Annotation of src/usr.bin/mandoc/read.c, Revision 1.145
1.145 ! schwarze 1: /* $OpenBSD: read.c,v 1.144 2017/06/07 23:29:31 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 */
52: const char *defos; /* default operating system */
53: mandocmsg mmsg; /* warning/error message handler */
54: enum mandoclevel file_status; /* status of current parse */
55: enum mandoclevel wlevel; /* ignore messages below this */
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:
1.137 schwarze 85: "generic style suggestion",
1.138 schwarze 86:
87: "useless macro",
1.139 schwarze 88: "consider using OS macro",
1.144 schwarze 89: "errnos out of order",
90: "duplicate errno",
1.140 schwarze 91: "description line ends with a full stop",
1.138 schwarze 92:
1.1 schwarze 93: "generic warning",
94:
95: /* related to the prologue */
1.57 schwarze 96: "missing manual title, using UNTITLED",
97: "missing manual title, using \"\"",
1.32 schwarze 98: "lower case character in document title",
1.57 schwarze 99: "missing manual section, using \"\"",
1.1 schwarze 100: "unknown manual section",
1.32 schwarze 101: "missing date, using today's date",
1.1 schwarze 102: "cannot parse date, using it verbatim",
1.57 schwarze 103: "missing Os macro, using \"\"",
104: "duplicate prologue macro",
105: "late prologue macro",
106: "skipping late title macro",
1.1 schwarze 107: "prologue macros out of order",
108:
109: /* related to document structure */
110: ".so is fragile, better use ln(1)",
1.28 schwarze 111: "no document body",
1.32 schwarze 112: "content before first section header",
113: "first section is not \"NAME\"",
1.129 schwarze 114: "NAME section without Nm before Nd",
1.103 schwarze 115: "NAME section without description",
116: "description not at the end of NAME",
117: "bad NAME section content",
1.129 schwarze 118: "missing comma before name",
1.96 schwarze 119: "missing description line, using \"\"",
1.134 schwarze 120: "description line outside NAME section",
1.1 schwarze 121: "sections out of conventional order",
1.32 schwarze 122: "duplicate section title",
123: "unexpected section",
1.63 schwarze 124: "unusual Xr order",
125: "unusual Xr punctuation",
1.62 schwarze 126: "AUTHORS section without An macro",
1.1 schwarze 127:
128: /* related to macros and nesting */
1.33 schwarze 129: "obsolete macro",
1.78 schwarze 130: "macro neither callable nor escaped",
1.1 schwarze 131: "skipping paragraph macro",
1.10 schwarze 132: "moving paragraph macro out of list",
1.1 schwarze 133: "skipping no-space macro",
134: "blocks badly nested",
135: "nested displays are not portable",
1.35 schwarze 136: "moving content out of list",
1.56 schwarze 137: "fill mode already enabled, skipping",
138: "fill mode already disabled, skipping",
1.1 schwarze 139: "line scope broken",
1.141 schwarze 140: "skipping blank line in line scope",
1.1 schwarze 141:
142: /* related to missing macro arguments */
1.36 schwarze 143: "skipping empty request",
144: "conditional request controls empty scope",
1.1 schwarze 145: "skipping empty macro",
1.99 schwarze 146: "empty block",
1.40 schwarze 147: "empty argument, using 0n",
1.38 schwarze 148: "missing display type, using -ragged",
149: "list type is not the first argument",
1.125 schwarze 150: "missing -width in -tag list, using 6n",
1.56 schwarze 151: "missing utility name, using \"\"",
1.98 schwarze 152: "missing function name, using \"\"",
1.38 schwarze 153: "empty head in list item",
154: "empty list item",
1.39 schwarze 155: "missing font type, using \\fR",
156: "unknown font type, using \\fR",
1.79 schwarze 157: "nothing follows prefix",
1.97 schwarze 158: "empty reference block",
1.128 schwarze 159: "missing section argument",
1.38 schwarze 160: "missing -std argument, adding it",
1.100 schwarze 161: "missing option string, using \"\"",
162: "missing resource identifier, using \"\"",
1.66 schwarze 163: "missing eqn box, using \"\"",
1.1 schwarze 164:
165: /* related to bad macro arguments */
1.42 schwarze 166: "unterminated quoted argument",
1.1 schwarze 167: "duplicate argument",
1.54 schwarze 168: "skipping duplicate argument",
1.41 schwarze 169: "skipping duplicate display type",
