Annotation of src/usr.bin/mandoc/read.c, Revision 1.111
1.111 ! schwarze 1: /* $OpenBSD: read.c,v 1.110 2015/04/18 16:34:03 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.82 schwarze 4: * Copyright (c) 2010-2015 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>
1.58 schwarze 22: #include <sys/wait.h>
1.1 schwarze 23:
24: #include <assert.h>
25: #include <ctype.h>
1.18 schwarze 26: #include <errno.h>
1.1 schwarze 27: #include <fcntl.h>
1.107 schwarze 28: #include <signal.h>
1.1 schwarze 29: #include <stdarg.h>
30: #include <stdio.h>
31: #include <stdlib.h>
32: #include <string.h>
33: #include <unistd.h>
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"
1.1 schwarze 41:
42: #define REPARSE_LIMIT 1000
43:
44: struct mparse {
1.109 schwarze 45: struct roff_man *man; /* man parser */
1.1 schwarze 46: struct roff *roff; /* roff parser (!NULL) */
1.69 schwarze 47: const struct mchars *mchars; /* character table */
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 */
1.70 schwarze 57: int filenc; /* encoding of the current file */
1.1 schwarze 58: int reparse_count; /* finite interp. stack */
1.59 schwarze 59: int line; /* line number in the file */
1.73 schwarze 60: pid_t child; /* the gunzip(1) process */
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\"",
1.103 schwarze 106: "NAME section without name",
107: "NAME section without description",
108: "description not at the end of NAME",
109: "bad NAME section content",
1.96 schwarze 110: "missing description line, using \"\"",
1.1 schwarze 111: "sections out of conventional order",
1.32 schwarze 112: "duplicate section title",
113: "unexpected section",
1.63 schwarze 114: "unusual Xr order",
115: "unusual Xr punctuation",
1.62 schwarze 116: "AUTHORS section without An macro",
1.1 schwarze 117:
118: /* related to macros and nesting */
1.33 schwarze 119: "obsolete macro",
1.78 schwarze 120: "macro neither callable nor escaped",
1.1 schwarze 121: "skipping paragraph macro",
1.10 schwarze 122: "moving paragraph macro out of list",
1.1 schwarze 123: "skipping no-space macro",
124: "blocks badly nested",
125: "nested displays are not portable",
1.35 schwarze 126: "moving content out of list",
127: ".Vt block has child macro",
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",
140: "missing -width in -tag list, using 8n",
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",
1.38 schwarze 149: "missing -std argument, adding it",
1.100 schwarze 150: "missing option string, using \"\"",
151: "missing resource identifier, using \"\"",
1.66 schwarze 152: "missing eqn box, using \"\"",
1.1 schwarze 153:
154: /* related to bad macro arguments */
1.42 schwarze 155: "unterminated quoted argument",
1.1 schwarze 156: "duplicate argument",
1.54 schwarze 157: "skipping duplicate argument",
1.41 schwarze 158: "skipping duplicate display type",
159: "skipping duplicate list type",
1.54 schwarze 160: "skipping -width argument",
1.101 schwarze 161: "wrong number of cells",
1.1 schwarze 162: "unknown AT&T UNIX version",
1.64 schwarze 163: "comma in function argument",
1.65 schwarze 164: "parenthesis in function name",
1.45 schwarze 165: "invalid content in Rs block",
1.41 schwarze 166: "invalid Boolean argument",
167: "unknown font, skipping request",
1.101 schwarze 168: "odd number of characters in request",
1.1 schwarze 169:
170: /* related to plain text */
