Annotation of mandoc/read.c, Revision 1.125
1.125 ! schwarze 1: /* $Id: read.c,v 1.124 2015/02/06 07:13:14 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",
1.121 schwarze 113: "missing description line, using \"\"",
1.7 kristaps 114: "sections out of conventional order",
1.54 schwarze 115: "duplicate section title",
116: "unexpected section",
1.87 schwarze 117: "unusual Xr order",
118: "unusual Xr punctuation",
1.86 schwarze 119: "AUTHORS section without An macro",
1.7 kristaps 120:
121: /* related to macros and nesting */
1.55 schwarze 122: "obsolete macro",
1.102 schwarze 123: "macro neither callable nor escaped",
1.7 kristaps 124: "skipping paragraph macro",
1.31 schwarze 125: "moving paragraph macro out of list",
1.7 kristaps 126: "skipping no-space macro",
127: "blocks badly nested",
128: "nested displays are not portable",
1.57 schwarze 129: "moving content out of list",
130: ".Vt block has child macro",
1.78 schwarze 131: "fill mode already enabled, skipping",
132: "fill mode already disabled, skipping",
1.7 kristaps 133: "line scope broken",
134:
135: /* related to missing macro arguments */
1.58 schwarze 136: "skipping empty request",
137: "conditional request controls empty scope",
1.7 kristaps 138: "skipping empty macro",
1.124 schwarze 139: "empty block",
1.62 schwarze 140: "empty argument, using 0n",
1.60 schwarze 141: "missing display type, using -ragged",
142: "list type is not the first argument",
143: "missing -width in -tag list, using 8n",
1.78 schwarze 144: "missing utility name, using \"\"",
1.123 schwarze 145: "missing function name, using \"\"",
1.60 schwarze 146: "empty head in list item",
147: "empty list item",
1.61 schwarze 148: "missing font type, using \\fR",
149: "unknown font type, using \\fR",
1.103 schwarze 150: "nothing follows prefix",
1.122 schwarze 151: "empty reference block",
1.60 schwarze 152: "missing -std argument, adding it",
1.125 ! schwarze 153: "missing option string, using \"\"",
! 154: "missing resource identifier, using \"\"",
1.90 schwarze 155: "missing eqn box, using \"\"",
1.7 kristaps 156:
157: /* related to bad macro arguments */
1.64 schwarze 158: "unterminated quoted argument",
1.7 kristaps 159: "duplicate argument",
1.76 schwarze 160: "skipping duplicate argument",
1.63 schwarze 161: "skipping duplicate display type",
162: "skipping duplicate list type",
1.76 schwarze 163: "skipping -width argument",
1.7 kristaps 164: "unknown AT&T UNIX version",
1.88 schwarze 165: "comma in function argument",
1.89 schwarze 166: "parenthesis in function name",
1.67 schwarze 167: "invalid content in Rs block",
1.63 schwarze 168: "invalid Boolean argument",
169: "unknown font, skipping request",
1.7 kristaps 170:
171: /* related to plain text */
1.64 schwarze 172: "blank line in fill mode, using .sp",
173: "tab in filled text",
174: "whitespace at end of input line",
1.7 kristaps 175: "bad comment style",
1.64 schwarze 176: "invalid escape sequence",
177: "undefined string, using \"\"",
1.16 kristaps 178:
1.118 schwarze 179: /* related to tables */
180: "tbl line starts with span",
181: "tbl column starts with span",
182: "skipping vertical bar in tbl layout",
183:
1.7 kristaps 184: "generic error",
185:
186: /* related to tables */
1.116 schwarze 187: "non-alphabetic character in tbl options",
188: "skipping unknown tbl option",
189: "missing tbl option argument",
190: "wrong tbl option argument size",
1.118 schwarze 191: "empty tbl layout",
192: "invalid character in tbl layout",
193: "unmatched parenthesis in tbl layout",
1.119 schwarze 194: "tbl without any data cells",
195: "ignoring data in spanned tbl cell",
196: "ignoring extra tbl data cells",
197: "data block open at end of tbl",
1.7 kristaps 198:
1.68 schwarze 199: /* related to document structure and macros */
1.108 schwarze 200: NULL,
