Annotation of src/usr.bin/mandoc/read.c, Revision 1.154
1.154 ! schwarze 1: /* $OpenBSD: read.c,v 1.153 2017/06/24 18:58:09 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.129 schwarze 4: * Copyright (c) 2010-2017 Ingo Schwarze <schwarze@openbsd.org>
1.20 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 schwarze 6: *
7: * Permission to use, copy, modify, and distribute this software for any
8: * purpose with or without fee is hereby granted, provided that the above
9: * copyright notice and this permission notice appear in all copies.
10: *
1.108 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.108 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18: */
1.58 schwarze 19: #include <sys/types.h>
20: #include <sys/mman.h>
1.1 schwarze 21: #include <sys/stat.h>
22:
23: #include <assert.h>
24: #include <ctype.h>
1.118 schwarze 25: #include <err.h>
1.18 schwarze 26: #include <errno.h>
1.1 schwarze 27: #include <fcntl.h>
28: #include <stdarg.h>
29: #include <stdio.h>
30: #include <stdlib.h>
31: #include <string.h>
32: #include <unistd.h>
1.115 schwarze 33: #include <zlib.h>
1.1 schwarze 34:
1.108 schwarze 35: #include "mandoc_aux.h"
1.1 schwarze 36: #include "mandoc.h"
1.108 schwarze 37: #include "roff.h"
1.1 schwarze 38: #include "mdoc.h"
39: #include "man.h"
1.108 schwarze 40: #include "libmandoc.h"
1.114 schwarze 41: #include "roff_int.h"
1.1 schwarze 42:
43: #define REPARSE_LIMIT 1000
44:
45: struct mparse {
1.133 schwarze 46: struct roff *roff; /* roff parser (!NULL) */
1.109 schwarze 47: struct roff_man *man; /* man parser */
1.23 schwarze 48: char *sodest; /* filename pointed to by .so */
1.59 schwarze 49: const char *file; /* filename of current input file */
50: struct buf *primary; /* buffer currently being parsed */
51: struct buf *secondary; /* preprocessed copy of input */
1.151 schwarze 52: const char *os_s; /* default operating system */
1.59 schwarze 53: mandocmsg mmsg; /* warning/error message handler */
54: enum mandoclevel file_status; /* status of current parse */
1.151 schwarze 55: enum mandocerr mmin; /* ignore messages below this */
1.59 schwarze 56: int options; /* parser options */
1.115 schwarze 57: int gzip; /* current input file is gzipped */
1.70 schwarze 58: int filenc; /* encoding of the current file */
1.1 schwarze 59: int reparse_count; /* finite interp. stack */
1.59 schwarze 60: int line; /* line number in the file */
1.1 schwarze 61: };
62:
1.60 schwarze 63: static void choose_parser(struct mparse *);
1.1 schwarze 64: static void resize_buf(struct buf *, size_t);
1.135 schwarze 65: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.18 schwarze 66: static int read_whole_file(struct mparse *, const char *, int,
67: struct buf *, int *);
1.1 schwarze 68: static void mparse_end(struct mparse *);
1.15 schwarze 69: static void mparse_parse_buffer(struct mparse *, struct buf,
70: const char *);
1.1 schwarze 71:
72: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
73: MANDOCERR_OK,
1.137 schwarze 74: MANDOCERR_STYLE,
1.1 schwarze 75: MANDOCERR_WARNING,
76: MANDOCERR_ERROR,
1.87 schwarze 77: MANDOCERR_UNSUPP,
1.1 schwarze 78: MANDOCERR_MAX,
79: MANDOCERR_MAX
80: };
81:
82: static const char * const mandocerrs[MANDOCERR_MAX] = {
83: "ok",
84:
1.151 schwarze 85: "base system convention",
1.138 schwarze 86:
1.148 schwarze 87: "Mdocdate found",
88: "Mdocdate missing",
1.153 schwarze 89: "unknown architecture",
1.152 schwarze 90: "operating system explicitly specified",
1.151 schwarze 91: "RCS id missing",
92:
93: "generic style suggestion",
94:
1.148 schwarze 95: "legacy man(7) date format",
1.150 schwarze 96: "duplicate RCS id",
1.154 ! schwarze 97: "typo in section name",
1.138 schwarze 98: "useless macro",
1.139 schwarze 99: "consider using OS macro",
1.144 schwarze 100: "errnos out of order",
101: "duplicate errno",
