Annotation of src/usr.bin/mandoc/read.c, Revision 1.129
1.129 ! schwarze 1: /* $OpenBSD: read.c,v 1.128 2016/12/28 17:21:17 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 {
1.109 schwarze 46: struct roff_man *man; /* man parser */
1.1 schwarze 47: struct roff *roff; /* roff parser (!NULL) */
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);
1.71 schwarze 65: static void 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,
74: MANDOCERR_WARNING,
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:
85: "generic warning",
86:
87: /* related to the prologue */
1.57 schwarze 88: "missing manual title, using UNTITLED",
89: "missing manual title, using \"\"",
1.32 schwarze 90: "lower case character in document title",
1.57 schwarze 91: "missing manual section, using \"\"",
1.1 schwarze 92: "unknown manual section",
1.32 schwarze 93: "missing date, using today's date",
1.1 schwarze 94: "cannot parse date, using it verbatim",
1.57 schwarze 95: "missing Os macro, using \"\"",
96: "duplicate prologue macro",
97: "late prologue macro",
98: "skipping late title macro",
1.1 schwarze 99: "prologue macros out of order",
100:
101: /* related to document structure */
102: ".so is fragile, better use ln(1)",
1.28 schwarze 103: "no document body",
1.32 schwarze 104: "content before first section header",
105: "first section is not \"NAME\"",
1.129 ! schwarze 106: "NAME section without Nm before Nd",
1.103 schwarze 107: "NAME section without description",
108: "description not at the end of NAME",
109: "bad NAME section content",
1.129 ! schwarze 110: "missing comma before name",
1.96 schwarze 111: "missing description line, using \"\"",
1.1 schwarze 112: "sections out of conventional order",
1.32 schwarze 113: "duplicate section title",
114: "unexpected section",
1.63 schwarze 115: "unusual Xr order",
116: "unusual Xr punctuation",
1.62 schwarze 117: "AUTHORS section without An macro",
1.1 schwarze 118:
119: /* related to macros and nesting */
1.33 schwarze 120: "obsolete macro",
1.78 schwarze 121: "macro neither callable nor escaped",
1.1 schwarze 122: "skipping paragraph macro",
1.10 schwarze 123: "moving paragraph macro out of list",
1.1 schwarze 124: "skipping no-space macro",
125: "blocks badly nested",
126: "nested displays are not portable",
1.35 schwarze 127: "moving content out of list",
1.56 schwarze 128: "fill mode already enabled, skipping",
129: "fill mode already disabled, skipping",
1.1 schwarze 130: "line scope broken",
131:
132: /* related to missing macro arguments */
1.36 schwarze 133: "skipping empty request",
134: "conditional request controls empty scope",
1.1 schwarze 135: "skipping empty macro",
1.99 schwarze 136: "empty block",
1.40 schwarze 137: "empty argument, using 0n",
1.38 schwarze 138: "missing display type, using -ragged",
139: "list type is not the first argument",
1.125 schwarze 140: "missing -width in -tag list, using 6n",
1.56 schwarze 141: "missing utility name, using \"\"",
1.98 schwarze 142: "missing function name, using \"\"",
1.38 schwarze 143: "empty head in list item",
144: "empty list item",
1.39 schwarze 145: "missing font type, using \\fR",
146: "unknown font type, using \\fR",
1.79 schwarze 147: "nothing follows prefix",
1.97 schwarze 148: "empty reference block",
1.128 schwarze 149: "missing section argument",
1.38 schwarze 150: "missing -std argument, adding it",
1.100 schwarze 151: "missing option string, using \"\"",
152: "missing resource identifier, using \"\"",
1.66 schwarze 153: "missing eqn box, using \"\"",
1.1 schwarze 154:
155: /* related to bad macro arguments */
1.42 schwarze 156: "unterminated quoted argument",
1.1 schwarze 157: "duplicate argument",
1.54 schwarze 158: "skipping duplicate argument",
1.41 schwarze 159: "skipping duplicate display type",
160: "skipping duplicate list type",
1.54 schwarze 161: "skipping -width argument",
