Annotation of mandoc/read.c, Revision 1.150.2.4
1.150.2.4! schwarze 1: /* $Id: read.c,v 1.156 2017/01/08 00:11:24 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.150.2.4! schwarze 4: * Copyright (c) 2010-2017 Ingo Schwarze <schwarze@openbsd.org>
1.42 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 kristaps 6: *
7: * Permission to use, copy, modify, and distribute this software for any
8: * purpose with or without fee is hereby granted, provided that the above
9: * copyright notice and this permission notice appear in all copies.
10: *
1.133 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 kristaps 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.133 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 kristaps 14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18: */
1.11 kristaps 19: #include "config.h"
20:
1.80 schwarze 21: #include <sys/types.h>
1.81 schwarze 22: #if HAVE_MMAP
1.82 schwarze 23: #include <sys/mman.h>
1.80 schwarze 24: #include <sys/stat.h>
1.15 kristaps 25: #endif
1.1 kristaps 26:
27: #include <assert.h>
28: #include <ctype.h>
1.146 schwarze 29: #if HAVE_ERR
1.143 schwarze 30: #include <err.h>
1.146 schwarze 31: #endif
1.40 schwarze 32: #include <errno.h>
1.1 kristaps 33: #include <fcntl.h>
1.3 kristaps 34: #include <stdarg.h>
1.28 joerg 35: #include <stdint.h>
1.1 kristaps 36: #include <stdio.h>
37: #include <stdlib.h>
38: #include <string.h>
39: #include <unistd.h>
1.140 schwarze 40: #include <zlib.h>
1.1 kristaps 41:
1.133 schwarze 42: #include "mandoc_aux.h"
1.1 kristaps 43: #include "mandoc.h"
1.133 schwarze 44: #include "roff.h"
1.1 kristaps 45: #include "mdoc.h"
46: #include "man.h"
1.133 schwarze 47: #include "libmandoc.h"
1.139 schwarze 48: #include "roff_int.h"
1.1 kristaps 49:
50: #define REPARSE_LIMIT 1000
51:
52: struct mparse {
1.134 schwarze 53: struct roff_man *man; /* man parser */
1.1 kristaps 54: struct roff *roff; /* roff parser (!NULL) */
1.45 schwarze 55: char *sodest; /* filename pointed to by .so */
1.83 schwarze 56: const char *file; /* filename of current input file */
57: struct buf *primary; /* buffer currently being parsed */
58: struct buf *secondary; /* preprocessed copy of input */
59: const char *defos; /* default operating system */
60: mandocmsg mmsg; /* warning/error message handler */
61: enum mandoclevel file_status; /* status of current parse */
62: enum mandoclevel wlevel; /* ignore messages below this */
63: int options; /* parser options */
1.140 schwarze 64: int gzip; /* current input file is gzipped */
1.93 schwarze 65: int filenc; /* encoding of the current file */
1.1 kristaps 66: int reparse_count; /* finite interp. stack */
1.83 schwarze 67: int line; /* line number in the file */
1.1 kristaps 68: };
69:
1.84 schwarze 70: static void choose_parser(struct mparse *);
1.1 kristaps 71: static void resize_buf(struct buf *, size_t);
1.95 schwarze 72: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 73: static int read_whole_file(struct mparse *, const char *, int,
74: struct buf *, int *);
1.1 kristaps 75: static void mparse_end(struct mparse *);
1.37 schwarze 76: static void mparse_parse_buffer(struct mparse *, struct buf,
77: const char *);
1.1 kristaps 78:
1.3 kristaps 79: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
80: MANDOCERR_OK,
81: MANDOCERR_WARNING,
82: MANDOCERR_WARNING,
83: MANDOCERR_ERROR,
1.112 schwarze 84: MANDOCERR_UNSUPP,
1.3 kristaps 85: MANDOCERR_MAX,
86: MANDOCERR_MAX
87: };
88:
1.7 kristaps 89: static const char * const mandocerrs[MANDOCERR_MAX] = {
90: "ok",
91:
92: "generic warning",
93:
94: /* related to the prologue */
1.79 schwarze 95: "missing manual title, using UNTITLED",
96: "missing manual title, using \"\"",