1.140 schwarze 102: "description line ends with a full stop",
1.146 schwarze 103: "no blank before trailing delimiter",
1.147 schwarze 104: "function name without markup",
1.138 schwarze 105:
1.1 schwarze 106: "generic warning",
107:
108: /* related to the prologue */
1.57 schwarze 109: "missing manual title, using UNTITLED",
110: "missing manual title, using \"\"",
1.32 schwarze 111: "lower case character in document title",
1.57 schwarze 112: "missing manual section, using \"\"",
1.1 schwarze 113: "unknown manual section",
1.32 schwarze 114: "missing date, using today's date",
1.1 schwarze 115: "cannot parse date, using it verbatim",
1.57 schwarze 116: "missing Os macro, using \"\"",
117: "duplicate prologue macro",
118: "late prologue macro",
119: "skipping late title macro",
1.1 schwarze 120: "prologue macros out of order",
121:
122: /* related to document structure */
123: ".so is fragile, better use ln(1)",
1.28 schwarze 124: "no document body",
1.32 schwarze 125: "content before first section header",
126: "first section is not \"NAME\"",
1.129 schwarze 127: "NAME section without Nm before Nd",
1.103 schwarze 128: "NAME section without description",
129: "description not at the end of NAME",
130: "bad NAME section content",
1.129 schwarze 131: "missing comma before name",
1.96 schwarze 132: "missing description line, using \"\"",
1.134 schwarze 133: "description line outside NAME section",
1.1 schwarze 134: "sections out of conventional order",
1.32 schwarze 135: "duplicate section title",
136: "unexpected section",
1.63 schwarze 137: "unusual Xr order",
138: "unusual Xr punctuation",
1.62 schwarze 139: "AUTHORS section without An macro",
1.1 schwarze 140:
141: /* related to macros and nesting */
1.33 schwarze 142: "obsolete macro",
1.78 schwarze 143: "macro neither callable nor escaped",
1.1 schwarze 144: "skipping paragraph macro",
1.10 schwarze 145: "moving paragraph macro out of list",
1.1 schwarze 146: "skipping no-space macro",
147: "blocks badly nested",
148: "nested displays are not portable",
1.35 schwarze 149: "moving content out of list",
1.56 schwarze 150: "fill mode already enabled, skipping",
151: "fill mode already disabled, skipping",
1.1 schwarze 152: "line scope broken",
1.141 schwarze 153: "skipping blank line in line scope",
1.1 schwarze 154:
155: /* related to missing macro arguments */
1.36 schwarze 156: "skipping empty request",
157: "conditional request controls empty scope",
1.1 schwarze 158: "skipping empty macro",
1.99 schwarze 159: "empty block",
1.40 schwarze 160: "empty argument, using 0n",
1.38 schwarze 161: "missing display type, using -ragged",
162: "list type is not the first argument",
1.125 schwarze 163: "missing -width in -tag list, using 6n",
1.56 schwarze 164: "missing utility name, using \"\"",
1.98 schwarze 165: "missing function name, using \"\"",
1.38 schwarze 166: "empty head in list item",
167: "empty list item",
1.39 schwarze 168: "missing font type, using \\fR",
169: "unknown font type, using \\fR",
1.79 schwarze 170: "nothing follows prefix",
1.97 schwarze 171: "empty reference block",
1.128 schwarze 172: "missing section argument",
1.38 schwarze 173: "missing -std argument, adding it",
1.100 schwarze 174: "missing option string, using \"\"",
175: "missing resource identifier, using \"\"",
1.66 schwarze 176: "missing eqn box, using \"\"",
1.1 schwarze 177:
178: /* related to bad macro arguments */
1.42 schwarze 179: "unterminated quoted argument",
1.1 schwarze 180: "duplicate argument",
1.54 schwarze 181: "skipping duplicate argument",
1.41 schwarze 182: "skipping duplicate display type",
183: "skipping duplicate list type",
1.54 schwarze 184: "skipping -width argument",
1.101 schwarze 185: "wrong number of cells",
1.1 schwarze 186: "unknown AT&T UNIX version",
1.64 schwarze 187: "comma in function argument",
1.65 schwarze 188: "parenthesis in function name",
