Annotation of mandoc/read.c, Revision 1.150.2.2
1.150.2.2! schwarze 1: /* $Id: read.c,v 1.153 2016/11/10 12:47:50 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.148 schwarze 4: * Copyright (c) 2010-2016 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\"",
1.128 schwarze 113: "NAME section without name",
114: "NAME section without description",
115: "description not at the end of NAME",
116: "bad NAME section content",
1.121 schwarze 117: "missing description line, using \"\"",
1.7 kristaps 118: "sections out of conventional order",
1.54 schwarze 119: "duplicate section title",
120: "unexpected section",
1.87 schwarze 121: "unusual Xr order",
122: "unusual Xr punctuation",
1.86 schwarze 123: "AUTHORS section without An macro",
1.7 kristaps 124:
125: /* related to macros and nesting */
1.55 schwarze 126: "obsolete macro",
1.102 schwarze 127: "macro neither callable nor escaped",
1.7 kristaps 128: "skipping paragraph macro",
1.31 schwarze 129: "moving paragraph macro out of list",
1.7 kristaps 130: "skipping no-space macro",
131: "blocks badly nested",
132: "nested displays are not portable",
1.57 schwarze 133: "moving content out of list",
1.78 schwarze 134: "fill mode already enabled, skipping",
135: "fill mode already disabled, skipping",
1.7 kristaps 136: "line scope broken",
137:
138: /* related to missing macro arguments */
1.58 schwarze 139: "skipping empty request",
140: "conditional request controls empty scope",
1.7 kristaps 141: "skipping empty macro",
1.124 schwarze 142: "empty block",
1.62 schwarze 143: "empty argument, using 0n",
1.60 schwarze 144: "missing display type, using -ragged",
145: "list type is not the first argument",
1.150.2.1 schwarze 146: "missing -width in -tag list, using 6n",
1.78 schwarze 147: "missing utility name, using \"\"",
1.123 schwarze 148: "missing function name, using \"\"",
1.60 schwarze 149: "empty head in list item",
150: "empty list item",
1.61 schwarze 151: "missing font type, using \\fR",
152: "unknown font type, using \\fR",
1.103 schwarze 153: "nothing follows prefix",
1.122 schwarze 154: "empty reference block",
1.60 schwarze 155: "missing -std argument, adding it",
1.125 schwarze 156: "missing option string, using \"\"",
157: "missing resource identifier, using \"\"",
1.90 schwarze 158: "missing eqn box, using \"\"",
1.7 kristaps 159:
160: /* related to bad macro arguments */
1.64 schwarze 161: "unterminated quoted argument",
1.7 kristaps 162: "duplicate argument",
1.76 schwarze 163: "skipping duplicate argument",
1.63 schwarze 164: "skipping duplicate display type",
165: "skipping duplicate list type",
1.76 schwarze 166: "skipping -width argument",
1.126 schwarze 167: "wrong number of cells",
1.7 kristaps 168: "unknown AT&T UNIX version",
1.88 schwarze 169: "comma in function argument",
1.89 schwarze 170: "parenthesis in function name",
1.67 schwarze 171: "invalid content in Rs block",
1.63 schwarze 172: "invalid Boolean argument",
173: "unknown font, skipping request",
1.126 schwarze 174: "odd number of characters in request",
1.7 kristaps 175:
176: /* related to plain text */
1.64 schwarze 177: "blank line in fill mode, using .sp",
178: "tab in filled text",
179: "whitespace at end of input line",
1.7 kristaps 180: "bad comment style",
1.64 schwarze 181: "invalid escape sequence",
182: "undefined string, using \"\"",
1.16 kristaps 183:
1.118 schwarze 184: /* related to tables */
185: "tbl line starts with span",
186: "tbl column starts with span",
187: "skipping vertical bar in tbl layout",
188:
1.7 kristaps 189: "generic error",
190:
191: /* related to tables */
1.116 schwarze 192: "non-alphabetic character in tbl options",
193: "skipping unknown tbl option",
194: "missing tbl option argument",
195: "wrong tbl option argument size",