1.7 kristaps 201: "input stack limit exceeded, infinite loop?",
202: "skipping bad character",
1.68 schwarze 203: "skipping unknown macro",
1.112 schwarze 204: "skipping insecure request",
1.73 schwarze 205: "skipping item outside list",
1.68 schwarze 206: "skipping column outside column list",
207: "skipping end of block that is not open",
1.115 schwarze 208: "fewer RS blocks open, skipping",
1.68 schwarze 209: "inserting missing end of block",
210: "appending missing end of block",
211:
212: /* related to request and macro arguments */
1.7 kristaps 213: "escaped character not allowed in a name",
214: "argument count wrong",
1.99 schwarze 215: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 216: "missing list type, using -item",
1.70 schwarze 217: "missing manual name, using \"\"",
1.71 schwarze 218: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 219: "unknown standard specifier",
1.71 schwarze 220: "skipping request without numeric argument",
1.109 schwarze 221: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 222: ".so request failed",
1.61 schwarze 223: "skipping all arguments",
224: "skipping excess arguments",
1.92 kristaps 225: "divide by zero",
1.112 schwarze 226:
227: "unsupported feature",
228: "input too large",
1.114 schwarze 229: "unsupported control character",
1.112 schwarze 230: "unsupported roff request",
1.119 schwarze 231: "eqn delim option in tbl",
1.118 schwarze 232: "unsupported tbl layout modifier",
1.112 schwarze 233: "ignoring macro in table",
1.7 kristaps 234: };
235:
236: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
237: "SUCCESS",
238: "RESERVED",
239: "WARNING",
240: "ERROR",
1.112 schwarze 241: "UNSUPP",
1.7 kristaps 242: "BADARG",
243: "SYSERR"
244: };
245:
1.47 schwarze 246:
1.1 kristaps 247: static void
248: resize_buf(struct buf *buf, size_t initial)
249: {
250:
251: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
252: buf->buf = mandoc_realloc(buf->buf, buf->sz);
253: }
254:
255: static void
1.84 schwarze 256: choose_parser(struct mparse *curp)
1.1 kristaps 257: {
1.83 schwarze 258: char *cp, *ep;
259: int format;
1.1 kristaps 260:
1.83 schwarze 261: /*
262: * If neither command line arguments -mdoc or -man select
263: * a parser nor the roff parser found a .Dd or .TH macro
264: * yet, look ahead in the main input buffer.
265: */
266:
267: if ((format = roff_getformat(curp->roff)) == 0) {
268: cp = curp->primary->buf;
269: ep = cp + curp->primary->sz;
270: while (cp < ep) {
1.85 schwarze 271: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 272: cp++;
273: if (cp[0] == 'D' && cp[1] == 'd') {
274: format = MPARSE_MDOC;
275: break;
276: }
277: if (cp[0] == 'T' && cp[1] == 'H') {
278: format = MPARSE_MAN;
279: break;
280: }
281: }
282: cp = memchr(cp, '\n', ep - cp);
283: if (cp == NULL)
284: break;
285: cp++;
286: }
1.1 kristaps 287: }
288:
1.83 schwarze 289: if (format == MPARSE_MDOC) {
1.47 schwarze 290: if (NULL == curp->pmdoc)
1.44 schwarze 291: curp->pmdoc = mdoc_alloc(
292: curp->roff, curp, curp->defos,
293: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 294: assert(curp->pmdoc);
295: curp->mdoc = curp->pmdoc;
296: return;
1.47 schwarze 297: }
1.1 kristaps 298:
1.83 schwarze 299: /* Fall back to man(7) as a last resort. */
300:
1.47 schwarze 301: if (NULL == curp->pman)
1.106 schwarze 302: curp->pman = man_alloc(
303: curp->roff, curp, curp->defos,
1.44 schwarze 304: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 305: assert(curp->pman);
306: curp->man = curp->pman;
307: }
308:
309: /*
1.95 schwarze 310: * Main parse routine for a buffer.
311: * It assumes encoding and line numbering are already set up.
312: * It can recurse directly (for invocations of user-defined
313: * macros, inline equations, and input line traps)
314: * and indirectly (for .so file inclusion).
