 224: "missing manual name, using \"\"",
1.49 schwarze 225: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 226: "unknown standard specifier",
1.49 schwarze 227: "skipping request without numeric argument",
1.84 schwarze 228: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 229: ".so request failed",
1.39 schwarze 230: "skipping all arguments",
231: "skipping excess arguments",
1.68 schwarze 232: "divide by zero",
1.87 schwarze 233:
234: "unsupported feature",
235: "input too large",
1.89 schwarze 236: "unsupported control character",
1.87 schwarze 237: "unsupported roff request",
1.94 schwarze 238: "eqn delim option in tbl",
1.93 schwarze 239: "unsupported tbl layout modifier",
1.87 schwarze 240: "ignoring macro in table",
1.1 schwarze 241: };
242:
243: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
244: "SUCCESS",
1.137 schwarze 245: "STYLE",
1.1 schwarze 246: "WARNING",
247: "ERROR",
1.87 schwarze 248: "UNSUPP",
1.1 schwarze 249: "BADARG",
250: "SYSERR"
251: };
252:
1.25 schwarze 253:
1.1 schwarze 254: static void
255: resize_buf(struct buf *buf, size_t initial)
256: {
257:
258: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
259: buf->buf = mandoc_realloc(buf->buf, buf->sz);
260: }
261:
262: static void
1.60 schwarze 263: choose_parser(struct mparse *curp)
1.1 schwarze 264: {
1.59 schwarze 265: char *cp, *ep;
266: int format;
1.1 schwarze 267:
1.59 schwarze 268: /*
269: * If neither command line arguments -mdoc or -man select
270: * a parser nor the roff parser found a .Dd or .TH macro
271: * yet, look ahead in the main input buffer.
272: */
273:
274: if ((format = roff_getformat(curp->roff)) == 0) {
275: cp = curp->primary->buf;
276: ep = cp + curp->primary->sz;
277: while (cp < ep) {
1.61 schwarze 278: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 279: cp++;
280: if (cp[0] == 'D' && cp[1] == 'd') {
281: format = MPARSE_MDOC;
282: break;
283: }
284: if (cp[0] == 'T' && cp[1] == 'H') {
285: format = MPARSE_MAN;
286: break;
287: }
288: }
289: cp = memchr(cp, '\n', ep - cp);
290: if (cp == NULL)
291: break;
292: cp++;
293: }
1.1 schwarze 294: }
295:
1.59 schwarze 296: if (format == MPARSE_MDOC) {
1.112 schwarze 297: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 298: if (curp->man->mdocmac == NULL)
299: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 300: } else {
301: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 302: if (curp->man->manmac == NULL)
303: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 304: }
1.136 schwarze 305: curp->man->first->tok = TOKEN_NONE;
1.1 schwarze 306: }
307:
308: /*
1.71 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 schwarze 314: */
1.135 schwarze 315: static int
1.71 schwarze 316: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 317: {
318: const struct tbl_span *span;
319: struct buf ln;
1.88 schwarze 320: const char *save_file;
1.85 schwarze 321: char *cp;
1.71 schwarze 322: size_t pos; /* byte number in the ln buffer */
1.127 schwarze 323: size_t j; /* auxiliary byte number in the blk buffer */
1.1 schwarze 324: enum rofferr rr;
1.76 schwarze 325: int of;
1.1 schwarze 326: int lnn; /* line number in the real file */
1.85 schwarze 327: int fd;
1.1 schwarze 328: unsigned char c;
329:
1.71 schwarze 330: memset(&ln, 0, sizeof(ln));
1.1 schwarze 331:
1.25 schwarze 332: lnn = curp->line;
333: pos = 0;
1.1 schwarze 334:
1.71 schwarze 335: while (i < blk.sz) {
1.1 schwarze 336: if (0 == pos && '\0' == blk.buf[i])
337: break;
338:
339: if (start) {
340: curp->line = lnn;
341: curp->reparse_count = 0;
1.70 schwarze 342:
343: if (lnn < 3 &&
344: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 345: curp->filenc & MPARSE_LATIN1)
346: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 347: }
348:
1.71 schwarze 349: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 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:
1.71 schwarze 357: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 358: '\n' == blk.buf[i + 1])
359: ++i;
360: if ('\n' == blk.buf[i]) {
361: ++i;
362: ++lnn;
363: break;
364: }
365:
1.13 schwarze 366: /*
1.70 schwarze 367: * Make sure we have space for the worst
368: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 369: */
370:
1.71 schwarze 371: if (pos + 11 > ln.sz)
1.13 schwarze 372: resize_buf(&ln, 256);
373:
1.25 schwarze 374: /*
1.70 schwarze 375: * Encode 8-bit input.