1.118 schwarze 196: "empty tbl layout",
197: "invalid character in tbl layout",
198: "unmatched parenthesis in tbl layout",
1.119 schwarze 199: "tbl without any data cells",
200: "ignoring data in spanned tbl cell",
201: "ignoring extra tbl data cells",
202: "data block open at end of tbl",
1.7 kristaps 203:
1.68 schwarze 204: /* related to document structure and macros */
1.108 schwarze 205: NULL,
1.7 kristaps 206: "input stack limit exceeded, infinite loop?",
207: "skipping bad character",
1.68 schwarze 208: "skipping unknown macro",
1.112 schwarze 209: "skipping insecure request",
1.73 schwarze 210: "skipping item outside list",
1.68 schwarze 211: "skipping column outside column list",
212: "skipping end of block that is not open",
1.115 schwarze 213: "fewer RS blocks open, skipping",
1.68 schwarze 214: "inserting missing end of block",
215: "appending missing end of block",
216:
217: /* related to request and macro arguments */
1.7 kristaps 218: "escaped character not allowed in a name",
1.99 schwarze 219: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 220: "skipping display without arguments",
1.71 schwarze 221: "missing list type, using -item",
1.70 schwarze 222: "missing manual name, using \"\"",
1.71 schwarze 223: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 224: "unknown standard specifier",
1.71 schwarze 225: "skipping request without numeric argument",
1.109 schwarze 226: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 227: ".so request failed",
1.61 schwarze 228: "skipping all arguments",
229: "skipping excess arguments",
1.92 kristaps 230: "divide by zero",
1.112 schwarze 231:
232: "unsupported feature",
233: "input too large",
1.114 schwarze 234: "unsupported control character",
1.112 schwarze 235: "unsupported roff request",
1.119 schwarze 236: "eqn delim option in tbl",
1.118 schwarze 237: "unsupported tbl layout modifier",
1.112 schwarze 238: "ignoring macro in table",
1.7 kristaps 239: };
240:
241: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
242: "SUCCESS",
243: "RESERVED",
244: "WARNING",
245: "ERROR",
1.112 schwarze 246: "UNSUPP",
1.7 kristaps 247: "BADARG",
248: "SYSERR"
249: };
250:
1.47 schwarze 251:
1.1 kristaps 252: static void
253: resize_buf(struct buf *buf, size_t initial)
254: {
255:
256: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
257: buf->buf = mandoc_realloc(buf->buf, buf->sz);
258: }
259:
260: static void
1.84 schwarze 261: choose_parser(struct mparse *curp)
1.1 kristaps 262: {
1.83 schwarze 263: char *cp, *ep;
264: int format;
1.1 kristaps 265:
1.83 schwarze 266: /*
267: * If neither command line arguments -mdoc or -man select
268: * a parser nor the roff parser found a .Dd or .TH macro
269: * yet, look ahead in the main input buffer.
270: */
271:
272: if ((format = roff_getformat(curp->roff)) == 0) {
273: cp = curp->primary->buf;
274: ep = cp + curp->primary->sz;
275: while (cp < ep) {
1.85 schwarze 276: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 277: cp++;
278: if (cp[0] == 'D' && cp[1] == 'd') {
279: format = MPARSE_MDOC;
280: break;
281: }
282: if (cp[0] == 'T' && cp[1] == 'H') {
283: format = MPARSE_MAN;
284: break;
285: }
286: }
287: cp = memchr(cp, '\n', ep - cp);
288: if (cp == NULL)
289: break;
290: cp++;
291: }
1.1 kristaps 292: }
293:
1.83 schwarze 294: if (format == MPARSE_MDOC) {
1.136 schwarze 295: mdoc_hash_init();
1.137 schwarze 296: curp->man->macroset = MACROSET_MDOC;
1.138 schwarze 297: curp->man->first->tok = TOKEN_NONE;
1.137 schwarze 298: } else {
299: man_hash_init();
300: curp->man->macroset = MACROSET_MAN;
1.138 schwarze 301: curp->man->first->tok = TOKEN_NONE;
1.47 schwarze 302: }
1.1 kristaps 303: }
304:
305: /*
1.95 schwarze 306: * Main parse routine for a buffer.