1.101 schwarze 162: "wrong number of cells",
1.1 schwarze 163: "unknown AT&T UNIX version",
1.64 schwarze 164: "comma in function argument",
1.65 schwarze 165: "parenthesis in function name",
1.45 schwarze 166: "invalid content in Rs block",
1.41 schwarze 167: "invalid Boolean argument",
168: "unknown font, skipping request",
1.101 schwarze 169: "odd number of characters in request",
1.1 schwarze 170:
171: /* related to plain text */
1.42 schwarze 172: "blank line in fill mode, using .sp",
173: "tab in filled text",
174: "whitespace at end of input line",
1.1 schwarze 175: "bad comment style",
1.42 schwarze 176: "invalid escape sequence",
177: "undefined string, using \"\"",
1.3 schwarze 178:
1.93 schwarze 179: /* related to tables */
180: "tbl line starts with span",
181: "tbl column starts with span",
182: "skipping vertical bar in tbl layout",
183:
1.1 schwarze 184: "generic error",
1.3 schwarze 185:
1.1 schwarze 186: /* related to tables */
1.91 schwarze 187: "non-alphabetic character in tbl options",
188: "skipping unknown tbl option",
189: "missing tbl option argument",
190: "wrong tbl option argument size",
1.93 schwarze 191: "empty tbl layout",
192: "invalid character in tbl layout",
193: "unmatched parenthesis in tbl layout",
1.94 schwarze 194: "tbl without any data cells",
195: "ignoring data in spanned tbl cell",
196: "ignoring extra tbl data cells",
197: "data block open at end of tbl",
1.1 schwarze 198:
1.46 schwarze 199: /* related to document structure and macros */
1.83 schwarze 200: NULL,
1.1 schwarze 201: "input stack limit exceeded, infinite loop?",
202: "skipping bad character",
1.46 schwarze 203: "skipping unknown macro",
1.87 schwarze 204: "skipping insecure request",
1.51 schwarze 205: "skipping item outside list",
1.46 schwarze 206: "skipping column outside column list",
207: "skipping end of block that is not open",
1.90 schwarze 208: "fewer RS blocks open, skipping",
1.46 schwarze 209: "inserting missing end of block",
210: "appending missing end of block",
211:
212: /* related to request and macro arguments */
1.1 schwarze 213: "escaped character not allowed in a name",
1.75 schwarze 214: "NOT IMPLEMENTED: Bd -file",
1.120 schwarze 215: "skipping display without arguments",
1.49 schwarze 216: "missing list type, using -item",
1.48 schwarze 217: "missing manual name, using \"\"",
1.49 schwarze 218: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 219: "unknown standard specifier",
1.49 schwarze 220: "skipping request without numeric argument",
1.84 schwarze 221: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 222: ".so request failed",
1.39 schwarze 223: "skipping all arguments",
224: "skipping excess arguments",
1.68 schwarze 225: "divide by zero",
1.87 schwarze 226:
227: "unsupported feature",
228: "input too large",
1.89 schwarze 229: "unsupported control character",
1.87 schwarze 230: "unsupported roff request",
1.94 schwarze 231: "eqn delim option in tbl",
1.93 schwarze 232: "unsupported tbl layout modifier",
1.87 schwarze 233: "ignoring macro in table",
1.1 schwarze 234: };
235:
236: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
237: "SUCCESS",
238: "RESERVED",
239: "WARNING",
240: "ERROR",
1.87 schwarze 241: "UNSUPP",
1.1 schwarze 242: "BADARG",
243: "SYSERR"
244: };
245:
1.25 schwarze 246:
1.1 schwarze 247: static void
248: resize_buf(struct buf *buf, size_t initial)
249: {
250:
251: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
252: buf->buf = mandoc_realloc(buf->buf, buf->sz);
253: }
254:
255: static void
1.60 schwarze 256: choose_parser(struct mparse *curp)
1.1 schwarze 257: {
1.59 schwarze 258: char *cp, *ep;
259: int format;
1.1 schwarze 260:
1.59 schwarze 261: /*
262: * If neither command line arguments -mdoc or -man select
263: * a parser nor the roff parser found a .Dd or .TH macro
264: * yet, look ahead in the main input buffer.
