Annotation of src/usr.bin/mandoc/read.c, Revision 1.168
1.168 ! schwarze 1: /* $OpenBSD: read.c,v 1.167 2018/03/16 15:05:33 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.168 ! schwarze 4: * Copyright (c) 2010-2018 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>
1.18 schwarze 25: #include <errno.h>
1.1 schwarze 26: #include <fcntl.h>
27: #include <stdarg.h>
28: #include <stdio.h>
29: #include <stdlib.h>
30: #include <string.h>
31: #include <unistd.h>
1.115 schwarze 32: #include <zlib.h>
1.1 schwarze 33:
1.108 schwarze 34: #include "mandoc_aux.h"
1.1 schwarze 35: #include "mandoc.h"
1.108 schwarze 36: #include "roff.h"
1.1 schwarze 37: #include "mdoc.h"
38: #include "man.h"
1.108 schwarze 39: #include "libmandoc.h"
1.1 schwarze 40:
41: #define REPARSE_LIMIT 1000
42:
43: struct mparse {
1.133 schwarze 44: struct roff *roff; /* roff parser (!NULL) */
1.109 schwarze 45: struct roff_man *man; /* man parser */
1.23 schwarze 46: char *sodest; /* filename pointed to by .so */
1.59 schwarze 47: const char *file; /* filename of current input file */
48: struct buf *primary; /* buffer currently being parsed */
49: struct buf *secondary; /* preprocessed copy of input */
1.151 schwarze 50: const char *os_s; /* default operating system */
1.59 schwarze 51: mandocmsg mmsg; /* warning/error message handler */
52: enum mandoclevel file_status; /* status of current parse */
1.151 schwarze 53: enum mandocerr mmin; /* ignore messages below this */
1.59 schwarze 54: int options; /* parser options */
1.115 schwarze 55: int gzip; /* current input file is gzipped */
1.70 schwarze 56: int filenc; /* encoding of the current file */
1.1 schwarze 57: int reparse_count; /* finite interp. stack */
1.59 schwarze 58: int line; /* line number in the file */
1.1 schwarze 59: };
60:
1.60 schwarze 61: static void choose_parser(struct mparse *);
1.1 schwarze 62: static void resize_buf(struct buf *, size_t);
1.135 schwarze 63: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.18 schwarze 64: static int read_whole_file(struct mparse *, const char *, int,
65: struct buf *, int *);
1.1 schwarze 66: static void mparse_end(struct mparse *);
1.15 schwarze 67: static void mparse_parse_buffer(struct mparse *, struct buf,
68: const char *);
1.1 schwarze 69:
70: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
71: MANDOCERR_OK,
1.160 schwarze 72: MANDOCERR_OK,
1.1 schwarze 73: MANDOCERR_WARNING,
74: MANDOCERR_ERROR,
1.87 schwarze 75: MANDOCERR_UNSUPP,
1.1 schwarze 76: MANDOCERR_MAX,
77: MANDOCERR_MAX
78: };
79:
80: static const char * const mandocerrs[MANDOCERR_MAX] = {
81: "ok",
82:
1.151 schwarze 83: "base system convention",
1.138 schwarze 84:
1.148 schwarze 85: "Mdocdate found",
86: "Mdocdate missing",
1.153 schwarze 87: "unknown architecture",
1.152 schwarze 88: "operating system explicitly specified",
1.151 schwarze 89: "RCS id missing",
1.156 schwarze 90: "referenced manual not found",
1.151 schwarze 91:
92: "generic style suggestion",
93:
1.148 schwarze 94: "legacy man(7) date format",
1.168 ! schwarze 95: "normalizing date format to",
1.161 schwarze 96: "lower case character in document title",
1.150 schwarze 97: "duplicate RCS id",
1.165 jca 98: "possible typo in section name",
1.161 schwarze 99: "unterminated quoted argument",
1.138 schwarze 100: "useless macro",
1.139 schwarze 101: "consider using OS macro",
1.144 schwarze 102: "errnos out of order",
103: "duplicate errno",
1.159 schwarze 104: "trailing delimiter",
1.146 schwarze 105: "no blank before trailing delimiter",
1.161 schwarze 106: "fill mode already enabled, skipping",
107: "fill mode already disabled, skipping",
1.167 schwarze 108: "verbatim \"--\", maybe consider using \\(em",
1.147 schwarze 109: "function name without markup",
