Annotation of mandoc/read.c, Revision 1.112
1.112 ! schwarze 1: /* $Id: read.c,v 1.111 2015/01/15 04:26:40 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>
1.3 kristaps 32: #include <stdarg.h>
1.28 joerg 33: #include <stdint.h>
1.1 kristaps 34: #include <stdio.h>
35: #include <stdlib.h>
36: #include <string.h>
37: #include <unistd.h>
38:
39: #include "mandoc.h"
1.46 schwarze 40: #include "mandoc_aux.h"
1.3 kristaps 41: #include "libmandoc.h"
1.1 kristaps 42: #include "mdoc.h"
43: #include "man.h"
44:
45: #define REPARSE_LIMIT 1000
46:
47: struct mparse {
48: struct man *pman; /* persistent man parser */
49: struct mdoc *pmdoc; /* persistent mdoc parser */
50: struct man *man; /* man parser */
51: struct mdoc *mdoc; /* mdoc parser */
52: struct roff *roff; /* roff parser (!NULL) */
1.94 schwarze 53: const struct mchars *mchars; /* character table */
1.45 schwarze 54: char *sodest; /* filename pointed to by .so */
1.83 schwarze 55: const char *file; /* filename of current input file */
56: struct buf *primary; /* buffer currently being parsed */
57: struct buf *secondary; /* preprocessed copy of input */
58: const char *defos; /* default operating system */
59: mandocmsg mmsg; /* warning/error message handler */
60: enum mandoclevel file_status; /* status of current parse */
61: enum mandoclevel wlevel; /* ignore messages below this */
62: int options; /* parser options */
1.93 schwarze 63: int filenc; /* encoding of the current file */
1.1 kristaps 64: int reparse_count; /* finite interp. stack */
1.83 schwarze 65: int line; /* line number in the file */
1.97 schwarze 66: pid_t child; /* the gunzip(1) process */
1.1 kristaps 67: };
68:
1.84 schwarze 69: static void choose_parser(struct mparse *);
1.1 kristaps 70: static void resize_buf(struct buf *, size_t);
1.95 schwarze 71: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 72: static int read_whole_file(struct mparse *, const char *, int,
73: struct buf *, int *);
1.1 kristaps 74: static void mparse_end(struct mparse *);
1.37 schwarze 75: static void mparse_parse_buffer(struct mparse *, struct buf,
76: const char *);
1.1 kristaps 77:
1.3 kristaps 78: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
79: MANDOCERR_OK,
80: MANDOCERR_WARNING,
81: MANDOCERR_WARNING,
82: MANDOCERR_ERROR,
1.112 ! schwarze 83: MANDOCERR_UNSUPP,
1.3 kristaps 84: MANDOCERR_MAX,
85: MANDOCERR_MAX
86: };
87:
1.7 kristaps 88: static const char * const mandocerrs[MANDOCERR_MAX] = {
89: "ok",
90:
91: "generic warning",
92:
93: /* related to the prologue */
1.79 schwarze 94: "missing manual title, using UNTITLED",
95: "missing manual title, using \"\"",
1.54 schwarze 96: "lower case character in document title",
1.79 schwarze 97: "missing manual section, using \"\"",
1.7 kristaps 98: "unknown manual section",
1.54 schwarze 99: "missing date, using today's date",
1.7 kristaps 100: "cannot parse date, using it verbatim",
1.79 schwarze 101: "missing Os macro, using \"\"",
102: "duplicate prologue macro",
103: "late prologue macro",
104: "skipping late title macro",
1.7 kristaps 105: "prologue macros out of order",
106:
107: /* related to document structure */
108: ".so is fragile, better use ln(1)",
1.50 schwarze 109: "no document body",
1.54 schwarze 110: "content before first section header",
111: "first section is not \"NAME\"",
1.7 kristaps 112: "bad NAME section contents",
113: "sections out of conventional order",
1.54 schwarze 114: "duplicate section title",
115: "unexpected section",
1.87 schwarze 116: "unusual Xr order",
117: "unusual Xr punctuation",
1.86 schwarze 118: "AUTHORS section without An macro",
1.7 kristaps 119:
120: /* related to macros and nesting */
1.55 schwarze 121: "obsolete macro",
1.102 schwarze 122: "macro neither callable nor escaped",
1.7 kristaps 123: "skipping paragraph macro",
1.31 schwarze 124: "moving paragraph macro out of list",
1.7 kristaps 125: "skipping no-space macro",
126: "blocks badly nested",
127: "nested displays are not portable",
1.57 schwarze 128: "moving content out of list",
129: ".Vt block has child macro",
1.78 schwarze 130: "fill mode already enabled, skipping",
131: "fill mode already disabled, skipping",
1.7 kristaps 132: "line scope broken",
133:
134: /* related to missing macro arguments */
1.58 schwarze 135: "skipping empty request",
136: "conditional request controls empty scope",