1.54 schwarze 97: "lower case character in document title",
1.79 schwarze 98: "missing manual section, using \"\"",
1.7 kristaps 99: "unknown manual section",
1.54 schwarze 100: "missing date, using today's date",
1.7 kristaps 101: "cannot parse date, using it verbatim",
1.79 schwarze 102: "missing Os macro, using \"\"",
103: "duplicate prologue macro",
104: "late prologue macro",
105: "skipping late title macro",
1.7 kristaps 106: "prologue macros out of order",
107:
108: /* related to document structure */
109: ".so is fragile, better use ln(1)",
1.50 schwarze 110: "no document body",
1.54 schwarze 111: "content before first section header",
112: "first section is not \"NAME\"",
1.150.2.4! schwarze 113: "NAME section without Nm before Nd",
1.128 schwarze 114: "NAME section without description",
115: "description not at the end of NAME",
116: "bad NAME section content",
1.150.2.4! schwarze 117: "missing comma before name",
1.121 schwarze 118: "missing description line, using \"\"",
1.7 kristaps 119: "sections out of conventional order",
1.54 schwarze 120: "duplicate section title",
121: "unexpected section",
1.87 schwarze 122: "unusual Xr order",
123: "unusual Xr punctuation",
1.86 schwarze 124: "AUTHORS section without An macro",
1.7 kristaps 125:
126: /* related to macros and nesting */
1.55 schwarze 127: "obsolete macro",
1.102 schwarze 128: "macro neither callable nor escaped",
1.7 kristaps 129: "skipping paragraph macro",
1.31 schwarze 130: "moving paragraph macro out of list",
1.7 kristaps 131: "skipping no-space macro",
132: "blocks badly nested",
133: "nested displays are not portable",
1.57 schwarze 134: "moving content out of list",
1.78 schwarze 135: "fill mode already enabled, skipping",
136: "fill mode already disabled, skipping",
1.7 kristaps 137: "line scope broken",
138:
139: /* related to missing macro arguments */
1.58 schwarze 140: "skipping empty request",
141: "conditional request controls empty scope",
1.7 kristaps 142: "skipping empty macro",
1.124 schwarze 143: "empty block",
1.62 schwarze 144: "empty argument, using 0n",
1.60 schwarze 145: "missing display type, using -ragged",
146: "list type is not the first argument",
1.150.2.1 schwarze 147: "missing -width in -tag list, using 6n",
1.78 schwarze 148: "missing utility name, using \"\"",
1.123 schwarze 149: "missing function name, using \"\"",
1.60 schwarze 150: "empty head in list item",
151: "empty list item",
1.61 schwarze 152: "missing font type, using \\fR",
153: "unknown font type, using \\fR",
1.103 schwarze 154: "nothing follows prefix",
1.122 schwarze 155: "empty reference block",
1.150.2.4! schwarze 156: "missing section argument",
1.60 schwarze 157: "missing -std argument, adding it",
1.125 schwarze 158: "missing option string, using \"\"",
159: "missing resource identifier, using \"\"",
1.90 schwarze 160: "missing eqn box, using \"\"",
1.7 kristaps 161:
162: /* related to bad macro arguments */
1.64 schwarze 163: "unterminated quoted argument",
1.7 kristaps 164: "duplicate argument",
1.76 schwarze 165: "skipping duplicate argument",
1.63 schwarze 166: "skipping duplicate display type",
167: "skipping duplicate list type",
1.76 schwarze 168: "skipping -width argument",
1.126 schwarze 169: "wrong number of cells",
1.7 kristaps 170: "unknown AT&T UNIX version",
1.88 schwarze 171: "comma in function argument",
1.89 schwarze 172: "parenthesis in function name",
1.67 schwarze 173: "invalid content in Rs block",
1.63 schwarze 174: "invalid Boolean argument",
175: "unknown font, skipping request",
1.126 schwarze 176: "odd number of characters in request",
1.7 kristaps 177:
178: /* related to plain text */
1.64 schwarze 179: "blank line in fill mode, using .sp",