1.42 schwarze 171: "blank line in fill mode, using .sp",
172: "tab in filled text",
173: "whitespace at end of input line",
1.1 schwarze 174: "bad comment style",
1.42 schwarze 175: "invalid escape sequence",
176: "undefined string, using \"\"",
1.3 schwarze 177:
1.93 schwarze 178: /* related to tables */
179: "tbl line starts with span",
180: "tbl column starts with span",
181: "skipping vertical bar in tbl layout",
182:
1.1 schwarze 183: "generic error",
1.3 schwarze 184:
1.1 schwarze 185: /* related to tables */
1.91 schwarze 186: "non-alphabetic character in tbl options",
187: "skipping unknown tbl option",
188: "missing tbl option argument",
189: "wrong tbl option argument size",
1.93 schwarze 190: "empty tbl layout",
191: "invalid character in tbl layout",
192: "unmatched parenthesis in tbl layout",
1.94 schwarze 193: "tbl without any data cells",
194: "ignoring data in spanned tbl cell",
195: "ignoring extra tbl data cells",
196: "data block open at end of tbl",
1.1 schwarze 197:
1.46 schwarze 198: /* related to document structure and macros */
1.83 schwarze 199: NULL,
1.1 schwarze 200: "input stack limit exceeded, infinite loop?",
201: "skipping bad character",
1.46 schwarze 202: "skipping unknown macro",
1.87 schwarze 203: "skipping insecure request",
1.51 schwarze 204: "skipping item outside list",
1.46 schwarze 205: "skipping column outside column list",
206: "skipping end of block that is not open",
1.90 schwarze 207: "fewer RS blocks open, skipping",
1.46 schwarze 208: "inserting missing end of block",
209: "appending missing end of block",
210:
211: /* related to request and macro arguments */
1.1 schwarze 212: "escaped character not allowed in a name",
1.75 schwarze 213: "NOT IMPLEMENTED: Bd -file",
1.49 schwarze 214: "missing list type, using -item",
1.48 schwarze 215: "missing manual name, using \"\"",
1.49 schwarze 216: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 217: "unknown standard specifier",
1.49 schwarze 218: "skipping request without numeric argument",
1.84 schwarze 219: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 220: ".so request failed",
1.39 schwarze 221: "skipping all arguments",
222: "skipping excess arguments",
1.68 schwarze 223: "divide by zero",
1.87 schwarze 224:
225: "unsupported feature",
226: "input too large",
1.89 schwarze 227: "unsupported control character",
1.87 schwarze 228: "unsupported roff request",
1.94 schwarze 229: "eqn delim option in tbl",
1.93 schwarze 230: "unsupported tbl layout modifier",
1.87 schwarze 231: "ignoring macro in table",
1.1 schwarze 232: };
233:
234: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
235: "SUCCESS",
236: "RESERVED",
237: "WARNING",
238: "ERROR",
1.87 schwarze 239: "UNSUPP",
1.1 schwarze 240: "BADARG",
241: "SYSERR"
242: };
243:
1.25 schwarze 244:
1.1 schwarze 245: static void
246: resize_buf(struct buf *buf, size_t initial)
247: {
248:
249: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
250: buf->buf = mandoc_realloc(buf->buf, buf->sz);
251: }
252:
253: static void
1.60 schwarze 254: choose_parser(struct mparse *curp)
1.1 schwarze 255: {
1.59 schwarze 256: char *cp, *ep;
257: int format;
1.1 schwarze 258:
1.59 schwarze 259: /*
260: * If neither command line arguments -mdoc or -man select
261: * a parser nor the roff parser found a .Dd or .TH macro
262: * yet, look ahead in the main input buffer.
