Annotation of mandoc/read.c, Revision 1.113
1.113 ! schwarze 1: /* $Id: read.c,v 1.112 2015/01/20 21:16:51 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;
1.113 ! schwarze 310: const char *save_file;
1.110 schwarze 311: char *cp;
1.95 schwarze 312: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 313: enum rofferr rr;
1.100 schwarze 314: int of;
1.1 kristaps 315: int lnn; /* line number in the real file */
1.110 schwarze 316: int fd;
317: pid_t save_child;
1.1 kristaps 318: unsigned char c;
319:
1.95 schwarze 320: memset(&ln, 0, sizeof(ln));
1.1 kristaps 321:
1.47 schwarze 322: lnn = curp->line;
323: pos = 0;
1.1 kristaps 324:
1.95 schwarze 325: while (i < blk.sz) {
1.1 kristaps 326: if (0 == pos && '\0' == blk.buf[i])
327: break;
328:
329: if (start) {
330: curp->line = lnn;
331: curp->reparse_count = 0;
1.93 schwarze 332:
333: if (lnn < 3 &&
334: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 335: curp->filenc & MPARSE_LATIN1)
336: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 337: }
338:
1.95 schwarze 339: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 340:
341: /*
342: * When finding an unescaped newline character,
343: * leave the character loop to process the line.
344: * Skip a preceding carriage return, if any.
345: */
346:
1.95 schwarze 347: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 348: '\n' == blk.buf[i + 1])
349: ++i;
350: if ('\n' == blk.buf[i]) {
351: ++i;
352: ++lnn;
353: break;
354: }
355:
1.35 schwarze 356: /*
1.93 schwarze 357: * Make sure we have space for the worst
358: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 359: */
360:
1.95 schwarze 361: if (pos + 11 > ln.sz)
1.35 schwarze 362: resize_buf(&ln, 256);
363:
1.47 schwarze 364: /*
1.93 schwarze 365: * Encode 8-bit input.
1.1 kristaps 366: */
367:
1.93 schwarze 368: c = blk.buf[i];
369: if (c & 0x80) {
1.95 schwarze 370: if ( ! (curp->filenc && preconv_encode(
371: &blk, &i, &ln, &pos, &curp->filenc))) {
1.93 schwarze 372: mandoc_vmsg(MANDOCERR_BADCHAR,
373: curp, curp->line, pos,
374: "0x%x", c);
375: ln.buf[pos++] = '?';
376: i++;
377: }
378: continue;
379: }
380:
381: /*
382: * Exclude control characters.
383: */
1.1 kristaps 384:
1.93 schwarze 385: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.78 schwarze 386: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
387: curp->line, pos, "0x%x", c);
1.1 kristaps 388: i++;
1.27 joerg 389: ln.buf[pos++] = '?';
1.1 kristaps 390: continue;
391: }
392:
393: /* Trailing backslash = a plain char. */
394:
1.95 schwarze 395: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 396: ln.buf[pos++] = blk.buf[i++];
397: continue;
398: }
399:
400: /*
401: * Found escape and at least one other character.
402: * When it's a newline character, skip it.
403: * When there is a carriage return in between,
404: * skip that one as well.
405: */
406:
1.95 schwarze 407: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 408: '\n' == blk.buf[i + 2])
409: ++i;
410: if ('\n' == blk.buf[i + 1]) {
411: i += 2;
412: ++lnn;
413: continue;
414: }
415:
1.13 kristaps 416: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 417: i += 2;
418: /* Comment, skip to end of line */
1.95 schwarze 419: for (; i < blk.sz; ++i) {
1.1 kristaps 420: if ('\n' == blk.buf[i]) {
421: ++i;
422: ++lnn;
423: break;
424: }
425: }
426:
427: /* Backout trailing whitespaces */
428: for (; pos > 0; --pos) {
429: if (ln.buf[pos - 1] != ' ')
430: break;
431: if (pos > 2 && ln.buf[pos - 2] == '\\')
432: break;
433: }
434: break;
435: }
436:
1.35 schwarze 437: /* Catch escaped bogus characters. */
438:
439: c = (unsigned char) blk.buf[i+1];
440:
1.47 schwarze 441: if ( ! (isascii(c) &&
442: (isgraph(c) || isblank(c)))) {
1.78 schwarze 443: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
444: curp->line, pos, "0x%x", c);
1.35 schwarze 445: i += 2;
446: ln.buf[pos++] = '?';
447: continue;
448: }
449:
1.1 kristaps 450: /* Some other escape sequence, copy & cont. */
451:
452: ln.buf[pos++] = blk.buf[i++];
453: ln.buf[pos++] = blk.buf[i++];
454: }
455:
1.95 schwarze 456: if (pos >= ln.sz)
1.1 kristaps 457: resize_buf(&ln, 256);
458:
459: ln.buf[pos] = '\0';
460:
461: /*
462: * A significant amount of complexity is contained by
463: * the roff preprocessor. It's line-oriented but can be
464: * expressed on one line, so we need at times to
465: * readjust our starting point and re-run it. The roff
466: * preprocessor can also readjust the buffers with new
467: * data, so we pass them in wholesale.
