Annotation of mandoc/read.c, Revision 1.138
1.138 ! schwarze 1: /* $Id: read.c,v 1.137 2015/04/18 17:28:36 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: *
1.133 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 kristaps 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.133 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 kristaps 14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18: */
1.11 kristaps 19: #include "config.h"
20:
1.80 schwarze 21: #include <sys/types.h>
1.81 schwarze 22: #if HAVE_MMAP
1.82 schwarze 23: #include <sys/mman.h>
1.80 schwarze 24: #include <sys/stat.h>
1.15 kristaps 25: #endif
1.82 schwarze 26: #include <sys/wait.h>
1.1 kristaps 27:
28: #include <assert.h>
29: #include <ctype.h>
1.40 schwarze 30: #include <errno.h>
1.1 kristaps 31: #include <fcntl.h>
1.132 schwarze 32: #include <signal.h>
1.3 kristaps 33: #include <stdarg.h>
1.28 joerg 34: #include <stdint.h>
1.1 kristaps 35: #include <stdio.h>
36: #include <stdlib.h>
37: #include <string.h>
38: #include <unistd.h>
39:
1.133 schwarze 40: #include "mandoc_aux.h"
1.1 kristaps 41: #include "mandoc.h"
1.133 schwarze 42: #include "roff.h"
1.1 kristaps 43: #include "mdoc.h"
44: #include "man.h"
1.133 schwarze 45: #include "libmandoc.h"
1.1 kristaps 46:
47: #define REPARSE_LIMIT 1000
48:
49: struct mparse {
1.134 schwarze 50: struct roff_man *man; /* man parser */
1.1 kristaps 51: struct roff *roff; /* roff parser (!NULL) */
1.94 schwarze 52: const struct mchars *mchars; /* character table */
1.45 schwarze 53: char *sodest; /* filename pointed to by .so */
1.83 schwarze 54: const char *file; /* filename of current input file */
55: struct buf *primary; /* buffer currently being parsed */
56: struct buf *secondary; /* preprocessed copy of input */
57: const char *defos; /* default operating system */
58: mandocmsg mmsg; /* warning/error message handler */
59: enum mandoclevel file_status; /* status of current parse */
60: enum mandoclevel wlevel; /* ignore messages below this */
61: int options; /* parser options */
1.93 schwarze 62: int filenc; /* encoding of the current file */
1.1 kristaps 63: int reparse_count; /* finite interp. stack */
1.83 schwarze 64: int line; /* line number in the file */
1.97 schwarze 65: pid_t child; /* the gunzip(1) process */
1.1 kristaps 66: };
67:
1.84 schwarze 68: static void choose_parser(struct mparse *);
1.1 kristaps 69: static void resize_buf(struct buf *, size_t);
1.95 schwarze 70: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 71: static int read_whole_file(struct mparse *, const char *, int,
72: struct buf *, int *);
1.1 kristaps 73: static void mparse_end(struct mparse *);
1.37 schwarze 74: static void mparse_parse_buffer(struct mparse *, struct buf,
75: const char *);
1.1 kristaps 76:
1.3 kristaps 77: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
78: MANDOCERR_OK,
79: MANDOCERR_WARNING,
80: MANDOCERR_WARNING,
81: MANDOCERR_ERROR,
1.112 schwarze 82: MANDOCERR_UNSUPP,
1.3 kristaps 83: MANDOCERR_MAX,
84: MANDOCERR_MAX
85: };
86:
1.7 kristaps 87: static const char * const mandocerrs[MANDOCERR_MAX] = {
88: "ok",
89:
90: "generic warning",
91:
92: /* related to the prologue */
1.79 schwarze 93: "missing manual title, using UNTITLED",
94: "missing manual title, using \"\"",
1.54 schwarze 95: "lower case character in document title",
1.79 schwarze 96: "missing manual section, using \"\"",