1.145 schwarze 189: "unknown library name",
1.45 schwarze 190: "invalid content in Rs block",
1.41 schwarze 191: "invalid Boolean argument",
192: "unknown font, skipping request",
1.101 schwarze 193: "odd number of characters in request",
1.1 schwarze 194:
195: /* related to plain text */
1.42 schwarze 196: "blank line in fill mode, using .sp",
197: "tab in filled text",
198: "whitespace at end of input line",
1.131 schwarze 199: "new sentence, new line",
1.1 schwarze 200: "bad comment style",
1.42 schwarze 201: "invalid escape sequence",
202: "undefined string, using \"\"",
1.3 schwarze 203:
1.93 schwarze 204: /* related to tables */
205: "tbl line starts with span",
206: "tbl column starts with span",
207: "skipping vertical bar in tbl layout",
208:
1.1 schwarze 209: "generic error",
1.3 schwarze 210:
1.1 schwarze 211: /* related to tables */
1.91 schwarze 212: "non-alphabetic character in tbl options",
213: "skipping unknown tbl option",
214: "missing tbl option argument",
215: "wrong tbl option argument size",
1.93 schwarze 216: "empty tbl layout",
217: "invalid character in tbl layout",
218: "unmatched parenthesis in tbl layout",
1.94 schwarze 219: "tbl without any data cells",
220: "ignoring data in spanned tbl cell",
221: "ignoring extra tbl data cells",
222: "data block open at end of tbl",
1.1 schwarze 223:
1.46 schwarze 224: /* related to document structure and macros */
1.83 schwarze 225: NULL,
1.1 schwarze 226: "input stack limit exceeded, infinite loop?",
227: "skipping bad character",
1.46 schwarze 228: "skipping unknown macro",
1.87 schwarze 229: "skipping insecure request",
1.51 schwarze 230: "skipping item outside list",
1.46 schwarze 231: "skipping column outside column list",
232: "skipping end of block that is not open",
1.90 schwarze 233: "fewer RS blocks open, skipping",
1.46 schwarze 234: "inserting missing end of block",
235: "appending missing end of block",
236:
237: /* related to request and macro arguments */
1.1 schwarze 238: "escaped character not allowed in a name",
1.75 schwarze 239: "NOT IMPLEMENTED: Bd -file",
1.120 schwarze 240: "skipping display without arguments",
1.49 schwarze 241: "missing list type, using -item",
1.143 schwarze 242: "argument is not numeric, using 1",
1.48 schwarze 243: "missing manual name, using \"\"",
1.49 schwarze 244: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 245: "unknown standard specifier",
1.49 schwarze 246: "skipping request without numeric argument",
1.84 schwarze 247: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 248: ".so request failed",
1.39 schwarze 249: "skipping all arguments",
250: "skipping excess arguments",
1.68 schwarze 251: "divide by zero",
1.87 schwarze 252:
253: "unsupported feature",
254: "input too large",
1.89 schwarze 255: "unsupported control character",
1.87 schwarze 256: "unsupported roff request",
1.94 schwarze 257: "eqn delim option in tbl",
1.93 schwarze 258: "unsupported tbl layout modifier",
1.87 schwarze 259: "ignoring macro in table",
1.1 schwarze 260: };
261:
262: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
263: "SUCCESS",
1.137 schwarze 264: "STYLE",
1.1 schwarze 265: "WARNING",
266: "ERROR",
1.87 schwarze 267: "UNSUPP",
1.1 schwarze 268: "BADARG",
269: "SYSERR"
270: };
271:
1.25 schwarze 272:
1.1 schwarze 273: static void
274: resize_buf(struct buf *buf, size_t initial)
275: {
276:
277: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
278: buf->buf = mandoc_realloc(buf->buf, buf->sz);
279: }
280:
281: static void
1.60 schwarze 282: choose_parser(struct mparse *curp)
1.1 schwarze 283: {
1.59 schwarze 284: char *cp, *ep;
285: int format;
1.1 schwarze 286:
1.59 schwarze 287: /*
288: * If neither command line arguments -mdoc or -man select
289: * a parser nor the roff parser found a .Dd or .TH macro
290: * yet, look ahead in the main input buffer.