1.7 kristaps 137: "skipping empty macro",
1.62 schwarze 138: "empty argument, using 0n",
1.7 kristaps 139: "argument count wrong",
1.60 schwarze 140: "missing display type, using -ragged",
141: "list type is not the first argument",
142: "missing -width in -tag list, using 8n",
1.78 schwarze 143: "missing utility name, using \"\"",
1.60 schwarze 144: "empty head in list item",
145: "empty list item",
1.61 schwarze 146: "missing font type, using \\fR",
147: "unknown font type, using \\fR",
1.103 schwarze 148: "nothing follows prefix",
1.60 schwarze 149: "missing -std argument, adding it",
1.90 schwarze 150: "missing eqn box, using \"\"",
1.7 kristaps 151:
152: /* related to bad macro arguments */
1.64 schwarze 153: "unterminated quoted argument",
1.7 kristaps 154: "duplicate argument",
1.76 schwarze 155: "skipping duplicate argument",
1.63 schwarze 156: "skipping duplicate display type",
157: "skipping duplicate list type",
1.76 schwarze 158: "skipping -width argument",
1.7 kristaps 159: "unknown AT&T UNIX version",
1.88 schwarze 160: "comma in function argument",
1.89 schwarze 161: "parenthesis in function name",
1.67 schwarze 162: "invalid content in Rs block",
1.63 schwarze 163: "invalid Boolean argument",
164: "unknown font, skipping request",
1.7 kristaps 165:
166: /* related to plain text */
1.64 schwarze 167: "blank line in fill mode, using .sp",
168: "tab in filled text",
169: "whitespace at end of input line",
1.7 kristaps 170: "bad comment style",
1.64 schwarze 171: "invalid escape sequence",
172: "undefined string, using \"\"",
1.16 kristaps 173:
1.7 kristaps 174: "generic error",
1.17 kristaps 175:
176: /* related to equations */
1.20 kristaps 177: "unexpected equation scope closure",
178: "equation scope open on exit",
1.21 kristaps 179: "overlapping equation scopes",
180: "unexpected end of equation",
1.7 kristaps 181:
182: /* related to tables */
183: "no table layout cells specified",
184: "no table data cells specified",
185: "ignore data in cell",
186: "data block still open",
187: "ignoring extra data cells",
188:
1.68 schwarze 189: /* related to document structure and macros */
1.108 schwarze 190: NULL,
1.7 kristaps 191: "input stack limit exceeded, infinite loop?",
192: "skipping bad character",
1.68 schwarze 193: "skipping unknown macro",
1.112 ! schwarze 194: "skipping insecure request",
1.73 schwarze 195: "skipping item outside list",
1.68 schwarze 196: "skipping column outside column list",
197: "skipping end of block that is not open",
198: "inserting missing end of block",
199: "appending missing end of block",
200:
201: /* related to request and macro arguments */
1.7 kristaps 202: "escaped character not allowed in a name",
203: "argument count wrong",
1.99 schwarze 204: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 205: "missing list type, using -item",
1.70 schwarze 206: "missing manual name, using \"\"",
1.71 schwarze 207: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 208: "unknown standard specifier",
1.71 schwarze 209: "skipping request without numeric argument",
1.109 schwarze 210: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 211: ".so request failed",
1.61 schwarze 212: "skipping all arguments",
213: "skipping excess arguments",
1.92 kristaps 214: "divide by zero",
1.112 ! schwarze 215:
! 216: "unsupported feature",
! 217: "input too large",
! 218: "unsupported roff request",
! 219: "unsupported table syntax",
! 220: "unsupported table option",
! 221: "unsupported table layout",
! 222: "ignoring macro in table",
1.7 kristaps 223: };
224:
225: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
226: "SUCCESS",
227: "RESERVED",
228: "WARNING",
229: "ERROR",
1.112 ! schwarze 230: "UNSUPP",
1.7 kristaps 231: "BADARG",
232: "SYSERR"
233: };
234:
1.47 schwarze 235:
1.1 kristaps 236: static void
237: resize_buf(struct buf *buf, size_t initial)
238: {
239:
240: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
241: buf->buf = mandoc_realloc(buf->buf, buf->sz);
242: }
243:
244: static void
1.84 schwarze 245: choose_parser(struct mparse *curp)
1.1 kristaps 246: {
1.83 schwarze 247: char *cp, *ep;
248: int format;
1.1 kristaps 249:
1.83 schwarze 250: /*
251: * If neither command line arguments -mdoc or -man select
252: * a parser nor the roff parser found a .Dd or .TH macro
253: * yet, look ahead in the main input buffer.
