Annotation of mandoc/read.c, Revision 1.136
1.136 ! schwarze 1: /* $Id: read.c,v 1.135 2015/04/18 16:34:25 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;
1.136 ! schwarze 299: mdoc_hash_init();
1.1 kristaps 300: return;
1.47 schwarze 301: }
1.1 kristaps 302:
1.83 schwarze 303: /* Fall back to man(7) as a last resort. */
304:
1.135 schwarze 305: if (curp->man == NULL)
306: curp->man = man_alloc(
1.106 schwarze 307: curp->roff, curp, curp->defos,
1.44 schwarze 308: MPARSE_QUICK & curp->options ? 1 : 0);
1.135 schwarze 309: else
310: curp->man->macroset = MACROSET_MAN;
1.136 ! schwarze 311: man_hash_init();
1.1 kristaps 312: }
313:
314: /*
1.95 schwarze 315: * Main parse routine for a buffer.
316: * It assumes encoding and line numbering are already set up.
317: * It can recurse directly (for invocations of user-defined
318: * macros, inline equations, and input line traps)
319: * and indirectly (for .so file inclusion).
1.1 kristaps 320: */
321: static void
1.95 schwarze 322: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 323: {
324: const struct tbl_span *span;
325: struct buf ln;
1.113 schwarze 326: const char *save_file;
1.110 schwarze 327: char *cp;
1.95 schwarze 328: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 329: enum rofferr rr;
1.100 schwarze 330: int of;
1.1 kristaps 331: int lnn; /* line number in the real file */
1.110 schwarze 332: int fd;
333: pid_t save_child;
1.1 kristaps 334: unsigned char c;
335:
1.95 schwarze 336: memset(&ln, 0, sizeof(ln));
1.1 kristaps 337:
1.47 schwarze 338: lnn = curp->line;
339: pos = 0;
1.1 kristaps 340:
1.95 schwarze 341: while (i < blk.sz) {
1.1 kristaps 342: if (0 == pos && '\0' == blk.buf[i])
343: break;
344:
345: if (start) {
346: curp->line = lnn;
347: curp->reparse_count = 0;
1.93 schwarze 348:
349: if (lnn < 3 &&
350: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 351: curp->filenc & MPARSE_LATIN1)
352: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 353: }
354:
1.95 schwarze 355: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 356:
357: /*
358: * When finding an unescaped newline character,
359: * leave the character loop to process the line.
360: * Skip a preceding carriage return, if any.
361: */
362:
1.95 schwarze 363: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 364: '\n' == blk.buf[i + 1])
365: ++i;
366: if ('\n' == blk.buf[i]) {
367: ++i;
368: ++lnn;
369: break;
370: }
371:
1.35 schwarze 372: /*
1.93 schwarze 373: * Make sure we have space for the worst
374: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 375: */
376:
1.95 schwarze 377: if (pos + 11 > ln.sz)
1.35 schwarze 378: resize_buf(&ln, 256);
379:
1.47 schwarze 380: /*
1.93 schwarze 381: * Encode 8-bit input.
1.1 kristaps 382: */
383:
1.93 schwarze 384: c = blk.buf[i];
385: if (c & 0x80) {
1.95 schwarze 386: if ( ! (curp->filenc && preconv_encode(
387: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 388: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
389: curp->line, pos, "0x%x", c);
1.93 schwarze 390: ln.buf[pos++] = '?';
391: i++;
392: }
393: continue;
394: }
395:
396: /*
397: * Exclude control characters.
398: */
1.1 kristaps 399:
1.93 schwarze 400: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 401: mandoc_vmsg(c == 0x00 || c == 0x04 ||
402: c > 0x0a ? MANDOCERR_CHAR_BAD :
403: MANDOCERR_CHAR_UNSUPP,
404: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 405: i++;
1.127 schwarze 406: if (c != '\r')
407: ln.buf[pos++] = '?';
1.1 kristaps 408: continue;
409: }
410:
411: /* Trailing backslash = a plain char. */
412:
1.95 schwarze 413: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 414: ln.buf[pos++] = blk.buf[i++];
415: continue;
416: }
417:
418: /*
419: * Found escape and at least one other character.