291: */
292:
293: if ((format = roff_getformat(curp->roff)) == 0) {
294: cp = curp->primary->buf;
295: ep = cp + curp->primary->sz;
296: while (cp < ep) {
1.61 schwarze 297: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 298: cp++;
299: if (cp[0] == 'D' && cp[1] == 'd') {
300: format = MPARSE_MDOC;
301: break;
302: }
303: if (cp[0] == 'T' && cp[1] == 'H') {
304: format = MPARSE_MAN;
305: break;
306: }
307: }
308: cp = memchr(cp, '\n', ep - cp);
309: if (cp == NULL)
310: break;
311: cp++;
312: }
1.1 schwarze 313: }
314:
1.59 schwarze 315: if (format == MPARSE_MDOC) {
1.112 schwarze 316: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 317: if (curp->man->mdocmac == NULL)
318: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 319: } else {
320: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 321: if (curp->man->manmac == NULL)
322: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 323: }
1.136 schwarze 324: curp->man->first->tok = TOKEN_NONE;
1.1 schwarze 325: }
326:
327: /*
1.71 schwarze 328: * Main parse routine for a buffer.
329: * It assumes encoding and line numbering are already set up.
330: * It can recurse directly (for invocations of user-defined
331: * macros, inline equations, and input line traps)
332: * and indirectly (for .so file inclusion).
1.1 schwarze 333: */
1.135 schwarze 334: static int
1.71 schwarze 335: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 336: {
337: const struct tbl_span *span;
338: struct buf ln;
1.88 schwarze 339: const char *save_file;
1.85 schwarze 340: char *cp;
1.71 schwarze 341: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 342: enum rofferr rr;
1.76 schwarze 343: int of;
1.1 schwarze 344: int lnn; /* line number in the real file */
1.85 schwarze 345: int fd;
1.1 schwarze 346: unsigned char c;
347:
1.71 schwarze 348: memset(&ln, 0, sizeof(ln));
1.1 schwarze 349:
1.25 schwarze 350: lnn = curp->line;
351: pos = 0;
1.1 schwarze 352:
1.71 schwarze 353: while (i < blk.sz) {
1.1 schwarze 354: if (0 == pos && '\0' == blk.buf[i])
355: break;
356:
357: if (start) {
358: curp->line = lnn;
359: curp->reparse_count = 0;
1.70 schwarze 360:
361: if (lnn < 3 &&
362: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 363: curp->filenc & MPARSE_LATIN1)
364: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 365: }
366:
1.71 schwarze 367: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 368:
369: /*
370: * When finding an unescaped newline character,
371: * leave the character loop to process the line.
372: * Skip a preceding carriage return, if any.
373: */
374:
1.71 schwarze 375: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 376: '\n' == blk.buf[i + 1])
377: ++i;
378: if ('\n' == blk.buf[i]) {
379: ++i;
380: ++lnn;
381: break;
382: }
383:
1.13 schwarze 384: /*
1.70 schwarze 385: * Make sure we have space for the worst
386: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 387: */
388:
1.71 schwarze 389: if (pos + 11 > ln.sz)
1.13 schwarze 390: resize_buf(&ln, 256);
391:
1.25 schwarze 392: /*
1.70 schwarze 393: * Encode 8-bit input.
1.1 schwarze 394: */
395:
1.70 schwarze 396: c = blk.buf[i];
397: if (c & 0x80) {
1.71 schwarze 398: if ( ! (curp->filenc && preconv_encode(
399: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 400: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
401: curp->line, pos, "0x%x", c);
1.70 schwarze 402: ln.buf[pos++] = '?';
403: i++;
404: }
405: continue;
406: }
407:
408: /*
409: * Exclude control characters.