1.7 kristaps 97: "unknown manual section",
1.54 schwarze 98: "missing date, using today's date",
1.7 kristaps 99: "cannot parse date, using it verbatim",
1.79 schwarze 100: "missing Os macro, using \"\"",
101: "duplicate prologue macro",
102: "late prologue macro",
103: "skipping late title macro",
1.7 kristaps 104: "prologue macros out of order",
105:
106: /* related to document structure */
107: ".so is fragile, better use ln(1)",
1.50 schwarze 108: "no document body",
1.54 schwarze 109: "content before first section header",
110: "first section is not \"NAME\"",
1.128 schwarze 111: "NAME section without name",
112: "NAME section without description",
113: "description not at the end of NAME",
114: "bad NAME section content",
1.121 schwarze 115: "missing description line, using \"\"",
1.7 kristaps 116: "sections out of conventional order",
1.54 schwarze 117: "duplicate section title",
118: "unexpected section",
1.87 schwarze 119: "unusual Xr order",
120: "unusual Xr punctuation",
1.86 schwarze 121: "AUTHORS section without An macro",
1.7 kristaps 122:
123: /* related to macros and nesting */
1.55 schwarze 124: "obsolete macro",
1.102 schwarze 125: "macro neither callable nor escaped",
1.7 kristaps 126: "skipping paragraph macro",
1.31 schwarze 127: "moving paragraph macro out of list",
1.7 kristaps 128: "skipping no-space macro",
129: "blocks badly nested",
130: "nested displays are not portable",
1.57 schwarze 131: "moving content out of list",
132: ".Vt block has child macro",
1.78 schwarze 133: "fill mode already enabled, skipping",
134: "fill mode already disabled, skipping",
1.7 kristaps 135: "line scope broken",
136:
137: /* related to missing macro arguments */
1.58 schwarze 138: "skipping empty request",
139: "conditional request controls empty scope",
1.7 kristaps 140: "skipping empty macro",
1.124 schwarze 141: "empty block",
1.62 schwarze 142: "empty argument, using 0n",
1.60 schwarze 143: "missing display type, using -ragged",
144: "list type is not the first argument",
145: "missing -width in -tag list, using 8n",
1.78 schwarze 146: "missing utility name, using \"\"",
1.123 schwarze 147: "missing function name, using \"\"",
1.60 schwarze 148: "empty head in list item",
149: "empty list item",
1.61 schwarze 150: "missing font type, using \\fR",
151: "unknown font type, using \\fR",
1.103 schwarze 152: "nothing follows prefix",
1.122 schwarze 153: "empty reference block",
1.60 schwarze 154: "missing -std argument, adding it",
1.125 schwarze 155: "missing option string, using \"\"",
156: "missing resource identifier, using \"\"",
1.90 schwarze 157: "missing eqn box, using \"\"",
1.7 kristaps 158:
159: /* related to bad macro arguments */
1.64 schwarze 160: "unterminated quoted argument",
1.7 kristaps 161: "duplicate argument",
1.76 schwarze 162: "skipping duplicate argument",
1.63 schwarze 163: "skipping duplicate display type",
164: "skipping duplicate list type",
1.76 schwarze 165: "skipping -width argument",
1.126 schwarze 166: "wrong number of cells",
1.7 kristaps 167: "unknown AT&T UNIX version",
1.88 schwarze 168: "comma in function argument",
1.89 schwarze 169: "parenthesis in function name",
1.67 schwarze 170: "invalid content in Rs block",
1.63 schwarze 171: "invalid Boolean argument",
172: "unknown font, skipping request",
1.126 schwarze 173: "odd number of characters in request",
1.7 kristaps 174:
175: /* related to plain text */
1.64 schwarze 176: "blank line in fill mode, using .sp",
177: "tab in filled text",
178: "whitespace at end of input line",
1.7 kristaps 179: "bad comment style",
1.64 schwarze 180: "invalid escape sequence",
181: "undefined string, using \"\"",
1.16 kristaps 182:
1.118 schwarze 183: /* related to tables */
184: "tbl line starts with span",
185: "tbl column starts with span",
186: "skipping vertical bar in tbl layout",
187:
1.7 kristaps 188: "generic error",
189:
190: /* related to tables */
1.116 schwarze 191: "non-alphabetic character in tbl options",
192: "skipping unknown tbl option",
193: "missing tbl option argument",
194: "wrong tbl option argument size",
1.118 schwarze 195: "empty tbl layout",
196: "invalid character in tbl layout",
197: "unmatched parenthesis in tbl layout",
1.119 schwarze 198: "tbl without any data cells",
199: "ignoring data in spanned tbl cell",
200: "ignoring extra tbl data cells",
201: "data block open at end of tbl",
1.7 kristaps 202:
1.68 schwarze 203: /* related to document structure and macros */
1.108 schwarze 204: NULL,
1.7 kristaps 205: "input stack limit exceeded, infinite loop?",
206: "skipping bad character",
1.68 schwarze 207: "skipping unknown macro",
1.112 schwarze 208: "skipping insecure request",
1.73 schwarze 209: "skipping item outside list",
1.68 schwarze 210: "skipping column outside column list",
211: "skipping end of block that is not open",
1.115 schwarze 212: "fewer RS blocks open, skipping",
1.68 schwarze 213: "inserting missing end of block",
214: "appending missing end of block",
215:
216: /* related to request and macro arguments */
1.7 kristaps 217: "escaped character not allowed in a name",
1.99 schwarze 218: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 219: "missing list type, using -item",
1.70 schwarze 220: "missing manual name, using \"\"",
1.71 schwarze 221: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 222: "unknown standard specifier",
1.71 schwarze 223: "skipping request without numeric argument",
1.109 schwarze 224: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 225: ".so request failed",
1.61 schwarze 226: "skipping all arguments",
227: "skipping excess arguments",
1.92 kristaps 228: "divide by zero",
1.112 schwarze 229:
230: "unsupported feature",
231: "input too large",
1.114 schwarze 232: "unsupported control character",
1.112 schwarze 233: "unsupported roff request",
1.119 schwarze 234: "eqn delim option in tbl",
1.118 schwarze 235: "unsupported tbl layout modifier",
1.112 schwarze 236: "ignoring macro in table",
1.7 kristaps 237: };
238:
239: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
240: "SUCCESS",
241: "RESERVED",
242: "WARNING",
243: "ERROR",
1.112 schwarze 244: "UNSUPP",
1.7 kristaps 245: "BADARG",
246: "SYSERR"
247: };
248:
1.47 schwarze 249:
1.1 kristaps 250: static void
251: resize_buf(struct buf *buf, size_t initial)
252: {
253:
254: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
255: buf->buf = mandoc_realloc(buf->buf, buf->sz);
256: }
257:
258: static void
1.84 schwarze 259: choose_parser(struct mparse *curp)
1.1 kristaps 260: {
1.83 schwarze 261: char *cp, *ep;
262: int format;
1.1 kristaps 263:
1.83 schwarze 264: /*
265: * If neither command line arguments -mdoc or -man select
266: * a parser nor the roff parser found a .Dd or .TH macro
267: * yet, look ahead in the main input buffer.