1.1 kristaps 315: */
316: static void
1.95 schwarze 317: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 318: {
319: const struct tbl_span *span;
320: struct buf ln;
1.113 schwarze 321: const char *save_file;
1.110 schwarze 322: char *cp;
1.95 schwarze 323: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 324: enum rofferr rr;
1.100 schwarze 325: int of;
1.1 kristaps 326: int lnn; /* line number in the real file */
1.110 schwarze 327: int fd;
328: pid_t save_child;
1.1 kristaps 329: unsigned char c;
330:
1.95 schwarze 331: memset(&ln, 0, sizeof(ln));
1.1 kristaps 332:
1.47 schwarze 333: lnn = curp->line;
334: pos = 0;
1.1 kristaps 335:
1.95 schwarze 336: while (i < blk.sz) {
1.1 kristaps 337: if (0 == pos && '\0' == blk.buf[i])
338: break;
339:
340: if (start) {
341: curp->line = lnn;
342: curp->reparse_count = 0;
1.93 schwarze 343:
344: if (lnn < 3 &&
345: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 346: curp->filenc & MPARSE_LATIN1)
347: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 348: }
349:
1.95 schwarze 350: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 351:
352: /*
353: * When finding an unescaped newline character,
354: * leave the character loop to process the line.
355: * Skip a preceding carriage return, if any.
356: */
357:
1.95 schwarze 358: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 359: '\n' == blk.buf[i + 1])
360: ++i;
361: if ('\n' == blk.buf[i]) {
362: ++i;
363: ++lnn;
364: break;
365: }
366:
1.35 schwarze 367: /*
1.93 schwarze 368: * Make sure we have space for the worst
369: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 370: */
371:
1.95 schwarze 372: if (pos + 11 > ln.sz)
1.35 schwarze 373: resize_buf(&ln, 256);
374:
1.47 schwarze 375: /*
1.93 schwarze 376: * Encode 8-bit input.
1.1 kristaps 377: */
378:
1.93 schwarze 379: c = blk.buf[i];
380: if (c & 0x80) {
1.95 schwarze 381: if ( ! (curp->filenc && preconv_encode(
382: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 383: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
384: curp->line, pos, "0x%x", c);
1.93 schwarze 385: ln.buf[pos++] = '?';
386: i++;
387: }
388: continue;
389: }
390:
391: /*
392: * Exclude control characters.
393: */
1.1 kristaps 394:
1.93 schwarze 395: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 396: mandoc_vmsg(c == 0x00 || c == 0x04 ||
397: c > 0x0a ? MANDOCERR_CHAR_BAD :
398: MANDOCERR_CHAR_UNSUPP,
399: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 400: i++;
1.27 joerg 401: ln.buf[pos++] = '?';
1.1 kristaps 402: continue;
403: }
404:
405: /* Trailing backslash = a plain char. */
406:
1.95 schwarze 407: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 408: ln.buf[pos++] = blk.buf[i++];
409: continue;
410: }
411:
412: /*
413: * Found escape and at least one other character.
414: * When it's a newline character, skip it.
415: * When there is a carriage return in between,
416: * skip that one as well.
417: */
418:
1.95 schwarze 419: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 420: '\n' == blk.buf[i + 2])
421: ++i;
422: if ('\n' == blk.buf[i + 1]) {
423: i += 2;
424: ++lnn;
425: continue;
426: }
427:
1.13 kristaps 428: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 429: i += 2;
430: /* Comment, skip to end of line */
1.95 schwarze 431: for (; i < blk.sz; ++i) {
1.1 kristaps 432: if ('\n' == blk.buf[i]) {
433: ++i;
434: ++lnn;
435: break;
436: }
437: }
438:
439: /* Backout trailing whitespaces */
440: for (; pos > 0; --pos) {
441: if (ln.buf[pos - 1] != ' ')
442: break;
443: if (pos > 2 && ln.buf[pos - 2] == '\\')
444: break;
445: }
446: break;
447: }
448:
1.35 schwarze 449: /* Catch escaped bogus characters. */
450:
451: c = (unsigned char) blk.buf[i+1];
452:
1.47 schwarze 453: if ( ! (isascii(c) &&
454: (isgraph(c) || isblank(c)))) {
1.114 schwarze 455: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 456: curp->line, pos, "0x%x", c);
1.35 schwarze 457: i += 2;
458: ln.buf[pos++] = '?';
459: continue;
460: }
461:
1.1 kristaps 462: /* Some other escape sequence, copy & cont. */
463:
464: ln.buf[pos++] = blk.buf[i++];
465: ln.buf[pos++] = blk.buf[i++];
466: }
467:
1.95 schwarze 468: if (pos >= ln.sz)
1.1 kristaps 469: resize_buf(&ln, 256);
470:
471: ln.buf[pos] = '\0';
472:
473: /*
474: * A significant amount of complexity is contained by
475: * the roff preprocessor. It's line-oriented but can be
476: * expressed on one line, so we need at times to
477: * readjust our starting point and re-run it. The roff
478: * preprocessor can also readjust the buffers with new
479: * data, so we pass them in wholesale.