1.161 schwarze 110: "whitespace at end of input line",
111: "bad comment style",
1.138 schwarze 112:
1.1 schwarze 113: "generic warning",
114:
115: /* related to the prologue */
1.57 schwarze 116: "missing manual title, using UNTITLED",
117: "missing manual title, using \"\"",
118: "missing manual section, using \"\"",
1.1 schwarze 119: "unknown manual section",
1.32 schwarze 120: "missing date, using today's date",
1.1 schwarze 121: "cannot parse date, using it verbatim",
1.158 schwarze 122: "date in the future, using it anyway",
1.57 schwarze 123: "missing Os macro, using \"\"",
124: "late prologue macro",
1.1 schwarze 125: "prologue macros out of order",
126:
127: /* related to document structure */
128: ".so is fragile, better use ln(1)",
1.28 schwarze 129: "no document body",
1.32 schwarze 130: "content before first section header",
131: "first section is not \"NAME\"",
1.129 schwarze 132: "NAME section without Nm before Nd",
1.103 schwarze 133: "NAME section without description",
134: "description not at the end of NAME",
135: "bad NAME section content",
1.129 schwarze 136: "missing comma before name",
1.96 schwarze 137: "missing description line, using \"\"",
1.134 schwarze 138: "description line outside NAME section",
1.1 schwarze 139: "sections out of conventional order",
1.32 schwarze 140: "duplicate section title",
141: "unexpected section",
1.157 schwarze 142: "cross reference to self",
1.63 schwarze 143: "unusual Xr order",
144: "unusual Xr punctuation",
1.62 schwarze 145: "AUTHORS section without An macro",
1.1 schwarze 146:
147: /* related to macros and nesting */
1.33 schwarze 148: "obsolete macro",
1.78 schwarze 149: "macro neither callable nor escaped",
1.1 schwarze 150: "skipping paragraph macro",
1.10 schwarze 151: "moving paragraph macro out of list",
1.1 schwarze 152: "skipping no-space macro",
153: "blocks badly nested",
154: "nested displays are not portable",
1.35 schwarze 155: "moving content out of list",
1.155 schwarze 156: "first macro on line",
1.1 schwarze 157: "line scope broken",
1.141 schwarze 158: "skipping blank line in line scope",
1.1 schwarze 159:
160: /* related to missing macro arguments */
1.36 schwarze 161: "skipping empty request",
162: "conditional request controls empty scope",
1.1 schwarze 163: "skipping empty macro",
1.99 schwarze 164: "empty block",
1.40 schwarze 165: "empty argument, using 0n",
1.38 schwarze 166: "missing display type, using -ragged",
167: "list type is not the first argument",
1.125 schwarze 168: "missing -width in -tag list, using 6n",
1.56 schwarze 169: "missing utility name, using \"\"",
1.98 schwarze 170: "missing function name, using \"\"",
1.38 schwarze 171: "empty head in list item",
172: "empty list item",
1.155 schwarze 173: "missing argument, using next line",
1.39 schwarze 174: "missing font type, using \\fR",
175: "unknown font type, using \\fR",
1.79 schwarze 176: "nothing follows prefix",
1.97 schwarze 177: "empty reference block",
1.128 schwarze 178: "missing section argument",
1.38 schwarze 179: "missing -std argument, adding it",
1.100 schwarze 180: "missing option string, using \"\"",
181: "missing resource identifier, using \"\"",
1.66 schwarze 182: "missing eqn box, using \"\"",
1.1 schwarze 183:
184: /* related to bad macro arguments */
185: "duplicate argument",
1.54 schwarze 186: "skipping duplicate argument",
1.41 schwarze 187: "skipping duplicate display type",
188: "skipping duplicate list type",
1.54 schwarze 189: "skipping -width argument",
1.101 schwarze 190: "wrong number of cells",
1.1 schwarze 191: "unknown AT&T UNIX version",
1.64 schwarze 192: "comma in function argument",
1.65 schwarze 193: "parenthesis in function name",
1.145 schwarze 194: "unknown library name",
1.45 schwarze 195: "invalid content in Rs block",
