Annotation of mandoc/read.c, Revision 1.132
1.132 ! schwarze 1: /* $Id: read.c,v 1.131 2015/03/11 13:05:20 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.107 schwarze 4: * Copyright (c) 2010-2015 Ingo Schwarze <schwarze@openbsd.org>
1.42 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 kristaps 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: *
11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
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: */
1.11 kristaps 19: #include "config.h"
20:
1.80 schwarze 21: #include <sys/types.h>
1.81 schwarze 22: #if HAVE_MMAP
1.82 schwarze 23: #include <sys/mman.h>
1.80 schwarze 24: #include <sys/stat.h>
1.15 kristaps 25: #endif
1.82 schwarze 26: #include <sys/wait.h>
1.1 kristaps 27:
28: #include <assert.h>
29: #include <ctype.h>
1.40 schwarze 30: #include <errno.h>
1.1 kristaps 31: #include <fcntl.h>
1.132 ! schwarze 32: #include <signal.h>
1.3 kristaps 33: #include <stdarg.h>
1.28 joerg 34: #include <stdint.h>
1.1 kristaps 35: #include <stdio.h>
36: #include <stdlib.h>
37: #include <string.h>
38: #include <unistd.h>
39:
40: #include "mandoc.h"
1.46 schwarze 41: #include "mandoc_aux.h"
1.3 kristaps 42: #include "libmandoc.h"
1.1 kristaps 43: #include "mdoc.h"
44: #include "man.h"
45:
46: #define REPARSE_LIMIT 1000
47:
48: struct mparse {
49: struct man *pman; /* persistent man parser */
50: struct mdoc *pmdoc; /* persistent mdoc parser */
51: struct man *man; /* man parser */
52: struct mdoc *mdoc; /* mdoc parser */
53: struct roff *roff; /* roff parser (!NULL) */
1.94 schwarze 54: const struct mchars *mchars; /* character table */
1.45 schwarze 55: char *sodest; /* filename pointed to by .so */
1.83 schwarze 56: const char *file; /* filename of current input file */
57: struct buf *primary; /* buffer currently being parsed */
58: struct buf *secondary; /* preprocessed copy of input */
59: const char *defos; /* default operating system */
60: mandocmsg mmsg; /* warning/error message handler */
61: enum mandoclevel file_status; /* status of current parse */
62: enum mandoclevel wlevel; /* ignore messages below this */
63: int options; /* parser options */
1.93 schwarze 64: int filenc; /* encoding of the current file */
1.1 kristaps 65: int reparse_count; /* finite interp. stack */
1.83 schwarze 66: int line; /* line number in the file */
1.97 schwarze 67: pid_t child; /* the gunzip(1) process */
1.1 kristaps 68: };
69:
1.84 schwarze 70: static void choose_parser(struct mparse *);
1.1 kristaps 71: static void resize_buf(struct buf *, size_t);
1.95 schwarze 72: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 73: static int read_whole_file(struct mparse *, const char *, int,
74: struct buf *, int *);
1.1 kristaps 75: static void mparse_end(struct mparse *);
1.37 schwarze 76: static void mparse_parse_buffer(struct mparse *, struct buf,
77: const char *);
1.1 kristaps 78:
1.3 kristaps 79: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
80: MANDOCERR_OK,
81: MANDOCERR_WARNING,
82: MANDOCERR_WARNING,
83: MANDOCERR_ERROR,
1.112 schwarze 84: MANDOCERR_UNSUPP,
1.3 kristaps 85: MANDOCERR_MAX,
86: MANDOCERR_MAX
87: };
88:
1.7 kristaps 89: static const char * const mandocerrs[MANDOCERR_MAX] = {
90: "ok",
91:
92: "generic warning",
93:
94: /* related to the prologue */
1.79 schwarze 95: "missing manual title, using UNTITLED",
96: "missing manual title, using \"\"",
1.54 schwarze 97: "lower case character in document title",
1.79 schwarze 98: "missing manual section, using \"\"",
1.7 kristaps 99: "unknown manual section",
1.54 schwarze 100: "missing date, using today's date",
1.7 kristaps 101: "cannot parse date, using it verbatim",
1.79 schwarze 102: "missing Os macro, using \"\"",
103: "duplicate prologue macro",
104: "late prologue macro",
105: "skipping late title macro",
1.7 kristaps 106: "prologue macros out of order",
107:
108: /* related to document structure */
109: ".so is fragile, better use ln(1)",