265: */
266:
267: if ((format = roff_getformat(curp->roff)) == 0) {
268: cp = curp->primary->buf;
269: ep = cp + curp->primary->sz;
270: while (cp < ep) {
1.61 schwarze 271: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 272: cp++;
273: if (cp[0] == 'D' && cp[1] == 'd') {
274: format = MPARSE_MDOC;
275: break;
276: }
277: if (cp[0] == 'T' && cp[1] == 'H') {
278: format = MPARSE_MAN;
279: break;
280: }
281: }
282: cp = memchr(cp, '\n', ep - cp);
283: if (cp == NULL)
284: break;
285: cp++;
286: }
1.1 schwarze 287: }
288:
1.59 schwarze 289: if (format == MPARSE_MDOC) {
1.111 schwarze 290: mdoc_hash_init();
1.112 schwarze 291: curp->man->macroset = MACROSET_MDOC;
1.113 schwarze 292: curp->man->first->tok = TOKEN_NONE;
1.112 schwarze 293: } else {
294: man_hash_init();
295: curp->man->macroset = MACROSET_MAN;
1.113 schwarze 296: curp->man->first->tok = TOKEN_NONE;
1.25 schwarze 297: }
1.1 schwarze 298: }
299:
300: /*
1.71 schwarze 301: * Main parse routine for a buffer.
302: * It assumes encoding and line numbering are already set up.
303: * It can recurse directly (for invocations of user-defined
304: * macros, inline equations, and input line traps)
305: * and indirectly (for .so file inclusion).
1.1 schwarze 306: */
307: static void
1.71 schwarze 308: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 309: {
310: const struct tbl_span *span;
311: struct buf ln;
1.88 schwarze 312: const char *save_file;
1.85 schwarze 313: char *cp;
1.71 schwarze 314: size_t pos; /* byte number in the ln buffer */
1.127 schwarze 315: size_t j; /* auxiliary byte number in the blk buffer */
1.1 schwarze 316: enum rofferr rr;
1.76 schwarze 317: int of;
1.1 schwarze 318: int lnn; /* line number in the real file */
1.85 schwarze 319: int fd;
1.1 schwarze 320: unsigned char c;
321:
1.71 schwarze 322: memset(&ln, 0, sizeof(ln));
1.1 schwarze 323:
1.25 schwarze 324: lnn = curp->line;
325: pos = 0;
1.1 schwarze 326:
1.71 schwarze 327: while (i < blk.sz) {
1.1 schwarze 328: if (0 == pos && '\0' == blk.buf[i])
329: break;
330:
331: if (start) {
332: curp->line = lnn;
333: curp->reparse_count = 0;
1.70 schwarze 334:
335: if (lnn < 3 &&
336: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 337: curp->filenc & MPARSE_LATIN1)
338: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 339: }
340:
1.71 schwarze 341: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 342:
343: /*
344: * When finding an unescaped newline character,
345: * leave the character loop to process the line.
346: * Skip a preceding carriage return, if any.
347: */
348:
1.71 schwarze 349: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 350: '\n' == blk.buf[i + 1])
351: ++i;
352: if ('\n' == blk.buf[i]) {
353: ++i;
354: ++lnn;
355: break;
356: }
357:
1.13 schwarze 358: /*
1.70 schwarze 359: * Make sure we have space for the worst
360: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 361: */
362:
1.71 schwarze 363: if (pos + 11 > ln.sz)
1.13 schwarze 364: resize_buf(&ln, 256);
365:
1.25 schwarze 366: /*
1.70 schwarze 367: * Encode 8-bit input.
1.1 schwarze 368: */
369:
1.70 schwarze 370: c = blk.buf[i];
371: if (c & 0x80) {
1.71 schwarze 372: if ( ! (curp->filenc && preconv_encode(
373: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 374: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
375: curp->line, pos, "0x%x", c);
1.70 schwarze 376: ln.buf[pos++] = '?';
377: i++;
378: }
379: continue;
380: }
381:
382: /*
383: * Exclude control characters.
384: */
1.1 schwarze 385:
1.70 schwarze 386: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 387: mandoc_vmsg(c == 0x00 || c == 0x04 ||
388: c > 0x0a ? MANDOCERR_CHAR_BAD :
389: MANDOCERR_CHAR_UNSUPP,
390: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 391: i++;
1.102 schwarze 392: if (c != '\r')
393: ln.buf[pos++] = '?';
1.1 schwarze 394: continue;
395: }
396:
397: /* Trailing backslash = a plain char. */
398:
1.71 schwarze 399: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 schwarze 400: ln.buf[pos++] = blk.buf[i++];
401: continue;
402: }
403:
404: /*
405: * Found escape and at least one other character.
