Annotation of mandoc/read.c, Revision 1.118
1.118 ! schwarze 1: /* $Id: read.c,v 1.117 2015/01/26 13:03:48 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:
1.118 ! schwarze 174: /* related to tables */
! 175: "tbl line starts with span",
! 176: "tbl column starts with span",
! 177: "skipping vertical bar in tbl layout",
! 178:
1.7 kristaps 179: "generic error",
1.17 kristaps 180:
181: /* related to equations */
1.20 kristaps 182: "unexpected equation scope closure",
183: "equation scope open on exit",
1.21 kristaps 184: "overlapping equation scopes",
185: "unexpected end of equation",
1.7 kristaps 186:
187: /* related to tables */
1.116 schwarze 188: "non-alphabetic character in tbl options",
189: "skipping unknown tbl option",
190: "missing tbl option argument",
191: "wrong tbl option argument size",
1.118 ! schwarze 192: "empty tbl layout",
! 193: "invalid character in tbl layout",
! 194: "unmatched parenthesis in tbl layout",
1.7 kristaps 195: "no table data cells specified",
196: "ignore data in cell",
197: "data block still open",
198: "ignoring extra data cells",
199:
1.68 schwarze 200: /* related to document structure and macros */
1.108 schwarze 201: NULL,
1.7 kristaps 202: "input stack limit exceeded, infinite loop?",
203: "skipping bad character",
1.68 schwarze 204: "skipping unknown macro",
1.112 schwarze 205: "skipping insecure request",
1.73 schwarze 206: "skipping item outside list",
1.68 schwarze 207: "skipping column outside column list",
208: "skipping end of block that is not open",
1.115 schwarze 209: "fewer RS blocks open, skipping",
1.68 schwarze 210: "inserting missing end of block",
211: "appending missing end of block",
212:
213: /* related to request and macro arguments */
1.7 kristaps 214: "escaped character not allowed in a name",
215: "argument count wrong",
1.99 schwarze 216: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 217: "missing list type, using -item",
1.70 schwarze 218: "missing manual name, using \"\"",
1.71 schwarze 219: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 220: "unknown standard specifier",
1.71 schwarze 221: "skipping request without numeric argument",
1.109 schwarze 222: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 223: ".so request failed",
1.61 schwarze 224: "skipping all arguments",
225: "skipping excess arguments",
1.92 kristaps 226: "divide by zero",
1.112 schwarze 227:
228: "unsupported feature",
229: "input too large",
1.114 schwarze 230: "unsupported control character",
1.112 schwarze 231: "unsupported roff request",
1.118 ! schwarze 232: "unsupported tbl layout modifier",
1.112 schwarze 233: "ignoring macro in table",
1.117 schwarze 234: "eqn in tbl",
1.7 kristaps 235: };
236:
237: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
238: "SUCCESS",
239: "RESERVED",
240: "WARNING",
241: "ERROR",
1.112 schwarze 242: "UNSUPP",
1.7 kristaps 243: "BADARG",
244: "SYSERR"
245: };
246:
1.47 schwarze 247:
1.1 kristaps 248: static void
249: resize_buf(struct buf *buf, size_t initial)
250: {
251:
252: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
253: buf->buf = mandoc_realloc(buf->buf, buf->sz);
254: }
255:
256: static void
1.84 schwarze 257: choose_parser(struct mparse *curp)
1.1 kristaps 258: {
1.83 schwarze 259: char *cp, *ep;
260: int format;
1.1 kristaps 261:
1.83 schwarze 262: /*
263: * If neither command line arguments -mdoc or -man select
264: * a parser nor the roff parser found a .Dd or .TH macro
265: * yet, look ahead in the main input buffer.