480: */
481:
482: of = 0;
483:
1.24 kristaps 484: /*
485: * Maintain a lookaside buffer of all parsed lines. We
486: * only do this if mparse_keep() has been invoked (the
487: * buffer may be accessed with mparse_getkeep()).
488: */
489:
490: if (curp->secondary) {
1.47 schwarze 491: curp->secondary->buf = mandoc_realloc(
492: curp->secondary->buf,
493: curp->secondary->sz + pos + 2);
494: memcpy(curp->secondary->buf +
495: curp->secondary->sz,
496: ln.buf, pos);
1.24 kristaps 497: curp->secondary->sz += pos;
498: curp->secondary->buf
499: [curp->secondary->sz] = '\n';
500: curp->secondary->sz++;
501: curp->secondary->buf
502: [curp->secondary->sz] = '\0';
503: }
1.1 kristaps 504: rerun:
1.96 schwarze 505: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 506:
507: switch (rr) {
1.47 schwarze 508: case ROFF_REPARSE:
1.1 kristaps 509: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 510: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 511: else
1.3 kristaps 512: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 513: curp->line, pos, NULL);
1.1 kristaps 514: pos = 0;
515: continue;
1.47 schwarze 516: case ROFF_APPEND:
1.95 schwarze 517: pos = strlen(ln.buf);
1.1 kristaps 518: continue;
1.47 schwarze 519: case ROFF_RERUN:
1.1 kristaps 520: goto rerun;
1.47 schwarze 521: case ROFF_IGN:
1.1 kristaps 522: pos = 0;
523: continue;
1.47 schwarze 524: case ROFF_SO:
1.95 schwarze 525: if ( ! (curp->options & MPARSE_SO) &&
526: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 527: curp->sodest = mandoc_strdup(ln.buf + of);
528: free(ln.buf);
529: return;
530: }
1.24 kristaps 531: /*
532: * We remove `so' clauses from our lookaside
533: * buffer because we're going to descend into
534: * the file recursively.
535: */
1.47 schwarze 536: if (curp->secondary)
1.25 kristaps 537: curp->secondary->sz -= pos + 1;
1.113 schwarze 538: save_file = curp->file;
1.110 schwarze 539: save_child = curp->child;
540: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 schwarze 541: MANDOCLEVEL_OK) {
1.110 schwarze 542: mparse_readfd(curp, fd, ln.buf + of);
1.113 schwarze 543: curp->file = save_file;
544: } else {
545: curp->file = save_file;
1.52 schwarze 546: mandoc_vmsg(MANDOCERR_SO_FAIL,
547: curp, curp->line, pos,
548: ".so %s", ln.buf + of);
1.110 schwarze 549: ln.sz = mandoc_asprintf(&cp,
550: ".sp\nSee the file %s.\n.sp",
551: ln.buf + of);
552: free(ln.buf);
553: ln.buf = cp;
554: of = 0;
555: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 556: }
1.110 schwarze 557: curp->child = save_child;
1.1 kristaps 558: pos = 0;
559: continue;
560: default:
561: break;
562: }
563:
564: /*
565: * If input parsers have not been allocated, do so now.
1.14 kristaps 566: * We keep these instanced between parsers, but set them
1.1 kristaps 567: * locally per parse routine since we can use different
568: * parsers with each one.
569: */
570:
571: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 572: choose_parser(curp);
1.1 kristaps 573:
1.47 schwarze 574: /*
1.84 schwarze 575: * Lastly, push down into the parsers themselves.
1.1 kristaps 576: * If libroff returns ROFF_TBL, then add it to the
577: * currently open parse. Since we only get here if
578: * there does exist data (see tbl_data.c), we're
579: * guaranteed that something's been allocated.
580: * Do the same for ROFF_EQN.
581: */
582:
1.100 schwarze 583: if (rr == ROFF_TBL) {
584: while ((span = roff_span(curp->roff)) != NULL)
585: if (curp->man == NULL)
586: mdoc_addspan(curp->mdoc, span);
587: else
588: man_addspan(curp->man, span);
589: } else if (rr == ROFF_EQN) {
590: if (curp->man == NULL)
591: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
592: else
593: man_addeqn(curp->man, roff_eqn(curp->roff));
594: } else if ((curp->man == NULL ?