406: * When it's a newline character, skip it.
407: * When there is a carriage return in between,
408: * skip that one as well.
409: */
410:
1.71 schwarze 411: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 schwarze 412: '\n' == blk.buf[i + 2])
413: ++i;
414: if ('\n' == blk.buf[i + 1]) {
415: i += 2;
416: ++lnn;
417: continue;
418: }
419:
420: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.127 schwarze 421: j = i;
1.1 schwarze 422: i += 2;
423: /* Comment, skip to end of line */
1.71 schwarze 424: for (; i < blk.sz; ++i) {
1.126 schwarze 425: if (blk.buf[i] != '\n')
426: continue;
427: if (blk.buf[i - 1] == ' ' ||
428: blk.buf[i - 1] == '\t')
429: mandoc_msg(
430: MANDOCERR_SPACE_EOL,
431: curp, curp->line,
1.127 schwarze 432: pos + i-1 - j, NULL);
1.126 schwarze 433: ++i;
434: ++lnn;
435: break;
1.1 schwarze 436: }
437:
438: /* Backout trailing whitespaces */
439: for (; pos > 0; --pos) {
440: if (ln.buf[pos - 1] != ' ')
441: break;
442: if (pos > 2 && ln.buf[pos - 2] == '\\')
443: break;
444: }
445: break;
446: }
447:
1.13 schwarze 448: /* Catch escaped bogus characters. */
449:
450: c = (unsigned char) blk.buf[i+1];
451:
1.25 schwarze 452: if ( ! (isascii(c) &&
453: (isgraph(c) || isblank(c)))) {
1.89 schwarze 454: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.56 schwarze 455: curp->line, pos, "0x%x", c);
1.13 schwarze 456: i += 2;
457: ln.buf[pos++] = '?';
458: continue;
459: }
460:
1.1 schwarze 461: /* Some other escape sequence, copy & cont. */
462:
463: ln.buf[pos++] = blk.buf[i++];
464: ln.buf[pos++] = blk.buf[i++];
465: }
466:
1.71 schwarze 467: if (pos >= ln.sz)
1.1 schwarze 468: resize_buf(&ln, 256);
469:
470: ln.buf[pos] = '\0';
471:
472: /*
473: * A significant amount of complexity is contained by
474: * the roff preprocessor. It's line-oriented but can be
475: * expressed on one line, so we need at times to
476: * readjust our starting point and re-run it. The roff
477: * preprocessor can also readjust the buffers with new
478: * data, so we pass them in wholesale.
479: */
480:
481: of = 0;
482:
1.4 schwarze 483: /*
484: * Maintain a lookaside buffer of all parsed lines. We
485: * only do this if mparse_keep() has been invoked (the
486: * buffer may be accessed with mparse_getkeep()).
487: */
488:
489: if (curp->secondary) {
1.25 schwarze 490: curp->secondary->buf = mandoc_realloc(
491: curp->secondary->buf,
492: curp->secondary->sz + pos + 2);
493: memcpy(curp->secondary->buf +
494: curp->secondary->sz,
495: ln.buf, pos);
1.4 schwarze 496: curp->secondary->sz += pos;
497: curp->secondary->buf
498: [curp->secondary->sz] = '\n';
499: curp->secondary->sz++;
500: curp->secondary->buf
501: [curp->secondary->sz] = '\0';
502: }
1.1 schwarze 503: rerun:
1.72 schwarze 504: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 505:
506: switch (rr) {
1.25 schwarze 507: case ROFF_REPARSE:
1.1 schwarze 508: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.71 schwarze 509: mparse_buf_r(curp, ln, of, 0);
1.1 schwarze 510: else
511: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 512: curp->line, pos, NULL);
1.1 schwarze 513: pos = 0;
514: continue;
1.25 schwarze 515: case ROFF_APPEND:
1.71 schwarze 516: pos = strlen(ln.buf);
1.1 schwarze 517: continue;
1.25 schwarze 518: case ROFF_RERUN:
1.1 schwarze 519: goto rerun;
1.25 schwarze 520: case ROFF_IGN:
1.1 schwarze 521: pos = 0;
522: continue;
1.25 schwarze 523: case ROFF_SO:
1.71 schwarze 524: if ( ! (curp->options & MPARSE_SO) &&
525: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 526: curp->sodest = mandoc_strdup(ln.buf + of);
527: free(ln.buf);
528: return;
529: }
1.4 schwarze 530: /*
531: * We remove `so' clauses from our lookaside
532: * buffer because we're going to descend into
533: * the file recursively.