263: */
264:
265: if ((format = roff_getformat(curp->roff)) == 0) {
266: cp = curp->primary->buf;
267: ep = cp + curp->primary->sz;
268: while (cp < ep) {
1.61 schwarze 269: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 270: cp++;
271: if (cp[0] == 'D' && cp[1] == 'd') {
272: format = MPARSE_MDOC;
273: break;
274: }
275: if (cp[0] == 'T' && cp[1] == 'H') {
276: format = MPARSE_MAN;
277: break;
278: }
279: }
280: cp = memchr(cp, '\n', ep - cp);
281: if (cp == NULL)
282: break;
283: cp++;
284: }
1.1 schwarze 285: }
286:
1.59 schwarze 287: if (format == MPARSE_MDOC) {
1.110 schwarze 288: if (curp->man == NULL)
289: curp->man = mdoc_alloc(
1.22 schwarze 290: curp->roff, curp, curp->defos,
291: MPARSE_QUICK & curp->options ? 1 : 0);
1.110 schwarze 292: else
293: curp->man->macroset = MACROSET_MDOC;
1.111 ! schwarze 294: mdoc_hash_init();
1.1 schwarze 295: return;
1.25 schwarze 296: }
1.1 schwarze 297:
1.59 schwarze 298: /* Fall back to man(7) as a last resort. */
299:
1.110 schwarze 300: if (curp->man == NULL)
301: curp->man = man_alloc(
1.81 schwarze 302: curp->roff, curp, curp->defos,
1.22 schwarze 303: MPARSE_QUICK & curp->options ? 1 : 0);
1.110 schwarze 304: else
305: curp->man->macroset = MACROSET_MAN;
1.111 ! schwarze 306: man_hash_init();
1.1 schwarze 307: }
308:
309: /*
1.71 schwarze 310: * Main parse routine for a buffer.
311: * It assumes encoding and line numbering are already set up.
312: * It can recurse directly (for invocations of user-defined
313: * macros, inline equations, and input line traps)
314: * and indirectly (for .so file inclusion).
1.1 schwarze 315: */
316: static void
1.71 schwarze 317: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 318: {
319: const struct tbl_span *span;
320: struct buf ln;
1.88 schwarze 321: const char *save_file;
1.85 schwarze 322: char *cp;
1.71 schwarze 323: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 324: enum rofferr rr;
1.76 schwarze 325: int of;
1.1 schwarze 326: int lnn; /* line number in the real file */
1.85 schwarze 327: int fd;
328: pid_t save_child;
1.1 schwarze 329: unsigned char c;
330:
1.71 schwarze 331: memset(&ln, 0, sizeof(ln));
1.1 schwarze 332:
1.25 schwarze 333: lnn = curp->line;
334: pos = 0;
1.1 schwarze 335:
1.71 schwarze 336: while (i < blk.sz) {
1.1 schwarze 337: if (0 == pos && '\0' == blk.buf[i])
338: break;
339:
340: if (start) {
341: curp->line = lnn;
342: curp->reparse_count = 0;
1.70 schwarze 343:
344: if (lnn < 3 &&
345: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 346: curp->filenc & MPARSE_LATIN1)
347: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 348: }
349:
1.71 schwarze 350: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 351:
352: /*
353: * When finding an unescaped newline character,
354: * leave the character loop to process the line.
355: * Skip a preceding carriage return, if any.
356: */
357:
1.71 schwarze 358: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 359: '\n' == blk.buf[i + 1])
360: ++i;
361: if ('\n' == blk.buf[i]) {
362: ++i;
363: ++lnn;
364: break;
365: }
366:
1.13 schwarze 367: /*
1.70 schwarze 368: * Make sure we have space for the worst
369: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 370: */
371:
1.71 schwarze 372: if (pos + 11 > ln.sz)
1.13 schwarze 373: resize_buf(&ln, 256);
374:
1.25 schwarze 375: /*
1.70 schwarze 376: * Encode 8-bit input.
1.1 schwarze 377: */
378:
1.70 schwarze 379: c = blk.buf[i];
380: if (c & 0x80) {
1.71 schwarze 381: if ( ! (curp->filenc && preconv_encode(
382: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 383: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
384: curp->line, pos, "0x%x", c);
1.70 schwarze 385: ln.buf[pos++] = '?';
386: i++;
387: }
388: continue;
389: }
390:
391: /*
392: * Exclude control characters.
393: */
1.1 schwarze 394:
1.70 schwarze 395: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 396: mandoc_vmsg(c == 0x00 || c == 0x04 ||
397: c > 0x0a ? MANDOCERR_CHAR_BAD :
398: MANDOCERR_CHAR_UNSUPP,
399: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 400: i++;
1.102 schwarze 401: if (c != '\r')
402: ln.buf[pos++] = '?';
1.1 schwarze 403: continue;
404: }
405:
406: /* Trailing backslash = a plain char. */
407:
1.71 schwarze 408: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 schwarze 409: ln.buf[pos++] = blk.buf[i++];
410: continue;
411: }
412:
413: /*
414: * Found escape and at least one other character.