180: "tab in filled text",
181: "whitespace at end of input line",
1.7 kristaps 182: "bad comment style",
1.64 schwarze 183: "invalid escape sequence",
184: "undefined string, using \"\"",
1.16 kristaps 185:
1.118 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:
1.7 kristaps 191: "generic error",
192:
193: /* related to tables */
1.116 schwarze 194: "non-alphabetic character in tbl options",
195: "skipping unknown tbl option",
196: "missing tbl option argument",
197: "wrong tbl option argument size",
1.118 schwarze 198: "empty tbl layout",
199: "invalid character in tbl layout",
200: "unmatched parenthesis in tbl layout",
1.119 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",
1.7 kristaps 205:
1.68 schwarze 206: /* related to document structure and macros */
1.108 schwarze 207: NULL,
1.7 kristaps 208: "input stack limit exceeded, infinite loop?",
209: "skipping bad character",
1.68 schwarze 210: "skipping unknown macro",
1.112 schwarze 211: "skipping insecure request",
1.73 schwarze 212: "skipping item outside list",
1.68 schwarze 213: "skipping column outside column list",
214: "skipping end of block that is not open",
1.115 schwarze 215: "fewer RS blocks open, skipping",
1.68 schwarze 216: "inserting missing end of block",
217: "appending missing end of block",
218:
219: /* related to request and macro arguments */
1.7 kristaps 220: "escaped character not allowed in a name",
1.99 schwarze 221: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 222: "skipping display without arguments",
1.71 schwarze 223: "missing list type, using -item",
1.70 schwarze 224: "missing manual name, using \"\"",
1.71 schwarze 225: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 226: "unknown standard specifier",
1.71 schwarze 227: "skipping request without numeric argument",
1.109 schwarze 228: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 229: ".so request failed",
1.61 schwarze 230: "skipping all arguments",
231: "skipping excess arguments",
1.92 kristaps 232: "divide by zero",
1.112 schwarze 233:
234: "unsupported feature",
235: "input too large",
1.114 schwarze 236: "unsupported control character",
1.112 schwarze 237: "unsupported roff request",
1.119 schwarze 238: "eqn delim option in tbl",
1.118 schwarze 239: "unsupported tbl layout modifier",
1.112 schwarze 240: "ignoring macro in table",
1.7 kristaps 241: };
242:
243: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
244: "SUCCESS",
245: "RESERVED",
246: "WARNING",
247: "ERROR",
1.112 schwarze 248: "UNSUPP",
1.7 kristaps 249: "BADARG",
250: "SYSERR"
251: };
252:
1.47 schwarze 253:
1.1 kristaps 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
1.84 schwarze 263: choose_parser(struct mparse *curp)
1.1 kristaps 264: {
1.83 schwarze 265: char *cp, *ep;
266: int format;
1.1 kristaps 267:
1.83 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) {
1.85 schwarze 278: if (*cp == '.' || *cp == '\'') {
1.83 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 kristaps 294: }
295:
1.83 schwarze 296: if (format == MPARSE_MDOC) {
1.136 schwarze 297: mdoc_hash_init();
1.137 schwarze 298: curp->man->macroset = MACROSET_MDOC;
1.138 schwarze 299: curp->man->first->tok = TOKEN_NONE;
1.137 schwarze 300: } else {
301: man_hash_init();
302: curp->man->macroset = MACROSET_MAN;
1.138 schwarze 303: curp->man->first->tok = TOKEN_NONE;
1.47 schwarze 304: }
1.1 kristaps 305: }
306:
307: /*
1.95 schwarze 308: * Main parse routine for a buffer.
309: * It assumes encoding and line numbering are already set up.
310: * It can recurse directly (for invocations of user-defined
311: * macros, inline equations, and input line traps)
312: * and indirectly (for .so file inclusion).