410: */
1.1 schwarze 411:
1.70 schwarze 412: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 413: mandoc_vmsg(c == 0x00 || c == 0x04 ||
414: c > 0x0a ? MANDOCERR_CHAR_BAD :
415: MANDOCERR_CHAR_UNSUPP,
416: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 417: i++;
1.102 schwarze 418: if (c != '\r')
419: ln.buf[pos++] = '?';
1.1 schwarze 420: continue;
421: }
422:
423: ln.buf[pos++] = blk.buf[i++];
424: }
425:
1.142 schwarze 426: if (pos + 1 >= ln.sz)
1.1 schwarze 427: resize_buf(&ln, 256);
428:
1.142 schwarze 429: if (i == blk.sz || blk.buf[i] == '\0')
430: ln.buf[pos++] = '\n';
1.1 schwarze 431: ln.buf[pos] = '\0';
432:
433: /*
434: * A significant amount of complexity is contained by
435: * the roff preprocessor. It's line-oriented but can be
436: * expressed on one line, so we need at times to
437: * readjust our starting point and re-run it. The roff
438: * preprocessor can also readjust the buffers with new
439: * data, so we pass them in wholesale.
440: */
441:
442: of = 0;
443:
1.4 schwarze 444: /*
445: * Maintain a lookaside buffer of all parsed lines. We
446: * only do this if mparse_keep() has been invoked (the
447: * buffer may be accessed with mparse_getkeep()).
448: */
449:
450: if (curp->secondary) {
1.25 schwarze 451: curp->secondary->buf = mandoc_realloc(
452: curp->secondary->buf,
453: curp->secondary->sz + pos + 2);
454: memcpy(curp->secondary->buf +
455: curp->secondary->sz,
456: ln.buf, pos);
1.4 schwarze 457: curp->secondary->sz += pos;
458: curp->secondary->buf
459: [curp->secondary->sz] = '\n';
460: curp->secondary->sz++;
461: curp->secondary->buf
462: [curp->secondary->sz] = '\0';
463: }
1.1 schwarze 464: rerun:
1.72 schwarze 465: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 466:
467: switch (rr) {
1.25 schwarze 468: case ROFF_REPARSE:
1.135 schwarze 469: if (++curp->reparse_count > REPARSE_LIMIT)
1.1 schwarze 470: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 471: curp->line, pos, NULL);
1.135 schwarze 472: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
473: start == 1) {
474: pos = 0;
475: continue;
476: }
477: free(ln.buf);
478: return 0;
1.25 schwarze 479: case ROFF_APPEND:
1.71 schwarze 480: pos = strlen(ln.buf);
1.1 schwarze 481: continue;
1.25 schwarze 482: case ROFF_RERUN:
1.1 schwarze 483: goto rerun;
1.25 schwarze 484: case ROFF_IGN:
1.1 schwarze 485: pos = 0;
486: continue;
1.25 schwarze 487: case ROFF_SO:
1.71 schwarze 488: if ( ! (curp->options & MPARSE_SO) &&
489: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 490: curp->sodest = mandoc_strdup(ln.buf + of);
491: free(ln.buf);
1.135 schwarze 492: return 1;
1.23 schwarze 493: }
1.4 schwarze 494: /*
495: * We remove `so' clauses from our lookaside
496: * buffer because we're going to descend into
497: * the file recursively.
498: */
1.25 schwarze 499: if (curp->secondary)
1.4 schwarze 500: curp->secondary->sz -= pos + 1;
1.88 schwarze 501: save_file = curp->file;
1.122 schwarze 502: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 503: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 504: close(fd);
1.88 schwarze 505: curp->file = save_file;
506: } else {
507: curp->file = save_file;
1.30 schwarze 508: mandoc_vmsg(MANDOCERR_SO_FAIL,
509: curp, curp->line, pos,
510: ".so %s", ln.buf + of);
1.85 schwarze 511: ln.sz = mandoc_asprintf(&cp,
512: ".sp\nSee the file %s.\n.sp",
513: ln.buf + of);
514: free(ln.buf);
515: ln.buf = cp;
516: of = 0;
517: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 518: }
1.1 schwarze 519: pos = 0;
520: continue;
521: default:
522: break;
523: }
524:
1.124 schwarze 525: if (curp->man->macroset == MACROSET_NONE)
1.60 schwarze 526: choose_parser(curp);
1.1 schwarze 527:
1.25 schwarze 528: /*
1.60 schwarze 529: * Lastly, push down into the parsers themselves.
