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