268: */
269:
270: if ((format = roff_getformat(curp->roff)) == 0) {
271: cp = curp->primary->buf;
272: ep = cp + curp->primary->sz;
273: while (cp < ep) {
1.85 schwarze 274: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 275: cp++;
276: if (cp[0] == 'D' && cp[1] == 'd') {
277: format = MPARSE_MDOC;
278: break;
279: }
280: if (cp[0] == 'T' && cp[1] == 'H') {
281: format = MPARSE_MAN;
282: break;
283: }
284: }
285: cp = memchr(cp, '\n', ep - cp);
286: if (cp == NULL)
287: break;
288: cp++;
289: }
1.1 kristaps 290: }
291:
1.137 schwarze 292: if (curp->man == NULL) {
293: curp->man = roff_man_alloc(curp->roff, curp, curp->defos,
294: curp->options & MPARSE_QUICK ? 1 : 0);
295: curp->man->macroset = MACROSET_MAN;
1.138 ! schwarze 296: curp->man->first->tok = TOKEN_NONE;
1.137 schwarze 297: }
298:
1.83 schwarze 299: if (format == MPARSE_MDOC) {
1.136 schwarze 300: mdoc_hash_init();
1.137 schwarze 301: curp->man->macroset = MACROSET_MDOC;
1.138 ! schwarze 302: curp->man->first->tok = TOKEN_NONE;
1.137 schwarze 303: } else {
304: man_hash_init();
305: curp->man->macroset = MACROSET_MAN;
1.138 ! schwarze 306: curp->man->first->tok = TOKEN_NONE;
1.47 schwarze 307: }
1.1 kristaps 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++;
1.127 schwarze 402: if (c != '\r')
403: ln.buf[pos++] = '?';
1.1 kristaps 404: continue;
405: }
406:
407: /* Trailing backslash = a plain char. */
408:
1.95 schwarze 409: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 410: ln.buf[pos++] = blk.buf[i++];
411: continue;
412: }
413:
414: /*
415: * Found escape and at least one other character.
416: * When it's a newline character, skip it.
417: * When there is a carriage return in between,
418: * skip that one as well.
419: */
420:
1.95 schwarze 421: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 422: '\n' == blk.buf[i + 2])
423: ++i;
424: if ('\n' == blk.buf[i + 1]) {
425: i += 2;
426: ++lnn;
427: continue;
428: }
429:
1.13 kristaps 430: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 431: i += 2;
432: /* Comment, skip to end of line */
1.95 schwarze 433: for (; i < blk.sz; ++i) {
1.1 kristaps 434: if ('\n' == blk.buf[i]) {
435: ++i;
436: ++lnn;
437: break;
438: }
439: }
440:
441: /* Backout trailing whitespaces */
442: for (; pos > 0; --pos) {
443: if (ln.buf[pos - 1] != ' ')
444: break;
445: if (pos > 2 && ln.buf[pos - 2] == '\\')
446: break;
447: }
448: break;
449: }
450:
1.35 schwarze 451: /* Catch escaped bogus characters. */
452:
453: c = (unsigned char) blk.buf[i+1];
454:
1.47 schwarze 455: if ( ! (isascii(c) &&
456: (isgraph(c) || isblank(c)))) {
1.114 schwarze 457: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 458: curp->line, pos, "0x%x", c);
1.35 schwarze 459: i += 2;
460: ln.buf[pos++] = '?';
461: continue;
462: }
463:
1.1 kristaps 464: /* Some other escape sequence, copy & cont. */
465:
466: ln.buf[pos++] = blk.buf[i++];
467: ln.buf[pos++] = blk.buf[i++];
468: }
469:
1.95 schwarze 470: if (pos >= ln.sz)
1.1 kristaps 471: resize_buf(&ln, 256);
472:
473: ln.buf[pos] = '\0';
474:
475: /*
476: * A significant amount of complexity is contained by
477: * the roff preprocessor. It's line-oriented but can be
478: * expressed on one line, so we need at times to
479: * readjust our starting point and re-run it. The roff
480: * preprocessor can also readjust the buffers with new
481: * data, so we pass them in wholesale.
482: */
483:
484: of = 0;
485:
1.24 kristaps 486: /*
487: * Maintain a lookaside buffer of all parsed lines. We
488: * only do this if mparse_keep() has been invoked (the
489: * buffer may be accessed with mparse_getkeep()).