415: * When it's a newline character, skip it.
416: * When there is a carriage return in between,
417: * skip that one as well.
418: */
419:
1.71 schwarze 420: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 schwarze 421: '\n' == blk.buf[i + 2])
422: ++i;
423: if ('\n' == blk.buf[i + 1]) {
424: i += 2;
425: ++lnn;
426: continue;
427: }
428:
429: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
430: i += 2;
431: /* Comment, skip to end of line */
1.71 schwarze 432: for (; i < blk.sz; ++i) {
1.1 schwarze 433: if ('\n' == blk.buf[i]) {
434: ++i;
435: ++lnn;
436: break;
437: }
438: }
439:
440: /* Backout trailing whitespaces */
441: for (; pos > 0; --pos) {
442: if (ln.buf[pos - 1] != ' ')
443: break;
444: if (pos > 2 && ln.buf[pos - 2] == '\\')
445: break;
446: }
447: break;
448: }
449:
1.13 schwarze 450: /* Catch escaped bogus characters. */
451:
452: c = (unsigned char) blk.buf[i+1];
453:
1.25 schwarze 454: if ( ! (isascii(c) &&
455: (isgraph(c) || isblank(c)))) {
1.89 schwarze 456: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.56 schwarze 457: curp->line, pos, "0x%x", c);
1.13 schwarze 458: i += 2;
459: ln.buf[pos++] = '?';
460: continue;
461: }
462:
1.1 schwarze 463: /* Some other escape sequence, copy & cont. */
464:
465: ln.buf[pos++] = blk.buf[i++];
466: ln.buf[pos++] = blk.buf[i++];
467: }
468:
1.71 schwarze 469: if (pos >= ln.sz)
1.1 schwarze 470: resize_buf(&ln, 256);
471:
472: ln.buf[pos] = '\0';
473:
474: /*
475: * A significant amount of complexity is contained by
476: * the roff preprocessor. It's line-oriented but can be
477: * expressed on one line, so we need at times to
478: * readjust our starting point and re-run it. The roff
479: * preprocessor can also readjust the buffers with new
480: * data, so we pass them in wholesale.
481: */
482:
483: of = 0;
484:
1.4 schwarze 485: /*
486: * Maintain a lookaside buffer of all parsed lines. We
487: * only do this if mparse_keep() has been invoked (the
488: * buffer may be accessed with mparse_getkeep()).
489: */
490:
491: if (curp->secondary) {
1.25 schwarze 492: curp->secondary->buf = mandoc_realloc(
493: curp->secondary->buf,
494: curp->secondary->sz + pos + 2);
495: memcpy(curp->secondary->buf +
496: curp->secondary->sz,
497: ln.buf, pos);
1.4 schwarze 498: curp->secondary->sz += pos;
499: curp->secondary->buf
500: [curp->secondary->sz] = '\n';
501: curp->secondary->sz++;
502: curp->secondary->buf
503: [curp->secondary->sz] = '\0';
504: }
1.1 schwarze 505: rerun:
1.72 schwarze 506: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 507:
508: switch (rr) {
1.25 schwarze 509: case ROFF_REPARSE:
1.1 schwarze 510: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.71 schwarze 511: mparse_buf_r(curp, ln, of, 0);
1.1 schwarze 512: else
513: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 514: curp->line, pos, NULL);
1.1 schwarze 515: pos = 0;
516: continue;
1.25 schwarze 517: case ROFF_APPEND:
1.71 schwarze 518: pos = strlen(ln.buf);
1.1 schwarze 519: continue;
1.25 schwarze 520: case ROFF_RERUN:
1.1 schwarze 521: goto rerun;
1.25 schwarze 522: case ROFF_IGN:
1.1 schwarze 523: pos = 0;
524: continue;
1.25 schwarze 525: case ROFF_SO:
1.71 schwarze 526: if ( ! (curp->options & MPARSE_SO) &&
527: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 528: curp->sodest = mandoc_strdup(ln.buf + of);
529: free(ln.buf);
530: return;
531: }
1.4 schwarze 532: /*
533: * We remove `so' clauses from our lookaside
534: * buffer because we're going to descend into
535: * the file recursively.