307: * It assumes encoding and line numbering are already set up.
308: * It can recurse directly (for invocations of user-defined
309: * macros, inline equations, and input line traps)
310: * and indirectly (for .so file inclusion).
1.1 kristaps 311: */
312: static void
1.95 schwarze 313: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 314: {
315: const struct tbl_span *span;
316: struct buf ln;
1.113 schwarze 317: const char *save_file;
1.110 schwarze 318: char *cp;
1.95 schwarze 319: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 320: enum rofferr rr;
1.100 schwarze 321: int of;
1.1 kristaps 322: int lnn; /* line number in the real file */
1.110 schwarze 323: int fd;
1.1 kristaps 324: unsigned char c;
325:
1.95 schwarze 326: memset(&ln, 0, sizeof(ln));
1.1 kristaps 327:
1.47 schwarze 328: lnn = curp->line;
329: pos = 0;
1.1 kristaps 330:
1.95 schwarze 331: while (i < blk.sz) {
1.1 kristaps 332: if (0 == pos && '\0' == blk.buf[i])
333: break;
334:
335: if (start) {
336: curp->line = lnn;
337: curp->reparse_count = 0;
1.93 schwarze 338:
339: if (lnn < 3 &&
340: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 341: curp->filenc & MPARSE_LATIN1)
342: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 343: }
344:
1.95 schwarze 345: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 346:
347: /*
348: * When finding an unescaped newline character,
349: * leave the character loop to process the line.
350: * Skip a preceding carriage return, if any.
351: */
352:
1.95 schwarze 353: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 354: '\n' == blk.buf[i + 1])
355: ++i;
356: if ('\n' == blk.buf[i]) {
357: ++i;
358: ++lnn;
359: break;
360: }
361:
1.35 schwarze 362: /*
1.93 schwarze 363: * Make sure we have space for the worst
364: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 365: */
366:
1.95 schwarze 367: if (pos + 11 > ln.sz)
1.35 schwarze 368: resize_buf(&ln, 256);
369:
1.47 schwarze 370: /*
1.93 schwarze 371: * Encode 8-bit input.
1.1 kristaps 372: */
373:
1.93 schwarze 374: c = blk.buf[i];
375: if (c & 0x80) {
1.95 schwarze 376: if ( ! (curp->filenc && preconv_encode(
377: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 378: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
379: curp->line, pos, "0x%x", c);
1.93 schwarze 380: ln.buf[pos++] = '?';
381: i++;
382: }
383: continue;
384: }
385:
386: /*
387: * Exclude control characters.
388: */
1.1 kristaps 389:
1.93 schwarze 390: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 391: mandoc_vmsg(c == 0x00 || c == 0x04 ||
392: c > 0x0a ? MANDOCERR_CHAR_BAD :
393: MANDOCERR_CHAR_UNSUPP,
394: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 395: i++;
1.127 schwarze 396: if (c != '\r')
397: ln.buf[pos++] = '?';
1.1 kristaps 398: continue;
399: }
400:
401: /* Trailing backslash = a plain char. */
402:
1.95 schwarze 403: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 404: ln.buf[pos++] = blk.buf[i++];
405: continue;
406: }
407:
408: /*
409: * Found escape and at least one other character.
410: * When it's a newline character, skip it.
411: * When there is a carriage return in between,
412: * skip that one as well.