595: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
596: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
597: break;
1.1 kristaps 598:
599: /* Temporary buffers typically are not full. */
600:
601: if (0 == start && '\0' == blk.buf[i])
602: break;
603:
604: /* Start the next input line. */
605:
606: pos = 0;
607: }
608:
609: free(ln.buf);
610: }
611:
612: static int
1.40 schwarze 613: read_whole_file(struct mparse *curp, const char *file, int fd,
614: struct buf *fb, int *with_mmap)
1.1 kristaps 615: {
616: size_t off;
617: ssize_t ssz;
618:
1.81 schwarze 619: #if HAVE_MMAP
1.15 kristaps 620: struct stat st;
1.1 kristaps 621: if (-1 == fstat(fd, &st)) {
1.107 schwarze 622: perror(file);
623: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 624: }
625:
626: /*
627: * If we're a regular file, try just reading in the whole entry
628: * via mmap(). This is faster than reading it into blocks, and
629: * since each file is only a few bytes to begin with, I'm not
630: * concerned that this is going to tank any machines.
631: */
632:
633: if (S_ISREG(st.st_mode)) {
634: if (st.st_size >= (1U << 31)) {
1.111 schwarze 635: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 636: return(0);
637: }
638: *with_mmap = 1;
639: fb->sz = (size_t)st.st_size;
1.37 schwarze 640: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 641: if (fb->buf != MAP_FAILED)
642: return(1);
643: }
1.15 kristaps 644: #endif
1.1 kristaps 645:
646: /*
647: * If this isn't a regular file (like, say, stdin), then we must
648: * go the old way and just read things in bit by bit.
649: */
650:
651: *with_mmap = 0;
652: off = 0;
653: fb->sz = 0;
654: fb->buf = NULL;
655: for (;;) {
656: if (off == fb->sz) {
657: if (fb->sz == (1U << 31)) {
1.111 schwarze 658: mandoc_msg(MANDOCERR_TOOLARGE, curp,
659: 0, 0, NULL);
1.1 kristaps 660: break;
661: }
662: resize_buf(fb, 65536);
663: }
664: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
665: if (ssz == 0) {
666: fb->sz = off;
667: return(1);
668: }
669: if (ssz == -1) {
1.107 schwarze 670: perror(file);
671: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 672: }
673: off += (size_t)ssz;
674: }
675:
676: free(fb->buf);
677: fb->buf = NULL;
678: return(0);
679: }
680:
681: static void
682: mparse_end(struct mparse *curp)
683: {
684:
1.72 schwarze 685: if (curp->mdoc == NULL &&
686: curp->man == NULL &&
687: curp->sodest == NULL) {
688: if (curp->options & MPARSE_MDOC)
689: curp->mdoc = curp->pmdoc;
690: else {
691: if (curp->pman == NULL)
1.106 schwarze 692: curp->pman = man_alloc(
693: curp->roff, curp, curp->defos,
1.72 schwarze 694: curp->options & MPARSE_QUICK ? 1 : 0);
695: curp->man = curp->pman;
696: }
697: }
1.111 schwarze 698: if (curp->mdoc)
699: mdoc_endparse(curp->mdoc);
700: if (curp->man)
701: man_endparse(curp->man);
1.1 kristaps 702: roff_endparse(curp->roff);
703: }
704:
705: static void
1.36 schwarze 706: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 707: {
1.85 schwarze 708: struct buf *svprimary;
1.28 joerg 709: const char *svfile;
1.95 schwarze 710: size_t offset;
1.36 schwarze 711: static int recursion_depth;
712:
713: if (64 < recursion_depth) {
714: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
715: return;
716: }
1.28 joerg 717:
718: /* Line number is per-file. */
719: svfile = curp->file;
720: curp->file = file;
1.85 schwarze 721: svprimary = curp->primary;
1.83 schwarze 722: curp->primary = &blk;
1.28 joerg 723: curp->line = 1;
1.36 schwarze 724: recursion_depth++;
1.28 joerg 725:
1.93 schwarze 726: /* Skip an UTF-8 byte order mark. */
727: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
728: (unsigned char)blk.buf[0] == 0xef &&
729: (unsigned char)blk.buf[1] == 0xbb &&
730: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 731: offset = 3;
1.93 schwarze 732: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 733: } else
734: offset = 0;
1.93 schwarze 735:
1.95 schwarze 736: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 737:
1.111 schwarze 738: if (--recursion_depth == 0)
1.28 joerg 739: mparse_end(curp);
740:
1.85 schwarze 741: curp->primary = svprimary;
1.28 joerg 742: curp->file = svfile;
743: }
744:
745: enum mandoclevel
1.104 schwarze 746: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 747: const char *file)
748: {
749: struct buf blk;
750:
1.104 schwarze 751: blk.buf = buf;
1.28 joerg 752: blk.sz = len;
753:
1.36 schwarze 754: mparse_parse_buffer(curp, blk, file);
1.28 joerg 755: return(curp->file_status);
756: }
757:
1.98 schwarze 758: /*
759: * Read the whole file into memory and call the parsers.