170: "skipping duplicate list type",
1.54 schwarze 171: "skipping -width argument",
1.101 schwarze 172: "wrong number of cells",
1.1 schwarze 173: "unknown AT&T UNIX version",
1.64 schwarze 174: "comma in function argument",
1.65 schwarze 175: "parenthesis in function name",
1.145 ! schwarze 176: "unknown library name",
1.45 schwarze 177: "invalid content in Rs block",
1.41 schwarze 178: "invalid Boolean argument",
179: "unknown font, skipping request",
1.101 schwarze 180: "odd number of characters in request",
1.1 schwarze 181:
182: /* related to plain text */
1.42 schwarze 183: "blank line in fill mode, using .sp",
184: "tab in filled text",
185: "whitespace at end of input line",
1.131 schwarze 186: "new sentence, new line",
1.1 schwarze 187: "bad comment style",
1.42 schwarze 188: "invalid escape sequence",
189: "undefined string, using \"\"",
1.3 schwarze 190:
1.93 schwarze 191: /* related to tables */
192: "tbl line starts with span",
193: "tbl column starts with span",
194: "skipping vertical bar in tbl layout",
195:
1.1 schwarze 196: "generic error",
1.3 schwarze 197:
1.1 schwarze 198: /* related to tables */
1.91 schwarze 199: "non-alphabetic character in tbl options",
200: "skipping unknown tbl option",
201: "missing tbl option argument",
202: "wrong tbl option argument size",
1.93 schwarze 203: "empty tbl layout",
204: "invalid character in tbl layout",
205: "unmatched parenthesis in tbl layout",
1.94 schwarze 206: "tbl without any data cells",
207: "ignoring data in spanned tbl cell",
208: "ignoring extra tbl data cells",
209: "data block open at end of tbl",
1.1 schwarze 210:
1.46 schwarze 211: /* related to document structure and macros */
1.83 schwarze 212: NULL,
1.1 schwarze 213: "input stack limit exceeded, infinite loop?",
214: "skipping bad character",
1.46 schwarze 215: "skipping unknown macro",
1.87 schwarze 216: "skipping insecure request",
1.51 schwarze 217: "skipping item outside list",
1.46 schwarze 218: "skipping column outside column list",
219: "skipping end of block that is not open",
1.90 schwarze 220: "fewer RS blocks open, skipping",
1.46 schwarze 221: "inserting missing end of block",
222: "appending missing end of block",
223:
224: /* related to request and macro arguments */
1.1 schwarze 225: "escaped character not allowed in a name",
1.75 schwarze 226: "NOT IMPLEMENTED: Bd -file",
1.120 schwarze 227: "skipping display without arguments",
1.49 schwarze 228: "missing list type, using -item",
1.143 schwarze 229: "argument is not numeric, using 1",
1.48 schwarze 230: "missing manual name, using \"\"",
1.49 schwarze 231: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 232: "unknown standard specifier",
1.49 schwarze 233: "skipping request without numeric argument",
1.84 schwarze 234: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 235: ".so request failed",
1.39 schwarze 236: "skipping all arguments",
237: "skipping excess arguments",
1.68 schwarze 238: "divide by zero",
1.87 schwarze 239:
240: "unsupported feature",
241: "input too large",
1.89 schwarze 242: "unsupported control character",
1.87 schwarze 243: "unsupported roff request",
1.94 schwarze 244: "eqn delim option in tbl",
1.93 schwarze 245: "unsupported tbl layout modifier",
1.87 schwarze 246: "ignoring macro in table",
1.1 schwarze 247: };
248:
249: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
250: "SUCCESS",
1.137 schwarze 251: "STYLE",
1.1 schwarze 252: "WARNING",
253: "ERROR",
1.87 schwarze 254: "UNSUPP",
1.1 schwarze 255: "BADARG",
256: "SYSERR"
257: };
258:
1.25 schwarze 259:
1.1 schwarze 260: static void
261: resize_buf(struct buf *buf, size_t initial)
262: {
263:
264: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
265: buf->buf = mandoc_realloc(buf->buf, buf->sz);
266: }
267:
268: static void
1.60 schwarze 269: choose_parser(struct mparse *curp)
1.1 schwarze 270: {
1.59 schwarze 271: char *cp, *ep;
272: int format;
1.1 schwarze 273:
1.59 schwarze 274: /*
275: * If neither command line arguments -mdoc or -man select
276: * a parser nor the roff parser found a .Dd or .TH macro
277: * yet, look ahead in the main input buffer.