1.41 schwarze 196: "invalid Boolean argument",
197: "unknown font, skipping request",
1.101 schwarze 198: "odd number of characters in request",
1.1 schwarze 199:
200: /* related to plain text */
1.42 schwarze 201: "blank line in fill mode, using .sp",
202: "tab in filled text",
1.131 schwarze 203: "new sentence, new line",
1.42 schwarze 204: "invalid escape sequence",
205: "undefined string, using \"\"",
1.3 schwarze 206:
1.93 schwarze 207: /* related to tables */
208: "tbl line starts with span",
209: "tbl column starts with span",
210: "skipping vertical bar in tbl layout",
211:
1.1 schwarze 212: "generic error",
1.3 schwarze 213:
1.1 schwarze 214: /* related to tables */
1.91 schwarze 215: "non-alphabetic character in tbl options",
216: "skipping unknown tbl option",
217: "missing tbl option argument",
218: "wrong tbl option argument size",
1.93 schwarze 219: "empty tbl layout",
220: "invalid character in tbl layout",
221: "unmatched parenthesis in tbl layout",
1.94 schwarze 222: "tbl without any data cells",
223: "ignoring data in spanned tbl cell",
224: "ignoring extra tbl data cells",
225: "data block open at end of tbl",
1.1 schwarze 226:
1.46 schwarze 227: /* related to document structure and macros */
1.83 schwarze 228: NULL,
1.161 schwarze 229: "duplicate prologue macro",
230: "skipping late title macro",
1.1 schwarze 231: "input stack limit exceeded, infinite loop?",
232: "skipping bad character",
1.46 schwarze 233: "skipping unknown macro",
1.87 schwarze 234: "skipping insecure request",
1.51 schwarze 235: "skipping item outside list",
1.46 schwarze 236: "skipping column outside column list",
237: "skipping end of block that is not open",
1.90 schwarze 238: "fewer RS blocks open, skipping",
1.46 schwarze 239: "inserting missing end of block",
240: "appending missing end of block",
241:
242: /* related to request and macro arguments */
1.1 schwarze 243: "escaped character not allowed in a name",
1.75 schwarze 244: "NOT IMPLEMENTED: Bd -file",
1.120 schwarze 245: "skipping display without arguments",
1.49 schwarze 246: "missing list type, using -item",
1.143 schwarze 247: "argument is not numeric, using 1",
1.48 schwarze 248: "missing manual name, using \"\"",
1.49 schwarze 249: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 250: "unknown standard specifier",
1.49 schwarze 251: "skipping request without numeric argument",
1.84 schwarze 252: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 253: ".so request failed",
1.39 schwarze 254: "skipping all arguments",
255: "skipping excess arguments",
1.68 schwarze 256: "divide by zero",
1.87 schwarze 257:
258: "unsupported feature",
259: "input too large",
1.89 schwarze 260: "unsupported control character",
1.87 schwarze 261: "unsupported roff request",
1.94 schwarze 262: "eqn delim option in tbl",
1.93 schwarze 263: "unsupported tbl layout modifier",
1.87 schwarze 264: "ignoring macro in table",
1.1 schwarze 265: };
266:
267: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
268: "SUCCESS",
1.137 schwarze 269: "STYLE",
1.1 schwarze 270: "WARNING",
271: "ERROR",
1.87 schwarze 272: "UNSUPP",
1.1 schwarze 273: "BADARG",
274: "SYSERR"
275: };
276:
1.25 schwarze 277:
1.1 schwarze 278: static void
279: resize_buf(struct buf *buf, size_t initial)
280: {
281:
282: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
283: buf->buf = mandoc_realloc(buf->buf, buf->sz);
284: }
285:
286: static void
1.60 schwarze 287: choose_parser(struct mparse *curp)
1.1 schwarze 288: {
1.59 schwarze 289: char *cp, *ep;
290: int format;
1.1 schwarze 291:
1.59 schwarze 292: /*
293: * If neither command line arguments -mdoc or -man select
294: * a parser nor the roff parser found a .Dd or .TH macro
295: * yet, look ahead in the main input buffer.