1.1 schwarze 376: */
377:
1.70 schwarze 378: c = blk.buf[i];
379: if (c & 0x80) {
1.71 schwarze 380: if ( ! (curp->filenc && preconv_encode(
381: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 382: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
383: curp->line, pos, "0x%x", c);
1.70 schwarze 384: ln.buf[pos++] = '?';
385: i++;
386: }
387: continue;
388: }
389:
390: /*
391: * Exclude control characters.
392: */
1.1 schwarze 393:
1.70 schwarze 394: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 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 schwarze 399: i++;
1.102 schwarze 400: if (c != '\r')
401: ln.buf[pos++] = '?';
1.1 schwarze 402: continue;
403: }
404:
405: /* Trailing backslash = a plain char. */
406:
1.71 schwarze 407: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 schwarze 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:
1.71 schwarze 419: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 schwarze 420: '\n' == blk.buf[i + 2])
421: ++i;
422: if ('\n' == blk.buf[i + 1]) {
423: i += 2;
424: ++lnn;
425: continue;
426: }
427:
428: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.127 schwarze 429: j = i;
1.1 schwarze 430: i += 2;
431: /* Comment, skip to end of line */
1.71 schwarze 432: for (; i < blk.sz; ++i) {
1.126 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,
1.127 schwarze 440: pos + i-1 - j, NULL);
1.126 schwarze 441: ++i;
442: ++lnn;
443: break;
1.1 schwarze 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:
1.13 schwarze 456: /* Catch escaped bogus characters. */
457:
458: c = (unsigned char) blk.buf[i+1];
459:
1.25 schwarze 460: if ( ! (isascii(c) &&
461: (isgraph(c) || isblank(c)))) {
1.89 schwarze 462: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.56 schwarze 463: curp->line, pos, "0x%x", c);
1.13 schwarze 464: i += 2;
465: ln.buf[pos++] = '?';
466: continue;
467: }
468:
1.1 schwarze 469: /* Some other escape sequence, copy & cont. */
470:
471: ln.buf[pos++] = blk.buf[i++];
472: ln.buf[pos++] = blk.buf[i++];
473: }
474:
1.71 schwarze 475: if (pos >= ln.sz)
1.1 schwarze 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:
1.4 schwarze 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) {
1.25 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);
1.4 schwarze 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 schwarze 511: rerun:
1.72 schwarze 512: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 513:
514: switch (rr) {
1.25 schwarze 515: case ROFF_REPARSE:
1.135 schwarze 516: if (++curp->reparse_count > REPARSE_LIMIT)
1.1 schwarze 517: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 518: curp->line, pos, NULL);
1.135 schwarze 519: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
520: start == 1) {
521: pos = 0;
522: continue;
523: }
524: free(ln.buf);
525: return 0;
1.25 schwarze 526: case ROFF_APPEND:
1.71 schwarze 527: pos = strlen(ln.buf);
1.1 schwarze 528: continue;
1.25 schwarze 529: case ROFF_RERUN:
1.1 schwarze 530: goto rerun;
1.25 schwarze 531: case ROFF_IGN:
1.1 schwarze 532: pos = 0;
533: continue;
1.25 schwarze 534: case ROFF_SO:
1.71 schwarze 535: if ( ! (curp->options & MPARSE_SO) &&
536: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 537: curp->sodest = mandoc_strdup(ln.buf + of);
538: free(ln.buf);
1.135 schwarze 539: return 1;
1.23 schwarze 540: }
1.4 schwarze 541: /*
542: * We remove `so' clauses from our lookaside
543: * buffer because we're going to descend into
544: * the file recursively.
