Annotation of mandoc/read.c, Revision 1.139
1.139 ! schwarze 1: /* $Id: read.c,v 1.138 2015/04/19 14:00:20 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"
1.139 ! schwarze 46: #include "roff_int.h"
1.1 kristaps 47:
48: #define REPARSE_LIMIT 1000
49:
50: struct mparse {
1.134 schwarze 51: struct roff_man *man; /* man parser */
1.1 kristaps 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\"",
1.128 schwarze 112: "NAME section without name",
113: "NAME section without description",
114: "description not at the end of NAME",
115: "bad NAME section content",
1.121 schwarze 116: "missing description line, using \"\"",
1.7 kristaps 117: "sections out of conventional order",
1.54 schwarze 118: "duplicate section title",
119: "unexpected section",
1.87 schwarze 120: "unusual Xr order",
121: "unusual Xr punctuation",
1.86 schwarze 122: "AUTHORS section without An macro",
1.7 kristaps 123:
124: /* related to macros and nesting */
1.55 schwarze 125: "obsolete macro",
1.102 schwarze 126: "macro neither callable nor escaped",
1.7 kristaps 127: "skipping paragraph macro",
1.31 schwarze 128: "moving paragraph macro out of list",
1.7 kristaps 129: "skipping no-space macro",
130: "blocks badly nested",
131: "nested displays are not portable",
1.57 schwarze 132: "moving content out of list",
133: ".Vt block has child macro",
1.78 schwarze 134: "fill mode already enabled, skipping",
135: "fill mode already disabled, skipping",
1.7 kristaps 136: "line scope broken",
137:
138: /* related to missing macro arguments */
1.58 schwarze 139: "skipping empty request",
140: "conditional request controls empty scope",
1.7 kristaps 141: "skipping empty macro",
1.124 schwarze 142: "empty block",
1.62 schwarze 143: "empty argument, using 0n",
1.60 schwarze 144: "missing display type, using -ragged",
145: "list type is not the first argument",
146: "missing -width in -tag list, using 8n",
1.78 schwarze 147: "missing utility name, using \"\"",
1.123 schwarze 148: "missing function name, using \"\"",
1.60 schwarze 149: "empty head in list item",
150: "empty list item",
1.61 schwarze 151: "missing font type, using \\fR",
152: "unknown font type, using \\fR",
1.103 schwarze 153: "nothing follows prefix",
1.122 schwarze 154: "empty reference block",
1.60 schwarze 155: "missing -std argument, adding it",
1.125 schwarze 156: "missing option string, using \"\"",
157: "missing resource identifier, using \"\"",
1.90 schwarze 158: "missing eqn box, using \"\"",
1.7 kristaps 159:
160: /* related to bad macro arguments */
1.64 schwarze 161: "unterminated quoted argument",
1.7 kristaps 162: "duplicate argument",
1.76 schwarze 163: "skipping duplicate argument",
1.63 schwarze 164: "skipping duplicate display type",
165: "skipping duplicate list type",
1.76 schwarze 166: "skipping -width argument",
1.126 schwarze 167: "wrong number of cells",
1.7 kristaps 168: "unknown AT&T UNIX version",
1.88 schwarze 169: "comma in function argument",
1.89 schwarze 170: "parenthesis in function name",
1.67 schwarze 171: "invalid content in Rs block",
1.63 schwarze 172: "invalid Boolean argument",
173: "unknown font, skipping request",
1.126 schwarze 174: "odd number of characters in request",
1.7 kristaps 175:
176: /* related to plain text */
1.64 schwarze 177: "blank line in fill mode, using .sp",
178: "tab in filled text",
179: "whitespace at end of input line",
1.7 kristaps 180: "bad comment style",
1.64 schwarze 181: "invalid escape sequence",
182: "undefined string, using \"\"",
1.16 kristaps 183:
1.118 schwarze 184: /* related to tables */
185: "tbl line starts with span",
186: "tbl column starts with span",
187: "skipping vertical bar in tbl layout",
188:
1.7 kristaps 189: "generic error",
190:
191: /* related to tables */
1.116 schwarze 192: "non-alphabetic character in tbl options",
193: "skipping unknown tbl option",
194: "missing tbl option argument",
195: "wrong tbl option argument size",
1.118 schwarze 196: "empty tbl layout",
197: "invalid character in tbl layout",
198: "unmatched parenthesis in tbl layout",
1.119 schwarze 199: "tbl without any data cells",
200: "ignoring data in spanned tbl cell",
201: "ignoring extra tbl data cells",
202: "data block open at end of tbl",
1.7 kristaps 203:
1.68 schwarze 204: /* related to document structure and macros */
1.108 schwarze 205: NULL,
1.7 kristaps 206: "input stack limit exceeded, infinite loop?",
207: "skipping bad character",
1.68 schwarze 208: "skipping unknown macro",
1.112 schwarze 209: "skipping insecure request",
1.73 schwarze 210: "skipping item outside list",
1.68 schwarze 211: "skipping column outside column list",
212: "skipping end of block that is not open",
1.115 schwarze 213: "fewer RS blocks open, skipping",
1.68 schwarze 214: "inserting missing end of block",
215: "appending missing end of block",
216:
217: /* related to request and macro arguments */
1.7 kristaps 218: "escaped character not allowed in a name",
1.99 schwarze 219: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 220: "missing list type, using -item",
1.70 schwarze 221: "missing manual name, using \"\"",
1.71 schwarze 222: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 223: "unknown standard specifier",
1.71 schwarze 224: "skipping request without numeric argument",
1.109 schwarze 225: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 226: ".so request failed",
1.61 schwarze 227: "skipping all arguments",
228: "skipping excess arguments",
1.92 kristaps 229: "divide by zero",
1.112 schwarze 230:
231: "unsupported feature",
232: "input too large",
1.114 schwarze 233: "unsupported control character",
1.112 schwarze 234: "unsupported roff request",
1.119 schwarze 235: "eqn delim option in tbl",
1.118 schwarze 236: "unsupported tbl layout modifier",
1.112 schwarze 237: "ignoring macro in table",
1.7 kristaps 238: };
239:
240: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
241: "SUCCESS",
242: "RESERVED",
243: "WARNING",
244: "ERROR",
1.112 schwarze 245: "UNSUPP",
1.7 kristaps 246: "BADARG",
247: "SYSERR"
248: };
249:
1.47 schwarze 250:
1.1 kristaps 251: static void
252: resize_buf(struct buf *buf, size_t initial)
253: {
254:
255: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
256: buf->buf = mandoc_realloc(buf->buf, buf->sz);
257: }
258:
259: static void
1.84 schwarze 260: choose_parser(struct mparse *curp)
1.1 kristaps 261: {
1.83 schwarze 262: char *cp, *ep;
263: int format;
1.1 kristaps 264:
1.83 schwarze 265: /*
266: * If neither command line arguments -mdoc or -man select
267: * a parser nor the roff parser found a .Dd or .TH macro
268: * yet, look ahead in the main input buffer.
269: */
270:
271: if ((format = roff_getformat(curp->roff)) == 0) {
272: cp = curp->primary->buf;
273: ep = cp + curp->primary->sz;
274: while (cp < ep) {
1.85 schwarze 275: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 276: cp++;
277: if (cp[0] == 'D' && cp[1] == 'd') {
278: format = MPARSE_MDOC;
279: break;
280: }
281: if (cp[0] == 'T' && cp[1] == 'H') {
282: format = MPARSE_MAN;
283: break;
284: }
285: }
286: cp = memchr(cp, '\n', ep - cp);
287: if (cp == NULL)
288: break;
289: cp++;
290: }
1.1 kristaps 291: }
292:
1.137 schwarze 293: if (curp->man == NULL) {
294: curp->man = roff_man_alloc(curp->roff, curp, curp->defos,
295: curp->options & MPARSE_QUICK ? 1 : 0);
296: curp->man->macroset = MACROSET_MAN;
1.138 schwarze 297: curp->man->first->tok = TOKEN_NONE;
1.137 schwarze 298: }
299:
1.83 schwarze 300: if (format == MPARSE_MDOC) {
1.136 schwarze 301: mdoc_hash_init();
1.137 schwarze 302: curp->man->macroset = MACROSET_MDOC;
1.138 schwarze 303: curp->man->first->tok = TOKEN_NONE;
1.137 schwarze 304: } else {
305: man_hash_init();
306: curp->man->macroset = MACROSET_MAN;
1.138 schwarze 307: curp->man->first->tok = TOKEN_NONE;
1.47 schwarze 308: }
1.1 kristaps 309: }
310:
311: /*
1.95 schwarze 312: * Main parse routine for a buffer.
313: * It assumes encoding and line numbering are already set up.
314: * It can recurse directly (for invocations of user-defined
315: * macros, inline equations, and input line traps)
316: * and indirectly (for .so file inclusion).
1.1 kristaps 317: */
318: static void
1.95 schwarze 319: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 320: {
321: const struct tbl_span *span;
322: struct buf ln;
1.113 schwarze 323: const char *save_file;
1.110 schwarze 324: char *cp;
1.95 schwarze 325: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 326: enum rofferr rr;
1.100 schwarze 327: int of;
1.1 kristaps 328: int lnn; /* line number in the real file */
1.110 schwarze 329: int fd;
330: pid_t save_child;
1.1 kristaps 331: unsigned char c;
332:
1.95 schwarze 333: memset(&ln, 0, sizeof(ln));
1.1 kristaps 334:
1.47 schwarze 335: lnn = curp->line;
336: pos = 0;
1.1 kristaps 337:
1.95 schwarze 338: while (i < blk.sz) {
1.1 kristaps 339: if (0 == pos && '\0' == blk.buf[i])
340: break;
341:
342: if (start) {
343: curp->line = lnn;
344: curp->reparse_count = 0;
1.93 schwarze 345:
346: if (lnn < 3 &&
347: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 348: curp->filenc & MPARSE_LATIN1)
349: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 350: }
351:
1.95 schwarze 352: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 353:
354: /*
355: * When finding an unescaped newline character,
356: * leave the character loop to process the line.
357: * Skip a preceding carriage return, if any.
358: */
359:
1.95 schwarze 360: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 361: '\n' == blk.buf[i + 1])
362: ++i;
363: if ('\n' == blk.buf[i]) {
364: ++i;
365: ++lnn;
366: break;
367: }
368:
1.35 schwarze 369: /*
1.93 schwarze 370: * Make sure we have space for the worst
371: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 372: */
373:
1.95 schwarze 374: if (pos + 11 > ln.sz)
1.35 schwarze 375: resize_buf(&ln, 256);
376:
1.47 schwarze 377: /*
1.93 schwarze 378: * Encode 8-bit input.
1.1 kristaps 379: */
380:
1.93 schwarze 381: c = blk.buf[i];
382: if (c & 0x80) {
1.95 schwarze 383: if ( ! (curp->filenc && preconv_encode(
384: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 385: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
386: curp->line, pos, "0x%x", c);
1.93 schwarze 387: ln.buf[pos++] = '?';
388: i++;
389: }
390: continue;
391: }
392:
393: /*
394: * Exclude control characters.
395: */
1.1 kristaps 396:
1.93 schwarze 397: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 398: mandoc_vmsg(c == 0x00 || c == 0x04 ||
399: c > 0x0a ? MANDOCERR_CHAR_BAD :
400: MANDOCERR_CHAR_UNSUPP,
401: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 402: i++;
1.127 schwarze 403: if (c != '\r')
404: ln.buf[pos++] = '?';
1.1 kristaps 405: continue;
406: }
407:
408: /* Trailing backslash = a plain char. */
409:
1.95 schwarze 410: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 411: ln.buf[pos++] = blk.buf[i++];
412: continue;
413: }
414:
415: /*
416: * Found escape and at least one other character.
417: * When it's a newline character, skip it.
418: * When there is a carriage return in between,
419: * skip that one as well.
420: */
421:
1.95 schwarze 422: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 423: '\n' == blk.buf[i + 2])
424: ++i;
425: if ('\n' == blk.buf[i + 1]) {
426: i += 2;
427: ++lnn;
428: continue;
429: }
430:
1.13 kristaps 431: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 432: i += 2;
433: /* Comment, skip to end of line */
1.95 schwarze 434: for (; i < blk.sz; ++i) {
1.1 kristaps 435: if ('\n' == blk.buf[i]) {
436: ++i;
437: ++lnn;
438: break;
439: }
440: }
441:
442: /* Backout trailing whitespaces */
443: for (; pos > 0; --pos) {
444: if (ln.buf[pos - 1] != ' ')
445: break;
446: if (pos > 2 && ln.buf[pos - 2] == '\\')
447: break;
448: }
449: break;
450: }
451:
1.35 schwarze 452: /* Catch escaped bogus characters. */
453:
454: c = (unsigned char) blk.buf[i+1];
455:
1.47 schwarze 456: if ( ! (isascii(c) &&
457: (isgraph(c) || isblank(c)))) {
1.114 schwarze 458: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 459: curp->line, pos, "0x%x", c);
1.35 schwarze 460: i += 2;
461: ln.buf[pos++] = '?';
462: continue;
463: }
464:
1.1 kristaps 465: /* Some other escape sequence, copy & cont. */
466:
467: ln.buf[pos++] = blk.buf[i++];
468: ln.buf[pos++] = blk.buf[i++];
469: }
470:
1.95 schwarze 471: if (pos >= ln.sz)
1.1 kristaps 472: resize_buf(&ln, 256);
473:
474: ln.buf[pos] = '\0';
475:
476: /*
477: * A significant amount of complexity is contained by
478: * the roff preprocessor. It's line-oriented but can be
479: * expressed on one line, so we need at times to
480: * readjust our starting point and re-run it. The roff
481: * preprocessor can also readjust the buffers with new
482: * data, so we pass them in wholesale.
483: */
484:
485: of = 0;
486:
1.24 kristaps 487: /*
488: * Maintain a lookaside buffer of all parsed lines. We
489: * only do this if mparse_keep() has been invoked (the
490: * buffer may be accessed with mparse_getkeep()).
491: */
492:
493: if (curp->secondary) {
1.47 schwarze 494: curp->secondary->buf = mandoc_realloc(
495: curp->secondary->buf,
496: curp->secondary->sz + pos + 2);
497: memcpy(curp->secondary->buf +
498: curp->secondary->sz,
499: ln.buf, pos);
1.24 kristaps 500: curp->secondary->sz += pos;
501: curp->secondary->buf
502: [curp->secondary->sz] = '\n';
503: curp->secondary->sz++;
504: curp->secondary->buf
505: [curp->secondary->sz] = '\0';
506: }
1.1 kristaps 507: rerun:
1.96 schwarze 508: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 509:
510: switch (rr) {
1.47 schwarze 511: case ROFF_REPARSE:
1.1 kristaps 512: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 513: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 514: else
1.3 kristaps 515: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 516: curp->line, pos, NULL);
1.1 kristaps 517: pos = 0;
518: continue;
1.47 schwarze 519: case ROFF_APPEND:
1.95 schwarze 520: pos = strlen(ln.buf);
1.1 kristaps 521: continue;
1.47 schwarze 522: case ROFF_RERUN:
1.1 kristaps 523: goto rerun;
1.47 schwarze 524: case ROFF_IGN:
1.1 kristaps 525: pos = 0;
526: continue;
1.47 schwarze 527: case ROFF_SO:
1.95 schwarze 528: if ( ! (curp->options & MPARSE_SO) &&
529: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 530: curp->sodest = mandoc_strdup(ln.buf + of);
531: free(ln.buf);
532: return;
533: }
1.24 kristaps 534: /*
535: * We remove `so' clauses from our lookaside
536: * buffer because we're going to descend into
537: * the file recursively.
538: */
1.47 schwarze 539: if (curp->secondary)
1.25 kristaps 540: curp->secondary->sz -= pos + 1;
1.113 schwarze 541: save_file = curp->file;
1.110 schwarze 542: save_child = curp->child;
543: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 schwarze 544: MANDOCLEVEL_OK) {
1.110 schwarze 545: mparse_readfd(curp, fd, ln.buf + of);
1.113 schwarze 546: curp->file = save_file;
547: } else {
548: curp->file = save_file;
1.52 schwarze 549: mandoc_vmsg(MANDOCERR_SO_FAIL,
550: curp, curp->line, pos,
551: ".so %s", ln.buf + of);
1.110 schwarze 552: ln.sz = mandoc_asprintf(&cp,
553: ".sp\nSee the file %s.\n.sp",
554: ln.buf + of);
555: free(ln.buf);
556: ln.buf = cp;
557: of = 0;
558: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 559: }
1.110 schwarze 560: curp->child = save_child;
1.1 kristaps 561: pos = 0;
562: continue;
563: default:
564: break;
565: }
566:
567: /*
568: * If input parsers have not been allocated, do so now.
1.14 kristaps 569: * We keep these instanced between parsers, but set them
1.1 kristaps 570: * locally per parse routine since we can use different
571: * parsers with each one.
572: */
573:
1.135 schwarze 574: if (curp->man == NULL ||
575: curp->man->macroset == MACROSET_NONE)
1.84 schwarze 576: choose_parser(curp);
1.1 kristaps 577:
1.47 schwarze 578: /*
1.84 schwarze 579: * Lastly, push down into the parsers themselves.
1.1 kristaps 580: * If libroff returns ROFF_TBL, then add it to the
581: * currently open parse. Since we only get here if
582: * there does exist data (see tbl_data.c), we're
583: * guaranteed that something's been allocated.
584: * Do the same for ROFF_EQN.
585: */
586:
1.139 ! schwarze 587: if (rr == ROFF_TBL)
1.100 schwarze 588: while ((span = roff_span(curp->roff)) != NULL)
1.139 ! schwarze 589: roff_addtbl(curp->man, span);
! 590: else if (rr == ROFF_EQN)
! 591: roff_addeqn(curp->man, roff_eqn(curp->roff));
! 592: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 593: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 594: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
595: break;
1.1 kristaps 596:
597: /* Temporary buffers typically are not full. */
598:
599: if (0 == start && '\0' == blk.buf[i])
600: break;
601:
602: /* Start the next input line. */
603:
604: pos = 0;
605: }
606:
607: free(ln.buf);
608: }
609:
610: static int
1.40 schwarze 611: read_whole_file(struct mparse *curp, const char *file, int fd,
612: struct buf *fb, int *with_mmap)
1.1 kristaps 613: {
614: size_t off;
615: ssize_t ssz;
616:
1.81 schwarze 617: #if HAVE_MMAP
1.15 kristaps 618: struct stat st;
1.1 kristaps 619: if (-1 == fstat(fd, &st)) {
1.107 schwarze 620: perror(file);
621: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 622: }
623:
624: /*
625: * If we're a regular file, try just reading in the whole entry
626: * via mmap(). This is faster than reading it into blocks, and
627: * since each file is only a few bytes to begin with, I'm not
628: * concerned that this is going to tank any machines.
629: */
630:
631: if (S_ISREG(st.st_mode)) {
1.131 schwarze 632: if (st.st_size > 0x7fffffff) {
1.111 schwarze 633: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 634: return(0);
635: }
636: *with_mmap = 1;
637: fb->sz = (size_t)st.st_size;
1.37 schwarze 638: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 639: if (fb->buf != MAP_FAILED)
640: return(1);
641: }
1.15 kristaps 642: #endif
1.1 kristaps 643:
644: /*
645: * If this isn't a regular file (like, say, stdin), then we must
646: * go the old way and just read things in bit by bit.
647: */
648:
649: *with_mmap = 0;
650: off = 0;
651: fb->sz = 0;
652: fb->buf = NULL;
653: for (;;) {
654: if (off == fb->sz) {
655: if (fb->sz == (1U << 31)) {
1.111 schwarze 656: mandoc_msg(MANDOCERR_TOOLARGE, curp,
657: 0, 0, NULL);
1.1 kristaps 658: break;
659: }
660: resize_buf(fb, 65536);
661: }
662: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
663: if (ssz == 0) {
664: fb->sz = off;
665: return(1);
666: }
667: if (ssz == -1) {
1.107 schwarze 668: perror(file);
669: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 670: }
671: off += (size_t)ssz;
672: }
673:
674: free(fb->buf);
675: fb->buf = NULL;
676: return(0);
677: }
678:
679: static void
680: mparse_end(struct mparse *curp)
681: {
682:
1.135 schwarze 683: if (curp->man == NULL && curp->sodest == NULL)
1.137 schwarze 684: curp->man = roff_man_alloc(curp->roff, curp, curp->defos,
1.135 schwarze 685: curp->options & MPARSE_QUICK ? 1 : 0);
686: if (curp->man->macroset == MACROSET_NONE)
687: curp->man->macroset = MACROSET_MAN;
688: if (curp->man->macroset == MACROSET_MDOC)
689: mdoc_endparse(curp->man);
690: else
1.111 schwarze 691: man_endparse(curp->man);
1.1 kristaps 692: roff_endparse(curp->roff);
693: }
694:
695: static void
1.36 schwarze 696: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 697: {
1.85 schwarze 698: struct buf *svprimary;
1.28 joerg 699: const char *svfile;
1.95 schwarze 700: size_t offset;
1.36 schwarze 701: static int recursion_depth;
702:
703: if (64 < recursion_depth) {
704: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
705: return;
706: }
1.28 joerg 707:
708: /* Line number is per-file. */
709: svfile = curp->file;
710: curp->file = file;
1.85 schwarze 711: svprimary = curp->primary;
1.83 schwarze 712: curp->primary = &blk;
1.28 joerg 713: curp->line = 1;
1.36 schwarze 714: recursion_depth++;
1.28 joerg 715:
1.93 schwarze 716: /* Skip an UTF-8 byte order mark. */
717: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
718: (unsigned char)blk.buf[0] == 0xef &&
719: (unsigned char)blk.buf[1] == 0xbb &&
720: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 721: offset = 3;
1.93 schwarze 722: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 723: } else
724: offset = 0;
1.93 schwarze 725:
1.95 schwarze 726: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 727:
1.111 schwarze 728: if (--recursion_depth == 0)
1.28 joerg 729: mparse_end(curp);
730:
1.85 schwarze 731: curp->primary = svprimary;
1.28 joerg 732: curp->file = svfile;
733: }
734:
735: enum mandoclevel
1.104 schwarze 736: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 737: const char *file)
738: {
739: struct buf blk;
740:
1.104 schwarze 741: blk.buf = buf;
1.28 joerg 742: blk.sz = len;
743:
1.36 schwarze 744: mparse_parse_buffer(curp, blk, file);
1.28 joerg 745: return(curp->file_status);
746: }
747:
1.98 schwarze 748: /*
749: * Read the whole file into memory and call the parsers.
750: * Called recursively when an .so request is encountered.
751: */
1.36 schwarze 752: enum mandoclevel
753: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 754: {
1.28 joerg 755: struct buf blk;
756: int with_mmap;
1.93 schwarze 757: int save_filenc;
1.1 kristaps 758:
1.91 schwarze 759: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 760: save_filenc = curp->filenc;
761: curp->filenc = curp->options &
762: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 763: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 764: curp->filenc = save_filenc;
1.81 schwarze 765: #if HAVE_MMAP
1.91 schwarze 766: if (with_mmap)
767: munmap(blk.buf, blk.sz);
768: else
1.28 joerg 769: #endif
1.91 schwarze 770: free(blk.buf);
771: }
1.1 kristaps 772:
1.98 schwarze 773: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 774: perror(file);
1.91 schwarze 775:
1.98 schwarze 776: mparse_wait(curp);
1.1 kristaps 777: return(curp->file_status);
1.82 schwarze 778: }
779:
780: enum mandoclevel
1.97 schwarze 781: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 782: {
783: int pfd[2];
1.98 schwarze 784: int save_errno;
1.82 schwarze 785: char *cp;
786:
787: curp->file = file;
1.98 schwarze 788:
789: /* Unless zipped, try to just open the file. */
790:
1.82 schwarze 791: if ((cp = strrchr(file, '.')) == NULL ||
792: strcmp(cp + 1, "gz")) {
1.97 schwarze 793: curp->child = 0;
1.98 schwarze 794: if ((*fd = open(file, O_RDONLY)) != -1)
795: return(MANDOCLEVEL_OK);
796:
797: /* Open failed; try to append ".gz". */
798:
799: mandoc_asprintf(&cp, "%s.gz", file);
800: file = cp;
801: } else
802: cp = NULL;
803:
804: /* Before forking, make sure the file can be read. */
805:
806: save_errno = errno;
807: if (access(file, R_OK) == -1) {
808: if (cp != NULL)
809: errno = save_errno;
1.108 schwarze 810: free(cp);
811: *fd = -1;
812: curp->child = 0;
813: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
814: return(MANDOCLEVEL_ERROR);
1.82 schwarze 815: }
816:
1.98 schwarze 817: /* Run gunzip(1). */
818:
1.82 schwarze 819: if (pipe(pfd) == -1) {
1.107 schwarze 820: perror("pipe");
821: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 822: }
823:
1.97 schwarze 824: switch (curp->child = fork()) {
1.82 schwarze 825: case -1:
1.107 schwarze 826: perror("fork");
827: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 828: case 0:
829: close(pfd[0]);
830: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 831: perror("dup");
832: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 833: }
1.132 schwarze 834: signal(SIGPIPE, SIG_DFL);
1.82 schwarze 835: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 836: perror("exec");
837: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 838: default:
839: close(pfd[1]);
840: *fd = pfd[0];
841: return(MANDOCLEVEL_OK);
842: }
843: }
844:
845: enum mandoclevel
1.97 schwarze 846: mparse_wait(struct mparse *curp)
1.82 schwarze 847: {
848: int status;
849:
1.97 schwarze 850: if (curp->child == 0)
851: return(MANDOCLEVEL_OK);
852:
853: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 854: perror("wait");
855: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 856: }
1.129 schwarze 857: curp->child = 0;
1.82 schwarze 858: if (WIFSIGNALED(status)) {
1.108 schwarze 859: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
860: "gunzip died from signal %d", WTERMSIG(status));
861: return(MANDOCLEVEL_ERROR);
1.82 schwarze 862: }
863: if (WEXITSTATUS(status)) {
1.108 schwarze 864: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
865: "gunzip failed with code %d", WEXITSTATUS(status));
866: return(MANDOCLEVEL_ERROR);
1.82 schwarze 867: }
868: return(MANDOCLEVEL_OK);
1.1 kristaps 869: }
870:
871: struct mparse *
1.94 schwarze 872: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
873: const struct mchars *mchars, const char *defos)
1.1 kristaps 874: {
875: struct mparse *curp;
1.10 kristaps 876:
1.1 kristaps 877: curp = mandoc_calloc(1, sizeof(struct mparse));
878:
1.44 schwarze 879: curp->options = options;
1.3 kristaps 880: curp->wlevel = wlevel;
1.1 kristaps 881: curp->mmsg = mmsg;
1.29 schwarze 882: curp->defos = defos;
1.1 kristaps 883:
1.94 schwarze 884: curp->mchars = mchars;
885: curp->roff = roff_alloc(curp, curp->mchars, options);
1.137 schwarze 886: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
887: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 888: if (curp->options & MPARSE_MDOC) {
889: mdoc_hash_init();
1.137 schwarze 890: curp->man->macroset = MACROSET_MDOC;
891: } else if (curp->options & MPARSE_MAN) {
1.136 schwarze 892: man_hash_init();
1.137 schwarze 893: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 894: }
1.138 schwarze 895: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 896: return(curp);
897: }
898:
899: void
900: mparse_reset(struct mparse *curp)
901: {
902:
903: roff_reset(curp->roff);
904:
1.137 schwarze 905: if (curp->man != NULL)
906: roff_man_reset(curp->man);
1.24 kristaps 907: if (curp->secondary)
908: curp->secondary->sz = 0;
1.1 kristaps 909:
910: curp->file_status = MANDOCLEVEL_OK;
1.45 schwarze 911:
912: free(curp->sodest);
913: curp->sodest = NULL;
1.1 kristaps 914: }
915:
916: void
917: mparse_free(struct mparse *curp)
918: {
919:
1.137 schwarze 920: roff_man_free(curp->man);
1.1 kristaps 921: if (curp->roff)
922: roff_free(curp->roff);
1.24 kristaps 923: if (curp->secondary)
924: free(curp->secondary->buf);
1.1 kristaps 925:
1.24 kristaps 926: free(curp->secondary);
1.45 schwarze 927: free(curp->sodest);
1.1 kristaps 928: free(curp);
929: }
930:
931: void
1.135 schwarze 932: mparse_result(struct mparse *curp, struct roff_man **man,
933: char **sodest)
1.1 kristaps 934: {
935:
1.45 schwarze 936: if (sodest && NULL != (*sodest = curp->sodest)) {
937: *man = NULL;
938: return;
939: }
1.9 kristaps 940: if (man)
941: *man = curp->man;
1.3 kristaps 942: }
943:
944: void
945: mandoc_vmsg(enum mandocerr t, struct mparse *m,
946: int ln, int pos, const char *fmt, ...)
947: {
948: char buf[256];
949: va_list ap;
950:
951: va_start(ap, fmt);
1.48 schwarze 952: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 953: va_end(ap);
954:
955: mandoc_msg(t, m, ln, pos, buf);
956: }
957:
958: void
1.47 schwarze 959: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 960: int ln, int col, const char *msg)
961: {
962: enum mandoclevel level;
963:
1.112 schwarze 964: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 965: while (er < mandoclimits[level])
966: level--;
967:
1.108 schwarze 968: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 969: return;
970:
1.8 kristaps 971: if (m->mmsg)
972: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 973:
974: if (m->file_status < level)
975: m->file_status = level;
1.7 kristaps 976: }
977:
978: const char *
979: mparse_strerror(enum mandocerr er)
980: {
981:
982: return(mandocerrs[er]);
983: }
984:
985: const char *
986: mparse_strlevel(enum mandoclevel lvl)
987: {
988: return(mandoclevels[lvl]);
1.24 kristaps 989: }
990:
991: void
992: mparse_keep(struct mparse *p)
993: {
994:
995: assert(NULL == p->secondary);
996: p->secondary = mandoc_calloc(1, sizeof(struct buf));
997: }
998:
999: const char *
1000: mparse_getkeep(const struct mparse *p)
1001: {
1002:
1003: assert(p->secondary);
1004: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1005: }
CVSweb