266: */
267:
268: if ((format = roff_getformat(curp->roff)) == 0) {
269: cp = curp->primary->buf;
270: ep = cp + curp->primary->sz;
271: while (cp < ep) {
1.85 schwarze 272: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 273: cp++;
274: if (cp[0] == 'D' && cp[1] == 'd') {
275: format = MPARSE_MDOC;
276: break;
277: }
278: if (cp[0] == 'T' && cp[1] == 'H') {
279: format = MPARSE_MAN;
280: break;
281: }
282: }
283: cp = memchr(cp, '\n', ep - cp);
284: if (cp == NULL)
285: break;
286: cp++;
287: }
1.1 kristaps 288: }
289:
1.83 schwarze 290: if (format == MPARSE_MDOC) {
1.47 schwarze 291: if (NULL == curp->pmdoc)
1.44 schwarze 292: curp->pmdoc = mdoc_alloc(
293: curp->roff, curp, curp->defos,
294: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 295: assert(curp->pmdoc);
296: curp->mdoc = curp->pmdoc;
297: return;
1.47 schwarze 298: }
1.1 kristaps 299:
1.83 schwarze 300: /* Fall back to man(7) as a last resort. */
301:
1.47 schwarze 302: if (NULL == curp->pman)
1.106 schwarze 303: curp->pman = man_alloc(
304: curp->roff, curp, curp->defos,
1.44 schwarze 305: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 306: assert(curp->pman);
307: curp->man = curp->pman;
308: }
309:
310: /*
1.95 schwarze 311: * Main parse routine for a buffer.
312: * It assumes encoding and line numbering are already set up.
313: * It can recurse directly (for invocations of user-defined
314: * macros, inline equations, and input line traps)
315: * and indirectly (for .so file inclusion).
1.1 kristaps 316: */
317: static void
1.95 schwarze 318: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 319: {
320: const struct tbl_span *span;
321: struct buf ln;
1.113 schwarze 322: const char *save_file;
1.110 schwarze 323: char *cp;
1.95 schwarze 324: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 325: enum rofferr rr;
1.100 schwarze 326: int of;
1.1 kristaps 327: int lnn; /* line number in the real file */
1.110 schwarze 328: int fd;
329: pid_t save_child;
1.1 kristaps 330: unsigned char c;
331:
1.95 schwarze 332: memset(&ln, 0, sizeof(ln));
1.1 kristaps 333:
1.47 schwarze 334: lnn = curp->line;
335: pos = 0;
1.1 kristaps 336:
1.95 schwarze 337: while (i < blk.sz) {
1.1 kristaps 338: if (0 == pos && '\0' == blk.buf[i])
339: break;
340:
341: if (start) {
342: curp->line = lnn;
343: curp->reparse_count = 0;
1.93 schwarze 344:
345: if (lnn < 3 &&
346: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 347: curp->filenc & MPARSE_LATIN1)
348: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 349: }
350:
1.95 schwarze 351: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 352:
353: /*
354: * When finding an unescaped newline character,
355: * leave the character loop to process the line.
356: * Skip a preceding carriage return, if any.
357: */
358:
1.95 schwarze 359: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 360: '\n' == blk.buf[i + 1])
361: ++i;
362: if ('\n' == blk.buf[i]) {
363: ++i;
364: ++lnn;
365: break;
366: }
367:
1.35 schwarze 368: /*
1.93 schwarze 369: * Make sure we have space for the worst
370: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 371: */
372:
1.95 schwarze 373: if (pos + 11 > ln.sz)
1.35 schwarze 374: resize_buf(&ln, 256);
375:
1.47 schwarze 376: /*
1.93 schwarze 377: * Encode 8-bit input.
1.1 kristaps 378: */
379:
1.93 schwarze 380: c = blk.buf[i];
381: if (c & 0x80) {
1.95 schwarze 382: if ( ! (curp->filenc && preconv_encode(
383: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 384: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
385: curp->line, pos, "0x%x", c);
1.93 schwarze 386: ln.buf[pos++] = '?';
387: i++;
388: }
389: continue;
390: }
391:
392: /*
393: * Exclude control characters.