1.50 schwarze 110: "no document body",
1.54 schwarze 111: "content before first section header",
112: "first section is not \"NAME\"",
1.128 schwarze 113: "NAME section without name",
114: "NAME section without description",
115: "description not at the end of NAME",
116: "bad NAME section content",
1.121 schwarze 117: "missing description line, using \"\"",
1.7 kristaps 118: "sections out of conventional order",
1.54 schwarze 119: "duplicate section title",
120: "unexpected section",
1.87 schwarze 121: "unusual Xr order",
122: "unusual Xr punctuation",
1.86 schwarze 123: "AUTHORS section without An macro",
1.7 kristaps 124:
125: /* related to macros and nesting */
1.55 schwarze 126: "obsolete macro",
1.102 schwarze 127: "macro neither callable nor escaped",
1.7 kristaps 128: "skipping paragraph macro",
1.31 schwarze 129: "moving paragraph macro out of list",
1.7 kristaps 130: "skipping no-space macro",
131: "blocks badly nested",
132: "nested displays are not portable",
1.57 schwarze 133: "moving content out of list",
134: ".Vt block has child macro",
1.78 schwarze 135: "fill mode already enabled, skipping",
136: "fill mode already disabled, skipping",
1.7 kristaps 137: "line scope broken",
138:
139: /* related to missing macro arguments */
1.58 schwarze 140: "skipping empty request",
141: "conditional request controls empty scope",
1.7 kristaps 142: "skipping empty macro",
1.124 schwarze 143: "empty block",
1.62 schwarze 144: "empty argument, using 0n",
1.60 schwarze 145: "missing display type, using -ragged",
146: "list type is not the first argument",
147: "missing -width in -tag list, using 8n",
1.78 schwarze 148: "missing utility name, using \"\"",
1.123 schwarze 149: "missing function name, using \"\"",
1.60 schwarze 150: "empty head in list item",
151: "empty list item",
1.61 schwarze 152: "missing font type, using \\fR",
153: "unknown font type, using \\fR",
1.103 schwarze 154: "nothing follows prefix",
1.122 schwarze 155: "empty reference block",
1.60 schwarze 156: "missing -std argument, adding it",
1.125 schwarze 157: "missing option string, using \"\"",
158: "missing resource identifier, using \"\"",
1.90 schwarze 159: "missing eqn box, using \"\"",
1.7 kristaps 160:
161: /* related to bad macro arguments */
1.64 schwarze 162: "unterminated quoted argument",
1.7 kristaps 163: "duplicate argument",
1.76 schwarze 164: "skipping duplicate argument",
1.63 schwarze 165: "skipping duplicate display type",
166: "skipping duplicate list type",
1.76 schwarze 167: "skipping -width argument",
1.126 schwarze 168: "wrong number of cells",
1.7 kristaps 169: "unknown AT&T UNIX version",
1.88 schwarze 170: "comma in function argument",
1.89 schwarze 171: "parenthesis in function name",
1.67 schwarze 172: "invalid content in Rs block",
1.63 schwarze 173: "invalid Boolean argument",
174: "unknown font, skipping request",
1.126 schwarze 175: "odd number of characters in request",
1.7 kristaps 176:
177: /* related to plain text */
1.64 schwarze 178: "blank line in fill mode, using .sp",
179: "tab in filled text",
180: "whitespace at end of input line",
1.7 kristaps 181: "bad comment style",
1.64 schwarze 182: "invalid escape sequence",
183: "undefined string, using \"\"",
1.16 kristaps 184:
1.118 schwarze 185: /* related to tables */
186: "tbl line starts with span",
187: "tbl column starts with span",
188: "skipping vertical bar in tbl layout",
189:
1.7 kristaps 190: "generic error",
191:
192: /* related to tables */
1.116 schwarze 193: "non-alphabetic character in tbl options",
194: "skipping unknown tbl option",
195: "missing tbl option argument",
196: "wrong tbl option argument size",
1.118 schwarze 197: "empty tbl layout",
198: "invalid character in tbl layout",
199: "unmatched parenthesis in tbl layout",
1.119 schwarze 200: "tbl without any data cells",
201: "ignoring data in spanned tbl cell",
202: "ignoring extra tbl data cells",