545: */
1.25 schwarze 546: if (curp->secondary)
1.4 schwarze 547: curp->secondary->sz -= pos + 1;
1.88 schwarze 548: save_file = curp->file;
1.122 schwarze 549: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 550: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 551: close(fd);
1.88 schwarze 552: curp->file = save_file;
553: } else {
554: curp->file = save_file;
1.30 schwarze 555: mandoc_vmsg(MANDOCERR_SO_FAIL,
556: curp, curp->line, pos,
557: ".so %s", ln.buf + of);
1.85 schwarze 558: ln.sz = mandoc_asprintf(&cp,
559: ".sp\nSee the file %s.\n.sp",
560: ln.buf + of);
561: free(ln.buf);
562: ln.buf = cp;
563: of = 0;
564: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 565: }
1.1 schwarze 566: pos = 0;
567: continue;
568: default:
569: break;
570: }
571:
1.124 schwarze 572: if (curp->man->macroset == MACROSET_NONE)
1.60 schwarze 573: choose_parser(curp);
1.1 schwarze 574:
1.25 schwarze 575: /*
1.60 schwarze 576: * Lastly, push down into the parsers themselves.
1.1 schwarze 577: * If libroff returns ROFF_TBL, then add it to the
578: * currently open parse. Since we only get here if
579: * there does exist data (see tbl_data.c), we're
580: * guaranteed that something's been allocated.
581: * Do the same for ROFF_EQN.
582: */
583:
1.114 schwarze 584: if (rr == ROFF_TBL)
1.76 schwarze 585: while ((span = roff_span(curp->roff)) != NULL)
1.114 schwarze 586: roff_addtbl(curp->man, span);
587: else if (rr == ROFF_EQN)
588: roff_addeqn(curp->man, roff_eqn(curp->roff));
589: else if ((curp->man->macroset == MACROSET_MDOC ?
1.110 schwarze 590: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 591: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
592: break;
1.1 schwarze 593:
594: /* Temporary buffers typically are not full. */
595:
596: if (0 == start && '\0' == blk.buf[i])
597: break;
598:
599: /* Start the next input line. */
600:
601: pos = 0;
602: }
603:
604: free(ln.buf);
1.135 schwarze 605: return 1;
1.1 schwarze 606: }
607:
608: static int
1.18 schwarze 609: read_whole_file(struct mparse *curp, const char *file, int fd,
610: struct buf *fb, int *with_mmap)
1.1 schwarze 611: {
612: struct stat st;
1.115 schwarze 613: gzFile gz;
1.1 schwarze 614: size_t off;
615: ssize_t ssz;
616:
1.118 schwarze 617: if (fstat(fd, &st) == -1)
618: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 619:
620: /*
621: * If we're a regular file, try just reading in the whole entry
622: * via mmap(). This is faster than reading it into blocks, and
623: * since each file is only a few bytes to begin with, I'm not
624: * concerned that this is going to tank any machines.
625: */
626:
1.115 schwarze 627: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 628: if (st.st_size > 0x7fffffff) {
1.86 schwarze 629: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.117 schwarze 630: return 0;
1.1 schwarze 631: }
632: *with_mmap = 1;
633: fb->sz = (size_t)st.st_size;
1.15 schwarze 634: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 635: if (fb->buf != MAP_FAILED)
1.117 schwarze 636: return 1;
1.1 schwarze 637: }
638:
1.115 schwarze 639: if (curp->gzip) {
1.118 schwarze 640: if ((gz = gzdopen(fd, "rb")) == NULL)
641: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.115 schwarze 642: } else
643: gz = NULL;
644:
1.1 schwarze 645: /*
646: * If this isn't a regular file (like, say, stdin), then we must
647: * go the old way and just read things in bit by bit.
648: */
649:
650: *with_mmap = 0;
651: off = 0;
652: fb->sz = 0;
653: fb->buf = NULL;
654: for (;;) {
655: if (off == fb->sz) {
656: if (fb->sz == (1U << 31)) {
1.86 schwarze 657: mandoc_msg(MANDOCERR_TOOLARGE, curp,
658: 0, 0, NULL);
1.1 schwarze 659: break;
660: }
661: resize_buf(fb, 65536);
662: }
1.115 schwarze 663: ssz = curp->gzip ?
