Annotation of mandoc/read.c, Revision 1.135
1.135 ! schwarze 1: /* $Id: read.c,v 1.134 2015/04/18 16:06:41 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:
1.83 schwarze 292: if (format == MPARSE_MDOC) {
1.135 ! schwarze 293: if (curp->man == NULL)
! 294: curp->man = mdoc_alloc(
1.44 schwarze 295: curp->roff, curp, curp->defos,
296: MPARSE_QUICK & curp->options ? 1 : 0);
1.135 ! schwarze 297: else
! 298: curp->man->macroset = MACROSET_MDOC;
1.1 kristaps 299: return;
1.47 schwarze 300: }
1.1 kristaps 301:
1.83 schwarze 302: /* Fall back to man(7) as a last resort. */
303:
1.135 ! schwarze 304: if (curp->man == NULL)
! 305: curp->man = man_alloc(
1.106 schwarze 306: curp->roff, curp, curp->defos,
1.44 schwarze 307: MPARSE_QUICK & curp->options ? 1 : 0);
1.135 ! schwarze 308: else
! 309: curp->man->macroset = MACROSET_MAN;
1.1 kristaps 310: }
311:
312: /*
1.95 schwarze 313: * Main parse routine for a buffer.
314: * It assumes encoding and line numbering are already set up.
315: * It can recurse directly (for invocations of user-defined
316: * macros, inline equations, and input line traps)
317: * and indirectly (for .so file inclusion).
1.1 kristaps 318: */
319: static void
1.95 schwarze 320: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 321: {
322: const struct tbl_span *span;
323: struct buf ln;
1.113 schwarze 324: const char *save_file;
1.110 schwarze 325: char *cp;
1.95 schwarze 326: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 327: enum rofferr rr;
1.100 schwarze 328: int of;
1.1 kristaps 329: int lnn; /* line number in the real file */
1.110 schwarze 330: int fd;
331: pid_t save_child;
1.1 kristaps 332: unsigned char c;
333:
1.95 schwarze 334: memset(&ln, 0, sizeof(ln));
1.1 kristaps 335:
1.47 schwarze 336: lnn = curp->line;
337: pos = 0;
1.1 kristaps 338:
1.95 schwarze 339: while (i < blk.sz) {
1.1 kristaps 340: if (0 == pos && '\0' == blk.buf[i])
341: break;
342:
343: if (start) {
344: curp->line = lnn;
345: curp->reparse_count = 0;
1.93 schwarze 346:
347: if (lnn < 3 &&
348: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 349: curp->filenc & MPARSE_LATIN1)
350: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 351: }
352:
1.95 schwarze 353: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 354:
355: /*
356: * When finding an unescaped newline character,
357: * leave the character loop to process the line.
358: * Skip a preceding carriage return, if any.
359: */
360:
1.95 schwarze 361: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 362: '\n' == blk.buf[i + 1])
363: ++i;
364: if ('\n' == blk.buf[i]) {
365: ++i;
366: ++lnn;
367: break;
368: }
369:
1.35 schwarze 370: /*
1.93 schwarze 371: * Make sure we have space for the worst
372: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 373: */
374:
1.95 schwarze 375: if (pos + 11 > ln.sz)
1.35 schwarze 376: resize_buf(&ln, 256);
377:
1.47 schwarze 378: /*
1.93 schwarze 379: * Encode 8-bit input.
1.1 kristaps 380: */
381:
1.93 schwarze 382: c = blk.buf[i];
383: if (c & 0x80) {
1.95 schwarze 384: if ( ! (curp->filenc && preconv_encode(
385: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 386: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
387: curp->line, pos, "0x%x", c);
1.93 schwarze 388: ln.buf[pos++] = '?';
389: i++;
390: }
391: continue;
392: }
393:
394: /*
395: * Exclude control characters.