203: "data block open at end of tbl",
1.7 kristaps 204:
1.68 schwarze 205: /* related to document structure and macros */
1.108 schwarze 206: NULL,
1.7 kristaps 207: "input stack limit exceeded, infinite loop?",
208: "skipping bad character",
1.68 schwarze 209: "skipping unknown macro",
1.112 schwarze 210: "skipping insecure request",
1.73 schwarze 211: "skipping item outside list",
1.68 schwarze 212: "skipping column outside column list",
213: "skipping end of block that is not open",
1.115 schwarze 214: "fewer RS blocks open, skipping",
1.68 schwarze 215: "inserting missing end of block",
216: "appending missing end of block",
217:
218: /* related to request and macro arguments */
1.7 kristaps 219: "escaped character not allowed in a name",
1.99 schwarze 220: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 221: "missing list type, using -item",
1.70 schwarze 222: "missing manual name, using \"\"",
1.71 schwarze 223: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 224: "unknown standard specifier",
1.71 schwarze 225: "skipping request without numeric argument",
1.109 schwarze 226: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 227: ".so request failed",
1.61 schwarze 228: "skipping all arguments",
229: "skipping excess arguments",
1.92 kristaps 230: "divide by zero",
1.112 schwarze 231:
232: "unsupported feature",
233: "input too large",
1.114 schwarze 234: "unsupported control character",
1.112 schwarze 235: "unsupported roff request",
1.119 schwarze 236: "eqn delim option in tbl",
1.118 schwarze 237: "unsupported tbl layout modifier",
1.112 schwarze 238: "ignoring macro in table",
1.7 kristaps 239: };
240:
241: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
242: "SUCCESS",
243: "RESERVED",
244: "WARNING",
245: "ERROR",
1.112 schwarze 246: "UNSUPP",
1.7 kristaps 247: "BADARG",
248: "SYSERR"
249: };
250:
1.47 schwarze 251:
1.1 kristaps 252: static void
253: resize_buf(struct buf *buf, size_t initial)
254: {
255:
256: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
257: buf->buf = mandoc_realloc(buf->buf, buf->sz);
258: }
259:
260: static void
1.84 schwarze 261: choose_parser(struct mparse *curp)
1.1 kristaps 262: {
1.83 schwarze 263: char *cp, *ep;
264: int format;
1.1 kristaps 265:
1.83 schwarze 266: /*
267: * If neither command line arguments -mdoc or -man select
268: * a parser nor the roff parser found a .Dd or .TH macro
269: * yet, look ahead in the main input buffer.
270: */
271:
272: if ((format = roff_getformat(curp->roff)) == 0) {
273: cp = curp->primary->buf;
274: ep = cp + curp->primary->sz;
275: while (cp < ep) {
1.85 schwarze 276: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 277: cp++;
278: if (cp[0] == 'D' && cp[1] == 'd') {
279: format = MPARSE_MDOC;
280: break;
281: }
282: if (cp[0] == 'T' && cp[1] == 'H') {
283: format = MPARSE_MAN;
284: break;
285: }
286: }
287: cp = memchr(cp, '\n', ep - cp);
288: if (cp == NULL)
289: break;
290: cp++;
291: }
1.1 kristaps 292: }
293:
1.83 schwarze 294: if (format == MPARSE_MDOC) {
1.47 schwarze 295: if (NULL == curp->pmdoc)
1.44 schwarze 296: curp->pmdoc = mdoc_alloc(
297: curp->roff, curp, curp->defos,
298: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 299: assert(curp->pmdoc);
300: curp->mdoc = curp->pmdoc;
301: return;
1.47 schwarze 302: }
1.1 kristaps 303:
1.83 schwarze 304: /* Fall back to man(7) as a last resort. */
305:
1.47 schwarze 306: if (NULL == curp->pman)
1.106 schwarze 307: curp->pman = man_alloc(
308: curp->roff, curp, curp->defos,
1.44 schwarze 309: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 310: assert(curp->pman);
311: curp->man = curp->pman;
312: }
313:
314: /*
1.95 schwarze 315: * Main parse routine for a buffer.
316: * It assumes encoding and line numbering are already set up.
317: * It can recurse directly (for invocations of user-defined
318: * macros, inline equations, and input line traps)
319: * and indirectly (for .so file inclusion).