278: */
279:
280: if ((format = roff_getformat(curp->roff)) == 0) {
281: cp = curp->primary->buf;
282: ep = cp + curp->primary->sz;
283: while (cp < ep) {
1.61 schwarze 284: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 285: cp++;
286: if (cp[0] == 'D' && cp[1] == 'd') {
287: format = MPARSE_MDOC;
288: break;
289: }
290: if (cp[0] == 'T' && cp[1] == 'H') {
291: format = MPARSE_MAN;
292: break;
293: }
294: }
295: cp = memchr(cp, '\n', ep - cp);
296: if (cp == NULL)
297: break;
298: cp++;
299: }
1.1 schwarze 300: }
301:
1.59 schwarze 302: if (format == MPARSE_MDOC) {
1.112 schwarze 303: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 304: if (curp->man->mdocmac == NULL)
305: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 306: } else {
307: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 308: if (curp->man->manmac == NULL)
309: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 310: }
1.136 schwarze 311: curp->man->first->tok = TOKEN_NONE;
1.1 schwarze 312: }
313:
314: /*
1.71 schwarze 315: * Main parse routine for a buffer.
316: * It assumes encoding and line numbering are already set up.
317: * It can recurse directly (for invocations of user-defined
318: * macros, inline equations, and input line traps)
319: * and indirectly (for .so file inclusion).
1.1 schwarze 320: */
1.135 schwarze 321: static int
1.71 schwarze 322: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 323: {
324: const struct tbl_span *span;
325: struct buf ln;
1.88 schwarze 326: const char *save_file;
1.85 schwarze 327: char *cp;
1.71 schwarze 328: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 329: enum rofferr rr;
1.76 schwarze 330: int of;
1.1 schwarze 331: int lnn; /* line number in the real file */
1.85 schwarze 332: int fd;
1.1 schwarze 333: unsigned char c;
334:
1.71 schwarze 335: memset(&ln, 0, sizeof(ln));
1.1 schwarze 336:
1.25 schwarze 337: lnn = curp->line;
338: pos = 0;
1.1 schwarze 339:
1.71 schwarze 340: while (i < blk.sz) {
1.1 schwarze 341: if (0 == pos && '\0' == blk.buf[i])
342: break;
343:
344: if (start) {
345: curp->line = lnn;
346: curp->reparse_count = 0;
1.70 schwarze 347:
348: if (lnn < 3 &&
349: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 350: curp->filenc & MPARSE_LATIN1)
351: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 352: }
353:
1.71 schwarze 354: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 355:
356: /*
357: * When finding an unescaped newline character,
358: * leave the character loop to process the line.
359: * Skip a preceding carriage return, if any.
360: */
361:
1.71 schwarze 362: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 363: '\n' == blk.buf[i + 1])
364: ++i;
365: if ('\n' == blk.buf[i]) {
366: ++i;
367: ++lnn;
368: break;
369: }
370:
1.13 schwarze 371: /*
1.70 schwarze 372: * Make sure we have space for the worst
373: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 374: */
375:
1.71 schwarze 376: if (pos + 11 > ln.sz)
1.13 schwarze 377: resize_buf(&ln, 256);
378:
1.25 schwarze 379: /*
1.70 schwarze 380: * Encode 8-bit input.
