Annotation of mandoc/read.c, Revision 1.109
1.109 ! schwarze 1: /* $Id: read.c,v 1.108 2015/01/14 22:02:50 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.107 schwarze 4: * Copyright (c) 2010-2015 Ingo Schwarze <schwarze@openbsd.org>
1.42 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 kristaps 6: *
7: * Permission to use, copy, modify, and distribute this software for any
8: * purpose with or without fee is hereby granted, provided that the above
9: * copyright notice and this permission notice appear in all copies.
10: *
11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18: */
1.11 kristaps 19: #include "config.h"
20:
1.80 schwarze 21: #include <sys/types.h>
1.81 schwarze 22: #if HAVE_MMAP
1.82 schwarze 23: #include <sys/mman.h>
1.80 schwarze 24: #include <sys/stat.h>
1.15 kristaps 25: #endif
1.82 schwarze 26: #include <sys/wait.h>
1.1 kristaps 27:
28: #include <assert.h>
29: #include <ctype.h>
1.40 schwarze 30: #include <errno.h>
1.1 kristaps 31: #include <fcntl.h>
1.3 kristaps 32: #include <stdarg.h>
1.28 joerg 33: #include <stdint.h>
1.1 kristaps 34: #include <stdio.h>
35: #include <stdlib.h>
36: #include <string.h>
37: #include <unistd.h>
38:
39: #include "mandoc.h"
1.46 schwarze 40: #include "mandoc_aux.h"
1.3 kristaps 41: #include "libmandoc.h"
1.1 kristaps 42: #include "mdoc.h"
43: #include "man.h"
44:
45: #define REPARSE_LIMIT 1000
46:
47: struct mparse {
48: struct man *pman; /* persistent man parser */
49: struct mdoc *pmdoc; /* persistent mdoc parser */
50: struct man *man; /* man parser */
51: struct mdoc *mdoc; /* mdoc parser */
52: struct roff *roff; /* roff parser (!NULL) */
1.94 schwarze 53: const struct mchars *mchars; /* character table */
1.45 schwarze 54: char *sodest; /* filename pointed to by .so */
1.83 schwarze 55: const char *file; /* filename of current input file */
56: struct buf *primary; /* buffer currently being parsed */
57: struct buf *secondary; /* preprocessed copy of input */
58: const char *defos; /* default operating system */
59: mandocmsg mmsg; /* warning/error message handler */
60: enum mandoclevel file_status; /* status of current parse */
61: enum mandoclevel wlevel; /* ignore messages below this */
62: int options; /* parser options */
1.93 schwarze 63: int filenc; /* encoding of the current file */
1.1 kristaps 64: int reparse_count; /* finite interp. stack */
1.83 schwarze 65: int line; /* line number in the file */
1.97 schwarze 66: pid_t child; /* the gunzip(1) process */
1.1 kristaps 67: };
68:
1.84 schwarze 69: static void choose_parser(struct mparse *);
1.1 kristaps 70: static void resize_buf(struct buf *, size_t);
1.95 schwarze 71: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 72: static int read_whole_file(struct mparse *, const char *, int,
73: struct buf *, int *);
1.1 kristaps 74: static void mparse_end(struct mparse *);
1.37 schwarze 75: static void mparse_parse_buffer(struct mparse *, struct buf,
76: const char *);
1.1 kristaps 77:
1.3 kristaps 78: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
79: MANDOCERR_OK,
80: MANDOCERR_WARNING,
81: MANDOCERR_WARNING,
82: MANDOCERR_ERROR,
83: MANDOCERR_FATAL,
84: MANDOCERR_MAX,
85: MANDOCERR_MAX
86: };
87:
1.7 kristaps 88: static const char * const mandocerrs[MANDOCERR_MAX] = {
89: "ok",
90:
91: "generic warning",
92:
93: /* related to the prologue */
1.79 schwarze 94: "missing manual title, using UNTITLED",
95: "missing manual title, using \"\"",
1.54 schwarze 96: "lower case character in document title",
1.79 schwarze 97: "missing manual section, using \"\"",