1.1 kristaps 313: */
314: static void
1.95 schwarze 315: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 316: {
317: const struct tbl_span *span;
318: struct buf ln;
1.113 schwarze 319: const char *save_file;
1.110 schwarze 320: char *cp;
1.95 schwarze 321: size_t pos; /* byte number in the ln buffer */
1.150.2.3 schwarze 322: size_t j; /* auxiliary byte number in the blk buffer */
1.1 kristaps 323: enum rofferr rr;
1.100 schwarze 324: int of;
1.1 kristaps 325: int lnn; /* line number in the real file */
1.110 schwarze 326: int fd;
1.1 kristaps 327: unsigned char c;
328:
1.95 schwarze 329: memset(&ln, 0, sizeof(ln));
1.1 kristaps 330:
1.47 schwarze 331: lnn = curp->line;
332: pos = 0;
1.1 kristaps 333:
1.95 schwarze 334: while (i < blk.sz) {
1.1 kristaps 335: if (0 == pos && '\0' == blk.buf[i])
336: break;
337:
338: if (start) {
339: curp->line = lnn;
340: curp->reparse_count = 0;
1.93 schwarze 341:
342: if (lnn < 3 &&
343: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 344: curp->filenc & MPARSE_LATIN1)
345: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 346: }
347:
1.95 schwarze 348: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 349:
350: /*
351: * When finding an unescaped newline character,
352: * leave the character loop to process the line.
353: * Skip a preceding carriage return, if any.
354: */
355:
1.95 schwarze 356: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 357: '\n' == blk.buf[i + 1])
358: ++i;
359: if ('\n' == blk.buf[i]) {
360: ++i;
361: ++lnn;
362: break;
363: }
364:
1.35 schwarze 365: /*
1.93 schwarze 366: * Make sure we have space for the worst
367: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 368: */
369:
1.95 schwarze 370: if (pos + 11 > ln.sz)
1.35 schwarze 371: resize_buf(&ln, 256);
372:
1.47 schwarze 373: /*
1.93 schwarze 374: * Encode 8-bit input.
1.1 kristaps 375: */
376:
1.93 schwarze 377: c = blk.buf[i];
378: if (c & 0x80) {
1.95 schwarze 379: if ( ! (curp->filenc && preconv_encode(
380: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 381: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
382: curp->line, pos, "0x%x", c);
1.93 schwarze 383: ln.buf[pos++] = '?';
384: i++;
385: }
386: continue;
387: }
388:
389: /*
390: * Exclude control characters.
391: */
1.1 kristaps 392:
1.93 schwarze 393: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 394: mandoc_vmsg(c == 0x00 || c == 0x04 ||
395: c > 0x0a ? MANDOCERR_CHAR_BAD :
396: MANDOCERR_CHAR_UNSUPP,
397: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 398: i++;
1.127 schwarze 399: if (c != '\r')
400: ln.buf[pos++] = '?';
1.1 kristaps 401: continue;
402: }
403:
404: /* Trailing backslash = a plain char. */
405:
1.95 schwarze 406: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 407: ln.buf[pos++] = blk.buf[i++];
408: continue;
409: }
410:
411: /*
412: * Found escape and at least one other character.
413: * When it's a newline character, skip it.
414: * When there is a carriage return in between,
415: * skip that one as well.