296: */
297:
298: if ((format = roff_getformat(curp->roff)) == 0) {
299: cp = curp->primary->buf;
300: ep = cp + curp->primary->sz;
301: while (cp < ep) {
1.61 schwarze 302: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 303: cp++;
304: if (cp[0] == 'D' && cp[1] == 'd') {
305: format = MPARSE_MDOC;
306: break;
307: }
308: if (cp[0] == 'T' && cp[1] == 'H') {
309: format = MPARSE_MAN;
310: break;
311: }
312: }
313: cp = memchr(cp, '\n', ep - cp);
314: if (cp == NULL)
315: break;
316: cp++;
317: }
1.1 schwarze 318: }
319:
1.59 schwarze 320: if (format == MPARSE_MDOC) {
1.112 schwarze 321: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 322: if (curp->man->mdocmac == NULL)
323: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 324: } else {
325: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 326: if (curp->man->manmac == NULL)
327: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 328: }
1.136 schwarze 329: curp->man->first->tok = TOKEN_NONE;
1.1 schwarze 330: }
331:
332: /*
1.71 schwarze 333: * Main parse routine for a buffer.
334: * It assumes encoding and line numbering are already set up.
335: * It can recurse directly (for invocations of user-defined
336: * macros, inline equations, and input line traps)
337: * and indirectly (for .so file inclusion).
1.1 schwarze 338: */
1.135 schwarze 339: static int
1.71 schwarze 340: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 341: {
342: struct buf ln;
1.88 schwarze 343: const char *save_file;
1.85 schwarze 344: char *cp;
1.71 schwarze 345: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 346: enum rofferr rr;
1.76 schwarze 347: int of;
1.1 schwarze 348: int lnn; /* line number in the real file */
1.85 schwarze 349: int fd;
1.1 schwarze 350: unsigned char c;
351:
1.71 schwarze 352: memset(&ln, 0, sizeof(ln));
1.1 schwarze 353:
1.25 schwarze 354: lnn = curp->line;
355: pos = 0;
1.1 schwarze 356:
1.71 schwarze 357: while (i < blk.sz) {
1.1 schwarze 358: if (0 == pos && '\0' == blk.buf[i])
359: break;
360:
361: if (start) {
362: curp->line = lnn;
363: curp->reparse_count = 0;
1.70 schwarze 364:
365: if (lnn < 3 &&
366: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 367: curp->filenc & MPARSE_LATIN1)
368: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 369: }
370:
1.71 schwarze 371: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 372:
373: /*
374: * When finding an unescaped newline character,
375: * leave the character loop to process the line.
376: * Skip a preceding carriage return, if any.
377: */
378:
1.71 schwarze 379: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 380: '\n' == blk.buf[i + 1])
381: ++i;
382: if ('\n' == blk.buf[i]) {
383: ++i;
384: ++lnn;
385: break;
386: }
387:
1.13 schwarze 388: /*
1.70 schwarze 389: * Make sure we have space for the worst
390: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 391: */
392:
1.71 schwarze 393: if (pos + 11 > ln.sz)
1.13 schwarze 394: resize_buf(&ln, 256);
395:
1.25 schwarze 396: /*
1.70 schwarze 397: * Encode 8-bit input.
1.1 schwarze 398: */
399:
1.70 schwarze 400: c = blk.buf[i];
401: if (c & 0x80) {
1.71 schwarze 402: if ( ! (curp->filenc && preconv_encode(
403: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 404: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
405: curp->line, pos, "0x%x", c);
1.70 schwarze 406: ln.buf[pos++] = '?';
407: i++;
408: }
409: continue;
410: }
411:
412: /*
413: * Exclude control characters.