1.1 kristaps 320: */
321: static void
1.95 schwarze 322: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 323: {
324: const struct tbl_span *span;
325: struct buf ln;
1.113 schwarze 326: const char *save_file;
1.110 schwarze 327: char *cp;
1.95 schwarze 328: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 329: enum rofferr rr;
1.100 schwarze 330: int of;
1.1 kristaps 331: int lnn; /* line number in the real file */
1.110 schwarze 332: int fd;
333: pid_t save_child;
1.1 kristaps 334: unsigned char c;
335:
1.95 schwarze 336: memset(&ln, 0, sizeof(ln));
1.1 kristaps 337:
1.47 schwarze 338: lnn = curp->line;
339: pos = 0;
1.1 kristaps 340:
1.95 schwarze 341: while (i < blk.sz) {
1.1 kristaps 342: if (0 == pos && '\0' == blk.buf[i])
343: break;
344:
345: if (start) {
346: curp->line = lnn;
347: curp->reparse_count = 0;
1.93 schwarze 348:
349: if (lnn < 3 &&
350: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 351: curp->filenc & MPARSE_LATIN1)
352: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 353: }
354:
1.95 schwarze 355: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 356:
357: /*
358: * When finding an unescaped newline character,
359: * leave the character loop to process the line.
360: * Skip a preceding carriage return, if any.
361: */
362:
1.95 schwarze 363: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 364: '\n' == blk.buf[i + 1])
365: ++i;
366: if ('\n' == blk.buf[i]) {
367: ++i;
368: ++lnn;
369: break;
370: }
371:
1.35 schwarze 372: /*
1.93 schwarze 373: * Make sure we have space for the worst
374: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 375: */
376:
1.95 schwarze 377: if (pos + 11 > ln.sz)
1.35 schwarze 378: resize_buf(&ln, 256);
379:
1.47 schwarze 380: /*
1.93 schwarze 381: * Encode 8-bit input.
1.1 kristaps 382: */
383:
1.93 schwarze 384: c = blk.buf[i];
385: if (c & 0x80) {
1.95 schwarze 386: if ( ! (curp->filenc && preconv_encode(
387: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 388: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
389: curp->line, pos, "0x%x", c);
1.93 schwarze 390: ln.buf[pos++] = '?';
391: i++;
392: }
393: continue;
394: }
395:
396: /*
397: * Exclude control characters.
398: */
1.1 kristaps 399:
1.93 schwarze 400: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 401: mandoc_vmsg(c == 0x00 || c == 0x04 ||
402: c > 0x0a ? MANDOCERR_CHAR_BAD :
403: MANDOCERR_CHAR_UNSUPP,
404: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 405: i++;
1.127 schwarze 406: if (c != '\r')
407: ln.buf[pos++] = '?';
1.1 kristaps 408: continue;
409: }
410:
411: /* Trailing backslash = a plain char. */
412:
1.95 schwarze 413: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 414: ln.buf[pos++] = blk.buf[i++];
415: continue;
416: }
417:
418: /*
419: * Found escape and at least one other character.
420: * When it's a newline character, skip it.
421: * When there is a carriage return in between,
422: * skip that one as well.
423: */
424:
1.95 schwarze 425: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 426: '\n' == blk.buf[i + 2])
427: ++i;
428: if ('\n' == blk.buf[i + 1]) {
429: i += 2;
430: ++lnn;
431: continue;
432: }
433:
1.13 kristaps 434: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 435: i += 2;
436: /* Comment, skip to end of line */
1.95 schwarze 437: for (; i < blk.sz; ++i) {
1.1 kristaps 438: if ('\n' == blk.buf[i]) {
439: ++i;
440: ++lnn;
441: break;
442: }
443: }
444:
445: /* Backout trailing whitespaces */
446: for (; pos > 0; --pos) {
447: if (ln.buf[pos - 1] != ' ')
448: break;
449: if (pos > 2 && ln.buf[pos - 2] == '\\')
450: break;
451: }
452: break;
453: }
454:
1.35 schwarze 455: /* Catch escaped bogus characters. */
456:
457: c = (unsigned char) blk.buf[i+1];
458:
1.47 schwarze 459: if ( ! (isascii(c) &&
460: (isgraph(c) || isblank(c)))) {
1.114 schwarze 461: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 462: curp->line, pos, "0x%x", c);
1.35 schwarze 463: i += 2;
464: ln.buf[pos++] = '?';
465: continue;
466: }
467:
1.1 kristaps 468: /* Some other escape sequence, copy & cont. */
469:
470: ln.buf[pos++] = blk.buf[i++];
471: ln.buf[pos++] = blk.buf[i++];
472: }
473:
1.95 schwarze 474: if (pos >= ln.sz)
1.1 kristaps 475: resize_buf(&ln, 256);
476:
477: ln.buf[pos] = '\0';
478:
479: /*
480: * A significant amount of complexity is contained by
481: * the roff preprocessor. It's line-oriented but can be
482: * expressed on one line, so we need at times to
483: * readjust our starting point and re-run it. The roff
484: * preprocessor can also readjust the buffers with new
485: * data, so we pass them in wholesale.