468: */
469:
470: of = 0;
471:
1.24 kristaps 472: /*
473: * Maintain a lookaside buffer of all parsed lines. We
474: * only do this if mparse_keep() has been invoked (the
475: * buffer may be accessed with mparse_getkeep()).
476: */
477:
478: if (curp->secondary) {
1.47 schwarze 479: curp->secondary->buf = mandoc_realloc(
480: curp->secondary->buf,
481: curp->secondary->sz + pos + 2);
482: memcpy(curp->secondary->buf +
483: curp->secondary->sz,
484: ln.buf, pos);
1.24 kristaps 485: curp->secondary->sz += pos;
486: curp->secondary->buf
487: [curp->secondary->sz] = '\n';
488: curp->secondary->sz++;
489: curp->secondary->buf
490: [curp->secondary->sz] = '\0';
491: }
1.1 kristaps 492: rerun:
1.96 schwarze 493: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 494:
495: switch (rr) {
1.47 schwarze 496: case ROFF_REPARSE:
1.1 kristaps 497: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 498: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 499: else
1.3 kristaps 500: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 501: curp->line, pos, NULL);
1.1 kristaps 502: pos = 0;
503: continue;
1.47 schwarze 504: case ROFF_APPEND:
1.95 schwarze 505: pos = strlen(ln.buf);
1.1 kristaps 506: continue;
1.47 schwarze 507: case ROFF_RERUN:
1.1 kristaps 508: goto rerun;
1.47 schwarze 509: case ROFF_IGN:
1.1 kristaps 510: pos = 0;
511: continue;
1.47 schwarze 512: case ROFF_SO:
1.95 schwarze 513: if ( ! (curp->options & MPARSE_SO) &&
514: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 515: curp->sodest = mandoc_strdup(ln.buf + of);
516: free(ln.buf);
517: return;
518: }
1.24 kristaps 519: /*
520: * We remove `so' clauses from our lookaside
521: * buffer because we're going to descend into
522: * the file recursively.
523: */
1.47 schwarze 524: if (curp->secondary)
1.25 kristaps 525: curp->secondary->sz -= pos + 1;
1.113 ! schwarze 526: save_file = curp->file;
1.110 schwarze 527: save_child = curp->child;
528: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 ! schwarze 529: MANDOCLEVEL_OK) {
1.110 schwarze 530: mparse_readfd(curp, fd, ln.buf + of);
1.113 ! schwarze 531: curp->file = save_file;
! 532: } else {
! 533: curp->file = save_file;
1.52 schwarze 534: mandoc_vmsg(MANDOCERR_SO_FAIL,
535: curp, curp->line, pos,
536: ".so %s", ln.buf + of);
1.110 schwarze 537: ln.sz = mandoc_asprintf(&cp,
538: ".sp\nSee the file %s.\n.sp",
539: ln.buf + of);
540: free(ln.buf);
541: ln.buf = cp;
542: of = 0;
543: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 544: }
1.110 schwarze 545: curp->child = save_child;
1.1 kristaps 546: pos = 0;
547: continue;
548: default:
549: break;
550: }
551:
552: /*
553: * If input parsers have not been allocated, do so now.
1.14 kristaps 554: * We keep these instanced between parsers, but set them
1.1 kristaps 555: * locally per parse routine since we can use different
556: * parsers with each one.
557: */
558:
559: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 560: choose_parser(curp);
1.1 kristaps 561:
1.47 schwarze 562: /*
1.84 schwarze 563: * Lastly, push down into the parsers themselves.
1.1 kristaps 564: * If libroff returns ROFF_TBL, then add it to the
565: * currently open parse. Since we only get here if
566: * there does exist data (see tbl_data.c), we're
567: * guaranteed that something's been allocated.
568: * Do the same for ROFF_EQN.