416: */
417:
1.95 schwarze 418: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 419: '\n' == blk.buf[i + 2])
420: ++i;
421: if ('\n' == blk.buf[i + 1]) {
422: i += 2;
423: ++lnn;
424: continue;
425: }
426:
1.13 kristaps 427: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.150.2.3 schwarze 428: j = i;
1.1 kristaps 429: i += 2;
430: /* Comment, skip to end of line */
1.95 schwarze 431: for (; i < blk.sz; ++i) {
1.150.2.2 schwarze 432: if (blk.buf[i] != '\n')
433: continue;
434: if (blk.buf[i - 1] == ' ' ||
435: blk.buf[i - 1] == '\t')
436: mandoc_msg(
437: MANDOCERR_SPACE_EOL,
438: curp, curp->line,
1.150.2.3 schwarze 439: pos + i-1 - j, NULL);
1.150.2.2 schwarze 440: ++i;
441: ++lnn;
442: break;
1.1 kristaps 443: }
444:
445: /* Backout trailing whitespaces */
446: for (; pos > 0; --pos) {
447: if (ln.buf[pos - 1] != ' ')
448: break;
449: if (pos > 2 && ln.buf[pos - 2] == '\\')
450: break;
451: }
452: break;
453: }
454:
1.35 schwarze 455: /* Catch escaped bogus characters. */
456:
457: c = (unsigned char) blk.buf[i+1];
458:
1.47 schwarze 459: if ( ! (isascii(c) &&
460: (isgraph(c) || isblank(c)))) {
1.114 schwarze 461: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 462: curp->line, pos, "0x%x", c);
1.35 schwarze 463: i += 2;
464: ln.buf[pos++] = '?';
465: continue;
466: }
467:
1.1 kristaps 468: /* Some other escape sequence, copy & cont. */
469:
470: ln.buf[pos++] = blk.buf[i++];
471: ln.buf[pos++] = blk.buf[i++];
472: }
473:
1.95 schwarze 474: if (pos >= ln.sz)
1.1 kristaps 475: resize_buf(&ln, 256);
476:
477: ln.buf[pos] = '\0';
478:
479: /*
480: * A significant amount of complexity is contained by
481: * the roff preprocessor. It's line-oriented but can be
482: * expressed on one line, so we need at times to
483: * readjust our starting point and re-run it. The roff
484: * preprocessor can also readjust the buffers with new
485: * data, so we pass them in wholesale.
486: */
487:
488: of = 0;
489:
1.24 kristaps 490: /*
491: * Maintain a lookaside buffer of all parsed lines. We
492: * only do this if mparse_keep() has been invoked (the
493: * buffer may be accessed with mparse_getkeep()).
494: */
495:
496: if (curp->secondary) {
1.47 schwarze 497: curp->secondary->buf = mandoc_realloc(
498: curp->secondary->buf,
499: curp->secondary->sz + pos + 2);
500: memcpy(curp->secondary->buf +
501: curp->secondary->sz,
502: ln.buf, pos);
1.24 kristaps 503: curp->secondary->sz += pos;
504: curp->secondary->buf
505: [curp->secondary->sz] = '\n';
506: curp->secondary->sz++;
507: curp->secondary->buf
508: [curp->secondary->sz] = '\0';
509: }
1.1 kristaps 510: rerun:
1.96 schwarze 511: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 512:
513: switch (rr) {
1.47 schwarze 514: case ROFF_REPARSE:
1.1 kristaps 515: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 516: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 517: else
1.3 kristaps 518: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 519: curp->line, pos, NULL);
1.1 kristaps 520: pos = 0;
521: continue;
1.47 schwarze 522: case ROFF_APPEND:
1.95 schwarze 523: pos = strlen(ln.buf);
1.1 kristaps 524: continue;
1.47 schwarze 525: case ROFF_RERUN:
1.1 kristaps 526: goto rerun;
1.47 schwarze 527: case ROFF_IGN:
1.1 kristaps 528: pos = 0;
529: continue;
1.47 schwarze 530: case ROFF_SO:
1.95 schwarze 531: if ( ! (curp->options & MPARSE_SO) &&
532: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 533: curp->sodest = mandoc_strdup(ln.buf + of);
534: free(ln.buf);
535: return;
536: }
1.24 kristaps 537: /*
538: * We remove `so' clauses from our lookaside
539: * buffer because we're going to descend into
540: * the file recursively.
541: */
1.47 schwarze 542: if (curp->secondary)
1.25 kristaps 543: curp->secondary->sz -= pos + 1;
1.113 schwarze 544: save_file = curp->file;
1.148 schwarze 545: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 546: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 547: close(fd);
1.113 schwarze 548: curp->file = save_file;
549: } else {
550: curp->file = save_file;
1.52 schwarze 551: mandoc_vmsg(MANDOCERR_SO_FAIL,
552: curp, curp->line, pos,
553: ".so %s", ln.buf + of);
1.110 schwarze 554: ln.sz = mandoc_asprintf(&cp,
555: ".sp\nSee the file %s.\n.sp",
556: ln.buf + of);
557: free(ln.buf);
558: ln.buf = cp;
559: of = 0;
560: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 561: }
1.1 kristaps 562: pos = 0;
563: continue;
564: default:
565: break;
566: }
567:
1.150 schwarze 568: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 569: choose_parser(curp);
1.1 kristaps 570:
1.47 schwarze 571: /*
1.84 schwarze 572: * Lastly, push down into the parsers themselves.