486: */
487:
488: of = 0;
489:
1.24 kristaps 490: /*
491: * Maintain a lookaside buffer of all parsed lines. We
492: * only do this if mparse_keep() has been invoked (the
493: * buffer may be accessed with mparse_getkeep()).
494: */
495:
496: if (curp->secondary) {
1.47 schwarze 497: curp->secondary->buf = mandoc_realloc(
498: curp->secondary->buf,
499: curp->secondary->sz + pos + 2);
500: memcpy(curp->secondary->buf +
501: curp->secondary->sz,
502: ln.buf, pos);
1.24 kristaps 503: curp->secondary->sz += pos;
504: curp->secondary->buf
505: [curp->secondary->sz] = '\n';
506: curp->secondary->sz++;
507: curp->secondary->buf
508: [curp->secondary->sz] = '\0';
509: }
1.1 kristaps 510: rerun:
1.96 schwarze 511: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 512:
513: switch (rr) {
1.47 schwarze 514: case ROFF_REPARSE:
1.1 kristaps 515: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 516: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 517: else
1.3 kristaps 518: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 519: curp->line, pos, NULL);
1.1 kristaps 520: pos = 0;
521: continue;
1.47 schwarze 522: case ROFF_APPEND:
1.95 schwarze 523: pos = strlen(ln.buf);
1.1 kristaps 524: continue;
1.47 schwarze 525: case ROFF_RERUN:
1.1 kristaps 526: goto rerun;
1.47 schwarze 527: case ROFF_IGN:
1.1 kristaps 528: pos = 0;
529: continue;
1.47 schwarze 530: case ROFF_SO:
1.95 schwarze 531: if ( ! (curp->options & MPARSE_SO) &&
532: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 533: curp->sodest = mandoc_strdup(ln.buf + of);
534: free(ln.buf);
535: return;
536: }
1.24 kristaps 537: /*
538: * We remove `so' clauses from our lookaside
539: * buffer because we're going to descend into
540: * the file recursively.
541: */
1.47 schwarze 542: if (curp->secondary)
1.25 kristaps 543: curp->secondary->sz -= pos + 1;
1.113 schwarze 544: save_file = curp->file;
1.110 schwarze 545: save_child = curp->child;
546: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 schwarze 547: MANDOCLEVEL_OK) {
1.110 schwarze 548: mparse_readfd(curp, fd, ln.buf + of);
1.113 schwarze 549: curp->file = save_file;
550: } else {
551: curp->file = save_file;
1.52 schwarze 552: mandoc_vmsg(MANDOCERR_SO_FAIL,
553: curp, curp->line, pos,
554: ".so %s", ln.buf + of);
1.110 schwarze 555: ln.sz = mandoc_asprintf(&cp,
556: ".sp\nSee the file %s.\n.sp",
557: ln.buf + of);
558: free(ln.buf);
559: ln.buf = cp;
560: of = 0;
561: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 562: }
1.110 schwarze 563: curp->child = save_child;
1.1 kristaps 564: pos = 0;
565: continue;
566: default:
567: break;
568: }
569:
570: /*
571: * If input parsers have not been allocated, do so now.
1.14 kristaps 572: * We keep these instanced between parsers, but set them
1.1 kristaps 573: * locally per parse routine since we can use different
574: * parsers with each one.
575: */
576:
577: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 578: choose_parser(curp);
1.1 kristaps 579:
1.47 schwarze 580: /*
1.84 schwarze 581: * Lastly, push down into the parsers themselves.
1.1 kristaps 582: * If libroff returns ROFF_TBL, then add it to the
583: * currently open parse. Since we only get here if
584: * there does exist data (see tbl_data.c), we're
585: * guaranteed that something's been allocated.
586: * Do the same for ROFF_EQN.
587: */
588:
1.100 schwarze 589: if (rr == ROFF_TBL) {
590: while ((span = roff_span(curp->roff)) != NULL)
591: if (curp->man == NULL)
592: mdoc_addspan(curp->mdoc, span);
593: else
594: man_addspan(curp->man, span);
595: } else if (rr == ROFF_EQN) {
596: if (curp->man == NULL)
597: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
598: else
599: man_addeqn(curp->man, roff_eqn(curp->roff));
600: } else if ((curp->man == NULL ?