536: */
1.25 schwarze 537: if (curp->secondary)
1.4 schwarze 538: curp->secondary->sz -= pos + 1;
1.88 schwarze 539: save_file = curp->file;
1.85 schwarze 540: save_child = curp->child;
541: if (mparse_open(curp, &fd, ln.buf + of) ==
1.88 schwarze 542: MANDOCLEVEL_OK) {
1.85 schwarze 543: mparse_readfd(curp, fd, ln.buf + of);
1.88 schwarze 544: curp->file = save_file;
545: } else {
546: curp->file = save_file;
1.30 schwarze 547: mandoc_vmsg(MANDOCERR_SO_FAIL,
548: curp, curp->line, pos,
549: ".so %s", ln.buf + of);
1.85 schwarze 550: ln.sz = mandoc_asprintf(&cp,
551: ".sp\nSee the file %s.\n.sp",
552: ln.buf + of);
553: free(ln.buf);
554: ln.buf = cp;
555: of = 0;
556: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 557: }
1.85 schwarze 558: curp->child = save_child;
1.1 schwarze 559: pos = 0;
560: continue;
561: default:
562: break;
563: }
564:
565: /*
566: * If input parsers have not been allocated, do so now.
1.2 schwarze 567: * We keep these instanced between parsers, but set them
1.1 schwarze 568: * locally per parse routine since we can use different
569: * parsers with each one.
570: */
571:
1.110 schwarze 572: if (curp->man == NULL ||
573: curp->man->macroset == MACROSET_NONE)
1.60 schwarze 574: choose_parser(curp);
1.1 schwarze 575:
1.25 schwarze 576: /*
1.60 schwarze 577: * Lastly, push down into the parsers themselves.
1.1 schwarze 578: * If libroff returns ROFF_TBL, then add it to the
579: * currently open parse. Since we only get here if
580: * there does exist data (see tbl_data.c), we're
581: * guaranteed that something's been allocated.
582: * Do the same for ROFF_EQN.
583: */
584:
1.76 schwarze 585: if (rr == ROFF_TBL) {
586: while ((span = roff_span(curp->roff)) != NULL)
1.110 schwarze 587: if (curp->man->macroset == MACROSET_MDOC)
588: mdoc_addspan(curp->man, span);
1.76 schwarze 589: else
590: man_addspan(curp->man, span);
591: } else if (rr == ROFF_EQN) {
1.110 schwarze 592: if (curp->man->macroset == MACROSET_MDOC)
593: mdoc_addeqn(curp->man, roff_eqn(curp->roff));
1.76 schwarze 594: else
595: man_addeqn(curp->man, roff_eqn(curp->roff));
1.110 schwarze 596: } else if ((curp->man->macroset == MACROSET_MDOC ?
597: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 598: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
599: break;
1.1 schwarze 600:
601: /* Temporary buffers typically are not full. */
602:
603: if (0 == start && '\0' == blk.buf[i])
604: break;
605:
606: /* Start the next input line. */
607:
608: pos = 0;
609: }
610:
611: free(ln.buf);
612: }
613:
614: static int
1.18 schwarze 615: read_whole_file(struct mparse *curp, const char *file, int fd,
616: struct buf *fb, int *with_mmap)
1.1 schwarze 617: {
618: struct stat st;
619: size_t off;
620: ssize_t ssz;
621:
622: if (-1 == fstat(fd, &st)) {
1.82 schwarze 623: perror(file);
624: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 625: }
626:
627: /*
628: * If we're a regular file, try just reading in the whole entry
629: * via mmap(). This is faster than reading it into blocks, and
630: * since each file is only a few bytes to begin with, I'm not
631: * concerned that this is going to tank any machines.