396: */
1.1 kristaps 397:
1.93 schwarze 398: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 399: mandoc_vmsg(c == 0x00 || c == 0x04 ||
400: c > 0x0a ? MANDOCERR_CHAR_BAD :
401: MANDOCERR_CHAR_UNSUPP,
402: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 403: i++;
1.127 schwarze 404: if (c != '\r')
405: ln.buf[pos++] = '?';
1.1 kristaps 406: continue;
407: }
408:
409: /* Trailing backslash = a plain char. */
410:
1.95 schwarze 411: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 412: ln.buf[pos++] = blk.buf[i++];
413: continue;
414: }
415:
416: /*
417: * Found escape and at least one other character.
418: * When it's a newline character, skip it.
419: * When there is a carriage return in between,
420: * skip that one as well.
421: */
422:
1.95 schwarze 423: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 424: '\n' == blk.buf[i + 2])
425: ++i;
426: if ('\n' == blk.buf[i + 1]) {
427: i += 2;
428: ++lnn;
429: continue;
430: }
431:
1.13 kristaps 432: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 433: i += 2;
434: /* Comment, skip to end of line */
1.95 schwarze 435: for (; i < blk.sz; ++i) {
1.1 kristaps 436: if ('\n' == blk.buf[i]) {
437: ++i;
438: ++lnn;
439: break;
440: }
441: }
442:
443: /* Backout trailing whitespaces */
444: for (; pos > 0; --pos) {
445: if (ln.buf[pos - 1] != ' ')
446: break;
447: if (pos > 2 && ln.buf[pos - 2] == '\\')
448: break;
449: }
450: break;
451: }
452:
1.35 schwarze 453: /* Catch escaped bogus characters. */
454:
455: c = (unsigned char) blk.buf[i+1];
456:
1.47 schwarze 457: if ( ! (isascii(c) &&
458: (isgraph(c) || isblank(c)))) {
1.114 schwarze 459: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 460: curp->line, pos, "0x%x", c);
1.35 schwarze 461: i += 2;
462: ln.buf[pos++] = '?';
463: continue;
464: }
465:
1.1 kristaps 466: /* Some other escape sequence, copy & cont. */
467:
468: ln.buf[pos++] = blk.buf[i++];
469: ln.buf[pos++] = blk.buf[i++];
470: }
471:
1.95 schwarze 472: if (pos >= ln.sz)
1.1 kristaps 473: resize_buf(&ln, 256);
474:
475: ln.buf[pos] = '\0';
476:
477: /*
478: * A significant amount of complexity is contained by
479: * the roff preprocessor. It's line-oriented but can be
480: * expressed on one line, so we need at times to
481: * readjust our starting point and re-run it. The roff
482: * preprocessor can also readjust the buffers with new
483: * data, so we pass them in wholesale.
484: */
485:
486: of = 0;
487:
1.24 kristaps 488: /*
489: * Maintain a lookaside buffer of all parsed lines. We
490: * only do this if mparse_keep() has been invoked (the
491: * buffer may be accessed with mparse_getkeep()).
492: */
493:
494: if (curp->secondary) {
1.47 schwarze 495: curp->secondary->buf = mandoc_realloc(
496: curp->secondary->buf,
497: curp->secondary->sz + pos + 2);
498: memcpy(curp->secondary->buf +
499: curp->secondary->sz,
500: ln.buf, pos);
1.24 kristaps 501: curp->secondary->sz += pos;
502: curp->secondary->buf
503: [curp->secondary->sz] = '\n';
504: curp->secondary->sz++;
505: curp->secondary->buf
506: [curp->secondary->sz] = '\0';
507: }
1.1 kristaps 508: rerun:
1.96 schwarze 509: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 510:
511: switch (rr) {
1.47 schwarze 512: case ROFF_REPARSE:
1.1 kristaps 513: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 514: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 515: else
1.3 kristaps 516: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 517: curp->line, pos, NULL);
1.1 kristaps 518: pos = 0;
519: continue;
1.47 schwarze 520: case ROFF_APPEND:
1.95 schwarze 521: pos = strlen(ln.buf);
1.1 kristaps 522: continue;
1.47 schwarze 523: case ROFF_RERUN:
1.1 kristaps 524: goto rerun;
1.47 schwarze 525: case ROFF_IGN:
1.1 kristaps 526: pos = 0;
527: continue;
1.47 schwarze 528: case ROFF_SO:
1.95 schwarze 529: if ( ! (curp->options & MPARSE_SO) &&
530: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 531: curp->sodest = mandoc_strdup(ln.buf + of);
532: free(ln.buf);
533: return;
534: }
1.24 kristaps 535: /*
536: * We remove `so' clauses from our lookaside
537: * buffer because we're going to descend into
538: * the file recursively.