254: */
255:
256: if ((format = roff_getformat(curp->roff)) == 0) {
257: cp = curp->primary->buf;
258: ep = cp + curp->primary->sz;
259: while (cp < ep) {
1.85 schwarze 260: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 261: cp++;
262: if (cp[0] == 'D' && cp[1] == 'd') {
263: format = MPARSE_MDOC;
264: break;
265: }
266: if (cp[0] == 'T' && cp[1] == 'H') {
267: format = MPARSE_MAN;
268: break;
269: }
270: }
271: cp = memchr(cp, '\n', ep - cp);
272: if (cp == NULL)
273: break;
274: cp++;
275: }
1.1 kristaps 276: }
277:
1.83 schwarze 278: if (format == MPARSE_MDOC) {
1.47 schwarze 279: if (NULL == curp->pmdoc)
1.44 schwarze 280: curp->pmdoc = mdoc_alloc(
281: curp->roff, curp, curp->defos,
282: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 283: assert(curp->pmdoc);
284: curp->mdoc = curp->pmdoc;
285: return;
1.47 schwarze 286: }
1.1 kristaps 287:
1.83 schwarze 288: /* Fall back to man(7) as a last resort. */
289:
1.47 schwarze 290: if (NULL == curp->pman)
1.106 schwarze 291: curp->pman = man_alloc(
292: curp->roff, curp, curp->defos,
1.44 schwarze 293: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 294: assert(curp->pman);
295: curp->man = curp->pman;
296: }
297:
298: /*
1.95 schwarze 299: * Main parse routine for a buffer.
300: * It assumes encoding and line numbering are already set up.
301: * It can recurse directly (for invocations of user-defined
302: * macros, inline equations, and input line traps)
303: * and indirectly (for .so file inclusion).
1.1 kristaps 304: */
305: static void
1.95 schwarze 306: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 307: {
308: const struct tbl_span *span;
309: struct buf ln;
1.110 schwarze 310: char *cp;
1.95 schwarze 311: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 312: enum rofferr rr;
1.100 schwarze 313: int of;
1.1 kristaps 314: int lnn; /* line number in the real file */
1.110 schwarze 315: int fd;
316: pid_t save_child;
1.1 kristaps 317: unsigned char c;
318:
1.95 schwarze 319: memset(&ln, 0, sizeof(ln));
1.1 kristaps 320:
1.47 schwarze 321: lnn = curp->line;
322: pos = 0;
1.1 kristaps 323:
1.95 schwarze 324: while (i < blk.sz) {
1.1 kristaps 325: if (0 == pos && '\0' == blk.buf[i])
326: break;
327:
328: if (start) {
329: curp->line = lnn;
330: curp->reparse_count = 0;
1.93 schwarze 331:
332: if (lnn < 3 &&
333: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 334: curp->filenc & MPARSE_LATIN1)
335: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 336: }
337:
1.95 schwarze 338: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 339:
340: /*
341: * When finding an unescaped newline character,
342: * leave the character loop to process the line.
343: * Skip a preceding carriage return, if any.