420: * When it's a newline character, skip it.
421: * When there is a carriage return in between,
422: * skip that one as well.
423: */
424:
1.95 schwarze 425: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 426: '\n' == blk.buf[i + 2])
427: ++i;
428: if ('\n' == blk.buf[i + 1]) {
429: i += 2;
430: ++lnn;
431: continue;
432: }
433:
1.13 kristaps 434: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 435: i += 2;
436: /* Comment, skip to end of line */
1.95 schwarze 437: for (; i < blk.sz; ++i) {
1.1 kristaps 438: if ('\n' == blk.buf[i]) {
439: ++i;
440: ++lnn;
441: break;
442: }
443: }
444:
445: /* Backout trailing whitespaces */
446: for (; pos > 0; --pos) {
447: if (ln.buf[pos - 1] != ' ')
448: break;
449: if (pos > 2 && ln.buf[pos - 2] == '\\')
450: break;
451: }
452: break;
453: }
454:
1.35 schwarze 455: /* Catch escaped bogus characters. */
456:
457: c = (unsigned char) blk.buf[i+1];
458:
1.47 schwarze 459: if ( ! (isascii(c) &&
460: (isgraph(c) || isblank(c)))) {
1.114 schwarze 461: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 462: curp->line, pos, "0x%x", c);
1.35 schwarze 463: i += 2;
464: ln.buf[pos++] = '?';
465: continue;
466: }
467:
1.1 kristaps 468: /* Some other escape sequence, copy & cont. */
469:
470: ln.buf[pos++] = blk.buf[i++];
471: ln.buf[pos++] = blk.buf[i++];
472: }
473:
1.95 schwarze 474: if (pos >= ln.sz)
1.1 kristaps 475: resize_buf(&ln, 256);
476:
477: ln.buf[pos] = '\0';
478:
479: /*
480: * A significant amount of complexity is contained by
481: * the roff preprocessor. It's line-oriented but can be
482: * expressed on one line, so we need at times to
483: * readjust our starting point and re-run it. The roff
484: * preprocessor can also readjust the buffers with new
485: * data, so we pass them in wholesale.
486: */
487:
488: of = 0;
489:
1.24 kristaps 490: /*
491: * Maintain a lookaside buffer of all parsed lines. We
492: * only do this if mparse_keep() has been invoked (the
493: * buffer may be accessed with mparse_getkeep()).
494: */
495:
496: if (curp->secondary) {
1.47 schwarze 497: curp->secondary->buf = mandoc_realloc(
498: curp->secondary->buf,
499: curp->secondary->sz + pos + 2);
500: memcpy(curp->secondary->buf +
501: curp->secondary->sz,
502: ln.buf, pos);
1.24 kristaps 503: curp->secondary->sz += pos;
504: curp->secondary->buf
505: [curp->secondary->sz] = '\n';
506: curp->secondary->sz++;
507: curp->secondary->buf
508: [curp->secondary->sz] = '\0';
509: }
1.1 kristaps 510: rerun:
1.96 schwarze 511: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 512:
513: switch (rr) {
1.47 schwarze 514: case ROFF_REPARSE:
1.1 kristaps 515: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 516: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 517: else
1.3 kristaps 518: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 519: curp->line, pos, NULL);
1.1 kristaps 520: pos = 0;
521: continue;
1.47 schwarze 522: case ROFF_APPEND:
1.95 schwarze 523: pos = strlen(ln.buf);
1.1 kristaps 524: continue;
1.47 schwarze 525: case ROFF_RERUN:
1.1 kristaps 526: goto rerun;
1.47 schwarze 527: case ROFF_IGN:
1.1 kristaps 528: pos = 0;
529: continue;
1.47 schwarze 530: case ROFF_SO:
1.95 schwarze 531: if ( ! (curp->options & MPARSE_SO) &&
532: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 533: curp->sodest = mandoc_strdup(ln.buf + of);
534: free(ln.buf);
535: return;
536: }
1.24 kristaps 537: /*
538: * We remove `so' clauses from our lookaside
539: * buffer because we're going to descend into
540: * the file recursively.