664: gzread(gz, fb->buf + (int)off, fb->sz - off) :
665: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 666: if (ssz == 0) {
667: fb->sz = off;
1.117 schwarze 668: return 1;
1.1 schwarze 669: }
1.118 schwarze 670: if (ssz == -1)
671: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 672: off += (size_t)ssz;
673: }
674:
675: free(fb->buf);
676: fb->buf = NULL;
1.117 schwarze 677: return 0;
1.1 schwarze 678: }
679:
680: static void
681: mparse_end(struct mparse *curp)
682: {
1.110 schwarze 683: if (curp->man->macroset == MACROSET_NONE)
684: curp->man->macroset = MACROSET_MAN;
685: if (curp->man->macroset == MACROSET_MDOC)
686: mdoc_endparse(curp->man);
687: else
1.86 schwarze 688: man_endparse(curp->man);
1.1 schwarze 689: roff_endparse(curp->roff);
690: }
691:
1.15 schwarze 692: static void
693: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 694: {
1.61 schwarze 695: struct buf *svprimary;
1.1 schwarze 696: const char *svfile;
1.71 schwarze 697: size_t offset;
1.14 schwarze 698: static int recursion_depth;
699:
700: if (64 < recursion_depth) {
701: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 702: return;
1.14 schwarze 703: }
1.1 schwarze 704:
1.15 schwarze 705: /* Line number is per-file. */
706: svfile = curp->file;
707: curp->file = file;
1.61 schwarze 708: svprimary = curp->primary;
1.59 schwarze 709: curp->primary = &blk;
1.15 schwarze 710: curp->line = 1;
711: recursion_depth++;
712:
1.70 schwarze 713: /* Skip an UTF-8 byte order mark. */
714: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
715: (unsigned char)blk.buf[0] == 0xef &&
716: (unsigned char)blk.buf[1] == 0xbb &&
717: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 718: offset = 3;
1.70 schwarze 719: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 720: } else
721: offset = 0;
1.70 schwarze 722:
1.71 schwarze 723: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 724:
1.86 schwarze 725: if (--recursion_depth == 0)
1.15 schwarze 726: mparse_end(curp);
727:
1.61 schwarze 728: curp->primary = svprimary;
1.15 schwarze 729: curp->file = svfile;
730: }
731:
1.74 schwarze 732: /*
733: * Read the whole file into memory and call the parsers.
734: * Called recursively when an .so request is encountered.
735: */
1.15 schwarze 736: enum mandoclevel
737: mparse_readfd(struct mparse *curp, int fd, const char *file)
738: {
739: struct buf blk;
740: int with_mmap;
1.70 schwarze 741: int save_filenc;
1.1 schwarze 742:
1.67 schwarze 743: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 744: save_filenc = curp->filenc;
745: curp->filenc = curp->options &
746: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 747: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 748: curp->filenc = save_filenc;
1.67 schwarze 749: if (with_mmap)
750: munmap(blk.buf, blk.sz);
751: else
752: free(blk.buf);
753: }
1.117 schwarze 754: return curp->file_status;
1.58 schwarze 755: }
756:
1.122 schwarze 757: int
758: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 759: {
760: char *cp;
1.122 schwarze 761: int fd;
1.58 schwarze 762:
763: curp->file = file;
1.115 schwarze 764: cp = strrchr(file, '.');
765: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 766:
1.115 schwarze 767: /* First try to use the filename as it is. */
1.74 schwarze 768:
1.122 schwarze 769: if ((fd = open(file, O_RDONLY)) != -1)
770: return fd;
1.74 schwarze 771:
1.115 schwarze 772: /*
773: * If that doesn't work and the filename doesn't
774: * already end in .gz, try appending .gz.