344: */
345:
1.95 schwarze 346: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 347: '\n' == blk.buf[i + 1])
348: ++i;
349: if ('\n' == blk.buf[i]) {
350: ++i;
351: ++lnn;
352: break;
353: }
354:
1.35 schwarze 355: /*
1.93 schwarze 356: * Make sure we have space for the worst
357: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 358: */
359:
1.95 schwarze 360: if (pos + 11 > ln.sz)
1.35 schwarze 361: resize_buf(&ln, 256);
362:
1.47 schwarze 363: /*
1.93 schwarze 364: * Encode 8-bit input.
1.1 kristaps 365: */
366:
1.93 schwarze 367: c = blk.buf[i];
368: if (c & 0x80) {
1.95 schwarze 369: if ( ! (curp->filenc && preconv_encode(
370: &blk, &i, &ln, &pos, &curp->filenc))) {
1.93 schwarze 371: mandoc_vmsg(MANDOCERR_BADCHAR,
372: curp, curp->line, pos,
373: "0x%x", c);
374: ln.buf[pos++] = '?';
375: i++;
376: }
377: continue;
378: }
379:
380: /*
381: * Exclude control characters.
382: */
1.1 kristaps 383:
1.93 schwarze 384: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.78 schwarze 385: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
386: curp->line, pos, "0x%x", c);
1.1 kristaps 387: i++;
1.27 joerg 388: ln.buf[pos++] = '?';
1.1 kristaps 389: continue;
390: }
391:
392: /* Trailing backslash = a plain char. */
393:
1.95 schwarze 394: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 395: ln.buf[pos++] = blk.buf[i++];
396: continue;
397: }
398:
399: /*
400: * Found escape and at least one other character.
401: * When it's a newline character, skip it.
402: * When there is a carriage return in between,
403: * skip that one as well.
404: */
405:
1.95 schwarze 406: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 407: '\n' == blk.buf[i + 2])
408: ++i;
409: if ('\n' == blk.buf[i + 1]) {
410: i += 2;
411: ++lnn;
412: continue;
413: }
414:
1.13 kristaps 415: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 416: i += 2;
417: /* Comment, skip to end of line */
1.95 schwarze 418: for (; i < blk.sz; ++i) {
1.1 kristaps 419: if ('\n' == blk.buf[i]) {
420: ++i;
421: ++lnn;
422: break;
423: }
424: }
425:
426: /* Backout trailing whitespaces */
427: for (; pos > 0; --pos) {
428: if (ln.buf[pos - 1] != ' ')
429: break;
430: if (pos > 2 && ln.buf[pos - 2] == '\\')
431: break;
432: }
433: break;
434: }
435:
1.35 schwarze 436: /* Catch escaped bogus characters. */
437:
438: c = (unsigned char) blk.buf[i+1];
439:
1.47 schwarze 440: if ( ! (isascii(c) &&
441: (isgraph(c) || isblank(c)))) {
1.78 schwarze 442: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
443: curp->line, pos, "0x%x", c);
1.35 schwarze 444: i += 2;
445: ln.buf[pos++] = '?';
446: continue;
447: }
448:
1.1 kristaps 449: /* Some other escape sequence, copy & cont. */
450:
451: ln.buf[pos++] = blk.buf[i++];
452: ln.buf[pos++] = blk.buf[i++];
453: }
454:
1.95 schwarze 455: if (pos >= ln.sz)
1.1 kristaps 456: resize_buf(&ln, 256);
457:
458: ln.buf[pos] = '\0';
459:
460: /*
461: * A significant amount of complexity is contained by
462: * the roff preprocessor. It's line-oriented but can be
463: * expressed on one line, so we need at times to
464: * readjust our starting point and re-run it. The roff
465: * preprocessor can also readjust the buffers with new
466: * data, so we pass them in wholesale.
467: */
468:
469: of = 0;
470:
1.24 kristaps 471: /*
472: * Maintain a lookaside buffer of all parsed lines. We
473: * only do this if mparse_keep() has been invoked (the
474: * buffer may be accessed with mparse_getkeep()).