414: */
1.1 schwarze 415:
1.70 schwarze 416: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 417: mandoc_vmsg(c == 0x00 || c == 0x04 ||
418: c > 0x0a ? MANDOCERR_CHAR_BAD :
419: MANDOCERR_CHAR_UNSUPP,
420: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 421: i++;
1.102 schwarze 422: if (c != '\r')
423: ln.buf[pos++] = '?';
1.1 schwarze 424: continue;
425: }
426:
427: ln.buf[pos++] = blk.buf[i++];
428: }
429:
1.142 schwarze 430: if (pos + 1 >= ln.sz)
1.1 schwarze 431: resize_buf(&ln, 256);
432:
1.142 schwarze 433: if (i == blk.sz || blk.buf[i] == '\0')
434: ln.buf[pos++] = '\n';
1.1 schwarze 435: ln.buf[pos] = '\0';
436:
437: /*
438: * A significant amount of complexity is contained by
439: * the roff preprocessor. It's line-oriented but can be
440: * expressed on one line, so we need at times to
441: * readjust our starting point and re-run it. The roff
442: * preprocessor can also readjust the buffers with new
443: * data, so we pass them in wholesale.
444: */
445:
446: of = 0;
447:
1.4 schwarze 448: /*
449: * Maintain a lookaside buffer of all parsed lines. We
450: * only do this if mparse_keep() has been invoked (the
451: * buffer may be accessed with mparse_getkeep()).
452: */
453:
454: if (curp->secondary) {
1.25 schwarze 455: curp->secondary->buf = mandoc_realloc(
456: curp->secondary->buf,
457: curp->secondary->sz + pos + 2);
458: memcpy(curp->secondary->buf +
459: curp->secondary->sz,
460: ln.buf, pos);
1.4 schwarze 461: curp->secondary->sz += pos;
462: curp->secondary->buf
463: [curp->secondary->sz] = '\n';
464: curp->secondary->sz++;
465: curp->secondary->buf
466: [curp->secondary->sz] = '\0';
467: }
1.1 schwarze 468: rerun:
1.72 schwarze 469: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 470:
471: switch (rr) {
1.25 schwarze 472: case ROFF_REPARSE:
1.135 schwarze 473: if (++curp->reparse_count > REPARSE_LIMIT)
1.1 schwarze 474: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 475: curp->line, pos, NULL);
1.135 schwarze 476: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
477: start == 1) {
478: pos = 0;
479: continue;
480: }
481: free(ln.buf);
482: return 0;
1.25 schwarze 483: case ROFF_APPEND:
1.71 schwarze 484: pos = strlen(ln.buf);
1.1 schwarze 485: continue;
1.25 schwarze 486: case ROFF_RERUN:
1.1 schwarze 487: goto rerun;
1.25 schwarze 488: case ROFF_IGN:
1.1 schwarze 489: pos = 0;
490: continue;
1.25 schwarze 491: case ROFF_SO:
1.71 schwarze 492: if ( ! (curp->options & MPARSE_SO) &&
493: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 494: curp->sodest = mandoc_strdup(ln.buf + of);
495: free(ln.buf);
1.135 schwarze 496: return 1;
1.23 schwarze 497: }
1.4 schwarze 498: /*
499: * We remove `so' clauses from our lookaside
500: * buffer because we're going to descend into
501: * the file recursively.
502: */
1.25 schwarze 503: if (curp->secondary)
1.4 schwarze 504: curp->secondary->sz -= pos + 1;
1.88 schwarze 505: save_file = curp->file;
1.122 schwarze 506: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 507: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 508: close(fd);
1.88 schwarze 509: curp->file = save_file;
510: } else {
511: curp->file = save_file;
1.30 schwarze 512: mandoc_vmsg(MANDOCERR_SO_FAIL,
513: curp, curp->line, pos,
514: ".so %s", ln.buf + of);
1.85 schwarze 515: ln.sz = mandoc_asprintf(&cp,
516: ".sp\nSee the file %s.\n.sp",
517: ln.buf + of);
518: free(ln.buf);
519: ln.buf = cp;
520: of = 0;
521: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 522: }
1.1 schwarze 523: pos = 0;
524: continue;
525: default:
526: break;
527: }
528:
1.124 schwarze 529: if (curp->man->macroset == MACROSET_NONE)
1.60 schwarze 530: choose_parser(curp);
1.1 schwarze 531:
1.163 schwarze 532: if ((curp->man->macroset == MACROSET_MDOC ?