413: */
414:
1.95 schwarze 415: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 416: '\n' == blk.buf[i + 2])
417: ++i;
418: if ('\n' == blk.buf[i + 1]) {
419: i += 2;
420: ++lnn;
421: continue;
422: }
423:
1.13 kristaps 424: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 425: i += 2;
426: /* Comment, skip to end of line */
1.95 schwarze 427: for (; i < blk.sz; ++i) {
1.150.2.2! schwarze 428: if (blk.buf[i] != '\n')
! 429: continue;
! 430: if (blk.buf[i - 1] == ' ' ||
! 431: blk.buf[i - 1] == '\t')
! 432: mandoc_msg(
! 433: MANDOCERR_SPACE_EOL,
! 434: curp, curp->line,
! 435: pos, NULL);
! 436: ++i;
! 437: ++lnn;
! 438: break;
1.1 kristaps 439: }
440:
441: /* Backout trailing whitespaces */
442: for (; pos > 0; --pos) {
443: if (ln.buf[pos - 1] != ' ')
444: break;
445: if (pos > 2 && ln.buf[pos - 2] == '\\')
446: break;
447: }
448: break;
449: }
450:
1.35 schwarze 451: /* Catch escaped bogus characters. */
452:
453: c = (unsigned char) blk.buf[i+1];
454:
1.47 schwarze 455: if ( ! (isascii(c) &&
456: (isgraph(c) || isblank(c)))) {
1.114 schwarze 457: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 458: curp->line, pos, "0x%x", c);
1.35 schwarze 459: i += 2;
460: ln.buf[pos++] = '?';
461: continue;
462: }
463:
1.1 kristaps 464: /* Some other escape sequence, copy & cont. */
465:
466: ln.buf[pos++] = blk.buf[i++];
467: ln.buf[pos++] = blk.buf[i++];
468: }
469:
1.95 schwarze 470: if (pos >= ln.sz)
1.1 kristaps 471: resize_buf(&ln, 256);
472:
473: ln.buf[pos] = '\0';
474:
475: /*
476: * A significant amount of complexity is contained by
477: * the roff preprocessor. It's line-oriented but can be
478: * expressed on one line, so we need at times to
479: * readjust our starting point and re-run it. The roff
480: * preprocessor can also readjust the buffers with new
481: * data, so we pass them in wholesale.
482: */
483:
484: of = 0;
485:
1.24 kristaps 486: /*
487: * Maintain a lookaside buffer of all parsed lines. We
488: * only do this if mparse_keep() has been invoked (the
489: * buffer may be accessed with mparse_getkeep()).
490: */
491:
492: if (curp->secondary) {
1.47 schwarze 493: curp->secondary->buf = mandoc_realloc(
494: curp->secondary->buf,
495: curp->secondary->sz + pos + 2);
496: memcpy(curp->secondary->buf +
497: curp->secondary->sz,
498: ln.buf, pos);
1.24 kristaps 499: curp->secondary->sz += pos;
500: curp->secondary->buf
501: [curp->secondary->sz] = '\n';
502: curp->secondary->sz++;
503: curp->secondary->buf
504: [curp->secondary->sz] = '\0';
505: }
1.1 kristaps 506: rerun:
1.96 schwarze 507: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 508:
509: switch (rr) {
1.47 schwarze 510: case ROFF_REPARSE:
1.1 kristaps 511: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 512: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 513: else
1.3 kristaps 514: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 515: curp->line, pos, NULL);
1.1 kristaps 516: pos = 0;
517: continue;
1.47 schwarze 518: case ROFF_APPEND:
1.95 schwarze 519: pos = strlen(ln.buf);
1.1 kristaps 520: continue;
1.47 schwarze 521: case ROFF_RERUN:
1.1 kristaps 522: goto rerun;
1.47 schwarze 523: case ROFF_IGN:
1.1 kristaps 524: pos = 0;
525: continue;
1.47 schwarze 526: case ROFF_SO:
1.95 schwarze 527: if ( ! (curp->options & MPARSE_SO) &&
528: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 529: curp->sodest = mandoc_strdup(ln.buf + of);
530: free(ln.buf);
531: return;
532: }
1.24 kristaps 533: /*
534: * We remove `so' clauses from our lookaside
535: * buffer because we're going to descend into
536: * the file recursively.
537: */
1.47 schwarze 538: if (curp->secondary)
1.25 kristaps 539: curp->secondary->sz -= pos + 1;
1.113 schwarze 540: save_file = curp->file;
1.148 schwarze 541: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 542: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 543: close(fd);
1.113 schwarze 544: curp->file = save_file;
545: } else {
546: curp->file = save_file;
1.52 schwarze 547: mandoc_vmsg(MANDOCERR_SO_FAIL,
548: curp, curp->line, pos,
549: ".so %s", ln.buf + of);
1.110 schwarze 550: ln.sz = mandoc_asprintf(&cp,
551: ".sp\nSee the file %s.\n.sp",
552: ln.buf + of);
553: free(ln.buf);
554: ln.buf = cp;
555: of = 0;
556: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 557: }
1.1 kristaps 558: pos = 0;
559: continue;
560: default:
561: break;
562: }
563:
1.150 schwarze 564: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 565: choose_parser(curp);
1.1 kristaps 566:
1.47 schwarze 567: /*
1.84 schwarze 568: * Lastly, push down into the parsers themselves.
1.1 kristaps 569: * If libroff returns ROFF_TBL, then add it to the
570: * currently open parse. Since we only get here if
571: * there does exist data (see tbl_data.c), we're
572: * guaranteed that something's been allocated.
573: * Do the same for ROFF_EQN.
574: */
575:
1.139 schwarze 576: if (rr == ROFF_TBL)
1.100 schwarze 577: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 578: roff_addtbl(curp->man, span);
579: else if (rr == ROFF_EQN)
580: roff_addeqn(curp->man, roff_eqn(curp->roff));
581: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 582: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 583: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
584: break;
1.1 kristaps 585:
586: /* Temporary buffers typically are not full. */
587:
588: if (0 == start && '\0' == blk.buf[i])
589: break;
590:
591: /* Start the next input line. */
592:
593: pos = 0;
594: }
595:
596: free(ln.buf);
597: }
598:
599: static int
1.40 schwarze 600: read_whole_file(struct mparse *curp, const char *file, int fd,
601: struct buf *fb, int *with_mmap)
1.1 kristaps 602: {
1.140 schwarze 603: gzFile gz;
1.1 kristaps 604: size_t off;
605: ssize_t ssz;
606:
1.81 schwarze 607: #if HAVE_MMAP
1.15 kristaps 608: struct stat st;
1.143 schwarze 609:
610: if (fstat(fd, &st) == -1)
611: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 612:
613: /*
614: * If we're a regular file, try just reading in the whole entry
615: * via mmap(). This is faster than reading it into blocks, and
616: * since each file is only a few bytes to begin with, I'm not
617: * concerned that this is going to tank any machines.
618: */
619:
1.140 schwarze 620: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 621: if (st.st_size > 0x7fffffff) {
1.111 schwarze 622: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 623: return 0;
1.1 kristaps 624: }
625: *with_mmap = 1;
626: fb->sz = (size_t)st.st_size;
1.37 schwarze 627: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 628: if (fb->buf != MAP_FAILED)
1.142 schwarze 629: return 1;
1.1 kristaps 630: }
1.15 kristaps 631: #endif
1.1 kristaps 632:
1.140 schwarze 633: if (curp->gzip) {
1.143 schwarze 634: if ((gz = gzdopen(fd, "rb")) == NULL)
635: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 636: } else
637: gz = NULL;
638:
1.1 kristaps 639: /*
640: * If this isn't a regular file (like, say, stdin), then we must
641: * go the old way and just read things in bit by bit.
642: */
643:
644: *with_mmap = 0;
645: off = 0;
646: fb->sz = 0;
647: fb->buf = NULL;
648: for (;;) {
649: if (off == fb->sz) {
650: if (fb->sz == (1U << 31)) {
1.111 schwarze 651: mandoc_msg(MANDOCERR_TOOLARGE, curp,
652: 0, 0, NULL);
1.1 kristaps 653: break;
654: }
655: resize_buf(fb, 65536);
656: }
1.140 schwarze 657: ssz = curp->gzip ?
658: gzread(gz, fb->buf + (int)off, fb->sz - off) :
659: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 660: if (ssz == 0) {
661: fb->sz = off;
1.142 schwarze 662: return 1;
1.1 kristaps 663: }
1.143 schwarze 664: if (ssz == -1)
665: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 666: off += (size_t)ssz;
667: }
668:
669: free(fb->buf);
670: fb->buf = NULL;
1.142 schwarze 671: return 0;
1.1 kristaps 672: }
673:
674: static void
675: mparse_end(struct mparse *curp)
676: {
1.135 schwarze 677: if (curp->man->macroset == MACROSET_NONE)
678: curp->man->macroset = MACROSET_MAN;
679: if (curp->man->macroset == MACROSET_MDOC)
680: mdoc_endparse(curp->man);
681: else
1.111 schwarze 682: man_endparse(curp->man);
1.1 kristaps 683: roff_endparse(curp->roff);
684: }
685:
686: static void
1.36 schwarze 687: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 688: {
1.85 schwarze 689: struct buf *svprimary;
1.28 joerg 690: const char *svfile;
1.95 schwarze 691: size_t offset;
1.36 schwarze 692: static int recursion_depth;
693:
694: if (64 < recursion_depth) {
695: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
696: return;
697: }
1.28 joerg 698:
699: /* Line number is per-file. */
700: svfile = curp->file;
701: curp->file = file;
1.85 schwarze 702: svprimary = curp->primary;
1.83 schwarze 703: curp->primary = &blk;
1.28 joerg 704: curp->line = 1;
1.36 schwarze 705: recursion_depth++;
1.28 joerg 706:
1.93 schwarze 707: /* Skip an UTF-8 byte order mark. */
708: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
709: (unsigned char)blk.buf[0] == 0xef &&
710: (unsigned char)blk.buf[1] == 0xbb &&
711: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 712: offset = 3;
1.93 schwarze 713: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 714: } else
715: offset = 0;
1.93 schwarze 716:
1.95 schwarze 717: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 718:
1.111 schwarze 719: if (--recursion_depth == 0)
1.28 joerg 720: mparse_end(curp);
721:
1.85 schwarze 722: curp->primary = svprimary;
1.28 joerg 723: curp->file = svfile;
724: }
725:
726: enum mandoclevel
1.104 schwarze 727: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 728: const char *file)
729: {
730: struct buf blk;
731:
1.104 schwarze 732: blk.buf = buf;
1.28 joerg 733: blk.sz = len;
734:
1.36 schwarze 735: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 736: return curp->file_status;
1.28 joerg 737: }
738:
1.98 schwarze 739: /*
740: * Read the whole file into memory and call the parsers.
741: * Called recursively when an .so request is encountered.
742: */
1.36 schwarze 743: enum mandoclevel
744: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 745: {
1.28 joerg 746: struct buf blk;
747: int with_mmap;
1.93 schwarze 748: int save_filenc;
1.1 kristaps 749:
1.91 schwarze 750: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 751: save_filenc = curp->filenc;
752: curp->filenc = curp->options &
753: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 754: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 755: curp->filenc = save_filenc;
1.81 schwarze 756: #if HAVE_MMAP
1.91 schwarze 757: if (with_mmap)
758: munmap(blk.buf, blk.sz);
759: else
1.28 joerg 760: #endif
1.91 schwarze 761: free(blk.buf);
762: }
1.142 schwarze 763: return curp->file_status;
1.82 schwarze 764: }
765:
1.148 schwarze 766: int
767: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 768: {
769: char *cp;
1.148 schwarze 770: int fd;
1.82 schwarze 771:
772: curp->file = file;
1.140 schwarze 773: cp = strrchr(file, '.');
774: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 775:
1.140 schwarze 776: /* First try to use the filename as it is. */
1.98 schwarze 777:
1.148 schwarze 778: if ((fd = open(file, O_RDONLY)) != -1)
779: return fd;
1.98 schwarze 780:
1.140 schwarze 781: /*
782: * If that doesn't work and the filename doesn't
783: * already end in .gz, try appending .gz.
784: */
1.98 schwarze 785:
1.140 schwarze 786: if ( ! curp->gzip) {
1.98 schwarze 787: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 788: fd = open(cp, O_RDONLY);
1.108 schwarze 789: free(cp);
1.148 schwarze 790: if (fd != -1) {
1.140 schwarze 791: curp->gzip = 1;
1.148 schwarze 792: return fd;
1.82 schwarze 793: }
794: }
795:
1.140 schwarze 796: /* Neither worked, give up. */
1.97 schwarze 797:
1.140 schwarze 798: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 799: return -1;
1.1 kristaps 800: }
801:
802: struct mparse *
1.94 schwarze 803: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.144 schwarze 804: const char *defos)
1.1 kristaps 805: {
806: struct mparse *curp;
1.10 kristaps 807:
1.1 kristaps 808: curp = mandoc_calloc(1, sizeof(struct mparse));
809:
1.44 schwarze 810: curp->options = options;
1.3 kristaps 811: curp->wlevel = wlevel;
1.1 kristaps 812: curp->mmsg = mmsg;
1.29 schwarze 813: curp->defos = defos;
1.1 kristaps 814:
1.144 schwarze 815: curp->roff = roff_alloc(curp, options);
1.137 schwarze 816: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
817: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 818: if (curp->options & MPARSE_MDOC) {
819: mdoc_hash_init();
1.137 schwarze 820: curp->man->macroset = MACROSET_MDOC;
821: } else if (curp->options & MPARSE_MAN) {
1.136 schwarze 822: man_hash_init();
1.137 schwarze 823: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 824: }
1.138 schwarze 825: curp->man->first->tok = TOKEN_NONE;
1.142 schwarze 826: return curp;
1.1 kristaps 827: }
828:
829: void
830: mparse_reset(struct mparse *curp)
831: {
832: roff_reset(curp->roff);
1.150 schwarze 833: roff_man_reset(curp->man);
1.24 kristaps 834: if (curp->secondary)
835: curp->secondary->sz = 0;
1.1 kristaps 836:
837: curp->file_status = MANDOCLEVEL_OK;
1.45 schwarze 838:
839: free(curp->sodest);
840: curp->sodest = NULL;
1.1 kristaps 841: }
842:
843: void
844: mparse_free(struct mparse *curp)
845: {
846:
1.137 schwarze 847: roff_man_free(curp->man);
1.1 kristaps 848: if (curp->roff)
849: roff_free(curp->roff);
1.24 kristaps 850: if (curp->secondary)
851: free(curp->secondary->buf);
1.1 kristaps 852:
1.24 kristaps 853: free(curp->secondary);
1.45 schwarze 854: free(curp->sodest);
1.1 kristaps 855: free(curp);
856: }
857:
858: void
1.135 schwarze 859: mparse_result(struct mparse *curp, struct roff_man **man,
860: char **sodest)
1.1 kristaps 861: {
862:
1.45 schwarze 863: if (sodest && NULL != (*sodest = curp->sodest)) {
864: *man = NULL;
865: return;
866: }
1.9 kristaps 867: if (man)
868: *man = curp->man;
1.3 kristaps 869: }
870:
871: void
872: mandoc_vmsg(enum mandocerr t, struct mparse *m,
873: int ln, int pos, const char *fmt, ...)
874: {
875: char buf[256];
876: va_list ap;
877:
878: va_start(ap, fmt);
1.48 schwarze 879: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 880: va_end(ap);
881:
882: mandoc_msg(t, m, ln, pos, buf);
883: }
884:
885: void
1.47 schwarze 886: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 887: int ln, int col, const char *msg)
888: {
889: enum mandoclevel level;
890:
1.112 schwarze 891: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 892: while (er < mandoclimits[level])
893: level--;
894:
1.108 schwarze 895: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 896: return;
897:
1.8 kristaps 898: if (m->mmsg)
899: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 900:
901: if (m->file_status < level)
902: m->file_status = level;
1.7 kristaps 903: }
904:
905: const char *
906: mparse_strerror(enum mandocerr er)
907: {
908:
1.142 schwarze 909: return mandocerrs[er];
1.7 kristaps 910: }
911:
912: const char *
913: mparse_strlevel(enum mandoclevel lvl)
914: {
1.142 schwarze 915: return mandoclevels[lvl];
1.24 kristaps 916: }
917:
918: void
919: mparse_keep(struct mparse *p)
920: {
921:
922: assert(NULL == p->secondary);
923: p->secondary = mandoc_calloc(1, sizeof(struct buf));
924: }
925:
926: const char *
927: mparse_getkeep(const struct mparse *p)
928: {
929:
930: assert(p->secondary);
1.142 schwarze 931: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 932: }
CVSweb