534: */
1.25 schwarze 535: if (curp->secondary)
1.4 schwarze 536: curp->secondary->sz -= pos + 1;
1.88 schwarze 537: save_file = curp->file;
1.122 schwarze 538: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 539: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 540: close(fd);
1.88 schwarze 541: curp->file = save_file;
542: } else {
543: curp->file = save_file;
1.30 schwarze 544: mandoc_vmsg(MANDOCERR_SO_FAIL,
545: curp, curp->line, pos,
546: ".so %s", ln.buf + of);
1.85 schwarze 547: ln.sz = mandoc_asprintf(&cp,
548: ".sp\nSee the file %s.\n.sp",
549: ln.buf + of);
550: free(ln.buf);
551: ln.buf = cp;
552: of = 0;
553: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 554: }
1.1 schwarze 555: pos = 0;
556: continue;
557: default:
558: break;
559: }
560:
1.124 schwarze 561: if (curp->man->macroset == MACROSET_NONE)
1.60 schwarze 562: choose_parser(curp);
1.1 schwarze 563:
1.25 schwarze 564: /*
1.60 schwarze 565: * Lastly, push down into the parsers themselves.
1.1 schwarze 566: * If libroff returns ROFF_TBL, then add it to the
567: * currently open parse. Since we only get here if
568: * there does exist data (see tbl_data.c), we're
569: * guaranteed that something's been allocated.
570: * Do the same for ROFF_EQN.
571: */
572:
1.114 schwarze 573: if (rr == ROFF_TBL)
1.76 schwarze 574: while ((span = roff_span(curp->roff)) != NULL)
1.114 schwarze 575: roff_addtbl(curp->man, span);
576: else if (rr == ROFF_EQN)
577: roff_addeqn(curp->man, roff_eqn(curp->roff));
578: else if ((curp->man->macroset == MACROSET_MDOC ?
1.110 schwarze 579: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 580: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
581: break;
1.1 schwarze 582:
583: /* Temporary buffers typically are not full. */
584:
585: if (0 == start && '\0' == blk.buf[i])
586: break;
587:
588: /* Start the next input line. */
589:
590: pos = 0;
591: }
592:
593: free(ln.buf);
594: }
595:
596: static int
1.18 schwarze 597: read_whole_file(struct mparse *curp, const char *file, int fd,
598: struct buf *fb, int *with_mmap)
1.1 schwarze 599: {
600: struct stat st;
1.115 schwarze 601: gzFile gz;
1.1 schwarze 602: size_t off;
603: ssize_t ssz;
604:
1.118 schwarze 605: if (fstat(fd, &st) == -1)
606: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 607:
608: /*
609: * If we're a regular file, try just reading in the whole entry
610: * via mmap(). This is faster than reading it into blocks, and
611: * since each file is only a few bytes to begin with, I'm not
612: * concerned that this is going to tank any machines.
613: */
614:
1.115 schwarze 615: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 616: if (st.st_size > 0x7fffffff) {
1.86 schwarze 617: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.117 schwarze 618: return 0;
1.1 schwarze 619: }
620: *with_mmap = 1;
621: fb->sz = (size_t)st.st_size;
1.15 schwarze 622: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 623: if (fb->buf != MAP_FAILED)
1.117 schwarze 624: return 1;
1.1 schwarze 625: }
626:
1.115 schwarze 627: if (curp->gzip) {
1.118 schwarze 628: if ((gz = gzdopen(fd, "rb")) == NULL)
629: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.115 schwarze 630: } else
631: gz = NULL;
632:
1.1 schwarze 633: /*
634: * If this isn't a regular file (like, say, stdin), then we must
635: * go the old way and just read things in bit by bit.
636: */
637:
638: *with_mmap = 0;
639: off = 0;
640: fb->sz = 0;
641: fb->buf = NULL;
642: for (;;) {
643: if (off == fb->sz) {
644: if (fb->sz == (1U << 31)) {
1.86 schwarze 645: mandoc_msg(MANDOCERR_TOOLARGE, curp,
646: 0, 0, NULL);
1.1 schwarze 647: break;
648: }
649: resize_buf(fb, 65536);
650: }
1.115 schwarze 651: ssz = curp->gzip ?