569: */
570:
1.100 schwarze 571: if (rr == ROFF_TBL) {
572: while ((span = roff_span(curp->roff)) != NULL)
573: if (curp->man == NULL)
574: mdoc_addspan(curp->mdoc, span);
575: else
576: man_addspan(curp->man, span);
577: } else if (rr == ROFF_EQN) {
578: if (curp->man == NULL)
579: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
580: else
581: man_addeqn(curp->man, roff_eqn(curp->roff));
582: } else if ((curp->man == NULL ?
583: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
584: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
585: break;
1.1 kristaps 586:
587: /* Temporary buffers typically are not full. */
588:
589: if (0 == start && '\0' == blk.buf[i])
590: break;
591:
592: /* Start the next input line. */
593:
594: pos = 0;
595: }
596:
597: free(ln.buf);
598: }
599:
600: static int
1.40 schwarze 601: read_whole_file(struct mparse *curp, const char *file, int fd,
602: struct buf *fb, int *with_mmap)
1.1 kristaps 603: {
604: size_t off;
605: ssize_t ssz;
606:
1.81 schwarze 607: #if HAVE_MMAP
1.15 kristaps 608: struct stat st;
1.1 kristaps 609: if (-1 == fstat(fd, &st)) {
1.107 schwarze 610: perror(file);
611: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 612: }
613:
614: /*
615: * If we're a regular file, try just reading in the whole entry
616: * via mmap(). This is faster than reading it into blocks, and
617: * since each file is only a few bytes to begin with, I'm not
618: * concerned that this is going to tank any machines.
619: */
620:
621: if (S_ISREG(st.st_mode)) {
622: if (st.st_size >= (1U << 31)) {
1.111 schwarze 623: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 624: return(0);
625: }
626: *with_mmap = 1;
627: fb->sz = (size_t)st.st_size;
1.37 schwarze 628: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 629: if (fb->buf != MAP_FAILED)
630: return(1);
631: }
1.15 kristaps 632: #endif
1.1 kristaps 633:
634: /*
635: * If this isn't a regular file (like, say, stdin), then we must
636: * go the old way and just read things in bit by bit.
637: */
638:
639: *with_mmap = 0;
640: off = 0;
641: fb->sz = 0;
642: fb->buf = NULL;
643: for (;;) {
644: if (off == fb->sz) {
645: if (fb->sz == (1U << 31)) {
1.111 schwarze 646: mandoc_msg(MANDOCERR_TOOLARGE, curp,
647: 0, 0, NULL);
1.1 kristaps 648: break;
649: }
650: resize_buf(fb, 65536);
651: }
652: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
653: if (ssz == 0) {
654: fb->sz = off;
655: return(1);
656: }
657: if (ssz == -1) {
1.107 schwarze 658: perror(file);
659: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 660: }
661: off += (size_t)ssz;
662: }
663:
664: free(fb->buf);
665: fb->buf = NULL;
666: return(0);
667: }
668:
669: static void
670: mparse_end(struct mparse *curp)
671: {
672:
1.72 schwarze 673: if (curp->mdoc == NULL &&
674: curp->man == NULL &&
675: curp->sodest == NULL) {
676: if (curp->options & MPARSE_MDOC)
677: curp->mdoc = curp->pmdoc;
678: else {
679: if (curp->pman == NULL)
1.106 schwarze 680: curp->pman = man_alloc(
681: curp->roff, curp, curp->defos,
1.72 schwarze 682: curp->options & MPARSE_QUICK ? 1 : 0);
683: curp->man = curp->pman;
684: }
685: }
1.111 schwarze 686: if (curp->mdoc)
687: mdoc_endparse(curp->mdoc);
688: if (curp->man)
689: man_endparse(curp->man);
1.1 kristaps 690: roff_endparse(curp->roff);
691: }
692:
693: static void
1.36 schwarze 694: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 695: {
1.85 schwarze 696: struct buf *svprimary;
1.28 joerg 697: const char *svfile;
1.95 schwarze 698: size_t offset;
1.36 schwarze 699: static int recursion_depth;
700:
701: if (64 < recursion_depth) {
702: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
703: return;
704: }
1.28 joerg 705:
706: /* Line number is per-file. */
707: svfile = curp->file;
708: curp->file = file;
1.85 schwarze 709: svprimary = curp->primary;
1.83 schwarze 710: curp->primary = &blk;
1.28 joerg 711: curp->line = 1;
1.36 schwarze 712: recursion_depth++;
1.28 joerg 713:
1.93 schwarze 714: /* Skip an UTF-8 byte order mark. */
715: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
716: (unsigned char)blk.buf[0] == 0xef &&
717: (unsigned char)blk.buf[1] == 0xbb &&
718: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 719: offset = 3;
1.93 schwarze 720: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 721: } else
722: offset = 0;
1.93 schwarze 723:
1.95 schwarze 724: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 725:
1.111 schwarze 726: if (--recursion_depth == 0)
1.28 joerg 727: mparse_end(curp);
728:
1.85 schwarze 729: curp->primary = svprimary;
1.28 joerg 730: curp->file = svfile;
731: }
732:
733: enum mandoclevel
1.104 schwarze 734: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 735: const char *file)
736: {
737: struct buf blk;
738:
1.104 schwarze 739: blk.buf = buf;
1.28 joerg 740: blk.sz = len;
741:
1.36 schwarze 742: mparse_parse_buffer(curp, blk, file);
1.28 joerg 743: return(curp->file_status);
744: }
745:
1.98 schwarze 746: /*
747: * Read the whole file into memory and call the parsers.