475: */
476:
477: if (curp->secondary) {
1.47 schwarze 478: curp->secondary->buf = mandoc_realloc(
479: curp->secondary->buf,
480: curp->secondary->sz + pos + 2);
481: memcpy(curp->secondary->buf +
482: curp->secondary->sz,
483: ln.buf, pos);
1.24 kristaps 484: curp->secondary->sz += pos;
485: curp->secondary->buf
486: [curp->secondary->sz] = '\n';
487: curp->secondary->sz++;
488: curp->secondary->buf
489: [curp->secondary->sz] = '\0';
490: }
1.1 kristaps 491: rerun:
1.96 schwarze 492: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 493:
494: switch (rr) {
1.47 schwarze 495: case ROFF_REPARSE:
1.1 kristaps 496: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 497: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 498: else
1.3 kristaps 499: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 500: curp->line, pos, NULL);
1.1 kristaps 501: pos = 0;
502: continue;
1.47 schwarze 503: case ROFF_APPEND:
1.95 schwarze 504: pos = strlen(ln.buf);
1.1 kristaps 505: continue;
1.47 schwarze 506: case ROFF_RERUN:
1.1 kristaps 507: goto rerun;
1.47 schwarze 508: case ROFF_IGN:
1.1 kristaps 509: pos = 0;
510: continue;
1.47 schwarze 511: case ROFF_SO:
1.95 schwarze 512: if ( ! (curp->options & MPARSE_SO) &&
513: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 514: curp->sodest = mandoc_strdup(ln.buf + of);
515: free(ln.buf);
516: return;
517: }
1.24 kristaps 518: /*
519: * We remove `so' clauses from our lookaside
520: * buffer because we're going to descend into
521: * the file recursively.
522: */
1.47 schwarze 523: if (curp->secondary)
1.25 kristaps 524: curp->secondary->sz -= pos + 1;
1.110 schwarze 525: save_child = curp->child;
526: if (mparse_open(curp, &fd, ln.buf + of) ==
527: MANDOCLEVEL_OK)
528: mparse_readfd(curp, fd, ln.buf + of);
529: else {
1.52 schwarze 530: mandoc_vmsg(MANDOCERR_SO_FAIL,
531: curp, curp->line, pos,
532: ".so %s", ln.buf + of);
1.110 schwarze 533: ln.sz = mandoc_asprintf(&cp,
534: ".sp\nSee the file %s.\n.sp",
535: ln.buf + of);
536: free(ln.buf);
537: ln.buf = cp;
538: of = 0;
539: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 540: }
1.110 schwarze 541: curp->child = save_child;
1.1 kristaps 542: pos = 0;
543: continue;
544: default:
545: break;
546: }
547:
548: /*
549: * If input parsers have not been allocated, do so now.
1.14 kristaps 550: * We keep these instanced between parsers, but set them
1.1 kristaps 551: * locally per parse routine since we can use different
552: * parsers with each one.
553: */
554:
555: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 556: choose_parser(curp);
1.1 kristaps 557:
1.47 schwarze 558: /*
1.84 schwarze 559: * Lastly, push down into the parsers themselves.
1.1 kristaps 560: * If libroff returns ROFF_TBL, then add it to the
561: * currently open parse. Since we only get here if
562: * there does exist data (see tbl_data.c), we're
563: * guaranteed that something's been allocated.
564: * Do the same for ROFF_EQN.
565: */
566:
1.100 schwarze 567: if (rr == ROFF_TBL) {
568: while ((span = roff_span(curp->roff)) != NULL)
569: if (curp->man == NULL)
570: mdoc_addspan(curp->mdoc, span);
571: else
572: man_addspan(curp->man, span);
573: } else if (rr == ROFF_EQN) {
574: if (curp->man == NULL)
575: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
576: else
577: man_addeqn(curp->man, roff_eqn(curp->roff));
578: } else if ((curp->man == NULL ?
579: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
580: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
581: break;
1.1 kristaps 582:
583: /* Temporary buffers typically are not full. */
584:
585: if (0 == start && '\0' == blk.buf[i])
586: break;
587:
588: /* Start the next input line. */
589:
590: pos = 0;
591: }
592:
593: free(ln.buf);
594: }
595:
596: static int
1.40 schwarze 597: read_whole_file(struct mparse *curp, const char *file, int fd,
598: struct buf *fb, int *with_mmap)
1.1 kristaps 599: {
600: size_t off;
601: ssize_t ssz;
602:
1.81 schwarze 603: #if HAVE_MMAP
1.15 kristaps 604: struct stat st;
1.1 kristaps 605: if (-1 == fstat(fd, &st)) {
1.107 schwarze 606: perror(file);
607: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 608: }
609:
610: /*
611: * If we're a regular file, try just reading in the whole entry
612: * via mmap(). This is faster than reading it into blocks, and
613: * since each file is only a few bytes to begin with, I'm not
614: * concerned that this is going to tank any machines.