394: */
1.1 kristaps 395:
1.93 schwarze 396: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 397: mandoc_vmsg(c == 0x00 || c == 0x04 ||
398: c > 0x0a ? MANDOCERR_CHAR_BAD :
399: MANDOCERR_CHAR_UNSUPP,
400: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 401: i++;
1.27 joerg 402: ln.buf[pos++] = '?';
1.1 kristaps 403: continue;
404: }
405:
406: /* Trailing backslash = a plain char. */
407:
1.95 schwarze 408: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 409: ln.buf[pos++] = blk.buf[i++];
410: continue;
411: }
412:
413: /*
414: * Found escape and at least one other character.
415: * When it's a newline character, skip it.
416: * When there is a carriage return in between,
417: * skip that one as well.
418: */
419:
1.95 schwarze 420: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 421: '\n' == blk.buf[i + 2])
422: ++i;
423: if ('\n' == blk.buf[i + 1]) {
424: i += 2;
425: ++lnn;
426: continue;
427: }
428:
1.13 kristaps 429: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 430: i += 2;
431: /* Comment, skip to end of line */
1.95 schwarze 432: for (; i < blk.sz; ++i) {
1.1 kristaps 433: if ('\n' == blk.buf[i]) {
434: ++i;
435: ++lnn;
436: break;
437: }
438: }
439:
440: /* Backout trailing whitespaces */
441: for (; pos > 0; --pos) {
442: if (ln.buf[pos - 1] != ' ')
443: break;
444: if (pos > 2 && ln.buf[pos - 2] == '\\')
445: break;
446: }
447: break;
448: }
449:
1.35 schwarze 450: /* Catch escaped bogus characters. */
451:
452: c = (unsigned char) blk.buf[i+1];
453:
1.47 schwarze 454: if ( ! (isascii(c) &&
455: (isgraph(c) || isblank(c)))) {
1.114 schwarze 456: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 457: curp->line, pos, "0x%x", c);
1.35 schwarze 458: i += 2;
459: ln.buf[pos++] = '?';
460: continue;
461: }
462:
1.1 kristaps 463: /* Some other escape sequence, copy & cont. */
464:
465: ln.buf[pos++] = blk.buf[i++];
466: ln.buf[pos++] = blk.buf[i++];
467: }
468:
1.95 schwarze 469: if (pos >= ln.sz)
1.1 kristaps 470: resize_buf(&ln, 256);
471:
472: ln.buf[pos] = '\0';
473:
474: /*
475: * A significant amount of complexity is contained by
476: * the roff preprocessor. It's line-oriented but can be
477: * expressed on one line, so we need at times to
478: * readjust our starting point and re-run it. The roff
479: * preprocessor can also readjust the buffers with new
480: * data, so we pass them in wholesale.
481: */
482:
483: of = 0;
484:
1.24 kristaps 485: /*
486: * Maintain a lookaside buffer of all parsed lines. We
487: * only do this if mparse_keep() has been invoked (the
488: * buffer may be accessed with mparse_getkeep()).
489: */
490:
491: if (curp->secondary) {
1.47 schwarze 492: curp->secondary->buf = mandoc_realloc(
493: curp->secondary->buf,
494: curp->secondary->sz + pos + 2);
495: memcpy(curp->secondary->buf +
496: curp->secondary->sz,
497: ln.buf, pos);
1.24 kristaps 498: curp->secondary->sz += pos;
499: curp->secondary->buf
500: [curp->secondary->sz] = '\n';
501: curp->secondary->sz++;
502: curp->secondary->buf
503: [curp->secondary->sz] = '\0';
504: }
1.1 kristaps 505: rerun:
1.96 schwarze 506: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 507:
508: switch (rr) {
1.47 schwarze 509: case ROFF_REPARSE:
1.1 kristaps 510: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 511: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 512: else
1.3 kristaps 513: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 514: curp->line, pos, NULL);
1.1 kristaps 515: pos = 0;
516: continue;
1.47 schwarze 517: case ROFF_APPEND:
1.95 schwarze 518: pos = strlen(ln.buf);
1.1 kristaps 519: continue;
1.47 schwarze 520: case ROFF_RERUN:
1.1 kristaps 521: goto rerun;
1.47 schwarze 522: case ROFF_IGN:
1.1 kristaps 523: pos = 0;
524: continue;
1.47 schwarze 525: case ROFF_SO:
1.95 schwarze 526: if ( ! (curp->options & MPARSE_SO) &&
527: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 528: curp->sodest = mandoc_strdup(ln.buf + of);
529: free(ln.buf);
530: return;
531: }
1.24 kristaps 532: /*
533: * We remove `so' clauses from our lookaside
534: * buffer because we're going to descend into
535: * the file recursively.