601: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
602: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
603: break;
1.1 kristaps 604:
605: /* Temporary buffers typically are not full. */
606:
607: if (0 == start && '\0' == blk.buf[i])
608: break;
609:
610: /* Start the next input line. */
611:
612: pos = 0;
613: }
614:
615: free(ln.buf);
616: }
617:
618: static int
1.40 schwarze 619: read_whole_file(struct mparse *curp, const char *file, int fd,
620: struct buf *fb, int *with_mmap)
1.1 kristaps 621: {
622: size_t off;
623: ssize_t ssz;
624:
1.81 schwarze 625: #if HAVE_MMAP
1.15 kristaps 626: struct stat st;
1.1 kristaps 627: if (-1 == fstat(fd, &st)) {
1.107 schwarze 628: perror(file);
629: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 630: }
631:
632: /*
633: * If we're a regular file, try just reading in the whole entry
634: * via mmap(). This is faster than reading it into blocks, and
635: * since each file is only a few bytes to begin with, I'm not
636: * concerned that this is going to tank any machines.
637: */
638:
639: if (S_ISREG(st.st_mode)) {
1.131 schwarze 640: if (st.st_size > 0x7fffffff) {
1.111 schwarze 641: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 642: return(0);
643: }
644: *with_mmap = 1;
645: fb->sz = (size_t)st.st_size;
1.37 schwarze 646: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 647: if (fb->buf != MAP_FAILED)
648: return(1);
649: }
1.15 kristaps 650: #endif
1.1 kristaps 651:
652: /*
653: * If this isn't a regular file (like, say, stdin), then we must
654: * go the old way and just read things in bit by bit.
655: */
656:
657: *with_mmap = 0;
658: off = 0;
659: fb->sz = 0;
660: fb->buf = NULL;
661: for (;;) {
662: if (off == fb->sz) {
663: if (fb->sz == (1U << 31)) {
1.111 schwarze 664: mandoc_msg(MANDOCERR_TOOLARGE, curp,
665: 0, 0, NULL);
1.1 kristaps 666: break;
667: }
668: resize_buf(fb, 65536);
669: }
670: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
671: if (ssz == 0) {
672: fb->sz = off;
673: return(1);
674: }
675: if (ssz == -1) {
1.107 schwarze 676: perror(file);
677: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 678: }
679: off += (size_t)ssz;
680: }
681:
682: free(fb->buf);
683: fb->buf = NULL;
684: return(0);
685: }
686:
687: static void
688: mparse_end(struct mparse *curp)
689: {
690:
1.72 schwarze 691: if (curp->mdoc == NULL &&
692: curp->man == NULL &&
693: curp->sodest == NULL) {
694: if (curp->options & MPARSE_MDOC)
695: curp->mdoc = curp->pmdoc;
696: else {
697: if (curp->pman == NULL)
1.106 schwarze 698: curp->pman = man_alloc(
699: curp->roff, curp, curp->defos,
1.72 schwarze 700: curp->options & MPARSE_QUICK ? 1 : 0);
701: curp->man = curp->pman;
702: }
703: }
1.111 schwarze 704: if (curp->mdoc)
705: mdoc_endparse(curp->mdoc);
706: if (curp->man)
707: man_endparse(curp->man);
1.1 kristaps 708: roff_endparse(curp->roff);
709: }
710:
711: static void
1.36 schwarze 712: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 713: {
1.85 schwarze 714: struct buf *svprimary;
1.28 joerg 715: const char *svfile;
1.95 schwarze 716: size_t offset;
1.36 schwarze 717: static int recursion_depth;
718:
719: if (64 < recursion_depth) {
720: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
721: return;
722: }
1.28 joerg 723:
724: /* Line number is per-file. */
725: svfile = curp->file;
726: curp->file = file;
1.85 schwarze 727: svprimary = curp->primary;
1.83 schwarze 728: curp->primary = &blk;
1.28 joerg 729: curp->line = 1;
1.36 schwarze 730: recursion_depth++;
1.28 joerg 731:
1.93 schwarze 732: /* Skip an UTF-8 byte order mark. */
733: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
734: (unsigned char)blk.buf[0] == 0xef &&
735: (unsigned char)blk.buf[1] == 0xbb &&
736: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 737: offset = 3;
1.93 schwarze 738: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 739: } else
740: offset = 0;
1.93 schwarze 741:
1.95 schwarze 742: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 743:
1.111 schwarze 744: if (--recursion_depth == 0)
1.28 joerg 745: mparse_end(curp);
746:
1.85 schwarze 747: curp->primary = svprimary;
1.28 joerg 748: curp->file = svfile;
749: }
750:
751: enum mandoclevel
1.104 schwarze 752: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 753: const char *file)
754: {
755: struct buf blk;
756:
1.104 schwarze 757: blk.buf = buf;
1.28 joerg 758: blk.sz = len;
759:
1.36 schwarze 760: mparse_parse_buffer(curp, blk, file);
1.28 joerg 761: return(curp->file_status);
762: }
763:
1.98 schwarze 764: /*
765: * Read the whole file into memory and call the parsers.
