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