536: */
1.47 schwarze 537: if (curp->secondary)
1.25 kristaps 538: curp->secondary->sz -= pos + 1;
1.113 schwarze 539: save_file = curp->file;
1.110 schwarze 540: save_child = curp->child;
541: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 schwarze 542: MANDOCLEVEL_OK) {
1.110 schwarze 543: mparse_readfd(curp, fd, ln.buf + of);
1.113 schwarze 544: curp->file = save_file;
545: } else {
546: curp->file = save_file;
1.52 schwarze 547: mandoc_vmsg(MANDOCERR_SO_FAIL,
548: curp, curp->line, pos,
549: ".so %s", ln.buf + of);
1.110 schwarze 550: ln.sz = mandoc_asprintf(&cp,
551: ".sp\nSee the file %s.\n.sp",
552: ln.buf + of);
553: free(ln.buf);
554: ln.buf = cp;
555: of = 0;
556: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 557: }
1.110 schwarze 558: curp->child = save_child;
1.1 kristaps 559: pos = 0;
560: continue;
561: default:
562: break;
563: }
564:
565: /*
566: * If input parsers have not been allocated, do so now.
1.14 kristaps 567: * We keep these instanced between parsers, but set them
1.1 kristaps 568: * locally per parse routine since we can use different
569: * parsers with each one.
570: */
571:
572: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 573: choose_parser(curp);
1.1 kristaps 574:
1.47 schwarze 575: /*
1.84 schwarze 576: * Lastly, push down into the parsers themselves.
1.1 kristaps 577: * If libroff returns ROFF_TBL, then add it to the
578: * currently open parse. Since we only get here if
579: * there does exist data (see tbl_data.c), we're
580: * guaranteed that something's been allocated.
581: * Do the same for ROFF_EQN.
582: */
583:
1.100 schwarze 584: if (rr == ROFF_TBL) {
585: while ((span = roff_span(curp->roff)) != NULL)
586: if (curp->man == NULL)
587: mdoc_addspan(curp->mdoc, span);
588: else
589: man_addspan(curp->man, span);
590: } else if (rr == ROFF_EQN) {
591: if (curp->man == NULL)
592: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
593: else
594: man_addeqn(curp->man, roff_eqn(curp->roff));
595: } else if ((curp->man == NULL ?
596: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
597: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
598: break;
1.1 kristaps 599:
600: /* Temporary buffers typically are not full. */
601:
602: if (0 == start && '\0' == blk.buf[i])
603: break;
604:
605: /* Start the next input line. */
606:
607: pos = 0;
608: }
609:
610: free(ln.buf);
611: }
612:
613: static int
1.40 schwarze 614: read_whole_file(struct mparse *curp, const char *file, int fd,
615: struct buf *fb, int *with_mmap)
1.1 kristaps 616: {
617: size_t off;
618: ssize_t ssz;
619:
1.81 schwarze 620: #if HAVE_MMAP
1.15 kristaps 621: struct stat st;
1.1 kristaps 622: if (-1 == fstat(fd, &st)) {
1.107 schwarze 623: perror(file);
624: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 625: }
626:
627: /*
628: * If we're a regular file, try just reading in the whole entry
629: * via mmap(). This is faster than reading it into blocks, and
630: * since each file is only a few bytes to begin with, I'm not
631: * concerned that this is going to tank any machines.