1.1 kristaps 573: * If libroff returns ROFF_TBL, then add it to the
574: * currently open parse. Since we only get here if
575: * there does exist data (see tbl_data.c), we're
576: * guaranteed that something's been allocated.
577: * Do the same for ROFF_EQN.
578: */
579:
1.139 schwarze 580: if (rr == ROFF_TBL)
1.100 schwarze 581: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 582: roff_addtbl(curp->man, span);
583: else if (rr == ROFF_EQN)
584: roff_addeqn(curp->man, roff_eqn(curp->roff));
585: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 586: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 587: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
588: break;
1.1 kristaps 589:
590: /* Temporary buffers typically are not full. */
591:
592: if (0 == start && '\0' == blk.buf[i])
593: break;
594:
595: /* Start the next input line. */
596:
597: pos = 0;
598: }
599:
600: free(ln.buf);
601: }
602:
603: static int
1.40 schwarze 604: read_whole_file(struct mparse *curp, const char *file, int fd,
605: struct buf *fb, int *with_mmap)
1.1 kristaps 606: {
1.140 schwarze 607: gzFile gz;
1.1 kristaps 608: size_t off;
609: ssize_t ssz;
610:
1.81 schwarze 611: #if HAVE_MMAP
1.15 kristaps 612: struct stat st;
1.143 schwarze 613:
614: if (fstat(fd, &st) == -1)
615: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 616:
617: /*
618: * If we're a regular file, try just reading in the whole entry
619: * via mmap(). This is faster than reading it into blocks, and
620: * since each file is only a few bytes to begin with, I'm not
621: * concerned that this is going to tank any machines.
622: */
623:
1.140 schwarze 624: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 625: if (st.st_size > 0x7fffffff) {
1.111 schwarze 626: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 627: return 0;
1.1 kristaps 628: }
629: *with_mmap = 1;
630: fb->sz = (size_t)st.st_size;
1.37 schwarze 631: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 632: if (fb->buf != MAP_FAILED)
1.142 schwarze 633: return 1;
1.1 kristaps 634: }
1.15 kristaps 635: #endif
1.1 kristaps 636:
1.140 schwarze 637: if (curp->gzip) {
1.143 schwarze 638: if ((gz = gzdopen(fd, "rb")) == NULL)
639: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 640: } else
641: gz = NULL;
642:
1.1 kristaps 643: /*
644: * If this isn't a regular file (like, say, stdin), then we must
645: * go the old way and just read things in bit by bit.
646: */
647:
648: *with_mmap = 0;
649: off = 0;
650: fb->sz = 0;
651: fb->buf = NULL;
652: for (;;) {
653: if (off == fb->sz) {
654: if (fb->sz == (1U << 31)) {
1.111 schwarze 655: mandoc_msg(MANDOCERR_TOOLARGE, curp,
656: 0, 0, NULL);
1.1 kristaps 657: break;
658: }
659: resize_buf(fb, 65536);
660: }
1.140 schwarze 661: ssz = curp->gzip ?