1.110 schwarze 533: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 534: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
535: break;
1.1 schwarze 536:
537: /* Temporary buffers typically are not full. */
538:
539: if (0 == start && '\0' == blk.buf[i])
540: break;
541:
542: /* Start the next input line. */
543:
544: pos = 0;
545: }
546:
547: free(ln.buf);
1.135 schwarze 548: return 1;
1.1 schwarze 549: }
550:
551: static int
1.18 schwarze 552: read_whole_file(struct mparse *curp, const char *file, int fd,
553: struct buf *fb, int *with_mmap)
1.1 schwarze 554: {
555: struct stat st;
1.115 schwarze 556: gzFile gz;
1.1 schwarze 557: size_t off;
558: ssize_t ssz;
1.166 schwarze 559: int gzerrnum, retval;
1.1 schwarze 560:
1.164 schwarze 561: if (fstat(fd, &st) == -1) {
562: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
563: "fstat: %s", strerror(errno));
564: return 0;
565: }
1.1 schwarze 566:
567: /*
568: * If we're a regular file, try just reading in the whole entry
569: * via mmap(). This is faster than reading it into blocks, and
570: * since each file is only a few bytes to begin with, I'm not
571: * concerned that this is going to tank any machines.
572: */
573:
1.115 schwarze 574: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 575: if (st.st_size > 0x7fffffff) {
1.86 schwarze 576: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.117 schwarze 577: return 0;
1.1 schwarze 578: }
579: *with_mmap = 1;
580: fb->sz = (size_t)st.st_size;
1.15 schwarze 581: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 582: if (fb->buf != MAP_FAILED)
1.117 schwarze 583: return 1;
1.1 schwarze 584: }
585:
1.115 schwarze 586: if (curp->gzip) {
1.166 schwarze 587: /*
588: * Duplicating the file descriptor is required
589: * because we will have to call gzclose(3)
590: * to free memory used internally by zlib,
591: * but that will also close the file descriptor,
592: * which this function must not do.
593: */
594: if ((fd = dup(fd)) == -1) {
595: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
596: "dup: %s", strerror(errno));
597: return 0;
598: }
1.164 schwarze 599: if ((gz = gzdopen(fd, "rb")) == NULL) {
600: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
601: "gzdopen: %s", strerror(errno));
1.166 schwarze 602: close(fd);
1.164 schwarze 603: return 0;
604: }
1.115 schwarze 605: } else
606: gz = NULL;
607:
1.1 schwarze 608: /*
609: * If this isn't a regular file (like, say, stdin), then we must
610: * go the old way and just read things in bit by bit.
611: */
612:
613: *with_mmap = 0;
614: off = 0;
1.166 schwarze 615: retval = 0;
1.1 schwarze 616: fb->sz = 0;
617: fb->buf = NULL;
618: for (;;) {
619: if (off == fb->sz) {
620: if (fb->sz == (1U << 31)) {
1.86 schwarze 621: mandoc_msg(MANDOCERR_TOOLARGE, curp,
622: 0, 0, NULL);
1.1 schwarze 623: break;
624: }
625: resize_buf(fb, 65536);
626: }
1.115 schwarze 627: ssz = curp->gzip ?
628: gzread(gz, fb->buf + (int)off, fb->sz - off) :
629: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 630: if (ssz == 0) {
631: fb->sz = off;
1.166 schwarze 632: retval = 1;
633: break;
1.1 schwarze 634: }
1.164 schwarze 635: if (ssz == -1) {
1.166 schwarze 636: if (curp->gzip)
637: (void)gzerror(gz, &gzerrnum);
638: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "read: %s",
639: curp->gzip && gzerrnum != Z_ERRNO ?