615: */
616:
617: if (S_ISREG(st.st_mode)) {
618: if (st.st_size >= (1U << 31)) {
1.111 schwarze 619: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 620: return(0);
621: }
622: *with_mmap = 1;
623: fb->sz = (size_t)st.st_size;
1.37 schwarze 624: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 625: if (fb->buf != MAP_FAILED)
626: return(1);
627: }
1.15 kristaps 628: #endif
1.1 kristaps 629:
630: /*
631: * If this isn't a regular file (like, say, stdin), then we must
632: * go the old way and just read things in bit by bit.
633: */
634:
635: *with_mmap = 0;
636: off = 0;
637: fb->sz = 0;
638: fb->buf = NULL;
639: for (;;) {
640: if (off == fb->sz) {
641: if (fb->sz == (1U << 31)) {
1.111 schwarze 642: mandoc_msg(MANDOCERR_TOOLARGE, curp,
643: 0, 0, NULL);
1.1 kristaps 644: break;
645: }
646: resize_buf(fb, 65536);
647: }
648: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
649: if (ssz == 0) {
650: fb->sz = off;
651: return(1);
652: }
653: if (ssz == -1) {
1.107 schwarze 654: perror(file);
655: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 656: }
657: off += (size_t)ssz;
658: }
659:
660: free(fb->buf);
661: fb->buf = NULL;
662: return(0);
663: }
664:
665: static void
666: mparse_end(struct mparse *curp)
667: {
668:
1.72 schwarze 669: if (curp->mdoc == NULL &&
670: curp->man == NULL &&
671: curp->sodest == NULL) {
672: if (curp->options & MPARSE_MDOC)
673: curp->mdoc = curp->pmdoc;
674: else {
675: if (curp->pman == NULL)
1.106 schwarze 676: curp->pman = man_alloc(
677: curp->roff, curp, curp->defos,
1.72 schwarze 678: curp->options & MPARSE_QUICK ? 1 : 0);
679: curp->man = curp->pman;
680: }
681: }
1.111 schwarze 682: if (curp->mdoc)
683: mdoc_endparse(curp->mdoc);
684: if (curp->man)
685: man_endparse(curp->man);
1.1 kristaps 686: roff_endparse(curp->roff);
687: }
688:
689: static void
1.36 schwarze 690: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 691: {
1.85 schwarze 692: struct buf *svprimary;
1.28 joerg 693: const char *svfile;
1.95 schwarze 694: size_t offset;
1.36 schwarze 695: static int recursion_depth;
696:
697: if (64 < recursion_depth) {
698: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
699: return;
700: }
1.28 joerg 701:
702: /* Line number is per-file. */
703: svfile = curp->file;
704: curp->file = file;
1.85 schwarze 705: svprimary = curp->primary;
1.83 schwarze 706: curp->primary = &blk;
1.28 joerg 707: curp->line = 1;
1.36 schwarze 708: recursion_depth++;
1.28 joerg 709:
1.93 schwarze 710: /* Skip an UTF-8 byte order mark. */
711: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
712: (unsigned char)blk.buf[0] == 0xef &&
713: (unsigned char)blk.buf[1] == 0xbb &&
714: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 715: offset = 3;
1.93 schwarze 716: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 717: } else
718: offset = 0;
1.93 schwarze 719:
1.95 schwarze 720: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 721:
1.111 schwarze 722: if (--recursion_depth == 0)
1.28 joerg 723: mparse_end(curp);
724:
1.85 schwarze 725: curp->primary = svprimary;
1.28 joerg 726: curp->file = svfile;
727: }
728:
729: enum mandoclevel
1.104 schwarze 730: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 731: const char *file)
732: {
733: struct buf blk;
734:
1.104 schwarze 735: blk.buf = buf;
1.28 joerg 736: blk.sz = len;
737:
1.36 schwarze 738: mparse_parse_buffer(curp, blk, file);
1.28 joerg 739: return(curp->file_status);
740: }
741:
1.98 schwarze 742: /*
743: * Read the whole file into memory and call the parsers.