1.1 schwarze 381: */
382:
1.70 schwarze 383: c = blk.buf[i];
384: if (c & 0x80) {
1.71 schwarze 385: if ( ! (curp->filenc && preconv_encode(
386: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 387: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
388: curp->line, pos, "0x%x", c);
1.70 schwarze 389: ln.buf[pos++] = '?';
390: i++;
391: }
392: continue;
393: }
394:
395: /*
396: * Exclude control characters.
397: */
1.1 schwarze 398:
1.70 schwarze 399: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 400: mandoc_vmsg(c == 0x00 || c == 0x04 ||
401: c > 0x0a ? MANDOCERR_CHAR_BAD :
402: MANDOCERR_CHAR_UNSUPP,
403: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 404: i++;
1.102 schwarze 405: if (c != '\r')
406: ln.buf[pos++] = '?';
1.1 schwarze 407: continue;
408: }
409:
410: ln.buf[pos++] = blk.buf[i++];
411: }
412:
1.142 schwarze 413: if (pos + 1 >= ln.sz)
1.1 schwarze 414: resize_buf(&ln, 256);
415:
1.142 schwarze 416: if (i == blk.sz || blk.buf[i] == '\0')
417: ln.buf[pos++] = '\n';
1.1 schwarze 418: ln.buf[pos] = '\0';
419:
420: /*
421: * A significant amount of complexity is contained by
422: * the roff preprocessor. It's line-oriented but can be
423: * expressed on one line, so we need at times to
424: * readjust our starting point and re-run it. The roff
425: * preprocessor can also readjust the buffers with new
426: * data, so we pass them in wholesale.
427: */
428:
429: of = 0;
430:
1.4 schwarze 431: /*
432: * Maintain a lookaside buffer of all parsed lines. We
433: * only do this if mparse_keep() has been invoked (the
434: * buffer may be accessed with mparse_getkeep()).
435: */
436:
437: if (curp->secondary) {
1.25 schwarze 438: curp->secondary->buf = mandoc_realloc(
439: curp->secondary->buf,
440: curp->secondary->sz + pos + 2);
441: memcpy(curp->secondary->buf +
442: curp->secondary->sz,
443: ln.buf, pos);
1.4 schwarze 444: curp->secondary->sz += pos;
445: curp->secondary->buf
446: [curp->secondary->sz] = '\n';
447: curp->secondary->sz++;
448: curp->secondary->buf
449: [curp->secondary->sz] = '\0';
450: }
1.1 schwarze 451: rerun:
1.72 schwarze 452: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 453:
454: switch (rr) {
1.25 schwarze 455: case ROFF_REPARSE:
1.135 schwarze 456: if (++curp->reparse_count > REPARSE_LIMIT)
1.1 schwarze 457: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 458: curp->line, pos, NULL);
1.135 schwarze 459: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
460: start == 1) {
461: pos = 0;
462: continue;
463: }
464: free(ln.buf);
465: return 0;
1.25 schwarze 466: case ROFF_APPEND:
1.71 schwarze 467: pos = strlen(ln.buf);
1.1 schwarze 468: continue;
1.25 schwarze 469: case ROFF_RERUN:
1.1 schwarze 470: goto rerun;
1.25 schwarze 471: case ROFF_IGN:
1.1 schwarze 472: pos = 0;
473: continue;
1.25 schwarze 474: case ROFF_SO:
1.71 schwarze 475: if ( ! (curp->options & MPARSE_SO) &&
476: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 477: curp->sodest = mandoc_strdup(ln.buf + of);
478: free(ln.buf);
1.135 schwarze 479: return 1;
1.23 schwarze 480: }
1.4 schwarze 481: /*
482: * We remove `so' clauses from our lookaside
483: * buffer because we're going to descend into
484: * the file recursively.