766: * Called recursively when an .so request is encountered.
767: */
1.36 schwarze 768: enum mandoclevel
769: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 770: {
1.28 joerg 771: struct buf blk;
772: int with_mmap;
1.93 schwarze 773: int save_filenc;
1.1 kristaps 774:
1.91 schwarze 775: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 776: save_filenc = curp->filenc;
777: curp->filenc = curp->options &
778: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 779: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 780: curp->filenc = save_filenc;
1.81 schwarze 781: #if HAVE_MMAP
1.91 schwarze 782: if (with_mmap)
783: munmap(blk.buf, blk.sz);
784: else
1.28 joerg 785: #endif
1.91 schwarze 786: free(blk.buf);
787: }
1.1 kristaps 788:
1.98 schwarze 789: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 790: perror(file);
1.91 schwarze 791:
1.98 schwarze 792: mparse_wait(curp);
1.1 kristaps 793: return(curp->file_status);
1.82 schwarze 794: }
795:
796: enum mandoclevel
1.97 schwarze 797: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 798: {
799: int pfd[2];
1.98 schwarze 800: int save_errno;
1.82 schwarze 801: char *cp;
802:
803: curp->file = file;
1.98 schwarze 804:
805: /* Unless zipped, try to just open the file. */
806:
1.82 schwarze 807: if ((cp = strrchr(file, '.')) == NULL ||
808: strcmp(cp + 1, "gz")) {
1.97 schwarze 809: curp->child = 0;
1.98 schwarze 810: if ((*fd = open(file, O_RDONLY)) != -1)
811: return(MANDOCLEVEL_OK);
812:
813: /* Open failed; try to append ".gz". */
814:
815: mandoc_asprintf(&cp, "%s.gz", file);
816: file = cp;
817: } else
818: cp = NULL;
819:
820: /* Before forking, make sure the file can be read. */
821:
822: save_errno = errno;
823: if (access(file, R_OK) == -1) {
824: if (cp != NULL)
825: errno = save_errno;
1.108 schwarze 826: free(cp);
827: *fd = -1;
828: curp->child = 0;
829: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
830: return(MANDOCLEVEL_ERROR);
1.82 schwarze 831: }
832:
1.98 schwarze 833: /* Run gunzip(1). */
834:
1.82 schwarze 835: if (pipe(pfd) == -1) {
1.107 schwarze 836: perror("pipe");
837: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 838: }
839:
1.97 schwarze 840: switch (curp->child = fork()) {
1.82 schwarze 841: case -1:
1.107 schwarze 842: perror("fork");
843: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 844: case 0:
845: close(pfd[0]);
846: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 847: perror("dup");
848: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 849: }
1.132 ! schwarze 850: signal(SIGPIPE, SIG_DFL);
1.82 schwarze 851: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 852: perror("exec");
853: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 854: default:
855: close(pfd[1]);
856: *fd = pfd[0];
857: return(MANDOCLEVEL_OK);
858: }
859: }
860:
861: enum mandoclevel
1.97 schwarze 862: mparse_wait(struct mparse *curp)
1.82 schwarze 863: {
864: int status;
865:
1.97 schwarze 866: if (curp->child == 0)
867: return(MANDOCLEVEL_OK);
868:
869: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 870: perror("wait");
871: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 872: }
1.129 schwarze 873: curp->child = 0;
1.82 schwarze 874: if (WIFSIGNALED(status)) {
1.108 schwarze 875: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
876: "gunzip died from signal %d", WTERMSIG(status));
877: return(MANDOCLEVEL_ERROR);
1.82 schwarze 878: }
879: if (WEXITSTATUS(status)) {
1.108 schwarze 880: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
881: "gunzip failed with code %d", WEXITSTATUS(status));
882: return(MANDOCLEVEL_ERROR);
1.82 schwarze 883: }
884: return(MANDOCLEVEL_OK);
1.1 kristaps 885: }