652: gzread(gz, fb->buf + (int)off, fb->sz - off) :
653: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 654: if (ssz == 0) {
655: fb->sz = off;
1.117 schwarze 656: return 1;
1.1 schwarze 657: }
1.118 schwarze 658: if (ssz == -1)
659: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 660: off += (size_t)ssz;
661: }
662:
663: free(fb->buf);
664: fb->buf = NULL;
1.117 schwarze 665: return 0;
1.1 schwarze 666: }
667:
668: static void
669: mparse_end(struct mparse *curp)
670: {
1.110 schwarze 671: if (curp->man->macroset == MACROSET_NONE)
672: curp->man->macroset = MACROSET_MAN;
673: if (curp->man->macroset == MACROSET_MDOC)
674: mdoc_endparse(curp->man);
675: else
1.86 schwarze 676: man_endparse(curp->man);
1.1 schwarze 677: roff_endparse(curp->roff);
678: }
679:
1.15 schwarze 680: static void
681: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 682: {
1.61 schwarze 683: struct buf *svprimary;
1.1 schwarze 684: const char *svfile;
1.71 schwarze 685: size_t offset;
1.14 schwarze 686: static int recursion_depth;
687:
688: if (64 < recursion_depth) {
689: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 690: return;
1.14 schwarze 691: }
1.1 schwarze 692:
1.15 schwarze 693: /* Line number is per-file. */
694: svfile = curp->file;
695: curp->file = file;
1.61 schwarze 696: svprimary = curp->primary;
1.59 schwarze 697: curp->primary = &blk;
1.15 schwarze 698: curp->line = 1;
699: recursion_depth++;
700:
1.70 schwarze 701: /* Skip an UTF-8 byte order mark. */
702: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
703: (unsigned char)blk.buf[0] == 0xef &&
704: (unsigned char)blk.buf[1] == 0xbb &&
705: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 706: offset = 3;
1.70 schwarze 707: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 708: } else
709: offset = 0;
1.70 schwarze 710:
1.71 schwarze 711: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 712:
1.86 schwarze 713: if (--recursion_depth == 0)
1.15 schwarze 714: mparse_end(curp);
715:
1.61 schwarze 716: curp->primary = svprimary;
1.15 schwarze 717: curp->file = svfile;
718: }
719:
1.74 schwarze 720: /*
721: * Read the whole file into memory and call the parsers.
722: * Called recursively when an .so request is encountered.
723: */
1.15 schwarze 724: enum mandoclevel
725: mparse_readfd(struct mparse *curp, int fd, const char *file)
726: {
727: struct buf blk;
728: int with_mmap;
1.70 schwarze 729: int save_filenc;
1.1 schwarze 730:
1.67 schwarze 731: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 732: save_filenc = curp->filenc;
733: curp->filenc = curp->options &
734: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 735: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 736: curp->filenc = save_filenc;
1.67 schwarze 737: if (with_mmap)
738: munmap(blk.buf, blk.sz);
739: else
740: free(blk.buf);
741: }
1.117 schwarze 742: return curp->file_status;
1.58 schwarze 743: }
744:
1.122 schwarze 745: int
746: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 747: {
748: char *cp;
1.122 schwarze 749: int fd;
1.58 schwarze 750:
751: curp->file = file;
1.115 schwarze 752: cp = strrchr(file, '.');
753: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 754:
1.115 schwarze 755: /* First try to use the filename as it is. */
1.74 schwarze 756:
1.122 schwarze 757: if ((fd = open(file, O_RDONLY)) != -1)
758: return fd;
1.74 schwarze 759:
1.115 schwarze 760: /*
761: * If that doesn't work and the filename doesn't
762: * already end in .gz, try appending .gz.