640: zError(gzerrnum) : strerror(errno));
1.164 schwarze 641: break;
642: }
1.1 schwarze 643: off += (size_t)ssz;
644: }
645:
1.166 schwarze 646: if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
647: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "gzclose: %s",
648: gzerrnum == Z_ERRNO ? strerror(errno) :
649: zError(gzerrnum));
650: if (retval == 0) {
651: free(fb->buf);
652: fb->buf = NULL;
653: }
654: return retval;
1.1 schwarze 655: }
656:
657: static void
658: mparse_end(struct mparse *curp)
659: {
1.110 schwarze 660: if (curp->man->macroset == MACROSET_NONE)
661: curp->man->macroset = MACROSET_MAN;
662: if (curp->man->macroset == MACROSET_MDOC)
663: mdoc_endparse(curp->man);
664: else
1.86 schwarze 665: man_endparse(curp->man);
1.1 schwarze 666: roff_endparse(curp->roff);
667: }
668:
1.15 schwarze 669: static void
670: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 671: {
1.61 schwarze 672: struct buf *svprimary;
1.1 schwarze 673: const char *svfile;
1.71 schwarze 674: size_t offset;
1.14 schwarze 675: static int recursion_depth;
676:
677: if (64 < recursion_depth) {
678: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 679: return;
1.14 schwarze 680: }
1.1 schwarze 681:
1.15 schwarze 682: /* Line number is per-file. */
683: svfile = curp->file;
684: curp->file = file;
1.61 schwarze 685: svprimary = curp->primary;
1.59 schwarze 686: curp->primary = &blk;
1.15 schwarze 687: curp->line = 1;
688: recursion_depth++;
689:
1.70 schwarze 690: /* Skip an UTF-8 byte order mark. */
691: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
692: (unsigned char)blk.buf[0] == 0xef &&
693: (unsigned char)blk.buf[1] == 0xbb &&
694: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 695: offset = 3;
1.70 schwarze 696: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 697: } else
698: offset = 0;
1.70 schwarze 699:
1.71 schwarze 700: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 701:
1.86 schwarze 702: if (--recursion_depth == 0)
1.15 schwarze 703: mparse_end(curp);
704:
1.61 schwarze 705: curp->primary = svprimary;
1.15 schwarze 706: curp->file = svfile;
707: }
708:
1.74 schwarze 709: /*
710: * Read the whole file into memory and call the parsers.
711: * Called recursively when an .so request is encountered.
712: */
1.15 schwarze 713: enum mandoclevel
714: mparse_readfd(struct mparse *curp, int fd, const char *file)
715: {
716: struct buf blk;
717: int with_mmap;
1.70 schwarze 718: int save_filenc;
1.1 schwarze 719:
1.67 schwarze 720: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 721: save_filenc = curp->filenc;
722: curp->filenc = curp->options &
723: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 724: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 725: curp->filenc = save_filenc;
1.67 schwarze 726: if (with_mmap)
727: munmap(blk.buf, blk.sz);
728: else
729: free(blk.buf);
730: }
1.117 schwarze 731: return curp->file_status;
1.58 schwarze 732: }
733:
1.122 schwarze 734: int
735: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 736: {
737: char *cp;
1.122 schwarze 738: int fd;
1.58 schwarze 739:
740: curp->file = file;
1.115 schwarze 741: cp = strrchr(file, '.');
742: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 743:
1.115 schwarze 744: /* First try to use the filename as it is. */
1.74 schwarze 745:
1.122 schwarze 746: if ((fd = open(file, O_RDONLY)) != -1)
747: return fd;
1.74 schwarze 748:
1.115 schwarze 749: /*
750: * If that doesn't work and the filename doesn't
751: * already end in .gz, try appending .gz.
752: */
1.74 schwarze 753:
1.115 schwarze 754: if ( ! curp->gzip) {
1.74 schwarze 755: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 756: fd = open(cp, O_RDONLY);
1.83 schwarze 757: free(cp);
1.122 schwarze 758: if (fd != -1) {
1.115 schwarze 759: curp->gzip = 1;
1.122 schwarze 760: return fd;
1.58 schwarze 761: }
762: }
763:
1.115 schwarze 764: /* Neither worked, give up. */
1.73 schwarze 765:
1.115 schwarze 766: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.122 schwarze 767: return -1;
1.1 schwarze 768: }
769:
770: struct mparse *
1.151 schwarze 771: mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
772: enum mandoc_os os_e, const char *os_s)
1.1 schwarze 773: {
774: struct mparse *curp;
775:
776: curp = mandoc_calloc(1, sizeof(struct mparse));
777:
1.22 schwarze 778: curp->options = options;
1.151 schwarze 779: curp->mmin = mmin;
1.1 schwarze 780: curp->mmsg = mmsg;
1.151 schwarze 781: curp->os_s = os_s;
1.1 schwarze 782:
1.119 schwarze 783: curp->roff = roff_alloc(curp, options);
1.151 schwarze 784: curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
1.112 schwarze 785: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 786: if (curp->options & MPARSE_MDOC) {
1.112 schwarze 787: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 788: if (curp->man->mdocmac == NULL)
789: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 790: } else if (curp->options & MPARSE_MAN) {
791: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 792: if (curp->man->manmac == NULL)
793: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 794: }
1.113 schwarze 795: curp->man->first->tok = TOKEN_NONE;
1.151 schwarze 796: curp->man->meta.os_e = os_e;
1.117 schwarze 797: return curp;
1.1 schwarze 798: }
799:
800: void
801: mparse_reset(struct mparse *curp)
802: {
803: roff_reset(curp->roff);
1.124 schwarze 804: roff_man_reset(curp->man);
1.133 schwarze 805:
806: free(curp->sodest);
807: curp->sodest = NULL;
808:
1.4 schwarze 809: if (curp->secondary)
810: curp->secondary->sz = 0;
1.1 schwarze 811:
812: curp->file_status = MANDOCLEVEL_OK;
1.132 schwarze 813: curp->gzip = 0;
1.1 schwarze 814: }
815:
816: void
817: mparse_free(struct mparse *curp)
818: {
819:
1.136 schwarze 820: roffhash_free(curp->man->mdocmac);
821: roffhash_free(curp->man->manmac);
1.112 schwarze 822: roff_man_free(curp->man);
1.133 schwarze 823: roff_free(curp->roff);
1.4 schwarze 824: if (curp->secondary)
825: free(curp->secondary->buf);
1.1 schwarze 826:
1.4 schwarze 827: free(curp->secondary);
1.23 schwarze 828: free(curp->sodest);
1.1 schwarze 829: free(curp);
830: }
831:
832: void
1.110 schwarze 833: mparse_result(struct mparse *curp, struct roff_man **man,
834: char **sodest)
1.1 schwarze 835: {
836:
1.23 schwarze 837: if (sodest && NULL != (*sodest = curp->sodest)) {
838: *man = NULL;
839: return;
840: }
1.1 schwarze 841: if (man)
842: *man = curp->man;
1.130 schwarze 843: }
844:
845: void
846: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
847: {
848: if (curp->file_status > *rc)
849: *rc = curp->file_status;
1.1 schwarze 850: }
851:
852: void
853: mandoc_vmsg(enum mandocerr t, struct mparse *m,
854: int ln, int pos, const char *fmt, ...)
855: {
856: char buf[256];
857: va_list ap;
858:
859: va_start(ap, fmt);
1.26 schwarze 860: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 861: va_end(ap);
862:
863: mandoc_msg(t, m, ln, pos, buf);
864: }
865:
866: void
1.25 schwarze 867: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 868: int ln, int col, const char *msg)
869: {
870: enum mandoclevel level;
871:
1.151 schwarze 872: if (er < m->mmin && er != MANDOCERR_FILE)
873: return;
874:
1.87 schwarze 875: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 876: while (er < mandoclimits[level])
877: level--;
878:
879: if (m->mmsg)
880: (*m->mmsg)(er, level, m->file, ln, col, msg);
881:
882: if (m->file_status < level)
883: m->file_status = level;
884: }
885:
886: const char *
887: mparse_strerror(enum mandocerr er)
888: {
889:
1.117 schwarze 890: return mandocerrs[er];
1.1 schwarze 891: }
892:
893: const char *
894: mparse_strlevel(enum mandoclevel lvl)
895: {
1.117 schwarze 896: return mandoclevels[lvl];
1.4 schwarze 897: }
898:
899: void
900: mparse_keep(struct mparse *p)
901: {
902:
903: assert(NULL == p->secondary);
904: p->secondary = mandoc_calloc(1, sizeof(struct buf));
905: }
906:
907: const char *
908: mparse_getkeep(const struct mparse *p)
909: {
910:
911: assert(p->secondary);
1.117 schwarze 912: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 schwarze 913: }