744: * Called recursively when an .so request is encountered.
745: */
1.36 schwarze 746: enum mandoclevel
747: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 748: {
1.28 joerg 749: struct buf blk;
750: int with_mmap;
1.93 schwarze 751: int save_filenc;
1.1 kristaps 752:
1.91 schwarze 753: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 754: save_filenc = curp->filenc;
755: curp->filenc = curp->options &
756: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 757: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 758: curp->filenc = save_filenc;
1.81 schwarze 759: #if HAVE_MMAP
1.91 schwarze 760: if (with_mmap)
761: munmap(blk.buf, blk.sz);
762: else
1.28 joerg 763: #endif
1.91 schwarze 764: free(blk.buf);
765: }
1.1 kristaps 766:
1.98 schwarze 767: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 768: perror(file);
1.91 schwarze 769:
1.98 schwarze 770: mparse_wait(curp);
1.1 kristaps 771: return(curp->file_status);
1.82 schwarze 772: }
773:
774: enum mandoclevel
1.97 schwarze 775: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 776: {
777: int pfd[2];
1.98 schwarze 778: int save_errno;
1.82 schwarze 779: char *cp;
780:
781: curp->file = file;
1.98 schwarze 782:
783: /* Unless zipped, try to just open the file. */
784:
1.82 schwarze 785: if ((cp = strrchr(file, '.')) == NULL ||
786: strcmp(cp + 1, "gz")) {
1.97 schwarze 787: curp->child = 0;
1.98 schwarze 788: if ((*fd = open(file, O_RDONLY)) != -1)
789: return(MANDOCLEVEL_OK);
790:
791: /* Open failed; try to append ".gz". */
792:
793: mandoc_asprintf(&cp, "%s.gz", file);
794: file = cp;
795: } else
796: cp = NULL;
797:
798: /* Before forking, make sure the file can be read. */
799:
800: save_errno = errno;
801: if (access(file, R_OK) == -1) {
802: if (cp != NULL)
803: errno = save_errno;
1.108 schwarze 804: free(cp);
805: *fd = -1;
806: curp->child = 0;
807: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
808: return(MANDOCLEVEL_ERROR);
1.82 schwarze 809: }
810:
1.98 schwarze 811: /* Run gunzip(1). */
812:
1.82 schwarze 813: if (pipe(pfd) == -1) {
1.107 schwarze 814: perror("pipe");
815: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 816: }
817:
1.97 schwarze 818: switch (curp->child = fork()) {
1.82 schwarze 819: case -1:
1.107 schwarze 820: perror("fork");
821: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 822: case 0:
823: close(pfd[0]);
824: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 825: perror("dup");
826: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 827: }
828: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 829: perror("exec");
830: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 831: default:
832: close(pfd[1]);
833: *fd = pfd[0];
834: return(MANDOCLEVEL_OK);
835: }
836: }
837:
838: enum mandoclevel
1.97 schwarze 839: mparse_wait(struct mparse *curp)
1.82 schwarze 840: {
841: int status;
842:
1.97 schwarze 843: if (curp->child == 0)
844: return(MANDOCLEVEL_OK);
845:
846: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 847: perror("wait");
848: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 849: }
850: if (WIFSIGNALED(status)) {
1.108 schwarze 851: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
852: "gunzip died from signal %d", WTERMSIG(status));
853: return(MANDOCLEVEL_ERROR);
1.82 schwarze 854: }
855: if (WEXITSTATUS(status)) {
1.108 schwarze 856: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
857: "gunzip failed with code %d", WEXITSTATUS(status));
858: return(MANDOCLEVEL_ERROR);
1.82 schwarze 859: }
860: return(MANDOCLEVEL_OK);
1.1 kristaps 861: }
862:
863: struct mparse *
1.94 schwarze 864: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
865: const struct mchars *mchars, const char *defos)
1.1 kristaps 866: {
867: struct mparse *curp;
1.10 kristaps 868:
1.1 kristaps 869: curp = mandoc_calloc(1, sizeof(struct mparse));
870:
1.44 schwarze 871: curp->options = options;
1.3 kristaps 872: curp->wlevel = wlevel;
1.1 kristaps 873: curp->mmsg = mmsg;
1.29 schwarze 874: curp->defos = defos;
1.1 kristaps 875:
1.94 schwarze 876: curp->mchars = mchars;
877: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 878: if (curp->options & MPARSE_MDOC)
879: curp->pmdoc = mdoc_alloc(
880: curp->roff, curp, curp->defos,
881: curp->options & MPARSE_QUICK ? 1 : 0);
882: if (curp->options & MPARSE_MAN)
1.106 schwarze 883: curp->pman = man_alloc(
884: curp->roff, curp, curp->defos,
1.72 schwarze 885: curp->options & MPARSE_QUICK ? 1 : 0);
886:
1.1 kristaps 887: return(curp);
888: }
889:
890: void
891: mparse_reset(struct mparse *curp)
892: {
893:
894: roff_reset(curp->roff);
895:
896: if (curp->mdoc)
897: mdoc_reset(curp->mdoc);
898: if (curp->man)
899: man_reset(curp->man);
1.24 kristaps 900: if (curp->secondary)
901: curp->secondary->sz = 0;
1.1 kristaps 902:
903: curp->file_status = MANDOCLEVEL_OK;
904: curp->mdoc = NULL;
905: curp->man = NULL;
1.45 schwarze 906:
907: free(curp->sodest);
908: curp->sodest = NULL;
1.1 kristaps 909: }
910:
911: void
912: mparse_free(struct mparse *curp)
913: {
914:
915: if (curp->pmdoc)
916: mdoc_free(curp->pmdoc);
917: if (curp->pman)
918: man_free(curp->pman);
919: if (curp->roff)
920: roff_free(curp->roff);
1.24 kristaps 921: if (curp->secondary)
922: free(curp->secondary->buf);
1.1 kristaps 923:
1.24 kristaps 924: free(curp->secondary);
1.45 schwarze 925: free(curp->sodest);
1.1 kristaps 926: free(curp);
927: }
928:
929: void
1.45 schwarze 930: mparse_result(struct mparse *curp,
931: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 932: {
933:
1.45 schwarze 934: if (sodest && NULL != (*sodest = curp->sodest)) {
935: *mdoc = NULL;
936: *man = NULL;
937: return;
938: }
1.9 kristaps 939: if (mdoc)
940: *mdoc = curp->mdoc;
941: if (man)
942: *man = curp->man;
1.3 kristaps 943: }
944:
945: void
946: mandoc_vmsg(enum mandocerr t, struct mparse *m,
947: int ln, int pos, const char *fmt, ...)
948: {
949: char buf[256];
950: va_list ap;
951:
952: va_start(ap, fmt);
1.48 schwarze 953: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 954: va_end(ap);
955:
956: mandoc_msg(t, m, ln, pos, buf);
957: }
958:
959: void
1.47 schwarze 960: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 961: int ln, int col, const char *msg)
962: {
963: enum mandoclevel level;
964:
1.112 ! schwarze 965: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 966: while (er < mandoclimits[level])
967: level--;
968:
1.108 schwarze 969: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 970: return;
971:
1.8 kristaps 972: if (m->mmsg)
973: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 974:
975: if (m->file_status < level)
976: m->file_status = level;
1.7 kristaps 977: }
978:
979: const char *
980: mparse_strerror(enum mandocerr er)
981: {
982:
983: return(mandocerrs[er]);
984: }
985:
986: const char *
987: mparse_strlevel(enum mandoclevel lvl)
988: {
989: return(mandoclevels[lvl]);
1.24 kristaps 990: }
991:
992: void
993: mparse_keep(struct mparse *p)
994: {
995:
996: assert(NULL == p->secondary);
997: p->secondary = mandoc_calloc(1, sizeof(struct buf));
998: }
999:
1000: const char *
1001: mparse_getkeep(const struct mparse *p)
1002: {
1003:
1004: assert(p->secondary);
1005: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1006: }
CVSweb