632: */
633:
634: if (S_ISREG(st.st_mode)) {
635: if (st.st_size >= (1U << 31)) {
1.111 schwarze 636: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 637: return(0);
638: }
639: *with_mmap = 1;
640: fb->sz = (size_t)st.st_size;
1.37 schwarze 641: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 642: if (fb->buf != MAP_FAILED)
643: return(1);
644: }
1.15 kristaps 645: #endif
1.1 kristaps 646:
647: /*
648: * If this isn't a regular file (like, say, stdin), then we must
649: * go the old way and just read things in bit by bit.
650: */
651:
652: *with_mmap = 0;
653: off = 0;
654: fb->sz = 0;
655: fb->buf = NULL;
656: for (;;) {
657: if (off == fb->sz) {
658: if (fb->sz == (1U << 31)) {
1.111 schwarze 659: mandoc_msg(MANDOCERR_TOOLARGE, curp,
660: 0, 0, NULL);
1.1 kristaps 661: break;
662: }
663: resize_buf(fb, 65536);
664: }
665: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
666: if (ssz == 0) {
667: fb->sz = off;
668: return(1);
669: }
670: if (ssz == -1) {
1.107 schwarze 671: perror(file);
672: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 673: }
674: off += (size_t)ssz;
675: }
676:
677: free(fb->buf);
678: fb->buf = NULL;
679: return(0);
680: }
681:
682: static void
683: mparse_end(struct mparse *curp)
684: {
685:
1.72 schwarze 686: if (curp->mdoc == NULL &&
687: curp->man == NULL &&
688: curp->sodest == NULL) {
689: if (curp->options & MPARSE_MDOC)
690: curp->mdoc = curp->pmdoc;
691: else {
692: if (curp->pman == NULL)
1.106 schwarze 693: curp->pman = man_alloc(
694: curp->roff, curp, curp->defos,
1.72 schwarze 695: curp->options & MPARSE_QUICK ? 1 : 0);
696: curp->man = curp->pman;
697: }
698: }
1.111 schwarze 699: if (curp->mdoc)
700: mdoc_endparse(curp->mdoc);
701: if (curp->man)
702: man_endparse(curp->man);
1.1 kristaps 703: roff_endparse(curp->roff);
704: }
705:
706: static void
1.36 schwarze 707: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 708: {
1.85 schwarze 709: struct buf *svprimary;
1.28 joerg 710: const char *svfile;
1.95 schwarze 711: size_t offset;
1.36 schwarze 712: static int recursion_depth;
713:
714: if (64 < recursion_depth) {
715: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
716: return;
717: }
1.28 joerg 718:
719: /* Line number is per-file. */
720: svfile = curp->file;
721: curp->file = file;
1.85 schwarze 722: svprimary = curp->primary;
1.83 schwarze 723: curp->primary = &blk;
1.28 joerg 724: curp->line = 1;
1.36 schwarze 725: recursion_depth++;
1.28 joerg 726:
1.93 schwarze 727: /* Skip an UTF-8 byte order mark. */
728: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
729: (unsigned char)blk.buf[0] == 0xef &&
730: (unsigned char)blk.buf[1] == 0xbb &&
731: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 732: offset = 3;
1.93 schwarze 733: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 734: } else
735: offset = 0;
1.93 schwarze 736:
1.95 schwarze 737: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 738:
1.111 schwarze 739: if (--recursion_depth == 0)
1.28 joerg 740: mparse_end(curp);
741:
1.85 schwarze 742: curp->primary = svprimary;
1.28 joerg 743: curp->file = svfile;
744: }
745:
746: enum mandoclevel
1.104 schwarze 747: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 748: const char *file)
749: {
750: struct buf blk;
751:
1.104 schwarze 752: blk.buf = buf;
1.28 joerg 753: blk.sz = len;
754:
1.36 schwarze 755: mparse_parse_buffer(curp, blk, file);
1.28 joerg 756: return(curp->file_status);
757: }
758:
1.98 schwarze 759: /*
760: * Read the whole file into memory and call the parsers.
761: * Called recursively when an .so request is encountered.