662: gzread(gz, fb->buf + (int)off, fb->sz - off) :
663: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 664: if (ssz == 0) {
665: fb->sz = off;
1.142 schwarze 666: return 1;
1.1 kristaps 667: }
1.143 schwarze 668: if (ssz == -1)
669: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 670: off += (size_t)ssz;
671: }
672:
673: free(fb->buf);
674: fb->buf = NULL;
1.142 schwarze 675: return 0;
1.1 kristaps 676: }
677:
678: static void
679: mparse_end(struct mparse *curp)
680: {
1.135 schwarze 681: if (curp->man->macroset == MACROSET_NONE)
682: curp->man->macroset = MACROSET_MAN;
683: if (curp->man->macroset == MACROSET_MDOC)
684: mdoc_endparse(curp->man);
685: else
1.111 schwarze 686: man_endparse(curp->man);
1.1 kristaps 687: roff_endparse(curp->roff);
688: }
689:
690: static void
1.36 schwarze 691: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 692: {
1.85 schwarze 693: struct buf *svprimary;
1.28 joerg 694: const char *svfile;
1.95 schwarze 695: size_t offset;
1.36 schwarze 696: static int recursion_depth;
697:
698: if (64 < recursion_depth) {
699: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
700: return;
701: }
1.28 joerg 702:
703: /* Line number is per-file. */
704: svfile = curp->file;
705: curp->file = file;
1.85 schwarze 706: svprimary = curp->primary;
1.83 schwarze 707: curp->primary = &blk;
1.28 joerg 708: curp->line = 1;
1.36 schwarze 709: recursion_depth++;
1.28 joerg 710:
1.93 schwarze 711: /* Skip an UTF-8 byte order mark. */
712: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
713: (unsigned char)blk.buf[0] == 0xef &&
714: (unsigned char)blk.buf[1] == 0xbb &&
715: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 716: offset = 3;
1.93 schwarze 717: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 718: } else
719: offset = 0;
1.93 schwarze 720:
1.95 schwarze 721: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 722:
1.111 schwarze 723: if (--recursion_depth == 0)
1.28 joerg 724: mparse_end(curp);
725:
1.85 schwarze 726: curp->primary = svprimary;
1.28 joerg 727: curp->file = svfile;
728: }
729:
730: enum mandoclevel
1.104 schwarze 731: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 732: const char *file)
733: {
734: struct buf blk;
735:
1.104 schwarze 736: blk.buf = buf;
1.28 joerg 737: blk.sz = len;
738:
1.36 schwarze 739: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 740: return curp->file_status;
1.28 joerg 741: }
742:
1.98 schwarze 743: /*
744: * Read the whole file into memory and call the parsers.
745: * Called recursively when an .so request is encountered.
746: */
1.36 schwarze 747: enum mandoclevel
748: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 749: {
1.28 joerg 750: struct buf blk;
751: int with_mmap;
1.93 schwarze 752: int save_filenc;
1.1 kristaps 753:
1.91 schwarze 754: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 755: save_filenc = curp->filenc;
756: curp->filenc = curp->options &
757: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 758: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 759: curp->filenc = save_filenc;
1.81 schwarze 760: #if HAVE_MMAP
1.91 schwarze 761: if (with_mmap)
762: munmap(blk.buf, blk.sz);
763: else
1.28 joerg 764: #endif
1.91 schwarze 765: free(blk.buf);
766: }
1.142 schwarze 767: return curp->file_status;
1.82 schwarze 768: }
769:
1.148 schwarze 770: int
771: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 772: {
773: char *cp;
1.148 schwarze 774: int fd;
1.82 schwarze 775:
776: curp->file = file;
1.140 schwarze 777: cp = strrchr(file, '.');
778: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 779:
1.140 schwarze 780: /* First try to use the filename as it is. */
1.98 schwarze 781:
1.148 schwarze 782: if ((fd = open(file, O_RDONLY)) != -1)
783: return fd;
1.98 schwarze 784:
1.140 schwarze 785: /*
786: * If that doesn't work and the filename doesn't
787: * already end in .gz, try appending .gz.