485: */
1.25 schwarze 486: if (curp->secondary)
1.4 schwarze 487: curp->secondary->sz -= pos + 1;
1.88 schwarze 488: save_file = curp->file;
1.122 schwarze 489: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 490: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 491: close(fd);
1.88 schwarze 492: curp->file = save_file;
493: } else {
494: curp->file = save_file;
1.30 schwarze 495: mandoc_vmsg(MANDOCERR_SO_FAIL,
496: curp, curp->line, pos,
497: ".so %s", ln.buf + of);
1.85 schwarze 498: ln.sz = mandoc_asprintf(&cp,
499: ".sp\nSee the file %s.\n.sp",
500: ln.buf + of);
501: free(ln.buf);
502: ln.buf = cp;
503: of = 0;
504: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 505: }
1.1 schwarze 506: pos = 0;
507: continue;
508: default:
509: break;
510: }
511:
1.124 schwarze 512: if (curp->man->macroset == MACROSET_NONE)
1.60 schwarze 513: choose_parser(curp);
1.1 schwarze 514:
1.25 schwarze 515: /*
1.60 schwarze 516: * Lastly, push down into the parsers themselves.
1.1 schwarze 517: * If libroff returns ROFF_TBL, then add it to the
518: * currently open parse. Since we only get here if
519: * there does exist data (see tbl_data.c), we're
520: * guaranteed that something's been allocated.
521: * Do the same for ROFF_EQN.
522: */
523:
1.114 schwarze 524: if (rr == ROFF_TBL)
1.76 schwarze 525: while ((span = roff_span(curp->roff)) != NULL)
1.114 schwarze 526: roff_addtbl(curp->man, span);
527: else if (rr == ROFF_EQN)
528: roff_addeqn(curp->man, roff_eqn(curp->roff));
529: else if ((curp->man->macroset == MACROSET_MDOC ?
1.110 schwarze 530: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 531: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
532: break;
1.1 schwarze 533:
534: /* Temporary buffers typically are not full. */
535:
536: if (0 == start && '\0' == blk.buf[i])
537: break;
538:
539: /* Start the next input line. */
540:
541: pos = 0;
542: }
543:
544: free(ln.buf);
1.135 schwarze 545: return 1;
1.1 schwarze 546: }
547:
548: static int
1.18 schwarze 549: read_whole_file(struct mparse *curp, const char *file, int fd,
550: struct buf *fb, int *with_mmap)
1.1 schwarze 551: {
552: struct stat st;
1.115 schwarze 553: gzFile gz;
1.1 schwarze 554: size_t off;
555: ssize_t ssz;
556:
1.118 schwarze 557: if (fstat(fd, &st) == -1)
558: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 559:
560: /*
561: * If we're a regular file, try just reading in the whole entry
562: * via mmap(). This is faster than reading it into blocks, and
563: * since each file is only a few bytes to begin with, I'm not
564: * concerned that this is going to tank any machines.
565: */
566:
1.115 schwarze 567: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 568: if (st.st_size > 0x7fffffff) {
1.86 schwarze 569: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.117 schwarze 570: return 0;
1.1 schwarze 571: }
572: *with_mmap = 1;
573: fb->sz = (size_t)st.st_size;
1.15 schwarze 574: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 575: if (fb->buf != MAP_FAILED)
1.117 schwarze 576: return 1;
1.1 schwarze 577: }
578:
1.115 schwarze 579: if (curp->gzip) {
1.118 schwarze 580: if ((gz = gzdopen(fd, "rb")) == NULL)
581: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.115 schwarze 582: } else
583: gz = NULL;
584:
1.1 schwarze 585: /*
586: * If this isn't a regular file (like, say, stdin), then we must
587: * go the old way and just read things in bit by bit.
588: */
589:
590: *with_mmap = 0;
591: off = 0;
592: fb->sz = 0;
593: fb->buf = NULL;
594: for (;;) {
595: if (off == fb->sz) {
596: if (fb->sz == (1U << 31)) {
1.86 schwarze 597: mandoc_msg(MANDOCERR_TOOLARGE, curp,
598: 0, 0, NULL);
1.1 schwarze 599: break;
600: }
601: resize_buf(fb, 65536);
602: }
1.115 schwarze 603: ssz = curp->gzip ?