1.1 schwarze 530: * If libroff returns ROFF_TBL, then add it to the
531: * currently open parse. Since we only get here if
532: * there does exist data (see tbl_data.c), we're
533: * guaranteed that something's been allocated.
534: * Do the same for ROFF_EQN.
535: */
536:
1.114 schwarze 537: if (rr == ROFF_TBL)
1.76 schwarze 538: while ((span = roff_span(curp->roff)) != NULL)
1.114 schwarze 539: roff_addtbl(curp->man, span);
540: else if (rr == ROFF_EQN)
541: roff_addeqn(curp->man, roff_eqn(curp->roff));
542: else if ((curp->man->macroset == MACROSET_MDOC ?
1.110 schwarze 543: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 544: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
545: break;
1.1 schwarze 546:
547: /* Temporary buffers typically are not full. */
548:
549: if (0 == start && '\0' == blk.buf[i])
550: break;
551:
552: /* Start the next input line. */
553:
554: pos = 0;
555: }
556:
557: free(ln.buf);
1.135 schwarze 558: return 1;
1.1 schwarze 559: }
560:
561: static int
1.18 schwarze 562: read_whole_file(struct mparse *curp, const char *file, int fd,
563: struct buf *fb, int *with_mmap)
1.1 schwarze 564: {
565: struct stat st;
1.115 schwarze 566: gzFile gz;
1.1 schwarze 567: size_t off;
568: ssize_t ssz;
569:
1.118 schwarze 570: if (fstat(fd, &st) == -1)
571: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 572:
573: /*
574: * If we're a regular file, try just reading in the whole entry
575: * via mmap(). This is faster than reading it into blocks, and
576: * since each file is only a few bytes to begin with, I'm not
577: * concerned that this is going to tank any machines.
578: */
579:
1.115 schwarze 580: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 581: if (st.st_size > 0x7fffffff) {
1.86 schwarze 582: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.117 schwarze 583: return 0;
1.1 schwarze 584: }
585: *with_mmap = 1;
586: fb->sz = (size_t)st.st_size;
1.15 schwarze 587: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 588: if (fb->buf != MAP_FAILED)
1.117 schwarze 589: return 1;
1.1 schwarze 590: }
591:
1.115 schwarze 592: if (curp->gzip) {
1.118 schwarze 593: if ((gz = gzdopen(fd, "rb")) == NULL)
594: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.115 schwarze 595: } else
596: gz = NULL;
597:
1.1 schwarze 598: /*
599: * If this isn't a regular file (like, say, stdin), then we must
600: * go the old way and just read things in bit by bit.
601: */
602:
603: *with_mmap = 0;
604: off = 0;
605: fb->sz = 0;
606: fb->buf = NULL;
607: for (;;) {
608: if (off == fb->sz) {
609: if (fb->sz == (1U << 31)) {
1.86 schwarze 610: mandoc_msg(MANDOCERR_TOOLARGE, curp,
611: 0, 0, NULL);
1.1 schwarze 612: break;
613: }
614: resize_buf(fb, 65536);
615: }
1.115 schwarze 616: ssz = curp->gzip ?