788: */
1.98 schwarze 789:
1.140 schwarze 790: if ( ! curp->gzip) {
1.98 schwarze 791: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 792: fd = open(cp, O_RDONLY);
1.108 schwarze 793: free(cp);
1.148 schwarze 794: if (fd != -1) {
1.140 schwarze 795: curp->gzip = 1;
1.148 schwarze 796: return fd;
1.82 schwarze 797: }
798: }
799:
1.140 schwarze 800: /* Neither worked, give up. */
1.97 schwarze 801:
1.140 schwarze 802: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 803: return -1;
1.1 kristaps 804: }
805:
806: struct mparse *
1.94 schwarze 807: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.144 schwarze 808: const char *defos)
1.1 kristaps 809: {
810: struct mparse *curp;
1.10 kristaps 811:
1.1 kristaps 812: curp = mandoc_calloc(1, sizeof(struct mparse));
813:
1.44 schwarze 814: curp->options = options;
1.3 kristaps 815: curp->wlevel = wlevel;
1.1 kristaps 816: curp->mmsg = mmsg;
1.29 schwarze 817: curp->defos = defos;
1.1 kristaps 818:
1.144 schwarze 819: curp->roff = roff_alloc(curp, options);
1.137 schwarze 820: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
821: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 822: if (curp->options & MPARSE_MDOC) {
823: mdoc_hash_init();
1.137 schwarze 824: curp->man->macroset = MACROSET_MDOC;
825: } else if (curp->options & MPARSE_MAN) {
1.136 schwarze 826: man_hash_init();
1.137 schwarze 827: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 828: }
1.138 schwarze 829: curp->man->first->tok = TOKEN_NONE;
1.142 schwarze 830: return curp;
1.1 kristaps 831: }
832:
833: void
834: mparse_reset(struct mparse *curp)
835: {
836: roff_reset(curp->roff);
1.150 schwarze 837: roff_man_reset(curp->man);
1.24 kristaps 838: if (curp->secondary)
839: curp->secondary->sz = 0;
1.1 kristaps 840:
841: curp->file_status = MANDOCLEVEL_OK;
1.45 schwarze 842:
843: free(curp->sodest);
844: curp->sodest = NULL;
1.1 kristaps 845: }
846:
847: void
848: mparse_free(struct mparse *curp)
849: {
850:
1.137 schwarze 851: roff_man_free(curp->man);
1.1 kristaps 852: if (curp->roff)
853: roff_free(curp->roff);
1.24 kristaps 854: if (curp->secondary)
855: free(curp->secondary->buf);
1.1 kristaps 856:
1.24 kristaps 857: free(curp->secondary);
1.45 schwarze 858: free(curp->sodest);
1.1 kristaps 859: free(curp);
860: }
861:
862: void
1.135 schwarze 863: mparse_result(struct mparse *curp, struct roff_man **man,
864: char **sodest)
1.1 kristaps 865: {
866:
1.45 schwarze 867: if (sodest && NULL != (*sodest = curp->sodest)) {
868: *man = NULL;
869: return;
870: }
1.9 kristaps 871: if (man)
872: *man = curp->man;
1.3 kristaps 873: }
874:
875: void
876: mandoc_vmsg(enum mandocerr t, struct mparse *m,
877: int ln, int pos, const char *fmt, ...)
878: {
879: char buf[256];
880: va_list ap;
881:
882: va_start(ap, fmt);
1.48 schwarze 883: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 884: va_end(ap);
885:
886: mandoc_msg(t, m, ln, pos, buf);
887: }
888:
889: void
1.47 schwarze 890: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 891: int ln, int col, const char *msg)
892: {
893: enum mandoclevel level;
894:
1.112 schwarze 895: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 896: while (er < mandoclimits[level])
897: level--;
898:
1.108 schwarze 899: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 900: return;
901:
1.8 kristaps 902: if (m->mmsg)
903: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 904:
905: if (m->file_status < level)
906: m->file_status = level;
1.7 kristaps 907: }
908:
909: const char *
910: mparse_strerror(enum mandocerr er)
911: {
912:
1.142 schwarze 913: return mandocerrs[er];
1.7 kristaps 914: }
915:
916: const char *
917: mparse_strlevel(enum mandoclevel lvl)
918: {
1.142 schwarze 919: return mandoclevels[lvl];
1.24 kristaps 920: }
921:
922: void
923: mparse_keep(struct mparse *p)
924: {
925:
926: assert(NULL == p->secondary);
927: p->secondary = mandoc_calloc(1, sizeof(struct buf));
928: }
929:
930: const char *
931: mparse_getkeep(const struct mparse *p)
932: {
933:
934: assert(p->secondary);
1.142 schwarze 935: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 936: }
CVSweb