490: */
491:
492: if (curp->secondary) {
1.47 schwarze 493: curp->secondary->buf = mandoc_realloc(
494: curp->secondary->buf,
495: curp->secondary->sz + pos + 2);
496: memcpy(curp->secondary->buf +
497: curp->secondary->sz,
498: ln.buf, pos);
1.24 kristaps 499: curp->secondary->sz += pos;
500: curp->secondary->buf
501: [curp->secondary->sz] = '\n';
502: curp->secondary->sz++;
503: curp->secondary->buf
504: [curp->secondary->sz] = '\0';
505: }
1.1 kristaps 506: rerun:
1.96 schwarze 507: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 508:
509: switch (rr) {
1.47 schwarze 510: case ROFF_REPARSE:
1.1 kristaps 511: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 512: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 513: else
1.3 kristaps 514: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 515: curp->line, pos, NULL);
1.1 kristaps 516: pos = 0;
517: continue;
1.47 schwarze 518: case ROFF_APPEND:
1.95 schwarze 519: pos = strlen(ln.buf);
1.1 kristaps 520: continue;
1.47 schwarze 521: case ROFF_RERUN:
1.1 kristaps 522: goto rerun;
1.47 schwarze 523: case ROFF_IGN:
1.1 kristaps 524: pos = 0;
525: continue;
1.47 schwarze 526: case ROFF_SO:
1.95 schwarze 527: if ( ! (curp->options & MPARSE_SO) &&
528: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 529: curp->sodest = mandoc_strdup(ln.buf + of);
530: free(ln.buf);
531: return;
532: }
1.24 kristaps 533: /*
534: * We remove `so' clauses from our lookaside
535: * buffer because we're going to descend into
536: * the file recursively.
537: */
1.47 schwarze 538: if (curp->secondary)
1.25 kristaps 539: curp->secondary->sz -= pos + 1;
1.113 schwarze 540: save_file = curp->file;
1.110 schwarze 541: save_child = curp->child;
542: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 schwarze 543: MANDOCLEVEL_OK) {
1.110 schwarze 544: mparse_readfd(curp, fd, ln.buf + of);
1.113 schwarze 545: curp->file = save_file;
546: } else {
547: curp->file = save_file;
1.52 schwarze 548: mandoc_vmsg(MANDOCERR_SO_FAIL,
549: curp, curp->line, pos,
550: ".so %s", ln.buf + of);
1.110 schwarze 551: ln.sz = mandoc_asprintf(&cp,
552: ".sp\nSee the file %s.\n.sp",
553: ln.buf + of);
554: free(ln.buf);
555: ln.buf = cp;
556: of = 0;
557: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 558: }
1.110 schwarze 559: curp->child = save_child;
1.1 kristaps 560: pos = 0;
561: continue;
562: default:
563: break;
564: }
565:
566: /*
567: * If input parsers have not been allocated, do so now.
1.14 kristaps 568: * We keep these instanced between parsers, but set them
1.1 kristaps 569: * locally per parse routine since we can use different
570: * parsers with each one.
571: */
572:
1.135 schwarze 573: if (curp->man == NULL ||
574: curp->man->macroset == MACROSET_NONE)
1.84 schwarze 575: choose_parser(curp);
1.1 kristaps 576:
1.47 schwarze 577: /*
1.84 schwarze 578: * Lastly, push down into the parsers themselves.
1.1 kristaps 579: * If libroff returns ROFF_TBL, then add it to the
580: * currently open parse. Since we only get here if
581: * there does exist data (see tbl_data.c), we're
582: * guaranteed that something's been allocated.
583: * Do the same for ROFF_EQN.
584: */
585:
1.100 schwarze 586: if (rr == ROFF_TBL) {
587: while ((span = roff_span(curp->roff)) != NULL)
1.135 schwarze 588: if (curp->man->macroset == MACROSET_MDOC)
589: mdoc_addspan(curp->man, span);
1.100 schwarze 590: else
591: man_addspan(curp->man, span);
592: } else if (rr == ROFF_EQN) {
1.135 schwarze 593: if (curp->man->macroset == MACROSET_MDOC)
594: mdoc_addeqn(curp->man, roff_eqn(curp->roff));
1.100 schwarze 595: else
596: man_addeqn(curp->man, roff_eqn(curp->roff));
1.135 schwarze 597: } else if ((curp->man->macroset == MACROSET_MDOC ?