604: gzread(gz, fb->buf + (int)off, fb->sz - off) :
605: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 606: if (ssz == 0) {
607: fb->sz = off;
1.117 schwarze 608: return 1;
1.1 schwarze 609: }
1.118 schwarze 610: if (ssz == -1)
611: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 612: off += (size_t)ssz;
613: }
614:
615: free(fb->buf);
616: fb->buf = NULL;
1.117 schwarze 617: return 0;
1.1 schwarze 618: }
619:
620: static void
621: mparse_end(struct mparse *curp)
622: {
1.110 schwarze 623: if (curp->man->macroset == MACROSET_NONE)
624: curp->man->macroset = MACROSET_MAN;
625: if (curp->man->macroset == MACROSET_MDOC)
626: mdoc_endparse(curp->man);
627: else
1.86 schwarze 628: man_endparse(curp->man);
1.1 schwarze 629: roff_endparse(curp->roff);
630: }
631:
1.15 schwarze 632: static void
633: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 634: {
1.61 schwarze 635: struct buf *svprimary;
1.1 schwarze 636: const char *svfile;
1.71 schwarze 637: size_t offset;
1.14 schwarze 638: static int recursion_depth;
639:
640: if (64 < recursion_depth) {
641: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 642: return;
1.14 schwarze 643: }
1.1 schwarze 644:
1.15 schwarze 645: /* Line number is per-file. */
646: svfile = curp->file;
647: curp->file = file;
1.61 schwarze 648: svprimary = curp->primary;
1.59 schwarze 649: curp->primary = &blk;
1.15 schwarze 650: curp->line = 1;
651: recursion_depth++;
652:
1.70 schwarze 653: /* Skip an UTF-8 byte order mark. */
654: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
655: (unsigned char)blk.buf[0] == 0xef &&
656: (unsigned char)blk.buf[1] == 0xbb &&
657: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 658: offset = 3;
1.70 schwarze 659: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 660: } else
661: offset = 0;
1.70 schwarze 662:
1.71 schwarze 663: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 664:
1.86 schwarze 665: if (--recursion_depth == 0)
1.15 schwarze 666: mparse_end(curp);
667:
1.61 schwarze 668: curp->primary = svprimary;
1.15 schwarze 669: curp->file = svfile;
670: }
671:
1.74 schwarze 672: /*
673: * Read the whole file into memory and call the parsers.
674: * Called recursively when an .so request is encountered.
675: */
1.15 schwarze 676: enum mandoclevel
677: mparse_readfd(struct mparse *curp, int fd, const char *file)
678: {
679: struct buf blk;
680: int with_mmap;
1.70 schwarze 681: int save_filenc;
1.1 schwarze 682:
1.67 schwarze 683: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 684: save_filenc = curp->filenc;
685: curp->filenc = curp->options &
686: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 687: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 688: curp->filenc = save_filenc;
1.67 schwarze 689: if (with_mmap)
690: munmap(blk.buf, blk.sz);
691: else
692: free(blk.buf);
693: }
1.117 schwarze 694: return curp->file_status;
1.58 schwarze 695: }
696:
1.122 schwarze 697: int
698: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 699: {
700: char *cp;
1.122 schwarze 701: int fd;
1.58 schwarze 702:
703: curp->file = file;
1.115 schwarze 704: cp = strrchr(file, '.');
705: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 706:
1.115 schwarze 707: /* First try to use the filename as it is. */
1.74 schwarze 708:
1.122 schwarze 709: if ((fd = open(file, O_RDONLY)) != -1)
710: return fd;
1.74 schwarze 711:
1.115 schwarze 712: /*
713: * If that doesn't work and the filename doesn't
714: * already end in .gz, try appending .gz.