617: gzread(gz, fb->buf + (int)off, fb->sz - off) :
618: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 619: if (ssz == 0) {
620: fb->sz = off;
1.117 schwarze 621: return 1;
1.1 schwarze 622: }
1.118 schwarze 623: if (ssz == -1)
624: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 625: off += (size_t)ssz;
626: }
627:
628: free(fb->buf);
629: fb->buf = NULL;
1.117 schwarze 630: return 0;
1.1 schwarze 631: }
632:
633: static void
634: mparse_end(struct mparse *curp)
635: {
1.110 schwarze 636: if (curp->man->macroset == MACROSET_NONE)
637: curp->man->macroset = MACROSET_MAN;
638: if (curp->man->macroset == MACROSET_MDOC)
639: mdoc_endparse(curp->man);
640: else
1.86 schwarze 641: man_endparse(curp->man);
1.1 schwarze 642: roff_endparse(curp->roff);
643: }
644:
1.15 schwarze 645: static void
646: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 647: {
1.61 schwarze 648: struct buf *svprimary;
1.1 schwarze 649: const char *svfile;
1.71 schwarze 650: size_t offset;
1.14 schwarze 651: static int recursion_depth;
652:
653: if (64 < recursion_depth) {
654: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 655: return;
1.14 schwarze 656: }
1.1 schwarze 657:
1.15 schwarze 658: /* Line number is per-file. */
659: svfile = curp->file;
660: curp->file = file;
1.61 schwarze 661: svprimary = curp->primary;
1.59 schwarze 662: curp->primary = &blk;
1.15 schwarze 663: curp->line = 1;
664: recursion_depth++;
665:
1.70 schwarze 666: /* Skip an UTF-8 byte order mark. */
667: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
668: (unsigned char)blk.buf[0] == 0xef &&
669: (unsigned char)blk.buf[1] == 0xbb &&
670: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 671: offset = 3;
1.70 schwarze 672: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 673: } else
674: offset = 0;
1.70 schwarze 675:
1.71 schwarze 676: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 677:
1.86 schwarze 678: if (--recursion_depth == 0)
1.15 schwarze 679: mparse_end(curp);
680:
1.61 schwarze 681: curp->primary = svprimary;
1.15 schwarze 682: curp->file = svfile;
683: }
684:
1.74 schwarze 685: /*
686: * Read the whole file into memory and call the parsers.
687: * Called recursively when an .so request is encountered.
688: */
1.15 schwarze 689: enum mandoclevel
690: mparse_readfd(struct mparse *curp, int fd, const char *file)
691: {
692: struct buf blk;
693: int with_mmap;
1.70 schwarze 694: int save_filenc;
1.1 schwarze 695:
1.67 schwarze 696: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 697: save_filenc = curp->filenc;
698: curp->filenc = curp->options &
699: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 700: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 701: curp->filenc = save_filenc;
1.67 schwarze 702: if (with_mmap)
703: munmap(blk.buf, blk.sz);
704: else
705: free(blk.buf);
706: }
1.117 schwarze 707: return curp->file_status;
1.58 schwarze 708: }
709:
1.122 schwarze 710: int
711: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 712: {
713: char *cp;
1.122 schwarze 714: int fd;
1.58 schwarze 715:
716: curp->file = file;
1.115 schwarze 717: cp = strrchr(file, '.');
718: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 719:
1.115 schwarze 720: /* First try to use the filename as it is. */
1.74 schwarze 721:
1.122 schwarze 722: if ((fd = open(file, O_RDONLY)) != -1)
723: return fd;
1.74 schwarze 724:
1.115 schwarze 725: /*
726: * If that doesn't work and the filename doesn't
727: * already end in .gz, try appending .gz.