760: * Called recursively when an .so request is encountered.
761: */
1.36 schwarze 762: enum mandoclevel
763: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 764: {
1.28 joerg 765: struct buf blk;
766: int with_mmap;
1.93 schwarze 767: int save_filenc;
1.1 kristaps 768:
1.91 schwarze 769: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 770: save_filenc = curp->filenc;
771: curp->filenc = curp->options &
772: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 773: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 774: curp->filenc = save_filenc;
1.81 schwarze 775: #if HAVE_MMAP
1.91 schwarze 776: if (with_mmap)
777: munmap(blk.buf, blk.sz);
778: else
1.28 joerg 779: #endif
1.91 schwarze 780: free(blk.buf);
781: }
1.1 kristaps 782:
1.98 schwarze 783: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 784: perror(file);
1.91 schwarze 785:
1.98 schwarze 786: mparse_wait(curp);
1.1 kristaps 787: return(curp->file_status);
1.82 schwarze 788: }
789:
790: enum mandoclevel
1.97 schwarze 791: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 792: {
793: int pfd[2];
1.98 schwarze 794: int save_errno;
1.82 schwarze 795: char *cp;
796:
797: curp->file = file;
1.98 schwarze 798:
799: /* Unless zipped, try to just open the file. */
800:
1.82 schwarze 801: if ((cp = strrchr(file, '.')) == NULL ||
802: strcmp(cp + 1, "gz")) {
1.97 schwarze 803: curp->child = 0;
1.98 schwarze 804: if ((*fd = open(file, O_RDONLY)) != -1)
805: return(MANDOCLEVEL_OK);
806:
807: /* Open failed; try to append ".gz". */
808:
809: mandoc_asprintf(&cp, "%s.gz", file);
810: file = cp;
811: } else
812: cp = NULL;
813:
814: /* Before forking, make sure the file can be read. */
815:
816: save_errno = errno;
817: if (access(file, R_OK) == -1) {
818: if (cp != NULL)
819: errno = save_errno;
1.108 schwarze 820: free(cp);
821: *fd = -1;
822: curp->child = 0;
823: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
824: return(MANDOCLEVEL_ERROR);
1.82 schwarze 825: }
826:
1.98 schwarze 827: /* Run gunzip(1). */
828:
1.82 schwarze 829: if (pipe(pfd) == -1) {
1.107 schwarze 830: perror("pipe");
831: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 832: }
833:
1.97 schwarze 834: switch (curp->child = fork()) {
1.82 schwarze 835: case -1:
1.107 schwarze 836: perror("fork");
837: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 838: case 0:
839: close(pfd[0]);
840: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 841: perror("dup");
842: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 843: }
844: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 845: perror("exec");
846: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 847: default:
848: close(pfd[1]);
849: *fd = pfd[0];
850: return(MANDOCLEVEL_OK);
851: }
852: }
853:
854: enum mandoclevel
1.97 schwarze 855: mparse_wait(struct mparse *curp)
1.82 schwarze 856: {
857: int status;
858:
1.97 schwarze 859: if (curp->child == 0)
860: return(MANDOCLEVEL_OK);
861:
862: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 863: perror("wait");
864: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 865: }
866: if (WIFSIGNALED(status)) {
1.108 schwarze 867: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
868: "gunzip died from signal %d", WTERMSIG(status));
869: return(MANDOCLEVEL_ERROR);
1.82 schwarze 870: }
871: if (WEXITSTATUS(status)) {
1.108 schwarze 872: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
873: "gunzip failed with code %d", WEXITSTATUS(status));
874: return(MANDOCLEVEL_ERROR);
1.82 schwarze 875: }
876: return(MANDOCLEVEL_OK);
1.1 kristaps 877: }
878:
879: struct mparse *
1.94 schwarze 880: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
881: const struct mchars *mchars, const char *defos)
1.1 kristaps 882: {
883: struct mparse *curp;
1.10 kristaps 884:
1.1 kristaps 885: curp = mandoc_calloc(1, sizeof(struct mparse));
886:
1.44 schwarze 887: curp->options = options;
1.3 kristaps 888: curp->wlevel = wlevel;
1.1 kristaps 889: curp->mmsg = mmsg;
1.29 schwarze 890: curp->defos = defos;
1.1 kristaps 891:
1.94 schwarze 892: curp->mchars = mchars;
893: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 894: if (curp->options & MPARSE_MDOC)
895: curp->pmdoc = mdoc_alloc(
896: curp->roff, curp, curp->defos,
897: curp->options & MPARSE_QUICK ? 1 : 0);
898: if (curp->options & MPARSE_MAN)
1.106 schwarze 899: curp->pman = man_alloc(
900: curp->roff, curp, curp->defos,
1.72 schwarze 901: curp->options & MPARSE_QUICK ? 1 : 0);
902:
1.1 kristaps 903: return(curp);
904: }
905:
906: void
907: mparse_reset(struct mparse *curp)
908: {
909:
910: roff_reset(curp->roff);
911:
912: if (curp->mdoc)
913: mdoc_reset(curp->mdoc);
914: if (curp->man)
915: man_reset(curp->man);
1.24 kristaps 916: if (curp->secondary)
917: curp->secondary->sz = 0;
1.1 kristaps 918:
919: curp->file_status = MANDOCLEVEL_OK;
920: curp->mdoc = NULL;
921: curp->man = NULL;
1.45 schwarze 922:
923: free(curp->sodest);
924: curp->sodest = NULL;
1.1 kristaps 925: }
926:
927: void
928: mparse_free(struct mparse *curp)
929: {
930:
931: if (curp->pmdoc)
932: mdoc_free(curp->pmdoc);
933: if (curp->pman)
934: man_free(curp->pman);
935: if (curp->roff)
936: roff_free(curp->roff);
1.24 kristaps 937: if (curp->secondary)
938: free(curp->secondary->buf);
1.1 kristaps 939:
1.24 kristaps 940: free(curp->secondary);
1.45 schwarze 941: free(curp->sodest);
1.1 kristaps 942: free(curp);
943: }
944:
945: void
1.45 schwarze 946: mparse_result(struct mparse *curp,
947: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 948: {
949:
1.45 schwarze 950: if (sodest && NULL != (*sodest = curp->sodest)) {
951: *mdoc = NULL;
952: *man = NULL;
953: return;
954: }
1.9 kristaps 955: if (mdoc)
956: *mdoc = curp->mdoc;
957: if (man)
958: *man = curp->man;
1.3 kristaps 959: }
960:
961: void
962: mandoc_vmsg(enum mandocerr t, struct mparse *m,
963: int ln, int pos, const char *fmt, ...)
964: {
965: char buf[256];
966: va_list ap;
967:
968: va_start(ap, fmt);
1.48 schwarze 969: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 970: va_end(ap);
971:
972: mandoc_msg(t, m, ln, pos, buf);
973: }
974:
975: void
1.47 schwarze 976: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 977: int ln, int col, const char *msg)
978: {
979: enum mandoclevel level;
980:
1.112 schwarze 981: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 982: while (er < mandoclimits[level])
983: level--;
984:
1.108 schwarze 985: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 986: return;
987:
1.8 kristaps 988: if (m->mmsg)
989: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 990:
991: if (m->file_status < level)
992: m->file_status = level;
1.7 kristaps 993: }
994:
995: const char *
996: mparse_strerror(enum mandocerr er)
997: {
998:
999: return(mandocerrs[er]);
1000: }
1001:
1002: const char *
1003: mparse_strlevel(enum mandoclevel lvl)
1004: {
1005: return(mandoclevels[lvl]);
1.24 kristaps 1006: }
1007:
1008: void
1009: mparse_keep(struct mparse *p)
1010: {
1011:
1012: assert(NULL == p->secondary);
1013: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1014: }
1015:
1016: const char *
1017: mparse_getkeep(const struct mparse *p)
1018: {
1019:
1020: assert(p->secondary);
1021: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1022: }
CVSweb