762: */
1.36 schwarze 763: enum mandoclevel
764: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 765: {
1.28 joerg 766: struct buf blk;
767: int with_mmap;
1.93 schwarze 768: int save_filenc;
1.1 kristaps 769:
1.91 schwarze 770: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 771: save_filenc = curp->filenc;
772: curp->filenc = curp->options &
773: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 774: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 775: curp->filenc = save_filenc;
1.81 schwarze 776: #if HAVE_MMAP
1.91 schwarze 777: if (with_mmap)
778: munmap(blk.buf, blk.sz);
779: else
1.28 joerg 780: #endif
1.91 schwarze 781: free(blk.buf);
782: }
1.1 kristaps 783:
1.98 schwarze 784: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 785: perror(file);
1.91 schwarze 786:
1.98 schwarze 787: mparse_wait(curp);
1.1 kristaps 788: return(curp->file_status);
1.82 schwarze 789: }
790:
791: enum mandoclevel
1.97 schwarze 792: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 793: {
794: int pfd[2];
1.98 schwarze 795: int save_errno;
1.82 schwarze 796: char *cp;
797:
798: curp->file = file;
1.98 schwarze 799:
800: /* Unless zipped, try to just open the file. */
801:
1.82 schwarze 802: if ((cp = strrchr(file, '.')) == NULL ||
803: strcmp(cp + 1, "gz")) {
1.97 schwarze 804: curp->child = 0;
1.98 schwarze 805: if ((*fd = open(file, O_RDONLY)) != -1)
806: return(MANDOCLEVEL_OK);
807:
808: /* Open failed; try to append ".gz". */
809:
810: mandoc_asprintf(&cp, "%s.gz", file);
811: file = cp;
812: } else
813: cp = NULL;
814:
815: /* Before forking, make sure the file can be read. */
816:
817: save_errno = errno;
818: if (access(file, R_OK) == -1) {
819: if (cp != NULL)
820: errno = save_errno;
1.108 schwarze 821: free(cp);
822: *fd = -1;
823: curp->child = 0;
824: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
825: return(MANDOCLEVEL_ERROR);
1.82 schwarze 826: }
827:
1.98 schwarze 828: /* Run gunzip(1). */
829:
1.82 schwarze 830: if (pipe(pfd) == -1) {
1.107 schwarze 831: perror("pipe");
832: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 833: }
834:
1.97 schwarze 835: switch (curp->child = fork()) {
1.82 schwarze 836: case -1:
1.107 schwarze 837: perror("fork");
838: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 839: case 0:
840: close(pfd[0]);
841: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 842: perror("dup");
843: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 844: }
845: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 846: perror("exec");
847: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 848: default:
849: close(pfd[1]);
850: *fd = pfd[0];
851: return(MANDOCLEVEL_OK);
852: }
853: }
854:
855: enum mandoclevel
1.97 schwarze 856: mparse_wait(struct mparse *curp)
1.82 schwarze 857: {
858: int status;
859:
1.97 schwarze 860: if (curp->child == 0)
861: return(MANDOCLEVEL_OK);
862:
863: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 864: perror("wait");
865: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 866: }
867: if (WIFSIGNALED(status)) {
1.108 schwarze 868: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
869: "gunzip died from signal %d", WTERMSIG(status));
870: return(MANDOCLEVEL_ERROR);
1.82 schwarze 871: }
872: if (WEXITSTATUS(status)) {
1.108 schwarze 873: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
874: "gunzip failed with code %d", WEXITSTATUS(status));
875: return(MANDOCLEVEL_ERROR);
1.82 schwarze 876: }
877: return(MANDOCLEVEL_OK);
1.1 kristaps 878: }
879:
880: struct mparse *
1.94 schwarze 881: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