539: */
1.47 schwarze 540: if (curp->secondary)
1.25 kristaps 541: curp->secondary->sz -= pos + 1;
1.113 schwarze 542: save_file = curp->file;
1.110 schwarze 543: save_child = curp->child;
544: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 schwarze 545: MANDOCLEVEL_OK) {
1.110 schwarze 546: mparse_readfd(curp, fd, ln.buf + of);
1.113 schwarze 547: curp->file = save_file;
548: } else {
549: curp->file = save_file;
1.52 schwarze 550: mandoc_vmsg(MANDOCERR_SO_FAIL,
551: curp, curp->line, pos,
552: ".so %s", ln.buf + of);
1.110 schwarze 553: ln.sz = mandoc_asprintf(&cp,
554: ".sp\nSee the file %s.\n.sp",
555: ln.buf + of);
556: free(ln.buf);
557: ln.buf = cp;
558: of = 0;
559: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 560: }
1.110 schwarze 561: curp->child = save_child;
1.1 kristaps 562: pos = 0;
563: continue;
564: default:
565: break;
566: }
567:
568: /*
569: * If input parsers have not been allocated, do so now.
1.14 kristaps 570: * We keep these instanced between parsers, but set them
1.1 kristaps 571: * locally per parse routine since we can use different
572: * parsers with each one.
573: */
574:
1.135 ! schwarze 575: if (curp->man == NULL ||
! 576: curp->man->macroset == MACROSET_NONE)
1.84 schwarze 577: choose_parser(curp);
1.1 kristaps 578:
1.47 schwarze 579: /*
1.84 schwarze 580: * Lastly, push down into the parsers themselves.
1.1 kristaps 581: * If libroff returns ROFF_TBL, then add it to the
582: * currently open parse. Since we only get here if
583: * there does exist data (see tbl_data.c), we're
584: * guaranteed that something's been allocated.
585: * Do the same for ROFF_EQN.
586: */
587:
1.100 schwarze 588: if (rr == ROFF_TBL) {
589: while ((span = roff_span(curp->roff)) != NULL)
1.135 ! schwarze 590: if (curp->man->macroset == MACROSET_MDOC)
! 591: mdoc_addspan(curp->man, span);
1.100 schwarze 592: else
593: man_addspan(curp->man, span);
594: } else if (rr == ROFF_EQN) {
1.135 ! schwarze 595: if (curp->man->macroset == MACROSET_MDOC)
! 596: mdoc_addeqn(curp->man, roff_eqn(curp->roff));
1.100 schwarze 597: else
598: man_addeqn(curp->man, roff_eqn(curp->roff));
1.135 ! schwarze 599: } else if ((curp->man->macroset == MACROSET_MDOC ?
! 600: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 601: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
602: break;
1.1 kristaps 603:
604: /* Temporary buffers typically are not full. */
605:
606: if (0 == start && '\0' == blk.buf[i])
607: break;
608:
609: /* Start the next input line. */
610:
611: pos = 0;
612: }
613:
614: free(ln.buf);
615: }
616:
617: static int
1.40 schwarze 618: read_whole_file(struct mparse *curp, const char *file, int fd,
619: struct buf *fb, int *with_mmap)
1.1 kristaps 620: {
621: size_t off;
622: ssize_t ssz;
623:
1.81 schwarze 624: #if HAVE_MMAP
1.15 kristaps 625: struct stat st;
1.1 kristaps 626: if (-1 == fstat(fd, &st)) {
1.107 schwarze 627: perror(file);
628: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 629: }
630:
631: /*
632: * If we're a regular file, try just reading in the whole entry
633: * via mmap(). This is faster than reading it into blocks, and
634: * since each file is only a few bytes to begin with, I'm not
635: * concerned that this is going to tank any machines.
636: */
637:
638: if (S_ISREG(st.st_mode)) {
1.131 schwarze 639: if (st.st_size > 0x7fffffff) {
1.111 schwarze 640: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 641: return(0);
642: }
643: *with_mmap = 1;
644: fb->sz = (size_t)st.st_size;
1.37 schwarze 645: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 646: if (fb->buf != MAP_FAILED)
647: return(1);
648: }
1.15 kristaps 649: #endif
1.1 kristaps 650:
651: /*
652: * If this isn't a regular file (like, say, stdin), then we must
653: * go the old way and just read things in bit by bit.
654: */
655:
656: *with_mmap = 0;
657: off = 0;
658: fb->sz = 0;
659: fb->buf = NULL;
660: for (;;) {
661: if (off == fb->sz) {
662: if (fb->sz == (1U << 31)) {
1.111 schwarze 663: mandoc_msg(MANDOCERR_TOOLARGE, curp,
664: 0, 0, NULL);
1.1 kristaps 665: break;
666: }
667: resize_buf(fb, 65536);
668: }
669: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
670: if (ssz == 0) {
671: fb->sz = off;
672: return(1);
673: }
674: if (ssz == -1) {
1.107 schwarze 675: perror(file);
676: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 677: }
678: off += (size_t)ssz;
679: }
680:
681: free(fb->buf);
682: fb->buf = NULL;
683: return(0);
684: }
685:
686: static void
687: mparse_end(struct mparse *curp)
688: {
689:
1.135 ! schwarze 690: if (curp->man == NULL && curp->sodest == NULL)
! 691: curp->man = man_alloc(curp->roff, curp, curp->defos,
! 692: curp->options & MPARSE_QUICK ? 1 : 0);
! 693: if (curp->man->macroset == MACROSET_NONE)
! 694: curp->man->macroset = MACROSET_MAN;
! 695: if (curp->man->macroset == MACROSET_MDOC)
! 696: mdoc_endparse(curp->man);
! 697: else
1.111 schwarze 698: man_endparse(curp->man);
1.1 kristaps 699: roff_endparse(curp->roff);
700: }
701:
702: static void
1.36 schwarze 703: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 704: {
1.85 schwarze 705: struct buf *svprimary;
1.28 joerg 706: const char *svfile;
1.95 schwarze 707: size_t offset;
1.36 schwarze 708: static int recursion_depth;
709:
710: if (64 < recursion_depth) {
711: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
712: return;
713: }
1.28 joerg 714:
715: /* Line number is per-file. */
716: svfile = curp->file;
717: curp->file = file;
1.85 schwarze 718: svprimary = curp->primary;
1.83 schwarze 719: curp->primary = &blk;
1.28 joerg 720: curp->line = 1;
1.36 schwarze 721: recursion_depth++;
1.28 joerg 722:
1.93 schwarze 723: /* Skip an UTF-8 byte order mark. */
724: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
725: (unsigned char)blk.buf[0] == 0xef &&
726: (unsigned char)blk.buf[1] == 0xbb &&
727: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 728: offset = 3;
1.93 schwarze 729: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 730: } else
731: offset = 0;
1.93 schwarze 732:
1.95 schwarze 733: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 734:
1.111 schwarze 735: if (--recursion_depth == 0)
1.28 joerg 736: mparse_end(curp);
737:
1.85 schwarze 738: curp->primary = svprimary;
1.28 joerg 739: curp->file = svfile;
740: }
741:
742: enum mandoclevel
1.104 schwarze 743: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 744: const char *file)
745: {
746: struct buf blk;
747:
1.104 schwarze 748: blk.buf = buf;
1.28 joerg 749: blk.sz = len;
750:
1.36 schwarze 751: mparse_parse_buffer(curp, blk, file);
1.28 joerg 752: return(curp->file_status);
753: }
754:
1.98 schwarze 755: /*
756: * Read the whole file into memory and call the parsers.
757: * Called recursively when an .so request is encountered.
758: */
1.36 schwarze 759: enum mandoclevel
760: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 761: {
1.28 joerg 762: struct buf blk;
763: int with_mmap;
1.93 schwarze 764: int save_filenc;
1.1 kristaps 765:
1.91 schwarze 766: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 767: save_filenc = curp->filenc;
768: curp->filenc = curp->options &
769: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 770: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 771: curp->filenc = save_filenc;
1.81 schwarze 772: #if HAVE_MMAP
1.91 schwarze 773: if (with_mmap)
774: munmap(blk.buf, blk.sz);
775: else
1.28 joerg 776: #endif
1.91 schwarze 777: free(blk.buf);
778: }
1.1 kristaps 779:
1.98 schwarze 780: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 781: perror(file);
1.91 schwarze 782:
1.98 schwarze 783: mparse_wait(curp);
1.1 kristaps 784: return(curp->file_status);
1.82 schwarze 785: }
786:
787: enum mandoclevel
1.97 schwarze 788: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 789: {
790: int pfd[2];
1.98 schwarze 791: int save_errno;
1.82 schwarze 792: char *cp;
793:
794: curp->file = file;
1.98 schwarze 795:
796: /* Unless zipped, try to just open the file. */
797:
1.82 schwarze 798: if ((cp = strrchr(file, '.')) == NULL ||
799: strcmp(cp + 1, "gz")) {
1.97 schwarze 800: curp->child = 0;
1.98 schwarze 801: if ((*fd = open(file, O_RDONLY)) != -1)
802: return(MANDOCLEVEL_OK);
803:
804: /* Open failed; try to append ".gz". */
805:
806: mandoc_asprintf(&cp, "%s.gz", file);
807: file = cp;
808: } else
809: cp = NULL;
810:
811: /* Before forking, make sure the file can be read. */
812:
813: save_errno = errno;
814: if (access(file, R_OK) == -1) {
815: if (cp != NULL)
816: errno = save_errno;
1.108 schwarze 817: free(cp);
818: *fd = -1;
819: curp->child = 0;
820: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
821: return(MANDOCLEVEL_ERROR);
1.82 schwarze 822: }
823:
1.98 schwarze 824: /* Run gunzip(1). */
825:
1.82 schwarze 826: if (pipe(pfd) == -1) {
1.107 schwarze 827: perror("pipe");
828: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 829: }
830:
1.97 schwarze 831: switch (curp->child = fork()) {
1.82 schwarze 832: case -1:
1.107 schwarze 833: perror("fork");
834: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 835: case 0:
836: close(pfd[0]);
837: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 838: perror("dup");
839: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 840: }
1.132 schwarze 841: signal(SIGPIPE, SIG_DFL);
1.82 schwarze 842: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 843: perror("exec");
844: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 845: default:
846: close(pfd[1]);
847: *fd = pfd[0];
848: return(MANDOCLEVEL_OK);
849: }
850: }
851:
852: enum mandoclevel
1.97 schwarze 853: mparse_wait(struct mparse *curp)
1.82 schwarze 854: {
855: int status;
856:
1.97 schwarze 857: if (curp->child == 0)
858: return(MANDOCLEVEL_OK);
859:
860: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 861: perror("wait");
862: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 863: }
1.129 schwarze 864: curp->child = 0;
1.82 schwarze 865: if (WIFSIGNALED(status)) {
1.108 schwarze 866: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
867: "gunzip died from signal %d", WTERMSIG(status));
868: return(MANDOCLEVEL_ERROR);
1.82 schwarze 869: }
870: if (WEXITSTATUS(status)) {
1.108 schwarze 871: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
872: "gunzip failed with code %d", WEXITSTATUS(status));