886:
887: struct mparse *
1.94 schwarze 888: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
889: const struct mchars *mchars, const char *defos)
1.1 kristaps 890: {
891: struct mparse *curp;
1.10 kristaps 892:
1.1 kristaps 893: curp = mandoc_calloc(1, sizeof(struct mparse));
894:
1.44 schwarze 895: curp->options = options;
1.3 kristaps 896: curp->wlevel = wlevel;
1.1 kristaps 897: curp->mmsg = mmsg;
1.29 schwarze 898: curp->defos = defos;
1.1 kristaps 899:
1.94 schwarze 900: curp->mchars = mchars;
901: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 902: if (curp->options & MPARSE_MDOC)
903: curp->pmdoc = mdoc_alloc(
904: curp->roff, curp, curp->defos,
905: curp->options & MPARSE_QUICK ? 1 : 0);
906: if (curp->options & MPARSE_MAN)
1.106 schwarze 907: curp->pman = man_alloc(
908: curp->roff, curp, curp->defos,
1.72 schwarze 909: curp->options & MPARSE_QUICK ? 1 : 0);
910:
1.1 kristaps 911: return(curp);
912: }
913:
914: void
915: mparse_reset(struct mparse *curp)
916: {
917:
918: roff_reset(curp->roff);
919:
920: if (curp->mdoc)
921: mdoc_reset(curp->mdoc);
922: if (curp->man)
923: man_reset(curp->man);
1.24 kristaps 924: if (curp->secondary)
925: curp->secondary->sz = 0;
1.1 kristaps 926:
927: curp->file_status = MANDOCLEVEL_OK;
928: curp->mdoc = NULL;
929: curp->man = NULL;
1.45 schwarze 930:
931: free(curp->sodest);
932: curp->sodest = NULL;
1.1 kristaps 933: }
934:
935: void
936: mparse_free(struct mparse *curp)
937: {
938:
939: if (curp->pmdoc)
940: mdoc_free(curp->pmdoc);
941: if (curp->pman)
942: man_free(curp->pman);
943: if (curp->roff)
944: roff_free(curp->roff);
1.24 kristaps 945: if (curp->secondary)
946: free(curp->secondary->buf);
1.1 kristaps 947:
1.24 kristaps 948: free(curp->secondary);
1.45 schwarze 949: free(curp->sodest);
1.1 kristaps 950: free(curp);
951: }
952:
953: void
1.45 schwarze 954: mparse_result(struct mparse *curp,
955: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 956: {
957:
1.45 schwarze 958: if (sodest && NULL != (*sodest = curp->sodest)) {
959: *mdoc = NULL;
960: *man = NULL;
961: return;
962: }
1.9 kristaps 963: if (mdoc)
964: *mdoc = curp->mdoc;
965: if (man)
966: *man = curp->man;
1.3 kristaps 967: }
968:
969: void
970: mandoc_vmsg(enum mandocerr t, struct mparse *m,
971: int ln, int pos, const char *fmt, ...)
972: {
973: char buf[256];
974: va_list ap;
975:
976: va_start(ap, fmt);
1.48 schwarze 977: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 978: va_end(ap);
979:
980: mandoc_msg(t, m, ln, pos, buf);
981: }
982:
983: void
1.47 schwarze 984: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 985: int ln, int col, const char *msg)
986: {
987: enum mandoclevel level;
988:
1.112 schwarze 989: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 990: while (er < mandoclimits[level])
991: level--;
992:
1.108 schwarze 993: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 994: return;
995:
1.8 kristaps 996: if (m->mmsg)
997: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 998:
999: if (m->file_status < level)
1000: m->file_status = level;
1.7 kristaps 1001: }
1002:
1003: const char *
1004: mparse_strerror(enum mandocerr er)
1005: {
1006:
1007: return(mandocerrs[er]);
1008: }
1009:
1010: const char *
1011: mparse_strlevel(enum mandoclevel lvl)
1012: {
1013: return(mandoclevels[lvl]);
1.24 kristaps 1014: }
1015:
1016: void
1017: mparse_keep(struct mparse *p)
1018: {
1019:
1020: assert(NULL == p->secondary);
1021: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1022: }
1023:
1024: const char *
1025: mparse_getkeep(const struct mparse *p)
1026: {
1027:
1028: assert(p->secondary);
1029: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1030: }
CVSweb