748: * Called recursively when an .so request is encountered.
749: */
1.36 schwarze 750: enum mandoclevel
751: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 752: {
1.28 joerg 753: struct buf blk;
754: int with_mmap;
1.93 schwarze 755: int save_filenc;
1.1 kristaps 756:
1.91 schwarze 757: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 758: save_filenc = curp->filenc;
759: curp->filenc = curp->options &
760: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 761: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 762: curp->filenc = save_filenc;
1.81 schwarze 763: #if HAVE_MMAP
1.91 schwarze 764: if (with_mmap)
765: munmap(blk.buf, blk.sz);
766: else
1.28 joerg 767: #endif
1.91 schwarze 768: free(blk.buf);
769: }
1.1 kristaps 770:
1.98 schwarze 771: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 772: perror(file);
1.91 schwarze 773:
1.98 schwarze 774: mparse_wait(curp);
1.1 kristaps 775: return(curp->file_status);
1.82 schwarze 776: }
777:
778: enum mandoclevel
1.97 schwarze 779: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 780: {
781: int pfd[2];
1.98 schwarze 782: int save_errno;
1.82 schwarze 783: char *cp;
784:
785: curp->file = file;
1.98 schwarze 786:
787: /* Unless zipped, try to just open the file. */
788:
1.82 schwarze 789: if ((cp = strrchr(file, '.')) == NULL ||
790: strcmp(cp + 1, "gz")) {
1.97 schwarze 791: curp->child = 0;
1.98 schwarze 792: if ((*fd = open(file, O_RDONLY)) != -1)
793: return(MANDOCLEVEL_OK);
794:
795: /* Open failed; try to append ".gz". */
796:
797: mandoc_asprintf(&cp, "%s.gz", file);
798: file = cp;
799: } else
800: cp = NULL;
801:
802: /* Before forking, make sure the file can be read. */
803:
804: save_errno = errno;
805: if (access(file, R_OK) == -1) {
806: if (cp != NULL)
807: errno = save_errno;
1.108 schwarze 808: free(cp);
809: *fd = -1;
810: curp->child = 0;
811: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
812: return(MANDOCLEVEL_ERROR);
1.82 schwarze 813: }
814:
1.98 schwarze 815: /* Run gunzip(1). */
816:
1.82 schwarze 817: if (pipe(pfd) == -1) {
1.107 schwarze 818: perror("pipe");
819: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 820: }
821:
1.97 schwarze 822: switch (curp->child = fork()) {
1.82 schwarze 823: case -1:
1.107 schwarze 824: perror("fork");
825: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 826: case 0:
827: close(pfd[0]);
828: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 829: perror("dup");
830: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 831: }
832: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 833: perror("exec");
834: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 835: default:
836: close(pfd[1]);
837: *fd = pfd[0];
838: return(MANDOCLEVEL_OK);
839: }
840: }
841:
842: enum mandoclevel
1.97 schwarze 843: mparse_wait(struct mparse *curp)
1.82 schwarze 844: {
845: int status;
846:
1.97 schwarze 847: if (curp->child == 0)
848: return(MANDOCLEVEL_OK);
849:
850: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 851: perror("wait");
852: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 853: }
854: if (WIFSIGNALED(status)) {
1.108 schwarze 855: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
856: "gunzip died from signal %d", WTERMSIG(status));
857: return(MANDOCLEVEL_ERROR);
1.82 schwarze 858: }
859: if (WEXITSTATUS(status)) {
1.108 schwarze 860: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
861: "gunzip failed with code %d", WEXITSTATUS(status));
862: return(MANDOCLEVEL_ERROR);
1.82 schwarze 863: }
864: return(MANDOCLEVEL_OK);
1.1 kristaps 865: }
866:
867: struct mparse *