541: */
1.47 schwarze 542: if (curp->secondary)
1.25 kristaps 543: curp->secondary->sz -= pos + 1;
1.113 schwarze 544: save_file = curp->file;
1.110 schwarze 545: save_child = curp->child;
546: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 schwarze 547: MANDOCLEVEL_OK) {
1.110 schwarze 548: mparse_readfd(curp, fd, ln.buf + of);
1.113 schwarze 549: curp->file = save_file;
550: } else {
551: curp->file = save_file;
1.52 schwarze 552: mandoc_vmsg(MANDOCERR_SO_FAIL,
553: curp, curp->line, pos,
554: ".so %s", ln.buf + of);
1.110 schwarze 555: ln.sz = mandoc_asprintf(&cp,
556: ".sp\nSee the file %s.\n.sp",
557: ln.buf + of);
558: free(ln.buf);
559: ln.buf = cp;
560: of = 0;
561: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 562: }
1.110 schwarze 563: curp->child = save_child;
1.1 kristaps 564: pos = 0;
565: continue;
566: default:
567: break;
568: }
569:
570: /*
571: * If input parsers have not been allocated, do so now.
1.14 kristaps 572: * We keep these instanced between parsers, but set them
1.1 kristaps 573: * locally per parse routine since we can use different
574: * parsers with each one.
575: */
576:
1.135 schwarze 577: if (curp->man == NULL ||
578: curp->man->macroset == MACROSET_NONE)
1.84 schwarze 579: choose_parser(curp);
1.1 kristaps 580:
1.47 schwarze 581: /*
1.84 schwarze 582: * Lastly, push down into the parsers themselves.
1.1 kristaps 583: * If libroff returns ROFF_TBL, then add it to the
584: * currently open parse. Since we only get here if
585: * there does exist data (see tbl_data.c), we're
586: * guaranteed that something's been allocated.
587: * Do the same for ROFF_EQN.
588: */
589:
1.100 schwarze 590: if (rr == ROFF_TBL) {
591: while ((span = roff_span(curp->roff)) != NULL)
1.135 schwarze 592: if (curp->man->macroset == MACROSET_MDOC)
593: mdoc_addspan(curp->man, span);
1.100 schwarze 594: else
595: man_addspan(curp->man, span);
596: } else if (rr == ROFF_EQN) {
1.135 schwarze 597: if (curp->man->macroset == MACROSET_MDOC)
598: mdoc_addeqn(curp->man, roff_eqn(curp->roff));
1.100 schwarze 599: else
600: man_addeqn(curp->man, roff_eqn(curp->roff));
1.135 schwarze 601: } else if ((curp->man->macroset == MACROSET_MDOC ?
602: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 603: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
604: break;
1.1 kristaps 605:
606: /* Temporary buffers typically are not full. */
607:
608: if (0 == start && '\0' == blk.buf[i])
609: break;
610:
611: /* Start the next input line. */
612:
613: pos = 0;
614: }
615:
616: free(ln.buf);
617: }
618:
619: static int
1.40 schwarze 620: read_whole_file(struct mparse *curp, const char *file, int fd,
621: struct buf *fb, int *with_mmap)
1.1 kristaps 622: {
623: size_t off;
624: ssize_t ssz;
625:
1.81 schwarze 626: #if HAVE_MMAP
1.15 kristaps 627: struct stat st;
1.1 kristaps 628: if (-1 == fstat(fd, &st)) {
1.107 schwarze 629: perror(file);
630: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 631: }
632:
633: /*
634: * If we're a regular file, try just reading in the whole entry
635: * via mmap(). This is faster than reading it into blocks, and
636: * since each file is only a few bytes to begin with, I'm not
637: * concerned that this is going to tank any machines.
638: */
639:
640: if (S_ISREG(st.st_mode)) {
1.131 schwarze 641: if (st.st_size > 0x7fffffff) {
1.111 schwarze 642: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 643: return(0);
644: }
645: *with_mmap = 1;
646: fb->sz = (size_t)st.st_size;
1.37 schwarze 647: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 648: if (fb->buf != MAP_FAILED)
649: return(1);
650: }
1.15 kristaps 651: #endif
1.1 kristaps 652:
653: /*
654: * If this isn't a regular file (like, say, stdin), then we must
655: * go the old way and just read things in bit by bit.
656: */
657:
658: *with_mmap = 0;
659: off = 0;
660: fb->sz = 0;
661: fb->buf = NULL;
662: for (;;) {
663: if (off == fb->sz) {
664: if (fb->sz == (1U << 31)) {
1.111 schwarze 665: mandoc_msg(MANDOCERR_TOOLARGE, curp,
666: 0, 0, NULL);
1.1 kristaps 667: break;
668: }
669: resize_buf(fb, 65536);
670: }
671: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
672: if (ssz == 0) {
673: fb->sz = off;
674: return(1);
675: }
676: if (ssz == -1) {
1.107 schwarze 677: perror(file);
678: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 679: }
680: off += (size_t)ssz;
681: }
682:
683: free(fb->buf);
684: fb->buf = NULL;
685: return(0);
686: }
687:
688: static void
689: mparse_end(struct mparse *curp)
690: {
691:
1.135 schwarze 692: if (curp->man == NULL && curp->sodest == NULL)
693: curp->man = man_alloc(curp->roff, curp, curp->defos,
694: curp->options & MPARSE_QUICK ? 1 : 0);
695: if (curp->man->macroset == MACROSET_NONE)
696: curp->man->macroset = MACROSET_MAN;
697: if (curp->man->macroset == MACROSET_MDOC)
698: mdoc_endparse(curp->man);
699: else
1.111 schwarze 700: man_endparse(curp->man);
1.1 kristaps 701: roff_endparse(curp->roff);
702: }
703:
704: static void
1.36 schwarze 705: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 706: {
1.85 schwarze 707: struct buf *svprimary;
1.28 joerg 708: const char *svfile;
1.95 schwarze 709: size_t offset;
1.36 schwarze 710: static int recursion_depth;
711:
712: if (64 < recursion_depth) {
713: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
714: return;
715: }
1.28 joerg 716:
717: /* Line number is per-file. */
718: svfile = curp->file;
719: curp->file = file;
1.85 schwarze 720: svprimary = curp->primary;
1.83 schwarze 721: curp->primary = &blk;
1.28 joerg 722: curp->line = 1;
1.36 schwarze 723: recursion_depth++;
1.28 joerg 724:
1.93 schwarze 725: /* Skip an UTF-8 byte order mark. */
726: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
727: (unsigned char)blk.buf[0] == 0xef &&
728: (unsigned char)blk.buf[1] == 0xbb &&
729: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 730: offset = 3;
1.93 schwarze 731: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 732: } else
733: offset = 0;
1.93 schwarze 734:
1.95 schwarze 735: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 736:
1.111 schwarze 737: if (--recursion_depth == 0)
1.28 joerg 738: mparse_end(curp);
739:
1.85 schwarze 740: curp->primary = svprimary;
1.28 joerg 741: curp->file = svfile;
742: }
743:
744: enum mandoclevel
1.104 schwarze 745: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 746: const char *file)
747: {
748: struct buf blk;
749:
1.104 schwarze 750: blk.buf = buf;
1.28 joerg 751: blk.sz = len;
752:
1.36 schwarze 753: mparse_parse_buffer(curp, blk, file);
1.28 joerg 754: return(curp->file_status);
755: }
756:
1.98 schwarze 757: /*
758: * Read the whole file into memory and call the parsers.
759: * Called recursively when an .so request is encountered.
760: */
1.36 schwarze 761: enum mandoclevel
762: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 763: {
1.28 joerg 764: struct buf blk;
765: int with_mmap;
1.93 schwarze 766: int save_filenc;
1.1 kristaps 767:
1.91 schwarze 768: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 769: save_filenc = curp->filenc;
770: curp->filenc = curp->options &
771: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 772: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 773: curp->filenc = save_filenc;
1.81 schwarze 774: #if HAVE_MMAP
1.91 schwarze 775: if (with_mmap)
776: munmap(blk.buf, blk.sz);
777: else
1.28 joerg 778: #endif
1.91 schwarze 779: free(blk.buf);
780: }
1.1 kristaps 781:
1.98 schwarze 782: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 783: perror(file);
1.91 schwarze 784:
1.98 schwarze 785: mparse_wait(curp);
1.1 kristaps 786: return(curp->file_status);
1.82 schwarze 787: }
788:
789: enum mandoclevel
1.97 schwarze 790: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 791: {
792: int pfd[2];
1.98 schwarze 793: int save_errno;
1.82 schwarze 794: char *cp;
795:
796: curp->file = file;
1.98 schwarze 797:
798: /* Unless zipped, try to just open the file. */
799:
1.82 schwarze 800: if ((cp = strrchr(file, '.')) == NULL ||
801: strcmp(cp + 1, "gz")) {
1.97 schwarze 802: curp->child = 0;
1.98 schwarze 803: if ((*fd = open(file, O_RDONLY)) != -1)
804: return(MANDOCLEVEL_OK);
805:
806: /* Open failed; try to append ".gz". */
807:
808: mandoc_asprintf(&cp, "%s.gz", file);
809: file = cp;
810: } else
811: cp = NULL;
812:
813: /* Before forking, make sure the file can be read. */
814:
815: save_errno = errno;
816: if (access(file, R_OK) == -1) {
817: if (cp != NULL)
818: errno = save_errno;
1.108 schwarze 819: free(cp);
820: *fd = -1;
821: curp->child = 0;
822: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
823: return(MANDOCLEVEL_ERROR);
1.82 schwarze 824: }
825:
1.98 schwarze 826: /* Run gunzip(1). */
827:
1.82 schwarze 828: if (pipe(pfd) == -1) {
1.107 schwarze 829: perror("pipe");
830: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 831: }
832:
1.97 schwarze 833: switch (curp->child = fork()) {
1.82 schwarze 834: case -1:
1.107 schwarze 835: perror("fork");
836: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 837: case 0:
838: close(pfd[0]);
839: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 840: perror("dup");
841: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 842: }
1.132 schwarze 843: signal(SIGPIPE, SIG_DFL);
1.82 schwarze 844: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 845: perror("exec");
846: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 847: default:
848: close(pfd[1]);
849: *fd = pfd[0];
850: return(MANDOCLEVEL_OK);
851: }
852: }
853:
854: enum mandoclevel
1.97 schwarze 855: mparse_wait(struct mparse *curp)
1.82 schwarze 856: {
857: int status;
858:
1.97 schwarze 859: if (curp->child == 0)
860: return(MANDOCLEVEL_OK);
861:
862: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 863: perror("wait");
864: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 865: }
1.129 schwarze 866: curp->child = 0;
1.82 schwarze 867: if (WIFSIGNALED(status)) {
1.108 schwarze 868: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
869: "gunzip died from signal %d", WTERMSIG(status));
870: return(MANDOCLEVEL_ERROR);
1.82 schwarze 871: }
872: if (WEXITSTATUS(status)) {
1.108 schwarze 873: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
874: "gunzip failed with code %d", WEXITSTATUS(status));
875: return(MANDOCLEVEL_ERROR);
1.82 schwarze 876: }
877: return(MANDOCLEVEL_OK);
1.1 kristaps 878: }
879:
880: struct mparse *
1.94 schwarze 881: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
882: const struct mchars *mchars, const char *defos)
1.1 kristaps 883: {