632: */
633:
634: if (S_ISREG(st.st_mode)) {
1.106 schwarze 635: if (st.st_size > 0x7fffffff) {
1.86 schwarze 636: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 schwarze 637: return(0);
638: }
639: *with_mmap = 1;
640: fb->sz = (size_t)st.st_size;
1.15 schwarze 641: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 642: if (fb->buf != MAP_FAILED)
643: return(1);
644: }
645:
646: /*
647: * If this isn't a regular file (like, say, stdin), then we must
648: * go the old way and just read things in bit by bit.
649: */
650:
651: *with_mmap = 0;
652: off = 0;
653: fb->sz = 0;
654: fb->buf = NULL;
655: for (;;) {
656: if (off == fb->sz) {
657: if (fb->sz == (1U << 31)) {
1.86 schwarze 658: mandoc_msg(MANDOCERR_TOOLARGE, curp,
659: 0, 0, NULL);
1.1 schwarze 660: break;
661: }
662: resize_buf(fb, 65536);
663: }
664: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
665: if (ssz == 0) {
666: fb->sz = off;
667: return(1);
668: }
669: if (ssz == -1) {
1.82 schwarze 670: perror(file);
671: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 672: }
673: off += (size_t)ssz;
674: }
675:
676: free(fb->buf);
677: fb->buf = NULL;
678: return(0);
679: }
680:
681: static void
682: mparse_end(struct mparse *curp)
683: {
684:
1.110 schwarze 685: if (curp->man == NULL && curp->sodest == NULL)
686: curp->man = man_alloc(curp->roff, curp, curp->defos,
687: curp->options & MPARSE_QUICK ? 1 : 0);
688: if (curp->man->macroset == MACROSET_NONE)
689: curp->man->macroset = MACROSET_MAN;
690: if (curp->man->macroset == MACROSET_MDOC)
691: mdoc_endparse(curp->man);
692: else
1.86 schwarze 693: man_endparse(curp->man);
1.1 schwarze 694: roff_endparse(curp->roff);
695: }
696:
1.15 schwarze 697: static void
698: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 699: {
1.61 schwarze 700: struct buf *svprimary;
1.1 schwarze 701: const char *svfile;
1.71 schwarze 702: size_t offset;
1.14 schwarze 703: static int recursion_depth;
704:
705: if (64 < recursion_depth) {
706: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 707: return;
1.14 schwarze 708: }
1.1 schwarze 709:
1.15 schwarze 710: /* Line number is per-file. */
711: svfile = curp->file;
712: curp->file = file;
1.61 schwarze 713: svprimary = curp->primary;
1.59 schwarze 714: curp->primary = &blk;
1.15 schwarze 715: curp->line = 1;
716: recursion_depth++;
717:
1.70 schwarze 718: /* Skip an UTF-8 byte order mark. */
719: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
720: (unsigned char)blk.buf[0] == 0xef &&
721: (unsigned char)blk.buf[1] == 0xbb &&
722: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 723: offset = 3;
1.70 schwarze 724: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 725: } else
726: offset = 0;
1.70 schwarze 727:
1.71 schwarze 728: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 729:
1.86 schwarze 730: if (--recursion_depth == 0)
1.15 schwarze 731: mparse_end(curp);
732:
1.61 schwarze 733: curp->primary = svprimary;
1.15 schwarze 734: curp->file = svfile;
735: }
736:
1.74 schwarze 737: /*
738: * Read the whole file into memory and call the parsers.
739: * Called recursively when an .so request is encountered.