775: */
1.74 schwarze 776:
1.115 schwarze 777: if ( ! curp->gzip) {
1.74 schwarze 778: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 779: fd = open(cp, O_RDONLY);
1.83 schwarze 780: free(cp);
1.122 schwarze 781: if (fd != -1) {
1.115 schwarze 782: curp->gzip = 1;
1.122 schwarze 783: return fd;
1.58 schwarze 784: }
785: }
786:
1.115 schwarze 787: /* Neither worked, give up. */
1.73 schwarze 788:
1.115 schwarze 789: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.122 schwarze 790: return -1;
1.1 schwarze 791: }
792:
793: struct mparse *
1.69 schwarze 794: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.119 schwarze 795: const char *defos)
1.1 schwarze 796: {
797: struct mparse *curp;
798:
799: curp = mandoc_calloc(1, sizeof(struct mparse));
800:
1.22 schwarze 801: curp->options = options;
1.1 schwarze 802: curp->wlevel = wlevel;
803: curp->mmsg = mmsg;
1.7 schwarze 804: curp->defos = defos;
1.1 schwarze 805:
1.119 schwarze 806: curp->roff = roff_alloc(curp, options);
1.112 schwarze 807: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
808: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 809: if (curp->options & MPARSE_MDOC) {
1.112 schwarze 810: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 811: if (curp->man->mdocmac == NULL)
812: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 813: } else if (curp->options & MPARSE_MAN) {
814: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 815: if (curp->man->manmac == NULL)
816: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 817: }
1.113 schwarze 818: curp->man->first->tok = TOKEN_NONE;
1.117 schwarze 819: return curp;
1.1 schwarze 820: }
821:
822: void
823: mparse_reset(struct mparse *curp)
824: {
825: roff_reset(curp->roff);
1.124 schwarze 826: roff_man_reset(curp->man);
1.133 schwarze 827:
828: free(curp->sodest);
829: curp->sodest = NULL;
830:
1.4 schwarze 831: if (curp->secondary)
832: curp->secondary->sz = 0;
1.1 schwarze 833:
834: curp->file_status = MANDOCLEVEL_OK;
1.132 schwarze 835: curp->gzip = 0;
1.1 schwarze 836: }
837:
838: void
839: mparse_free(struct mparse *curp)
840: {
841:
1.136 schwarze 842: roffhash_free(curp->man->mdocmac);
843: roffhash_free(curp->man->manmac);
1.112 schwarze 844: roff_man_free(curp->man);
1.133 schwarze 845: roff_free(curp->roff);
1.4 schwarze 846: if (curp->secondary)
847: free(curp->secondary->buf);
1.1 schwarze 848:
1.4 schwarze 849: free(curp->secondary);
1.23 schwarze 850: free(curp->sodest);
1.1 schwarze 851: free(curp);
852: }
853:
854: void
1.110 schwarze 855: mparse_result(struct mparse *curp, struct roff_man **man,
856: char **sodest)
1.1 schwarze 857: {
858:
1.23 schwarze 859: if (sodest && NULL != (*sodest = curp->sodest)) {
860: *man = NULL;
861: return;
862: }
1.1 schwarze 863: if (man)
864: *man = curp->man;
1.130 schwarze 865: }
866:
867: void
868: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
869: {
870: if (curp->file_status > *rc)
871: *rc = curp->file_status;
1.1 schwarze 872: }
873:
874: void
875: mandoc_vmsg(enum mandocerr t, struct mparse *m,
876: int ln, int pos, const char *fmt, ...)
877: {
878: char buf[256];
879: va_list ap;
880:
881: va_start(ap, fmt);
1.26 schwarze 882: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 883: va_end(ap);
884:
885: mandoc_msg(t, m, ln, pos, buf);
886: }
887:
888: void
1.25 schwarze 889: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 890: int ln, int col, const char *msg)
891: {
892: enum mandoclevel level;
893:
1.87 schwarze 894: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 895: while (er < mandoclimits[level])
896: level--;
897:
1.83 schwarze 898: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 899: return;
900:
901: if (m->mmsg)
902: (*m->mmsg)(er, level, m->file, ln, col, msg);
903:
904: if (m->file_status < level)
905: m->file_status = level;
906: }
907:
908: const char *
909: mparse_strerror(enum mandocerr er)
910: {
911:
1.117 schwarze 912: return mandocerrs[er];
1.1 schwarze 913: }
914:
915: const char *
916: mparse_strlevel(enum mandoclevel lvl)
917: {
1.117 schwarze 918: return mandoclevels[lvl];
1.4 schwarze 919: }
920:
921: void
922: mparse_keep(struct mparse *p)
923: {
924:
925: assert(NULL == p->secondary);
926: p->secondary = mandoc_calloc(1, sizeof(struct buf));
927: }
928:
929: const char *
930: mparse_getkeep(const struct mparse *p)
931: {
932:
933: assert(p->secondary);
1.117 schwarze 934: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 schwarze 935: }