884: struct mparse *curp;
1.10 kristaps 885:
1.1 kristaps 886: curp = mandoc_calloc(1, sizeof(struct mparse));
887:
1.44 schwarze 888: curp->options = options;
1.3 kristaps 889: curp->wlevel = wlevel;
1.1 kristaps 890: curp->mmsg = mmsg;
1.29 schwarze 891: curp->defos = defos;
1.1 kristaps 892:
1.94 schwarze 893: curp->mchars = mchars;
894: curp->roff = roff_alloc(curp, curp->mchars, options);
1.136 ! schwarze 895: if (curp->options & MPARSE_MDOC) {
1.135 schwarze 896: curp->man = mdoc_alloc(
1.72 schwarze 897: curp->roff, curp, curp->defos,
898: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 ! schwarze 899: mdoc_hash_init();
! 900: }
! 901: if (curp->options & MPARSE_MAN) {
1.135 schwarze 902: curp->man = man_alloc(
1.106 schwarze 903: curp->roff, curp, curp->defos,
1.72 schwarze 904: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 ! schwarze 905: man_hash_init();
! 906: }
1.72 schwarze 907:
1.1 kristaps 908: return(curp);
909: }
910:
911: void
912: mparse_reset(struct mparse *curp)
913: {
914:
915: roff_reset(curp->roff);
916:
1.135 schwarze 917: if (curp->man != NULL) {
918: if (curp->man->macroset == MACROSET_MDOC)
919: mdoc_reset(curp->man);
920: else
921: man_reset(curp->man);
922: curp->man->macroset = MACROSET_NONE;
923: }
1.24 kristaps 924: if (curp->secondary)
925: curp->secondary->sz = 0;
1.1 kristaps 926:
927: curp->file_status = MANDOCLEVEL_OK;
1.45 schwarze 928:
929: free(curp->sodest);
930: curp->sodest = NULL;
1.1 kristaps 931: }
932:
933: void
934: mparse_free(struct mparse *curp)
935: {
936:
1.135 schwarze 937: if (curp->man->macroset == MACROSET_MDOC)
938: mdoc_free(curp->man);
939: if (curp->man->macroset == MACROSET_MAN)
940: man_free(curp->man);
1.1 kristaps 941: if (curp->roff)
942: roff_free(curp->roff);
1.24 kristaps 943: if (curp->secondary)
944: free(curp->secondary->buf);
1.1 kristaps 945:
1.24 kristaps 946: free(curp->secondary);
1.45 schwarze 947: free(curp->sodest);
1.1 kristaps 948: free(curp);
949: }
950:
951: void
1.135 schwarze 952: mparse_result(struct mparse *curp, struct roff_man **man,
953: char **sodest)
1.1 kristaps 954: {
955:
1.45 schwarze 956: if (sodest && NULL != (*sodest = curp->sodest)) {
957: *man = NULL;
958: return;
959: }
1.9 kristaps 960: if (man)
961: *man = curp->man;
1.3 kristaps 962: }
963:
964: void
965: mandoc_vmsg(enum mandocerr t, struct mparse *m,
966: int ln, int pos, const char *fmt, ...)
967: {
968: char buf[256];
969: va_list ap;
970:
971: va_start(ap, fmt);
1.48 schwarze 972: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 973: va_end(ap);
974:
975: mandoc_msg(t, m, ln, pos, buf);
976: }
977:
978: void
1.47 schwarze 979: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 980: int ln, int col, const char *msg)
981: {
982: enum mandoclevel level;
983:
1.112 schwarze 984: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 985: while (er < mandoclimits[level])
986: level--;
987:
1.108 schwarze 988: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 989: return;
990:
1.8 kristaps 991: if (m->mmsg)
992: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 993:
994: if (m->file_status < level)
995: m->file_status = level;
1.7 kristaps 996: }
997:
998: const char *
999: mparse_strerror(enum mandocerr er)
1000: {
1001:
1002: return(mandocerrs[er]);
1003: }
1004:
1005: const char *
1006: mparse_strlevel(enum mandoclevel lvl)
1007: {
1008: return(mandoclevels[lvl]);
1.24 kristaps 1009: }
1010:
1011: void
1012: mparse_keep(struct mparse *p)
1013: {
1014:
1015: assert(NULL == p->secondary);
1016: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1017: }
1018:
1019: const char *
1020: mparse_getkeep(const struct mparse *p)
1021: {
1022:
1023: assert(p->secondary);
1024: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1025: }
CVSweb