728: */
1.74 schwarze 729:
1.115 schwarze 730: if ( ! curp->gzip) {
1.74 schwarze 731: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 732: fd = open(cp, O_RDONLY);
1.83 schwarze 733: free(cp);
1.122 schwarze 734: if (fd != -1) {
1.115 schwarze 735: curp->gzip = 1;
1.122 schwarze 736: return fd;
1.58 schwarze 737: }
738: }
739:
1.115 schwarze 740: /* Neither worked, give up. */
1.73 schwarze 741:
1.115 schwarze 742: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.122 schwarze 743: return -1;
1.1 schwarze 744: }
745:
746: struct mparse *
1.151 schwarze 747: mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
748: enum mandoc_os os_e, const char *os_s)
1.1 schwarze 749: {
750: struct mparse *curp;
751:
752: curp = mandoc_calloc(1, sizeof(struct mparse));
753:
1.22 schwarze 754: curp->options = options;
1.151 schwarze 755: curp->mmin = mmin;
1.1 schwarze 756: curp->mmsg = mmsg;
1.151 schwarze 757: curp->os_s = os_s;
1.1 schwarze 758:
1.119 schwarze 759: curp->roff = roff_alloc(curp, options);
1.151 schwarze 760: curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
1.112 schwarze 761: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 762: if (curp->options & MPARSE_MDOC) {
1.112 schwarze 763: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 764: if (curp->man->mdocmac == NULL)
765: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 766: } else if (curp->options & MPARSE_MAN) {
767: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 768: if (curp->man->manmac == NULL)
769: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 770: }
1.113 schwarze 771: curp->man->first->tok = TOKEN_NONE;
1.151 schwarze 772: curp->man->meta.os_e = os_e;
1.117 schwarze 773: return curp;
1.1 schwarze 774: }
775:
776: void
777: mparse_reset(struct mparse *curp)
778: {
779: roff_reset(curp->roff);
1.124 schwarze 780: roff_man_reset(curp->man);
1.133 schwarze 781:
782: free(curp->sodest);
783: curp->sodest = NULL;
784:
1.4 schwarze 785: if (curp->secondary)
786: curp->secondary->sz = 0;
1.1 schwarze 787:
788: curp->file_status = MANDOCLEVEL_OK;
1.132 schwarze 789: curp->gzip = 0;
1.1 schwarze 790: }
791:
792: void
793: mparse_free(struct mparse *curp)
794: {
795:
1.136 schwarze 796: roffhash_free(curp->man->mdocmac);
797: roffhash_free(curp->man->manmac);
1.112 schwarze 798: roff_man_free(curp->man);
1.133 schwarze 799: roff_free(curp->roff);
1.4 schwarze 800: if (curp->secondary)
801: free(curp->secondary->buf);
1.1 schwarze 802:
1.4 schwarze 803: free(curp->secondary);
1.23 schwarze 804: free(curp->sodest);
1.1 schwarze 805: free(curp);
806: }
807:
808: void
1.110 schwarze 809: mparse_result(struct mparse *curp, struct roff_man **man,
810: char **sodest)
1.1 schwarze 811: {
812:
1.23 schwarze 813: if (sodest && NULL != (*sodest = curp->sodest)) {
814: *man = NULL;
815: return;
816: }
1.1 schwarze 817: if (man)
818: *man = curp->man;
1.130 schwarze 819: }
820:
821: void
822: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
823: {
824: if (curp->file_status > *rc)
825: *rc = curp->file_status;
1.1 schwarze 826: }
827:
828: void
829: mandoc_vmsg(enum mandocerr t, struct mparse *m,
830: int ln, int pos, const char *fmt, ...)
831: {
832: char buf[256];
833: va_list ap;
834:
835: va_start(ap, fmt);
1.26 schwarze 836: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 837: va_end(ap);
838:
839: mandoc_msg(t, m, ln, pos, buf);
840: }
841:
842: void
1.25 schwarze 843: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 844: int ln, int col, const char *msg)
845: {
846: enum mandoclevel level;
847:
1.151 schwarze 848: if (er < m->mmin && er != MANDOCERR_FILE)
849: return;
850:
1.87 schwarze 851: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 852: while (er < mandoclimits[level])
853: level--;
854:
855: if (m->mmsg)
856: (*m->mmsg)(er, level, m->file, ln, col, msg);
857:
858: if (m->file_status < level)
859: m->file_status = level;
860: }
861:
862: const char *
863: mparse_strerror(enum mandocerr er)
864: {
865:
1.117 schwarze 866: return mandocerrs[er];
1.1 schwarze 867: }
868:
869: const char *
870: mparse_strlevel(enum mandoclevel lvl)
871: {
1.117 schwarze 872: return mandoclevels[lvl];
1.4 schwarze 873: }
874:
875: void
876: mparse_keep(struct mparse *p)
877: {
878:
879: assert(NULL == p->secondary);
880: p->secondary = mandoc_calloc(1, sizeof(struct buf));
881: }
882:
883: const char *
884: mparse_getkeep(const struct mparse *p)
885: {
886:
887: assert(p->secondary);
1.117 schwarze 888: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 schwarze 889: }