715: */
1.74 schwarze 716:
1.115 schwarze 717: if ( ! curp->gzip) {
1.74 schwarze 718: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 719: fd = open(cp, O_RDONLY);
1.83 schwarze 720: free(cp);
1.122 schwarze 721: if (fd != -1) {
1.115 schwarze 722: curp->gzip = 1;
1.122 schwarze 723: return fd;
1.58 schwarze 724: }
725: }
726:
1.115 schwarze 727: /* Neither worked, give up. */
1.73 schwarze 728:
1.115 schwarze 729: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.122 schwarze 730: return -1;
1.1 schwarze 731: }
732:
733: struct mparse *
1.69 schwarze 734: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.119 schwarze 735: const char *defos)
1.1 schwarze 736: {
737: struct mparse *curp;
738:
739: curp = mandoc_calloc(1, sizeof(struct mparse));
740:
1.22 schwarze 741: curp->options = options;
1.1 schwarze 742: curp->wlevel = wlevel;
743: curp->mmsg = mmsg;
1.7 schwarze 744: curp->defos = defos;
1.1 schwarze 745:
1.119 schwarze 746: curp->roff = roff_alloc(curp, options);
1.112 schwarze 747: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
748: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 749: if (curp->options & MPARSE_MDOC) {
1.112 schwarze 750: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 751: if (curp->man->mdocmac == NULL)
752: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 753: } else if (curp->options & MPARSE_MAN) {
754: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 755: if (curp->man->manmac == NULL)
756: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 757: }
1.113 schwarze 758: curp->man->first->tok = TOKEN_NONE;
1.117 schwarze 759: return curp;
1.1 schwarze 760: }
761:
762: void
763: mparse_reset(struct mparse *curp)
764: {
765: roff_reset(curp->roff);
1.124 schwarze 766: roff_man_reset(curp->man);
1.133 schwarze 767:
768: free(curp->sodest);
769: curp->sodest = NULL;
770:
1.4 schwarze 771: if (curp->secondary)
772: curp->secondary->sz = 0;
1.1 schwarze 773:
774: curp->file_status = MANDOCLEVEL_OK;
1.132 schwarze 775: curp->gzip = 0;
1.1 schwarze 776: }
777:
778: void
779: mparse_free(struct mparse *curp)
780: {
781:
1.136 schwarze 782: roffhash_free(curp->man->mdocmac);
783: roffhash_free(curp->man->manmac);
1.112 schwarze 784: roff_man_free(curp->man);
1.133 schwarze 785: roff_free(curp->roff);
1.4 schwarze 786: if (curp->secondary)
787: free(curp->secondary->buf);
1.1 schwarze 788:
1.4 schwarze 789: free(curp->secondary);
1.23 schwarze 790: free(curp->sodest);
1.1 schwarze 791: free(curp);
792: }
793:
794: void
1.110 schwarze 795: mparse_result(struct mparse *curp, struct roff_man **man,
796: char **sodest)
1.1 schwarze 797: {
798:
1.23 schwarze 799: if (sodest && NULL != (*sodest = curp->sodest)) {
800: *man = NULL;
801: return;
802: }
1.1 schwarze 803: if (man)
804: *man = curp->man;
1.130 schwarze 805: }
806:
807: void
808: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
809: {
810: if (curp->file_status > *rc)
811: *rc = curp->file_status;
1.1 schwarze 812: }
813:
814: void
815: mandoc_vmsg(enum mandocerr t, struct mparse *m,
816: int ln, int pos, const char *fmt, ...)
817: {
818: char buf[256];
819: va_list ap;
820:
821: va_start(ap, fmt);
1.26 schwarze 822: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 823: va_end(ap);
824:
825: mandoc_msg(t, m, ln, pos, buf);
826: }
827:
828: void
1.25 schwarze 829: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 830: int ln, int col, const char *msg)
831: {
832: enum mandoclevel level;
833:
1.87 schwarze 834: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 835: while (er < mandoclimits[level])
836: level--;
837:
1.83 schwarze 838: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 839: return;
840:
841: if (m->mmsg)
842: (*m->mmsg)(er, level, m->file, ln, col, msg);
843:
844: if (m->file_status < level)
845: m->file_status = level;
846: }
847:
848: const char *
849: mparse_strerror(enum mandocerr er)
850: {
851:
1.117 schwarze 852: return mandocerrs[er];
1.1 schwarze 853: }
854:
855: const char *
856: mparse_strlevel(enum mandoclevel lvl)
857: {
1.117 schwarze 858: return mandoclevels[lvl];
1.4 schwarze 859: }
860:
861: void
862: mparse_keep(struct mparse *p)
863: {
864:
865: assert(NULL == p->secondary);
866: p->secondary = mandoc_calloc(1, sizeof(struct buf));
867: }
868:
869: const char *
870: mparse_getkeep(const struct mparse *p)
871: {
872:
873: assert(p->secondary);
1.117 schwarze 874: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 schwarze 875: }