763: */
1.74 schwarze 764:
1.115 schwarze 765: if ( ! curp->gzip) {
1.74 schwarze 766: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 767: fd = open(cp, O_RDONLY);
1.83 schwarze 768: free(cp);
1.122 schwarze 769: if (fd != -1) {
1.115 schwarze 770: curp->gzip = 1;
1.122 schwarze 771: return fd;
1.58 schwarze 772: }
773: }
774:
1.115 schwarze 775: /* Neither worked, give up. */
1.73 schwarze 776:
1.115 schwarze 777: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.122 schwarze 778: return -1;
1.1 schwarze 779: }
780:
781: struct mparse *
1.69 schwarze 782: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.119 schwarze 783: const char *defos)
1.1 schwarze 784: {
785: struct mparse *curp;
786:
787: curp = mandoc_calloc(1, sizeof(struct mparse));
788:
1.22 schwarze 789: curp->options = options;
1.1 schwarze 790: curp->wlevel = wlevel;
791: curp->mmsg = mmsg;
1.7 schwarze 792: curp->defos = defos;
1.1 schwarze 793:
1.119 schwarze 794: curp->roff = roff_alloc(curp, options);
1.112 schwarze 795: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
796: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 797: if (curp->options & MPARSE_MDOC) {
798: mdoc_hash_init();
1.112 schwarze 799: curp->man->macroset = MACROSET_MDOC;
800: } else if (curp->options & MPARSE_MAN) {
1.111 schwarze 801: man_hash_init();
1.112 schwarze 802: curp->man->macroset = MACROSET_MAN;
1.111 schwarze 803: }
1.113 schwarze 804: curp->man->first->tok = TOKEN_NONE;
1.117 schwarze 805: return curp;
1.1 schwarze 806: }
807:
808: void
809: mparse_reset(struct mparse *curp)
810: {
811: roff_reset(curp->roff);
1.124 schwarze 812: roff_man_reset(curp->man);
1.4 schwarze 813: if (curp->secondary)
814: curp->secondary->sz = 0;
1.1 schwarze 815:
816: curp->file_status = MANDOCLEVEL_OK;
1.23 schwarze 817:
818: free(curp->sodest);
819: curp->sodest = NULL;
1.1 schwarze 820: }
821:
822: void
823: mparse_free(struct mparse *curp)
824: {
825:
1.112 schwarze 826: roff_man_free(curp->man);
1.1 schwarze 827: if (curp->roff)
828: roff_free(curp->roff);
1.4 schwarze 829: if (curp->secondary)
830: free(curp->secondary->buf);
1.1 schwarze 831:
1.4 schwarze 832: free(curp->secondary);
1.23 schwarze 833: free(curp->sodest);
1.1 schwarze 834: free(curp);
835: }
836:
837: void
1.110 schwarze 838: mparse_result(struct mparse *curp, struct roff_man **man,
839: char **sodest)
1.1 schwarze 840: {
841:
1.23 schwarze 842: if (sodest && NULL != (*sodest = curp->sodest)) {
843: *man = NULL;
844: return;
845: }
1.1 schwarze 846: if (man)
847: *man = curp->man;
848: }
849:
850: void
851: mandoc_vmsg(enum mandocerr t, struct mparse *m,
852: int ln, int pos, const char *fmt, ...)
853: {
854: char buf[256];
855: va_list ap;
856:
857: va_start(ap, fmt);
1.26 schwarze 858: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 859: va_end(ap);
860:
861: mandoc_msg(t, m, ln, pos, buf);
862: }
863:
864: void
1.25 schwarze 865: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 866: int ln, int col, const char *msg)
867: {
868: enum mandoclevel level;
869:
1.87 schwarze 870: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 871: while (er < mandoclimits[level])
872: level--;
873:
1.83 schwarze 874: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 875: return;
876:
877: if (m->mmsg)
878: (*m->mmsg)(er, level, m->file, ln, col, msg);
879:
880: if (m->file_status < level)
881: m->file_status = level;
882: }
883:
884: const char *
885: mparse_strerror(enum mandocerr er)
886: {
887:
1.117 schwarze 888: return mandocerrs[er];
1.1 schwarze 889: }
890:
891: const char *
892: mparse_strlevel(enum mandoclevel lvl)
893: {
1.117 schwarze 894: return mandoclevels[lvl];
1.4 schwarze 895: }
896:
897: void
898: mparse_keep(struct mparse *p)
899: {
900:
901: assert(NULL == p->secondary);
902: p->secondary = mandoc_calloc(1, sizeof(struct buf));
903: }
904:
905: const char *
906: mparse_getkeep(const struct mparse *p)
907: {
908:
909: assert(p->secondary);
1.117 schwarze 910: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 schwarze 911: }
![[BACK]](/icons/back.gif){kind=icon}
Return to read.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc