Annotation of src/usr.bin/mandoc/read.c, Revision 1.136
1.136 ! schwarze 1: /* $OpenBSD: read.c,v 1.135 2017/03/07 20:00:02 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,
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 \"\"",
1.134 schwarze 112: "description line outside NAME section",
1.1 schwarze 113: "sections out of conventional order",
1.32 schwarze 114: "duplicate section title",
115: "unexpected section",
1.63 schwarze 116: "unusual Xr order",
117: "unusual Xr punctuation",
1.62 schwarze 118: "AUTHORS section without An macro",
1.1 schwarze 119:
120: /* related to macros and nesting */
1.33 schwarze 121: "obsolete macro",
1.78 schwarze 122: "macro neither callable nor escaped",
1.1 schwarze 123: "skipping paragraph macro",
1.10 schwarze 124: "moving paragraph macro out of list",
1.1 schwarze 125: "skipping no-space macro",
126: "blocks badly nested",
127: "nested displays are not portable",
1.35 schwarze 128: "moving content out of list",
1.56 schwarze 129: "fill mode already enabled, skipping",
130: "fill mode already disabled, skipping",
1.1 schwarze 131: "line scope broken",
132:
133: /* related to missing macro arguments */
1.36 schwarze 134: "skipping empty request",
135: "conditional request controls empty scope",
1.1 schwarze 136: "skipping empty macro",
1.99 schwarze 137: "empty block",
1.40 schwarze 138: "empty argument, using 0n",
1.38 schwarze 139: "missing display type, using -ragged",
140: "list type is not the first argument",
1.125 schwarze 141: "missing -width in -tag list, using 6n",
1.56 schwarze 142: "missing utility name, using \"\"",
1.98 schwarze 143: "missing function name, using \"\"",
1.38 schwarze 144: "empty head in list item",
145: "empty list item",
1.39 schwarze 146: "missing font type, using \\fR",
147: "unknown font type, using \\fR",
1.79 schwarze 148: "nothing follows prefix",
1.97 schwarze 149: "empty reference block",
1.128 schwarze 150: "missing section argument",
1.38 schwarze 151: "missing -std argument, adding it",
1.100 schwarze 152: "missing option string, using \"\"",
153: "missing resource identifier, using \"\"",
1.66 schwarze 154: "missing eqn box, using \"\"",
1.1 schwarze 155:
156: /* related to bad macro arguments */
1.42 schwarze 157: "unterminated quoted argument",
1.1 schwarze 158: "duplicate argument",
1.54 schwarze 159: "skipping duplicate argument",
1.41 schwarze 160: "skipping duplicate display type",
161: "skipping duplicate list type",
1.54 schwarze 162: "skipping -width argument",
1.101 schwarze 163: "wrong number of cells",
1.1 schwarze 164: "unknown AT&T UNIX version",
1.64 schwarze 165: "comma in function argument",
1.65 schwarze 166: "parenthesis in function name",
1.45 schwarze 167: "invalid content in Rs block",
1.41 schwarze 168: "invalid Boolean argument",
169: "unknown font, skipping request",
1.101 schwarze 170: "odd number of characters in request",
1.1 schwarze 171:
172: /* related to plain text */
1.42 schwarze 173: "blank line in fill mode, using .sp",
174: "tab in filled text",
175: "whitespace at end of input line",
1.131 schwarze 176: "new sentence, new line",
1.1 schwarze 177: "bad comment style",
1.42 schwarze 178: "invalid escape sequence",
179: "undefined string, using \"\"",
1.3 schwarze 180:
1.93 schwarze 181: /* related to tables */
182: "tbl line starts with span",
183: "tbl column starts with span",
184: "skipping vertical bar in tbl layout",
185:
1.1 schwarze 186: "generic error",
1.3 schwarze 187:
1.1 schwarze 188: /* related to tables */
1.91 schwarze 189: "non-alphabetic character in tbl options",
190: "skipping unknown tbl option",
191: "missing tbl option argument",
192: "wrong tbl option argument size",
1.93 schwarze 193: "empty tbl layout",
194: "invalid character in tbl layout",
195: "unmatched parenthesis in tbl layout",
1.94 schwarze 196: "tbl without any data cells",
197: "ignoring data in spanned tbl cell",
198: "ignoring extra tbl data cells",
199: "data block open at end of tbl",
1.1 schwarze 200:
1.46 schwarze 201: /* related to document structure and macros */
1.83 schwarze 202: NULL,
1.1 schwarze 203: "input stack limit exceeded, infinite loop?",
204: "skipping bad character",
1.46 schwarze 205: "skipping unknown macro",
1.87 schwarze 206: "skipping insecure request",
1.51 schwarze 207: "skipping item outside list",
1.46 schwarze 208: "skipping column outside column list",
209: "skipping end of block that is not open",
1.90 schwarze 210: "fewer RS blocks open, skipping",
1.46 schwarze 211: "inserting missing end of block",
212: "appending missing end of block",
213:
214: /* related to request and macro arguments */
1.1 schwarze 215: "escaped character not allowed in a name",
1.75 schwarze 216: "NOT IMPLEMENTED: Bd -file",
1.120 schwarze 217: "skipping display without arguments",
1.49 schwarze 218: "missing list type, using -item",
1.48 schwarze 219: "missing manual name, using \"\"",
1.49 schwarze 220: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 221: "unknown standard specifier",
1.49 schwarze 222: "skipping request without numeric argument",
1.84 schwarze 223: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 224: ".so request failed",
1.39 schwarze 225: "skipping all arguments",
226: "skipping excess arguments",
1.68 schwarze 227: "divide by zero",
1.87 schwarze 228:
229: "unsupported feature",
230: "input too large",
1.89 schwarze 231: "unsupported control character",
1.87 schwarze 232: "unsupported roff request",
1.94 schwarze 233: "eqn delim option in tbl",
1.93 schwarze 234: "unsupported tbl layout modifier",
1.87 schwarze 235: "ignoring macro in table",
1.1 schwarze 236: };
237:
238: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
239: "SUCCESS",
240: "RESERVED",
241: "WARNING",
242: "ERROR",
1.87 schwarze 243: "UNSUPP",
1.1 schwarze 244: "BADARG",
245: "SYSERR"
246: };
247:
1.25 schwarze 248:
1.1 schwarze 249: static void
250: resize_buf(struct buf *buf, size_t initial)
251: {
252:
253: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
254: buf->buf = mandoc_realloc(buf->buf, buf->sz);
255: }
256:
257: static void
1.60 schwarze 258: choose_parser(struct mparse *curp)
1.1 schwarze 259: {
1.59 schwarze 260: char *cp, *ep;
261: int format;
1.1 schwarze 262:
1.59 schwarze 263: /*
264: * If neither command line arguments -mdoc or -man select
265: * a parser nor the roff parser found a .Dd or .TH macro
266: * yet, look ahead in the main input buffer.
267: */
268:
269: if ((format = roff_getformat(curp->roff)) == 0) {
270: cp = curp->primary->buf;
271: ep = cp + curp->primary->sz;
272: while (cp < ep) {
1.61 schwarze 273: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 274: cp++;
275: if (cp[0] == 'D' && cp[1] == 'd') {
276: format = MPARSE_MDOC;
277: break;
278: }
279: if (cp[0] == 'T' && cp[1] == 'H') {
280: format = MPARSE_MAN;
281: break;
282: }
283: }
284: cp = memchr(cp, '\n', ep - cp);
285: if (cp == NULL)
286: break;
287: cp++;
288: }
1.1 schwarze 289: }
290:
1.59 schwarze 291: if (format == MPARSE_MDOC) {
1.112 schwarze 292: curp->man->macroset = MACROSET_MDOC;
1.136 ! schwarze 293: if (curp->man->mdocmac == NULL)
! 294: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 295: } else {
296: curp->man->macroset = MACROSET_MAN;
1.136 ! schwarze 297: if (curp->man->manmac == NULL)
! 298: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 299: }
1.136 ! schwarze 300: curp->man->first->tok = TOKEN_NONE;
1.1 schwarze 301: }
302:
303: /*
1.71 schwarze 304: * Main parse routine for a buffer.
305: * It assumes encoding and line numbering are already set up.
306: * It can recurse directly (for invocations of user-defined
307: * macros, inline equations, and input line traps)
308: * and indirectly (for .so file inclusion).
1.1 schwarze 309: */
1.135 schwarze 310: static int
1.71 schwarze 311: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 312: {
313: const struct tbl_span *span;
314: struct buf ln;
1.88 schwarze 315: const char *save_file;
1.85 schwarze 316: char *cp;
1.71 schwarze 317: size_t pos; /* byte number in the ln buffer */
1.127 schwarze 318: size_t j; /* auxiliary byte number in the blk buffer */
1.1 schwarze 319: enum rofferr rr;
1.76 schwarze 320: int of;
1.1 schwarze 321: int lnn; /* line number in the real file */
1.85 schwarze 322: int fd;
1.1 schwarze 323: unsigned char c;
324:
1.71 schwarze 325: memset(&ln, 0, sizeof(ln));
1.1 schwarze 326:
1.25 schwarze 327: lnn = curp->line;
328: pos = 0;
1.1 schwarze 329:
1.71 schwarze 330: while (i < blk.sz) {
1.1 schwarze 331: if (0 == pos && '\0' == blk.buf[i])
332: break;
333:
334: if (start) {
335: curp->line = lnn;
336: curp->reparse_count = 0;
1.70 schwarze 337:
338: if (lnn < 3 &&
339: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 340: curp->filenc & MPARSE_LATIN1)
341: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 342: }
343:
1.71 schwarze 344: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 345:
346: /*
347: * When finding an unescaped newline character,
348: * leave the character loop to process the line.
349: * Skip a preceding carriage return, if any.
350: */
351:
1.71 schwarze 352: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 353: '\n' == blk.buf[i + 1])
354: ++i;
355: if ('\n' == blk.buf[i]) {
356: ++i;
357: ++lnn;
358: break;
359: }
360:
1.13 schwarze 361: /*
1.70 schwarze 362: * Make sure we have space for the worst
363: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 364: */
365:
1.71 schwarze 366: if (pos + 11 > ln.sz)
1.13 schwarze 367: resize_buf(&ln, 256);
368:
1.25 schwarze 369: /*
1.70 schwarze 370: * Encode 8-bit input.
1.1 schwarze 371: */
372:
1.70 schwarze 373: c = blk.buf[i];
374: if (c & 0x80) {
1.71 schwarze 375: if ( ! (curp->filenc && preconv_encode(
376: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 377: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
378: curp->line, pos, "0x%x", c);
1.70 schwarze 379: ln.buf[pos++] = '?';
380: i++;
381: }
382: continue;
383: }
384:
385: /*
386: * Exclude control characters.
387: */
1.1 schwarze 388:
1.70 schwarze 389: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 390: mandoc_vmsg(c == 0x00 || c == 0x04 ||
391: c > 0x0a ? MANDOCERR_CHAR_BAD :
392: MANDOCERR_CHAR_UNSUPP,
393: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 394: i++;
1.102 schwarze 395: if (c != '\r')
396: ln.buf[pos++] = '?';
1.1 schwarze 397: continue;
398: }
399:
400: /* Trailing backslash = a plain char. */
401:
1.71 schwarze 402: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 schwarze 403: ln.buf[pos++] = blk.buf[i++];
404: continue;
405: }
406:
407: /*
408: * Found escape and at least one other character.
409: * When it's a newline character, skip it.
410: * When there is a carriage return in between,
411: * skip that one as well.
412: */
413:
1.71 schwarze 414: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 schwarze 415: '\n' == blk.buf[i + 2])
416: ++i;
417: if ('\n' == blk.buf[i + 1]) {
418: i += 2;
419: ++lnn;
420: continue;
421: }
422:
423: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.127 schwarze 424: j = i;
1.1 schwarze 425: i += 2;
426: /* Comment, skip to end of line */
1.71 schwarze 427: for (; i < blk.sz; ++i) {
1.126 schwarze 428: if (blk.buf[i] != '\n')
429: continue;
430: if (blk.buf[i - 1] == ' ' ||
431: blk.buf[i - 1] == '\t')
432: mandoc_msg(
433: MANDOCERR_SPACE_EOL,
434: curp, curp->line,
1.127 schwarze 435: pos + i-1 - j, NULL);
1.126 schwarze 436: ++i;
437: ++lnn;
438: break;
1.1 schwarze 439: }
440:
441: /* Backout trailing whitespaces */
442: for (; pos > 0; --pos) {
443: if (ln.buf[pos - 1] != ' ')
444: break;
445: if (pos > 2 && ln.buf[pos - 2] == '\\')
446: break;
447: }
448: break;
449: }
450:
1.13 schwarze 451: /* Catch escaped bogus characters. */
452:
453: c = (unsigned char) blk.buf[i+1];
454:
1.25 schwarze 455: if ( ! (isascii(c) &&
456: (isgraph(c) || isblank(c)))) {
1.89 schwarze 457: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.56 schwarze 458: curp->line, pos, "0x%x", c);
1.13 schwarze 459: i += 2;
460: ln.buf[pos++] = '?';
461: continue;
462: }
463:
1.1 schwarze 464: /* Some other escape sequence, copy & cont. */
465:
466: ln.buf[pos++] = blk.buf[i++];
467: ln.buf[pos++] = blk.buf[i++];
468: }
469:
1.71 schwarze 470: if (pos >= ln.sz)
1.1 schwarze 471: resize_buf(&ln, 256);
472:
473: ln.buf[pos] = '\0';
474:
475: /*
476: * A significant amount of complexity is contained by
477: * the roff preprocessor. It's line-oriented but can be
478: * expressed on one line, so we need at times to
479: * readjust our starting point and re-run it. The roff
480: * preprocessor can also readjust the buffers with new
481: * data, so we pass them in wholesale.
482: */
483:
484: of = 0;
485:
1.4 schwarze 486: /*
487: * Maintain a lookaside buffer of all parsed lines. We
488: * only do this if mparse_keep() has been invoked (the
489: * buffer may be accessed with mparse_getkeep()).
490: */
491:
492: if (curp->secondary) {
1.25 schwarze 493: curp->secondary->buf = mandoc_realloc(
494: curp->secondary->buf,
495: curp->secondary->sz + pos + 2);
496: memcpy(curp->secondary->buf +
497: curp->secondary->sz,
498: ln.buf, pos);
1.4 schwarze 499: curp->secondary->sz += pos;
500: curp->secondary->buf
501: [curp->secondary->sz] = '\n';
502: curp->secondary->sz++;
503: curp->secondary->buf
504: [curp->secondary->sz] = '\0';
505: }
1.1 schwarze 506: rerun:
1.72 schwarze 507: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 508:
509: switch (rr) {
1.25 schwarze 510: case ROFF_REPARSE:
1.135 schwarze 511: if (++curp->reparse_count > REPARSE_LIMIT)
1.1 schwarze 512: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 513: curp->line, pos, NULL);
1.135 schwarze 514: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
515: start == 1) {
516: pos = 0;
517: continue;
518: }
519: free(ln.buf);
520: return 0;
1.25 schwarze 521: case ROFF_APPEND:
1.71 schwarze 522: pos = strlen(ln.buf);
1.1 schwarze 523: continue;
1.25 schwarze 524: case ROFF_RERUN:
1.1 schwarze 525: goto rerun;
1.25 schwarze 526: case ROFF_IGN:
1.1 schwarze 527: pos = 0;
528: continue;
1.25 schwarze 529: case ROFF_SO:
1.71 schwarze 530: if ( ! (curp->options & MPARSE_SO) &&
531: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 532: curp->sodest = mandoc_strdup(ln.buf + of);
533: free(ln.buf);
1.135 schwarze 534: return 1;
1.23 schwarze 535: }
1.4 schwarze 536: /*
537: * We remove `so' clauses from our lookaside
538: * buffer because we're going to descend into
539: * the file recursively.
540: */
1.25 schwarze 541: if (curp->secondary)
1.4 schwarze 542: curp->secondary->sz -= pos + 1;
1.88 schwarze 543: save_file = curp->file;
1.122 schwarze 544: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 545: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 546: close(fd);
1.88 schwarze 547: curp->file = save_file;
548: } else {
549: curp->file = save_file;
1.30 schwarze 550: mandoc_vmsg(MANDOCERR_SO_FAIL,
551: curp, curp->line, pos,
552: ".so %s", ln.buf + of);
1.85 schwarze 553: ln.sz = mandoc_asprintf(&cp,
554: ".sp\nSee the file %s.\n.sp",
555: ln.buf + of);
556: free(ln.buf);
557: ln.buf = cp;
558: of = 0;
559: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 560: }
1.1 schwarze 561: pos = 0;
562: continue;
563: default:
564: break;
565: }
566:
1.124 schwarze 567: if (curp->man->macroset == MACROSET_NONE)
1.60 schwarze 568: choose_parser(curp);
1.1 schwarze 569:
1.25 schwarze 570: /*
1.60 schwarze 571: * Lastly, push down into the parsers themselves.
1.1 schwarze 572: * If libroff returns ROFF_TBL, then add it to the
573: * currently open parse. Since we only get here if
574: * there does exist data (see tbl_data.c), we're
575: * guaranteed that something's been allocated.
576: * Do the same for ROFF_EQN.
577: */
578:
1.114 schwarze 579: if (rr == ROFF_TBL)
1.76 schwarze 580: while ((span = roff_span(curp->roff)) != NULL)
1.114 schwarze 581: roff_addtbl(curp->man, span);
582: else if (rr == ROFF_EQN)
583: roff_addeqn(curp->man, roff_eqn(curp->roff));
584: else if ((curp->man->macroset == MACROSET_MDOC ?
1.110 schwarze 585: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 586: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
587: break;
1.1 schwarze 588:
589: /* Temporary buffers typically are not full. */
590:
591: if (0 == start && '\0' == blk.buf[i])
592: break;
593:
594: /* Start the next input line. */
595:
596: pos = 0;
597: }
598:
599: free(ln.buf);
1.135 schwarze 600: return 1;
1.1 schwarze 601: }
602:
603: static int
1.18 schwarze 604: read_whole_file(struct mparse *curp, const char *file, int fd,
605: struct buf *fb, int *with_mmap)
1.1 schwarze 606: {
607: struct stat st;
1.115 schwarze 608: gzFile gz;
1.1 schwarze 609: size_t off;
610: ssize_t ssz;
611:
1.118 schwarze 612: if (fstat(fd, &st) == -1)
613: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 614:
615: /*
616: * If we're a regular file, try just reading in the whole entry
617: * via mmap(). This is faster than reading it into blocks, and
618: * since each file is only a few bytes to begin with, I'm not
619: * concerned that this is going to tank any machines.
620: */
621:
1.115 schwarze 622: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 623: if (st.st_size > 0x7fffffff) {
1.86 schwarze 624: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.117 schwarze 625: return 0;
1.1 schwarze 626: }
627: *with_mmap = 1;
628: fb->sz = (size_t)st.st_size;
1.15 schwarze 629: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 630: if (fb->buf != MAP_FAILED)
1.117 schwarze 631: return 1;
1.1 schwarze 632: }
633:
1.115 schwarze 634: if (curp->gzip) {
1.118 schwarze 635: if ((gz = gzdopen(fd, "rb")) == NULL)
636: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.115 schwarze 637: } else
638: gz = NULL;
639:
1.1 schwarze 640: /*
641: * If this isn't a regular file (like, say, stdin), then we must
642: * go the old way and just read things in bit by bit.
643: */
644:
645: *with_mmap = 0;
646: off = 0;
647: fb->sz = 0;
648: fb->buf = NULL;
649: for (;;) {
650: if (off == fb->sz) {
651: if (fb->sz == (1U << 31)) {
1.86 schwarze 652: mandoc_msg(MANDOCERR_TOOLARGE, curp,
653: 0, 0, NULL);
1.1 schwarze 654: break;
655: }
656: resize_buf(fb, 65536);
657: }
1.115 schwarze 658: ssz = curp->gzip ?
659: gzread(gz, fb->buf + (int)off, fb->sz - off) :
660: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 661: if (ssz == 0) {
662: fb->sz = off;
1.117 schwarze 663: return 1;
1.1 schwarze 664: }
1.118 schwarze 665: if (ssz == -1)
666: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 667: off += (size_t)ssz;
668: }
669:
670: free(fb->buf);
671: fb->buf = NULL;
1.117 schwarze 672: return 0;
1.1 schwarze 673: }
674:
675: static void
676: mparse_end(struct mparse *curp)
677: {
1.110 schwarze 678: if (curp->man->macroset == MACROSET_NONE)
679: curp->man->macroset = MACROSET_MAN;
680: if (curp->man->macroset == MACROSET_MDOC)
681: mdoc_endparse(curp->man);
682: else
1.86 schwarze 683: man_endparse(curp->man);
1.1 schwarze 684: roff_endparse(curp->roff);
685: }
686:
1.15 schwarze 687: static void
688: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 689: {
1.61 schwarze 690: struct buf *svprimary;
1.1 schwarze 691: const char *svfile;
1.71 schwarze 692: size_t offset;
1.14 schwarze 693: static int recursion_depth;
694:
695: if (64 < recursion_depth) {
696: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 697: return;
1.14 schwarze 698: }
1.1 schwarze 699:
1.15 schwarze 700: /* Line number is per-file. */
701: svfile = curp->file;
702: curp->file = file;
1.61 schwarze 703: svprimary = curp->primary;
1.59 schwarze 704: curp->primary = &blk;
1.15 schwarze 705: curp->line = 1;
706: recursion_depth++;
707:
1.70 schwarze 708: /* Skip an UTF-8 byte order mark. */
709: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
710: (unsigned char)blk.buf[0] == 0xef &&
711: (unsigned char)blk.buf[1] == 0xbb &&
712: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 713: offset = 3;
1.70 schwarze 714: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 715: } else
716: offset = 0;
1.70 schwarze 717:
1.71 schwarze 718: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 719:
1.86 schwarze 720: if (--recursion_depth == 0)
1.15 schwarze 721: mparse_end(curp);
722:
1.61 schwarze 723: curp->primary = svprimary;
1.15 schwarze 724: curp->file = svfile;
725: }
726:
1.74 schwarze 727: /*
728: * Read the whole file into memory and call the parsers.
729: * Called recursively when an .so request is encountered.
730: */
1.15 schwarze 731: enum mandoclevel
732: mparse_readfd(struct mparse *curp, int fd, const char *file)
733: {
734: struct buf blk;
735: int with_mmap;
1.70 schwarze 736: int save_filenc;
1.1 schwarze 737:
1.67 schwarze 738: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 739: save_filenc = curp->filenc;
740: curp->filenc = curp->options &
741: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 742: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 743: curp->filenc = save_filenc;
1.67 schwarze 744: if (with_mmap)
745: munmap(blk.buf, blk.sz);
746: else
747: free(blk.buf);
748: }
1.117 schwarze 749: return curp->file_status;
1.58 schwarze 750: }
751:
1.122 schwarze 752: int
753: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 754: {
755: char *cp;
1.122 schwarze 756: int fd;
1.58 schwarze 757:
758: curp->file = file;
1.115 schwarze 759: cp = strrchr(file, '.');
760: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 761:
1.115 schwarze 762: /* First try to use the filename as it is. */
1.74 schwarze 763:
1.122 schwarze 764: if ((fd = open(file, O_RDONLY)) != -1)
765: return fd;
1.74 schwarze 766:
1.115 schwarze 767: /*
768: * If that doesn't work and the filename doesn't
769: * already end in .gz, try appending .gz.
770: */
1.74 schwarze 771:
1.115 schwarze 772: if ( ! curp->gzip) {
1.74 schwarze 773: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 774: fd = open(cp, O_RDONLY);
1.83 schwarze 775: free(cp);
1.122 schwarze 776: if (fd != -1) {
1.115 schwarze 777: curp->gzip = 1;
1.122 schwarze 778: return fd;
1.58 schwarze 779: }
780: }
781:
1.115 schwarze 782: /* Neither worked, give up. */
1.73 schwarze 783:
1.115 schwarze 784: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.122 schwarze 785: return -1;
1.1 schwarze 786: }
787:
788: struct mparse *
1.69 schwarze 789: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.119 schwarze 790: const char *defos)
1.1 schwarze 791: {
792: struct mparse *curp;
793:
794: curp = mandoc_calloc(1, sizeof(struct mparse));
795:
1.22 schwarze 796: curp->options = options;
1.1 schwarze 797: curp->wlevel = wlevel;
798: curp->mmsg = mmsg;
1.7 schwarze 799: curp->defos = defos;
1.1 schwarze 800:
1.119 schwarze 801: curp->roff = roff_alloc(curp, options);
1.112 schwarze 802: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
803: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 804: if (curp->options & MPARSE_MDOC) {
1.112 schwarze 805: curp->man->macroset = MACROSET_MDOC;
1.136 ! schwarze 806: if (curp->man->mdocmac == NULL)
! 807: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 808: } else if (curp->options & MPARSE_MAN) {
809: curp->man->macroset = MACROSET_MAN;
1.136 ! schwarze 810: if (curp->man->manmac == NULL)
! 811: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 812: }
1.113 schwarze 813: curp->man->first->tok = TOKEN_NONE;
1.117 schwarze 814: return curp;
1.1 schwarze 815: }
816:
817: void
818: mparse_reset(struct mparse *curp)
819: {
820: roff_reset(curp->roff);
1.124 schwarze 821: roff_man_reset(curp->man);
1.133 schwarze 822:
823: free(curp->sodest);
824: curp->sodest = NULL;
825:
1.4 schwarze 826: if (curp->secondary)
827: curp->secondary->sz = 0;
1.1 schwarze 828:
829: curp->file_status = MANDOCLEVEL_OK;
1.132 schwarze 830: curp->gzip = 0;
1.1 schwarze 831: }
832:
833: void
834: mparse_free(struct mparse *curp)
835: {
836:
1.136 ! schwarze 837: roffhash_free(curp->man->mdocmac);
! 838: roffhash_free(curp->man->manmac);
1.112 schwarze 839: roff_man_free(curp->man);
1.133 schwarze 840: roff_free(curp->roff);
1.4 schwarze 841: if (curp->secondary)
842: free(curp->secondary->buf);
1.1 schwarze 843:
1.4 schwarze 844: free(curp->secondary);
1.23 schwarze 845: free(curp->sodest);
1.1 schwarze 846: free(curp);
847: }
848:
849: void
1.110 schwarze 850: mparse_result(struct mparse *curp, struct roff_man **man,
851: char **sodest)
1.1 schwarze 852: {
853:
1.23 schwarze 854: if (sodest && NULL != (*sodest = curp->sodest)) {
855: *man = NULL;
856: return;
857: }
1.1 schwarze 858: if (man)
859: *man = curp->man;
1.130 schwarze 860: }
861:
862: void
863: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
864: {
865: if (curp->file_status > *rc)
866: *rc = curp->file_status;
1.1 schwarze 867: }
868:
869: void
870: mandoc_vmsg(enum mandocerr t, struct mparse *m,
871: int ln, int pos, const char *fmt, ...)
872: {
873: char buf[256];
874: va_list ap;
875:
876: va_start(ap, fmt);
1.26 schwarze 877: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 878: va_end(ap);
879:
880: mandoc_msg(t, m, ln, pos, buf);
881: }
882:
883: void
1.25 schwarze 884: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 885: int ln, int col, const char *msg)
886: {
887: enum mandoclevel level;
888:
1.87 schwarze 889: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 890: while (er < mandoclimits[level])
891: level--;
892:
1.83 schwarze 893: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 894: return;
895:
896: if (m->mmsg)
897: (*m->mmsg)(er, level, m->file, ln, col, msg);
898:
899: if (m->file_status < level)
900: m->file_status = level;
901: }
902:
903: const char *
904: mparse_strerror(enum mandocerr er)
905: {
906:
1.117 schwarze 907: return mandocerrs[er];
1.1 schwarze 908: }
909:
910: const char *
911: mparse_strlevel(enum mandoclevel lvl)
912: {
1.117 schwarze 913: return mandoclevels[lvl];
1.4 schwarze 914: }
915:
916: void
917: mparse_keep(struct mparse *p)
918: {
919:
920: assert(NULL == p->secondary);
921: p->secondary = mandoc_calloc(1, sizeof(struct buf));
922: }
923:
924: const char *
925: mparse_getkeep(const struct mparse *p)
926: {
927:
928: assert(p->secondary);
1.117 schwarze 929: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 schwarze 930: }
![[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