598: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 599: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
600: break;
1.1 kristaps 601:
602: /* Temporary buffers typically are not full. */
603:
604: if (0 == start && '\0' == blk.buf[i])
605: break;
606:
607: /* Start the next input line. */
608:
609: pos = 0;
610: }
611:
612: free(ln.buf);
613: }
614:
615: static int
1.40 schwarze 616: read_whole_file(struct mparse *curp, const char *file, int fd,
617: struct buf *fb, int *with_mmap)
1.1 kristaps 618: {
619: size_t off;
620: ssize_t ssz;
621:
1.81 schwarze 622: #if HAVE_MMAP
1.15 kristaps 623: struct stat st;
1.1 kristaps 624: if (-1 == fstat(fd, &st)) {
1.107 schwarze 625: perror(file);
626: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 627: }
628:
629: /*
630: * If we're a regular file, try just reading in the whole entry
631: * via mmap(). This is faster than reading it into blocks, and
632: * since each file is only a few bytes to begin with, I'm not
633: * concerned that this is going to tank any machines.
634: */
635:
636: if (S_ISREG(st.st_mode)) {
1.131 schwarze 637: if (st.st_size > 0x7fffffff) {
1.111 schwarze 638: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 639: return(0);
640: }
641: *with_mmap = 1;
642: fb->sz = (size_t)st.st_size;
1.37 schwarze 643: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 644: if (fb->buf != MAP_FAILED)
645: return(1);
646: }
1.15 kristaps 647: #endif
1.1 kristaps 648:
649: /*
650: * If this isn't a regular file (like, say, stdin), then we must
651: * go the old way and just read things in bit by bit.
652: */
653:
654: *with_mmap = 0;
655: off = 0;
656: fb->sz = 0;
657: fb->buf = NULL;
658: for (;;) {
659: if (off == fb->sz) {
660: if (fb->sz == (1U << 31)) {
1.111 schwarze 661: mandoc_msg(MANDOCERR_TOOLARGE, curp,
662: 0, 0, NULL);
1.1 kristaps 663: break;
664: }
665: resize_buf(fb, 65536);
666: }
667: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
668: if (ssz == 0) {
669: fb->sz = off;
670: return(1);
671: }
672: if (ssz == -1) {
1.107 schwarze 673: perror(file);
674: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 675: }
676: off += (size_t)ssz;
677: }
678:
679: free(fb->buf);
680: fb->buf = NULL;
681: return(0);
682: }
683:
684: static void
685: mparse_end(struct mparse *curp)
686: {
687:
1.135 schwarze 688: if (curp->man == NULL && curp->sodest == NULL)
1.137 schwarze 689: curp->man = roff_man_alloc(curp->roff, curp, curp->defos,
1.135 schwarze 690: curp->options & MPARSE_QUICK ? 1 : 0);
691: if (curp->man->macroset == MACROSET_NONE)
692: curp->man->macroset = MACROSET_MAN;
693: if (curp->man->macroset == MACROSET_MDOC)
694: mdoc_endparse(curp->man);
695: else
1.111 schwarze 696: man_endparse(curp->man);
1.1 kristaps 697: roff_endparse(curp->roff);
698: }
699:
700: static void
1.36 schwarze 701: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 702: {
1.85 schwarze 703: struct buf *svprimary;
1.28 joerg 704: const char *svfile;
1.95 schwarze 705: size_t offset;
1.36 schwarze 706: static int recursion_depth;
707:
708: if (64 < recursion_depth) {
709: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
710: return;
711: }
1.28 joerg 712:
713: /* Line number is per-file. */
714: svfile = curp->file;
715: curp->file = file;
1.85 schwarze 716: svprimary = curp->primary;
1.83 schwarze 717: curp->primary = &blk;
1.28 joerg 718: curp->line = 1;
1.36 schwarze 719: recursion_depth++;
1.28 joerg 720:
1.93 schwarze 721: /* Skip an UTF-8 byte order mark. */
722: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
723: (unsigned char)blk.buf[0] == 0xef &&
724: (unsigned char)blk.buf[1] == 0xbb &&
725: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 726: offset = 3;
1.93 schwarze 727: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 728: } else
729: offset = 0;
1.93 schwarze 730:
1.95 schwarze 731: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 732:
1.111 schwarze 733: if (--recursion_depth == 0)
1.28 joerg 734: mparse_end(curp);
735:
1.85 schwarze 736: curp->primary = svprimary;
1.28 joerg 737: curp->file = svfile;
738: }
739:
740: enum mandoclevel
1.104 schwarze 741: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 742: const char *file)
743: {
744: struct buf blk;
745:
1.104 schwarze 746: blk.buf = buf;
1.28 joerg 747: blk.sz = len;
748:
1.36 schwarze 749: mparse_parse_buffer(curp, blk, file);
1.28 joerg 750: return(curp->file_status);
751: }
752:
1.98 schwarze 753: /*
754: * Read the whole file into memory and call the parsers.
755: * Called recursively when an .so request is encountered.
756: */
1.36 schwarze 757: enum mandoclevel
758: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 759: {
1.28 joerg 760: struct buf blk;
761: int with_mmap;
1.93 schwarze 762: int save_filenc;
1.1 kristaps 763:
1.91 schwarze 764: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 765: save_filenc = curp->filenc;
766: curp->filenc = curp->options &
767: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 768: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 769: curp->filenc = save_filenc;
1.81 schwarze 770: #if HAVE_MMAP
1.91 schwarze 771: if (with_mmap)
772: munmap(blk.buf, blk.sz);
773: else
1.28 joerg 774: #endif
1.91 schwarze 775: free(blk.buf);
776: }
1.1 kristaps 777:
1.98 schwarze 778: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 779: perror(file);
1.91 schwarze 780:
1.98 schwarze 781: mparse_wait(curp);
1.1 kristaps 782: return(curp->file_status);
1.82 schwarze 783: }
784:
785: enum mandoclevel
1.97 schwarze 786: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 787: {
788: int pfd[2];
1.98 schwarze 789: int save_errno;
1.82 schwarze 790: char *cp;
791:
792: curp->file = file;
1.98 schwarze 793:
794: /* Unless zipped, try to just open the file. */
795:
1.82 schwarze 796: if ((cp = strrchr(file, '.')) == NULL ||
797: strcmp(cp + 1, "gz")) {
1.97 schwarze 798: curp->child = 0;
1.98 schwarze 799: if ((*fd = open(file, O_RDONLY)) != -1)
800: return(MANDOCLEVEL_OK);
801:
802: /* Open failed; try to append ".gz". */
803:
804: mandoc_asprintf(&cp, "%s.gz", file);
805: file = cp;
806: } else
807: cp = NULL;
808:
809: /* Before forking, make sure the file can be read. */
810:
811: save_errno = errno;
812: if (access(file, R_OK) == -1) {
813: if (cp != NULL)
814: errno = save_errno;
1.108 schwarze 815: free(cp);
816: *fd = -1;
817: curp->child = 0;
818: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
819: return(MANDOCLEVEL_ERROR);
1.82 schwarze 820: }
821:
1.98 schwarze 822: /* Run gunzip(1). */
823:
1.82 schwarze 824: if (pipe(pfd) == -1) {
1.107 schwarze 825: perror("pipe");
826: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 827: }
828:
1.97 schwarze 829: switch (curp->child = fork()) {
1.82 schwarze 830: case -1:
1.107 schwarze 831: perror("fork");
832: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 833: case 0:
834: close(pfd[0]);
835: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 836: perror("dup");
837: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 838: }
1.132 schwarze 839: signal(SIGPIPE, SIG_DFL);
1.82 schwarze 840: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 841: perror("exec");
842: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 843: default:
844: close(pfd[1]);
845: *fd = pfd[0];
846: return(MANDOCLEVEL_OK);
847: }
848: }
849:
850: enum mandoclevel
1.97 schwarze 851: mparse_wait(struct mparse *curp)
1.82 schwarze 852: {
853: int status;
854:
1.97 schwarze 855: if (curp->child == 0)
856: return(MANDOCLEVEL_OK);
857:
858: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 859: perror("wait");
860: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 861: }
1.129 schwarze 862: curp->child = 0;
1.82 schwarze 863: if (WIFSIGNALED(status)) {
1.108 schwarze 864: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
865: "gunzip died from signal %d", WTERMSIG(status));
866: return(MANDOCLEVEL_ERROR);
1.82 schwarze 867: }
868: if (WEXITSTATUS(status)) {
1.108 schwarze 869: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
870: "gunzip failed with code %d", WEXITSTATUS(status));
871: return(MANDOCLEVEL_ERROR);
1.82 schwarze 872: }
873: return(MANDOCLEVEL_OK);
1.1 kristaps 874: }
875:
876: struct mparse *
1.94 schwarze 877: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
878: const struct mchars *mchars, const char *defos)
1.1 kristaps 879: {
880: struct mparse *curp;
1.10 kristaps 881:
1.1 kristaps 882: curp = mandoc_calloc(1, sizeof(struct mparse));
883:
1.44 schwarze 884: curp->options = options;
1.3 kristaps 885: curp->wlevel = wlevel;
1.1 kristaps 886: curp->mmsg = mmsg;
1.29 schwarze 887: curp->defos = defos;
1.1 kristaps 888:
1.94 schwarze 889: curp->mchars = mchars;
890: curp->roff = roff_alloc(curp, curp->mchars, options);
1.137 schwarze 891: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
892: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 893: if (curp->options & MPARSE_MDOC) {
894: mdoc_hash_init();
1.137 schwarze 895: curp->man->macroset = MACROSET_MDOC;
896: } else if (curp->options & MPARSE_MAN) {
1.136 schwarze 897: man_hash_init();
1.137 schwarze 898: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 899: }
1.138 ! schwarze 900: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 901: return(curp);
902: }
903:
904: void
905: mparse_reset(struct mparse *curp)
906: {
907:
908: roff_reset(curp->roff);
909:
1.137 schwarze 910: if (curp->man != NULL)
911: roff_man_reset(curp->man);
1.24 kristaps 912: if (curp->secondary)
913: curp->secondary->sz = 0;
1.1 kristaps 914:
915: curp->file_status = MANDOCLEVEL_OK;
1.45 schwarze 916:
917: free(curp->sodest);
918: curp->sodest = NULL;
1.1 kristaps 919: }
920:
921: void
922: mparse_free(struct mparse *curp)
923: {
924:
1.137 schwarze 925: roff_man_free(curp->man);
1.1 kristaps 926: if (curp->roff)
927: roff_free(curp->roff);
1.24 kristaps 928: if (curp->secondary)
929: free(curp->secondary->buf);
1.1 kristaps 930:
1.24 kristaps 931: free(curp->secondary);
1.45 schwarze 932: free(curp->sodest);
1.1 kristaps 933: free(curp);
934: }
935:
936: void
1.135 schwarze 937: mparse_result(struct mparse *curp, struct roff_man **man,
938: char **sodest)
1.1 kristaps 939: {
940:
1.45 schwarze 941: if (sodest && NULL != (*sodest = curp->sodest)) {
942: *man = NULL;
943: return;
944: }
1.9 kristaps 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