740: */
1.15 schwarze 741: enum mandoclevel
742: mparse_readfd(struct mparse *curp, int fd, const char *file)
743: {
744: struct buf blk;
745: int with_mmap;
1.70 schwarze 746: int save_filenc;
1.1 schwarze 747:
1.67 schwarze 748: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 749: save_filenc = curp->filenc;
750: curp->filenc = curp->options &
751: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 752: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 753: curp->filenc = save_filenc;
1.67 schwarze 754: if (with_mmap)
755: munmap(blk.buf, blk.sz);
756: else
757: free(blk.buf);
758: }
1.1 schwarze 759:
1.74 schwarze 760: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 schwarze 761: perror(file);
1.67 schwarze 762:
1.74 schwarze 763: mparse_wait(curp);
1.1 schwarze 764: return(curp->file_status);
1.58 schwarze 765: }
766:
767: enum mandoclevel
1.73 schwarze 768: mparse_open(struct mparse *curp, int *fd, const char *file)
1.58 schwarze 769: {
770: int pfd[2];
1.74 schwarze 771: int save_errno;
1.58 schwarze 772: char *cp;
773:
774: curp->file = file;
1.74 schwarze 775:
776: /* Unless zipped, try to just open the file. */
777:
1.58 schwarze 778: if ((cp = strrchr(file, '.')) == NULL ||
779: strcmp(cp + 1, "gz")) {
1.73 schwarze 780: curp->child = 0;
1.74 schwarze 781: if ((*fd = open(file, O_RDONLY)) != -1)
782: return(MANDOCLEVEL_OK);
783:
784: /* Open failed; try to append ".gz". */
785:
786: mandoc_asprintf(&cp, "%s.gz", file);
787: file = cp;
788: } else
789: cp = NULL;
790:
791: /* Before forking, make sure the file can be read. */
792:
793: save_errno = errno;
794: if (access(file, R_OK) == -1) {
795: if (cp != NULL)
796: errno = save_errno;
1.83 schwarze 797: free(cp);
798: *fd = -1;
799: curp->child = 0;
800: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
801: return(MANDOCLEVEL_ERROR);
1.58 schwarze 802: }
803:
1.74 schwarze 804: /* Run gunzip(1). */
805:
1.58 schwarze 806: if (pipe(pfd) == -1) {
1.82 schwarze 807: perror("pipe");
808: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 809: }
810:
1.73 schwarze 811: switch (curp->child = fork()) {
1.58 schwarze 812: case -1:
1.82 schwarze 813: perror("fork");
814: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 815: case 0:
816: close(pfd[0]);
817: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.82 schwarze 818: perror("dup");
819: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 820: }
1.107 schwarze 821: signal(SIGPIPE, SIG_DFL);
1.58 schwarze 822: execlp("gunzip", "gunzip", "-c", file, NULL);
1.82 schwarze 823: perror("exec");
824: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 825: default:
826: close(pfd[1]);
827: *fd = pfd[0];
828: return(MANDOCLEVEL_OK);
829: }
830: }
831:
832: enum mandoclevel
1.73 schwarze 833: mparse_wait(struct mparse *curp)
1.58 schwarze 834: {
835: int status;
836:
1.73 schwarze 837: if (curp->child == 0)
838: return(MANDOCLEVEL_OK);
839:
840: if (waitpid(curp->child, &status, 0) == -1) {
1.82 schwarze 841: perror("wait");
842: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 843: }
1.104 schwarze 844: curp->child = 0;
1.58 schwarze 845: if (WIFSIGNALED(status)) {
1.83 schwarze 846: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
847: "gunzip died from signal %d", WTERMSIG(status));
848: return(MANDOCLEVEL_ERROR);
1.58 schwarze 849: }
850: if (WEXITSTATUS(status)) {
1.83 schwarze 851: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
852: "gunzip failed with code %d", WEXITSTATUS(status));
853: return(MANDOCLEVEL_ERROR);
1.58 schwarze 854: }
855: return(MANDOCLEVEL_OK);
1.1 schwarze 856: }
857:
858: struct mparse *
1.69 schwarze 859: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
860: const struct mchars *mchars, const char *defos)
