Annotation of mandoc/read.c, Revision 1.150.2.8
1.150.2.8! schwarze 1: /* $Id: read.c,v 1.150.2.7 2017/02/03 18:22:30 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>
1.1 kristaps 35: #include <stdio.h>
36: #include <stdlib.h>
37: #include <string.h>
38: #include <unistd.h>
1.140 schwarze 39: #include <zlib.h>
1.1 kristaps 40:
1.133 schwarze 41: #include "mandoc_aux.h"
1.1 kristaps 42: #include "mandoc.h"
1.133 schwarze 43: #include "roff.h"
1.1 kristaps 44: #include "mdoc.h"
45: #include "man.h"
1.133 schwarze 46: #include "libmandoc.h"
1.139 schwarze 47: #include "roff_int.h"
1.1 kristaps 48:
49: #define REPARSE_LIMIT 1000
50:
51: struct mparse {
52: struct roff *roff; /* roff parser (!NULL) */
1.150.2.7 schwarze 53: struct roff_man *man; /* man parser */
1.45 schwarze 54: char *sodest; /* filename pointed to by .so */
1.83 schwarze 55: const char *file; /* filename of current input file */
56: struct buf *primary; /* buffer currently being parsed */
57: struct buf *secondary; /* preprocessed copy of input */
58: const char *defos; /* default operating system */
59: mandocmsg mmsg; /* warning/error message handler */
60: enum mandoclevel file_status; /* status of current parse */
61: enum mandoclevel wlevel; /* ignore messages below this */
62: int options; /* parser options */
1.140 schwarze 63: int gzip; /* current input file is gzipped */
1.93 schwarze 64: int filenc; /* encoding of the current file */
1.1 kristaps 65: int reparse_count; /* finite interp. stack */
1.83 schwarze 66: int line; /* line number in the file */
1.1 kristaps 67: };
68:
1.84 schwarze 69: static void choose_parser(struct mparse *);
1.1 kristaps 70: static void resize_buf(struct buf *, size_t);
1.95 schwarze 71: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 72: static int read_whole_file(struct mparse *, const char *, int,
73: struct buf *, int *);
1.1 kristaps 74: static void mparse_end(struct mparse *);
1.37 schwarze 75: static void mparse_parse_buffer(struct mparse *, struct buf,
76: const char *);
1.1 kristaps 77:
1.3 kristaps 78: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
79: MANDOCERR_OK,
80: MANDOCERR_WARNING,
81: MANDOCERR_WARNING,
82: MANDOCERR_ERROR,
1.112 schwarze 83: MANDOCERR_UNSUPP,
1.3 kristaps 84: MANDOCERR_MAX,
85: MANDOCERR_MAX
86: };
87:
1.7 kristaps 88: static const char * const mandocerrs[MANDOCERR_MAX] = {
89: "ok",
90:
91: "generic warning",
92:
93: /* related to the prologue */
1.79 schwarze 94: "missing manual title, using UNTITLED",
95: "missing manual title, using \"\"",
1.54 schwarze 96: "lower case character in document title",
1.79 schwarze 97: "missing manual section, using \"\"",
1.7 kristaps 98: "unknown manual section",
1.54 schwarze 99: "missing date, using today's date",
1.7 kristaps 100: "cannot parse date, using it verbatim",
1.79 schwarze 101: "missing Os macro, using \"\"",
102: "duplicate prologue macro",
103: "late prologue macro",
104: "skipping late title macro",
1.7 kristaps 105: "prologue macros out of order",
106:
107: /* related to document structure */
108: ".so is fragile, better use ln(1)",
1.50 schwarze 109: "no document body",
1.54 schwarze 110: "content before first section header",
111: "first section is not \"NAME\"",
1.150.2.4 schwarze 112: "NAME section without Nm before Nd",
1.128 schwarze 113: "NAME section without description",
114: "description not at the end of NAME",
115: "bad NAME section content",
1.150.2.4 schwarze 116: "missing comma before name",
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",
1.150.2.4 schwarze 155: "missing section argument",
1.60 schwarze 156: "missing -std argument, adding it",
1.125 schwarze 157: "missing option string, using \"\"",
158: "missing resource identifier, using \"\"",
1.90 schwarze 159: "missing eqn box, using \"\"",
1.7 kristaps 160:
161: /* related to bad macro arguments */
1.64 schwarze 162: "unterminated quoted argument",
1.7 kristaps 163: "duplicate argument",
1.76 schwarze 164: "skipping duplicate argument",
1.63 schwarze 165: "skipping duplicate display type",
166: "skipping duplicate list type",
1.76 schwarze 167: "skipping -width argument",
1.126 schwarze 168: "wrong number of cells",
1.7 kristaps 169: "unknown AT&T UNIX version",
1.88 schwarze 170: "comma in function argument",
1.89 schwarze 171: "parenthesis in function name",
1.67 schwarze 172: "invalid content in Rs block",
1.63 schwarze 173: "invalid Boolean argument",
174: "unknown font, skipping request",
1.126 schwarze 175: "odd number of characters in request",
1.7 kristaps 176:
177: /* related to plain text */
1.64 schwarze 178: "blank line in fill mode, using .sp",
179: "tab in filled text",
180: "whitespace at end of input line",
1.150.2.6 schwarze 181: "new sentence, new 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: {
1.150.2.8! schwarze 607: #if HAVE_MMAP
! 608: struct stat st;
! 609: #endif
1.140 schwarze 610: gzFile gz;
1.1 kristaps 611: size_t off;
612: ssize_t ssz;
613:
1.81 schwarze 614: #if HAVE_MMAP
1.143 schwarze 615: if (fstat(fd, &st) == -1)
616: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 617:
618: /*
619: * If we're a regular file, try just reading in the whole entry
620: * via mmap(). This is faster than reading it into blocks, and
621: * since each file is only a few bytes to begin with, I'm not
622: * concerned that this is going to tank any machines.
623: */
624:
1.140 schwarze 625: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 626: if (st.st_size > 0x7fffffff) {
1.111 schwarze 627: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 628: return 0;
1.1 kristaps 629: }
630: *with_mmap = 1;
631: fb->sz = (size_t)st.st_size;
1.37 schwarze 632: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 633: if (fb->buf != MAP_FAILED)
1.142 schwarze 634: return 1;
1.1 kristaps 635: }
1.15 kristaps 636: #endif
1.1 kristaps 637:
1.140 schwarze 638: if (curp->gzip) {
1.143 schwarze 639: if ((gz = gzdopen(fd, "rb")) == NULL)
640: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 641: } else
642: gz = NULL;
643:
1.1 kristaps 644: /*
645: * If this isn't a regular file (like, say, stdin), then we must
646: * go the old way and just read things in bit by bit.
647: */
648:
649: *with_mmap = 0;
650: off = 0;
651: fb->sz = 0;
652: fb->buf = NULL;
653: for (;;) {
654: if (off == fb->sz) {
655: if (fb->sz == (1U << 31)) {
1.111 schwarze 656: mandoc_msg(MANDOCERR_TOOLARGE, curp,
657: 0, 0, NULL);
1.1 kristaps 658: break;
659: }
660: resize_buf(fb, 65536);
661: }
1.140 schwarze 662: ssz = curp->gzip ?
663: gzread(gz, fb->buf + (int)off, fb->sz - off) :
664: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 665: if (ssz == 0) {
666: fb->sz = off;
1.142 schwarze 667: return 1;
1.1 kristaps 668: }
1.143 schwarze 669: if (ssz == -1)
670: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 671: off += (size_t)ssz;
672: }
673:
674: free(fb->buf);
675: fb->buf = NULL;
1.142 schwarze 676: return 0;
1.1 kristaps 677: }
678:
679: static void
680: mparse_end(struct mparse *curp)
681: {
1.135 schwarze 682: if (curp->man->macroset == MACROSET_NONE)
683: curp->man->macroset = MACROSET_MAN;
684: if (curp->man->macroset == MACROSET_MDOC)
685: mdoc_endparse(curp->man);
686: else
1.111 schwarze 687: man_endparse(curp->man);
1.1 kristaps 688: roff_endparse(curp->roff);
689: }
690:
691: static void
1.36 schwarze 692: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 693: {
1.85 schwarze 694: struct buf *svprimary;
1.28 joerg 695: const char *svfile;
1.95 schwarze 696: size_t offset;
1.36 schwarze 697: static int recursion_depth;
698:
699: if (64 < recursion_depth) {
700: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
701: return;
702: }
1.28 joerg 703:
704: /* Line number is per-file. */
705: svfile = curp->file;
706: curp->file = file;
1.85 schwarze 707: svprimary = curp->primary;
1.83 schwarze 708: curp->primary = &blk;
1.28 joerg 709: curp->line = 1;
1.36 schwarze 710: recursion_depth++;
1.28 joerg 711:
1.93 schwarze 712: /* Skip an UTF-8 byte order mark. */
713: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
714: (unsigned char)blk.buf[0] == 0xef &&
715: (unsigned char)blk.buf[1] == 0xbb &&
716: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 717: offset = 3;
1.93 schwarze 718: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 719: } else
720: offset = 0;
1.93 schwarze 721:
1.95 schwarze 722: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 723:
1.111 schwarze 724: if (--recursion_depth == 0)
1.28 joerg 725: mparse_end(curp);
726:
1.85 schwarze 727: curp->primary = svprimary;
1.28 joerg 728: curp->file = svfile;
729: }
730:
731: enum mandoclevel
1.104 schwarze 732: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 733: const char *file)
734: {
735: struct buf blk;
736:
1.104 schwarze 737: blk.buf = buf;
1.28 joerg 738: blk.sz = len;
739:
1.36 schwarze 740: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 741: return curp->file_status;
1.28 joerg 742: }
743:
1.98 schwarze 744: /*
745: * Read the whole file into memory and call the parsers.
746: * Called recursively when an .so request is encountered.
747: */
1.36 schwarze 748: enum mandoclevel
749: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 750: {
1.28 joerg 751: struct buf blk;
752: int with_mmap;
1.93 schwarze 753: int save_filenc;
1.1 kristaps 754:
1.91 schwarze 755: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 756: save_filenc = curp->filenc;
757: curp->filenc = curp->options &
758: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 759: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 760: curp->filenc = save_filenc;
1.81 schwarze 761: #if HAVE_MMAP
1.91 schwarze 762: if (with_mmap)
763: munmap(blk.buf, blk.sz);
764: else
1.28 joerg 765: #endif
1.91 schwarze 766: free(blk.buf);
767: }
1.142 schwarze 768: return curp->file_status;
1.82 schwarze 769: }
770:
1.148 schwarze 771: int
772: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 773: {
774: char *cp;
1.148 schwarze 775: int fd;
1.82 schwarze 776:
777: curp->file = file;
1.140 schwarze 778: cp = strrchr(file, '.');
779: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 780:
1.140 schwarze 781: /* First try to use the filename as it is. */
1.98 schwarze 782:
1.148 schwarze 783: if ((fd = open(file, O_RDONLY)) != -1)
784: return fd;
1.98 schwarze 785:
1.140 schwarze 786: /*
787: * If that doesn't work and the filename doesn't
788: * already end in .gz, try appending .gz.
789: */
1.98 schwarze 790:
1.140 schwarze 791: if ( ! curp->gzip) {
1.98 schwarze 792: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 793: fd = open(cp, O_RDONLY);
1.108 schwarze 794: free(cp);
1.148 schwarze 795: if (fd != -1) {
1.140 schwarze 796: curp->gzip = 1;
1.148 schwarze 797: return fd;
1.82 schwarze 798: }
799: }
800:
1.140 schwarze 801: /* Neither worked, give up. */
1.97 schwarze 802:
1.140 schwarze 803: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 804: return -1;
1.1 kristaps 805: }
806:
807: struct mparse *
1.94 schwarze 808: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.144 schwarze 809: const char *defos)
1.1 kristaps 810: {
811: struct mparse *curp;
1.10 kristaps 812:
1.1 kristaps 813: curp = mandoc_calloc(1, sizeof(struct mparse));
814:
1.44 schwarze 815: curp->options = options;
1.3 kristaps 816: curp->wlevel = wlevel;
1.1 kristaps 817: curp->mmsg = mmsg;
1.29 schwarze 818: curp->defos = defos;
1.1 kristaps 819:
1.144 schwarze 820: curp->roff = roff_alloc(curp, options);
1.137 schwarze 821: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
822: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 823: if (curp->options & MPARSE_MDOC) {
824: mdoc_hash_init();
1.137 schwarze 825: curp->man->macroset = MACROSET_MDOC;
826: } else if (curp->options & MPARSE_MAN) {
1.136 schwarze 827: man_hash_init();
1.137 schwarze 828: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 829: }
1.138 schwarze 830: curp->man->first->tok = TOKEN_NONE;
1.142 schwarze 831: return curp;
1.1 kristaps 832: }
833:
834: void
835: mparse_reset(struct mparse *curp)
836: {
837: roff_reset(curp->roff);
1.150 schwarze 838: roff_man_reset(curp->man);
1.150.2.7 schwarze 839:
840: free(curp->sodest);
841: curp->sodest = NULL;
842:
1.24 kristaps 843: if (curp->secondary)
844: curp->secondary->sz = 0;
1.1 kristaps 845:
846: curp->file_status = MANDOCLEVEL_OK;
1.150.2.7 schwarze 847: curp->gzip = 0;
1.1 kristaps 848: }
849:
850: void
851: mparse_free(struct mparse *curp)
852: {
853:
1.137 schwarze 854: roff_man_free(curp->man);
1.150.2.7 schwarze 855: roff_free(curp->roff);
1.24 kristaps 856: if (curp->secondary)
857: free(curp->secondary->buf);
1.1 kristaps 858:
1.24 kristaps 859: free(curp->secondary);
1.45 schwarze 860: free(curp->sodest);
1.1 kristaps 861: free(curp);
862: }
863:
864: void
1.135 schwarze 865: mparse_result(struct mparse *curp, struct roff_man **man,
866: char **sodest)
1.1 kristaps 867: {
868:
1.45 schwarze 869: if (sodest && NULL != (*sodest = curp->sodest)) {
870: *man = NULL;
871: return;
872: }
1.9 kristaps 873: if (man)
874: *man = curp->man;
1.150.2.5 schwarze 875: }
876:
877: void
878: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
879: {
880: if (curp->file_status > *rc)
881: *rc = curp->file_status;
1.3 kristaps 882: }
883:
884: void
885: mandoc_vmsg(enum mandocerr t, struct mparse *m,
886: int ln, int pos, const char *fmt, ...)
887: {
888: char buf[256];
889: va_list ap;
890:
891: va_start(ap, fmt);
1.48 schwarze 892: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 893: va_end(ap);
894:
895: mandoc_msg(t, m, ln, pos, buf);
896: }
897:
898: void
1.47 schwarze 899: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 900: int ln, int col, const char *msg)
901: {
902: enum mandoclevel level;
903:
1.112 schwarze 904: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 905: while (er < mandoclimits[level])
906: level--;
907:
1.108 schwarze 908: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 909: return;
910:
1.8 kristaps 911: if (m->mmsg)
912: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 913:
914: if (m->file_status < level)
915: m->file_status = level;
1.7 kristaps 916: }
917:
918: const char *
919: mparse_strerror(enum mandocerr er)
920: {
921:
1.142 schwarze 922: return mandocerrs[er];
1.7 kristaps 923: }
924:
925: const char *
926: mparse_strlevel(enum mandoclevel lvl)
927: {
1.142 schwarze 928: return mandoclevels[lvl];
1.24 kristaps 929: }
930:
931: void
932: mparse_keep(struct mparse *p)
933: {
934:
935: assert(NULL == p->secondary);
936: p->secondary = mandoc_calloc(1, sizeof(struct buf));
937: }
938:
939: const char *
940: mparse_getkeep(const struct mparse *p)
941: {
942:
943: assert(p->secondary);
1.142 schwarze 944: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 945: }
CVSweb