1.94 schwarze 868: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
869: const struct mchars *mchars, const char *defos)
1.1 kristaps 870: {
871: struct mparse *curp;
1.10 kristaps 872:
1.1 kristaps 873: curp = mandoc_calloc(1, sizeof(struct mparse));
874:
1.44 schwarze 875: curp->options = options;
1.3 kristaps 876: curp->wlevel = wlevel;
1.1 kristaps 877: curp->mmsg = mmsg;
1.29 schwarze 878: curp->defos = defos;
1.1 kristaps 879:
1.94 schwarze 880: curp->mchars = mchars;
881: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 882: if (curp->options & MPARSE_MDOC)
883: curp->pmdoc = mdoc_alloc(
884: curp->roff, curp, curp->defos,
885: curp->options & MPARSE_QUICK ? 1 : 0);
886: if (curp->options & MPARSE_MAN)
1.106 schwarze 887: curp->pman = man_alloc(
888: curp->roff, curp, curp->defos,
1.72 schwarze 889: curp->options & MPARSE_QUICK ? 1 : 0);
890:
1.1 kristaps 891: return(curp);
892: }
893:
894: void
895: mparse_reset(struct mparse *curp)
896: {
897:
898: roff_reset(curp->roff);
899:
900: if (curp->mdoc)
901: mdoc_reset(curp->mdoc);
902: if (curp->man)
903: man_reset(curp->man);
1.24 kristaps 904: if (curp->secondary)
905: curp->secondary->sz = 0;
1.1 kristaps 906:
907: curp->file_status = MANDOCLEVEL_OK;
908: curp->mdoc = NULL;
909: curp->man = NULL;
1.45 schwarze 910:
911: free(curp->sodest);
912: curp->sodest = NULL;
1.1 kristaps 913: }
914:
915: void
916: mparse_free(struct mparse *curp)
917: {
918:
919: if (curp->pmdoc)
920: mdoc_free(curp->pmdoc);
921: if (curp->pman)
922: man_free(curp->pman);
923: if (curp->roff)
924: roff_free(curp->roff);
1.24 kristaps 925: if (curp->secondary)
926: free(curp->secondary->buf);
1.1 kristaps 927:
1.24 kristaps 928: free(curp->secondary);
1.45 schwarze 929: free(curp->sodest);
1.1 kristaps 930: free(curp);
931: }
932:
933: void
1.45 schwarze 934: mparse_result(struct mparse *curp,
935: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 936: {
937:
1.45 schwarze 938: if (sodest && NULL != (*sodest = curp->sodest)) {
939: *mdoc = NULL;
940: *man = NULL;
941: return;
942: }
1.9 kristaps 943: if (mdoc)
944: *mdoc = curp->mdoc;
945: if (man)
946: *man = curp->man;
1.3 kristaps 947: }
948:
949: void
950: mandoc_vmsg(enum mandocerr t, struct mparse *m,
951: int ln, int pos, const char *fmt, ...)
952: {
953: char buf[256];
954: va_list ap;
955:
956: va_start(ap, fmt);
1.48 schwarze 957: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 958: va_end(ap);
959:
960: mandoc_msg(t, m, ln, pos, buf);
961: }
962:
963: void
1.47 schwarze 964: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 965: int ln, int col, const char *msg)
966: {
967: enum mandoclevel level;
968:
1.112 schwarze 969: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 970: while (er < mandoclimits[level])
971: level--;
972:
1.108 schwarze 973: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 974: return;
975:
1.8 kristaps 976: if (m->mmsg)
977: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 978:
979: if (m->file_status < level)
980: m->file_status = level;
1.7 kristaps 981: }
982:
983: const char *
984: mparse_strerror(enum mandocerr er)
985: {
986:
987: return(mandocerrs[er]);
988: }
989:
990: const char *
991: mparse_strlevel(enum mandoclevel lvl)
992: {
993: return(mandoclevels[lvl]);
1.24 kristaps 994: }
995:
996: void
997: mparse_keep(struct mparse *p)
998: {
999:
1000: assert(NULL == p->secondary);
1001: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1002: }
1003:
1004: const char *
1005: mparse_getkeep(const struct mparse *p)
1006: {
1007:
1008: assert(p->secondary);
1009: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1010: }
CVSweb