1.1 schwarze 861: {
862: struct mparse *curp;
863:
864: curp = mandoc_calloc(1, sizeof(struct mparse));
865:
1.22 schwarze 866: curp->options = options;
1.1 schwarze 867: curp->wlevel = wlevel;
868: curp->mmsg = mmsg;
1.7 schwarze 869: curp->defos = defos;
1.1 schwarze 870:
1.69 schwarze 871: curp->mchars = mchars;
872: curp->roff = roff_alloc(curp, curp->mchars, options);
1.111 ! schwarze 873: if (curp->options & MPARSE_MDOC) {
1.110 schwarze 874: curp->man = mdoc_alloc(
1.50 schwarze 875: curp->roff, curp, curp->defos,
876: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 ! schwarze 877: mdoc_hash_init();
! 878: }
! 879: if (curp->options & MPARSE_MAN) {
1.110 schwarze 880: curp->man = man_alloc(
1.81 schwarze 881: curp->roff, curp, curp->defos,
1.50 schwarze 882: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 ! schwarze 883: man_hash_init();
! 884: }
1.50 schwarze 885:
1.1 schwarze 886: return(curp);
887: }
888:
889: void
890: mparse_reset(struct mparse *curp)
891: {
892:
893: roff_reset(curp->roff);
894:
1.110 schwarze 895: if (curp->man != NULL) {
896: if (curp->man->macroset == MACROSET_MDOC)
897: mdoc_reset(curp->man);
898: else
899: man_reset(curp->man);
900: curp->man->macroset = MACROSET_NONE;
901: }
1.4 schwarze 902: if (curp->secondary)
903: curp->secondary->sz = 0;
1.1 schwarze 904:
905: curp->file_status = MANDOCLEVEL_OK;
1.23 schwarze 906:
907: free(curp->sodest);
908: curp->sodest = NULL;
1.1 schwarze 909: }
910:
911: void
912: mparse_free(struct mparse *curp)
913: {
914:
1.110 schwarze 915: if (curp->man->macroset == MACROSET_MDOC)
916: mdoc_free(curp->man);
917: if (curp->man->macroset == MACROSET_MAN)
918: man_free(curp->man);
1.1 schwarze 919: if (curp->roff)
920: roff_free(curp->roff);
1.4 schwarze 921: if (curp->secondary)
922: free(curp->secondary->buf);
1.1 schwarze 923:
1.4 schwarze 924: free(curp->secondary);
1.23 schwarze 925: free(curp->sodest);
1.1 schwarze 926: free(curp);
927: }
928:
929: void
1.110 schwarze 930: mparse_result(struct mparse *curp, struct roff_man **man,
931: char **sodest)
1.1 schwarze 932: {
933:
1.23 schwarze 934: if (sodest && NULL != (*sodest = curp->sodest)) {
935: *man = NULL;
936: return;
937: }
1.1 schwarze 938: if (man)
939: *man = curp->man;
940: }
941:
942: void
943: mandoc_vmsg(enum mandocerr t, struct mparse *m,
944: int ln, int pos, const char *fmt, ...)
945: {
946: char buf[256];
947: va_list ap;
948:
949: va_start(ap, fmt);
1.26 schwarze 950: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 951: va_end(ap);
952:
953: mandoc_msg(t, m, ln, pos, buf);
954: }
955:
956: void
1.25 schwarze 957: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 958: int ln, int col, const char *msg)
959: {
960: enum mandoclevel level;
961:
1.87 schwarze 962: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 963: while (er < mandoclimits[level])
964: level--;
965:
1.83 schwarze 966: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 967: return;
968:
969: if (m->mmsg)
970: (*m->mmsg)(er, level, m->file, ln, col, msg);
971:
972: if (m->file_status < level)
973: m->file_status = level;
974: }
975:
976: const char *
977: mparse_strerror(enum mandocerr er)
978: {
979:
980: return(mandocerrs[er]);
981: }
982:
983: const char *
984: mparse_strlevel(enum mandoclevel lvl)
985: {
986: return(mandoclevels[lvl]);
1.4 schwarze 987: }
988:
989: void
990: mparse_keep(struct mparse *p)
991: {
992:
993: assert(NULL == p->secondary);
994: p->secondary = mandoc_calloc(1, sizeof(struct buf));
995: }
996:
997: const char *
998: mparse_getkeep(const struct mparse *p)
999: {
1000:
1001: assert(p->secondary);
1002: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 schwarze 1003: }
![[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