873: return(MANDOCLEVEL_ERROR);
1.82 schwarze 874: }
875: return(MANDOCLEVEL_OK);
1.1 kristaps 876: }
877:
878: struct mparse *
1.94 schwarze 879: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
880: const struct mchars *mchars, const char *defos)
1.1 kristaps 881: {
882: struct mparse *curp;
1.10 kristaps 883:
1.1 kristaps 884: curp = mandoc_calloc(1, sizeof(struct mparse));
885:
1.44 schwarze 886: curp->options = options;
1.3 kristaps 887: curp->wlevel = wlevel;
1.1 kristaps 888: curp->mmsg = mmsg;
1.29 schwarze 889: curp->defos = defos;
1.1 kristaps 890:
1.94 schwarze 891: curp->mchars = mchars;
892: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 893: if (curp->options & MPARSE_MDOC)
1.135 ! schwarze 894: curp->man = mdoc_alloc(
1.72 schwarze 895: curp->roff, curp, curp->defos,
896: curp->options & MPARSE_QUICK ? 1 : 0);
897: if (curp->options & MPARSE_MAN)
1.135 ! schwarze 898: curp->man = man_alloc(
1.106 schwarze 899: curp->roff, curp, curp->defos,
1.72 schwarze 900: curp->options & MPARSE_QUICK ? 1 : 0);
901:
1.1 kristaps 902: return(curp);
903: }
904:
905: void
906: mparse_reset(struct mparse *curp)
907: {
908:
909: roff_reset(curp->roff);
910:
1.135 ! schwarze 911: if (curp->man != NULL) {
! 912: if (curp->man->macroset == MACROSET_MDOC)
! 913: mdoc_reset(curp->man);
! 914: else
! 915: man_reset(curp->man);
! 916: curp->man->macroset = MACROSET_NONE;
! 917: }
1.24 kristaps 918: if (curp->secondary)
919: curp->secondary->sz = 0;
1.1 kristaps 920:
921: curp->file_status = MANDOCLEVEL_OK;
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:
1.135 ! schwarze 931: if (curp->man->macroset == MACROSET_MDOC)
! 932: mdoc_free(curp->man);
! 933: if (curp->man->macroset == MACROSET_MAN)
! 934: man_free(curp->man);
1.1 kristaps 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
1.135 ! schwarze 946: mparse_result(struct mparse *curp, struct roff_man **man,
! 947: char **sodest)
1.1 kristaps 948: {
949:
1.45 schwarze 950: if (sodest && NULL != (*sodest = curp->sodest)) {
951: *man = NULL;
952: return;
953: }
1.9 kristaps 954: if (man)
955: *man = curp->man;
1.3 kristaps 956: }
957:
958: void
959: mandoc_vmsg(enum mandocerr t, struct mparse *m,
960: int ln, int pos, const char *fmt, ...)
961: {
962: char buf[256];
963: va_list ap;
964:
965: va_start(ap, fmt);
1.48 schwarze 966: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 967: va_end(ap);
968:
969: mandoc_msg(t, m, ln, pos, buf);
970: }
971:
972: void
1.47 schwarze 973: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 974: int ln, int col, const char *msg)
975: {
976: enum mandoclevel level;
977:
1.112 schwarze 978: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 979: while (er < mandoclimits[level])
980: level--;
981:
1.108 schwarze 982: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 983: return;
984:
1.8 kristaps 985: if (m->mmsg)
986: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 987:
988: if (m->file_status < level)
989: m->file_status = level;
1.7 kristaps 990: }
991:
992: const char *
993: mparse_strerror(enum mandocerr er)
994: {
995:
996: return(mandocerrs[er]);
997: }
998:
999: const char *
1000: mparse_strlevel(enum mandoclevel lvl)
1001: {
1002: return(mandoclevels[lvl]);
1.24 kristaps 1003: }
1004:
1005: void
1006: mparse_keep(struct mparse *p)
1007: {
1008:
1009: assert(NULL == p->secondary);
1010: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1011: }
1012:
1013: const char *
1014: mparse_getkeep(const struct mparse *p)
1015: {
1016:
1017: assert(p->secondary);
1018: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1019: }
CVSweb