882: const struct mchars *mchars, const char *defos)
1.1 kristaps 883: {
884: struct mparse *curp;
1.10 kristaps 885:
1.1 kristaps 886: curp = mandoc_calloc(1, sizeof(struct mparse));
887:
1.44 schwarze 888: curp->options = options;
1.3 kristaps 889: curp->wlevel = wlevel;
1.1 kristaps 890: curp->mmsg = mmsg;
1.29 schwarze 891: curp->defos = defos;
1.1 kristaps 892:
1.94 schwarze 893: curp->mchars = mchars;
894: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 895: if (curp->options & MPARSE_MDOC)
896: curp->pmdoc = mdoc_alloc(
897: curp->roff, curp, curp->defos,
898: curp->options & MPARSE_QUICK ? 1 : 0);
899: if (curp->options & MPARSE_MAN)
1.106 schwarze 900: curp->pman = man_alloc(
901: curp->roff, curp, curp->defos,
1.72 schwarze 902: curp->options & MPARSE_QUICK ? 1 : 0);
903:
1.1 kristaps 904: return(curp);
905: }
906:
907: void
908: mparse_reset(struct mparse *curp)
909: {
910:
911: roff_reset(curp->roff);
912:
913: if (curp->mdoc)
914: mdoc_reset(curp->mdoc);
915: if (curp->man)
916: man_reset(curp->man);
1.24 kristaps 917: if (curp->secondary)
918: curp->secondary->sz = 0;
1.1 kristaps 919:
920: curp->file_status = MANDOCLEVEL_OK;
921: curp->mdoc = NULL;
922: curp->man = NULL;
1.45 schwarze 923:
924: free(curp->sodest);
925: curp->sodest = NULL;
1.1 kristaps 926: }
927:
928: void
929: mparse_free(struct mparse *curp)
930: {
931:
932: if (curp->pmdoc)
933: mdoc_free(curp->pmdoc);
934: if (curp->pman)
935: man_free(curp->pman);
936: if (curp->roff)
937: roff_free(curp->roff);
1.24 kristaps 938: if (curp->secondary)
939: free(curp->secondary->buf);
1.1 kristaps 940:
1.24 kristaps 941: free(curp->secondary);
1.45 schwarze 942: free(curp->sodest);
1.1 kristaps 943: free(curp);
944: }
945:
946: void
1.45 schwarze 947: mparse_result(struct mparse *curp,
948: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 949: {
950:
1.45 schwarze 951: if (sodest && NULL != (*sodest = curp->sodest)) {
952: *mdoc = NULL;
953: *man = NULL;
954: return;
955: }
1.9 kristaps 956: if (mdoc)
957: *mdoc = curp->mdoc;
958: if (man)
959: *man = curp->man;
1.3 kristaps 960: }
961:
962: void
963: mandoc_vmsg(enum mandocerr t, struct mparse *m,
964: int ln, int pos, const char *fmt, ...)
965: {
966: char buf[256];
967: va_list ap;
968:
969: va_start(ap, fmt);
1.48 schwarze 970: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 971: va_end(ap);
972:
973: mandoc_msg(t, m, ln, pos, buf);
974: }
975:
976: void
1.47 schwarze 977: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 978: int ln, int col, const char *msg)
979: {
980: enum mandoclevel level;
981:
1.112 schwarze 982: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 983: while (er < mandoclimits[level])
984: level--;
985:
1.108 schwarze 986: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 987: return;
988:
1.8 kristaps 989: if (m->mmsg)
990: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 991:
992: if (m->file_status < level)
993: m->file_status = level;
1.7 kristaps 994: }
995:
996: const char *
997: mparse_strerror(enum mandocerr er)
998: {
999:
1000: return(mandocerrs[er]);
1001: }
1002:
1003: const char *
1004: mparse_strlevel(enum mandoclevel lvl)
1005: {
1006: return(mandoclevels[lvl]);
1.24 kristaps 1007: }
1008:
1009: void
1010: mparse_keep(struct mparse *p)
1011: {
1012:
1013: assert(NULL == p->secondary);
1014: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1015: }
1016:
1017: const char *
1018: mparse_getkeep(const struct mparse *p)
1019: {
1020:
1021: assert(p->secondary);
1022: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1023: }
CVSweb