1.7 kristaps 98: "unknown manual section",
1.54 schwarze 99: "missing date, using today's date",
1.7 kristaps 100: "cannot parse date, using it verbatim",
1.79 schwarze 101: "missing Os macro, using \"\"",
102: "duplicate prologue macro",
103: "late prologue macro",
104: "skipping late title macro",
1.7 kristaps 105: "prologue macros out of order",
106:
107: /* related to document structure */
108: ".so is fragile, better use ln(1)",
1.50 schwarze 109: "no document body",
1.54 schwarze 110: "content before first section header",
111: "first section is not \"NAME\"",
1.7 kristaps 112: "bad NAME section contents",
113: "sections out of conventional order",
1.54 schwarze 114: "duplicate section title",
115: "unexpected section",
1.87 schwarze 116: "unusual Xr order",
117: "unusual Xr punctuation",
1.86 schwarze 118: "AUTHORS section without An macro",
1.7 kristaps 119:
120: /* related to macros and nesting */
1.55 schwarze 121: "obsolete macro",
1.102 schwarze 122: "macro neither callable nor escaped",
1.7 kristaps 123: "skipping paragraph macro",
1.31 schwarze 124: "moving paragraph macro out of list",
1.7 kristaps 125: "skipping no-space macro",
126: "blocks badly nested",
127: "nested displays are not portable",
1.57 schwarze 128: "moving content out of list",
129: ".Vt block has child macro",
1.78 schwarze 130: "fill mode already enabled, skipping",
131: "fill mode already disabled, skipping",
1.7 kristaps 132: "line scope broken",
133:
134: /* related to missing macro arguments */
1.58 schwarze 135: "skipping empty request",
136: "conditional request controls empty scope",
1.7 kristaps 137: "skipping empty macro",
1.62 schwarze 138: "empty argument, using 0n",
1.7 kristaps 139: "argument count wrong",
1.60 schwarze 140: "missing display type, using -ragged",
141: "list type is not the first argument",
142: "missing -width in -tag list, using 8n",
1.78 schwarze 143: "missing utility name, using \"\"",
1.60 schwarze 144: "empty head in list item",
145: "empty list item",
1.61 schwarze 146: "missing font type, using \\fR",
147: "unknown font type, using \\fR",
1.103 schwarze 148: "nothing follows prefix",
1.60 schwarze 149: "missing -std argument, adding it",
1.90 schwarze 150: "missing eqn box, using \"\"",
1.7 kristaps 151:
152: /* related to bad macro arguments */
1.64 schwarze 153: "unterminated quoted argument",
1.7 kristaps 154: "duplicate argument",
1.76 schwarze 155: "skipping duplicate argument",
1.63 schwarze 156: "skipping duplicate display type",
157: "skipping duplicate list type",
1.76 schwarze 158: "skipping -width argument",
1.7 kristaps 159: "unknown AT&T UNIX version",
1.88 schwarze 160: "comma in function argument",
1.89 schwarze 161: "parenthesis in function name",
1.67 schwarze 162: "invalid content in Rs block",
1.63 schwarze 163: "invalid Boolean argument",
164: "unknown font, skipping request",
1.7 kristaps 165:
166: /* related to plain text */
1.64 schwarze 167: "blank line in fill mode, using .sp",
168: "tab in filled text",
169: "whitespace at end of input line",
1.7 kristaps 170: "bad comment style",
1.64 schwarze 171: "invalid escape sequence",
172: "undefined string, using \"\"",
1.16 kristaps 173:
1.7 kristaps 174: "generic error",
1.17 kristaps 175:
176: /* related to equations */
1.20 kristaps 177: "unexpected equation scope closure",
178: "equation scope open on exit",
1.21 kristaps 179: "overlapping equation scopes",
180: "unexpected end of equation",
1.7 kristaps 181:
182: /* related to tables */
183: "bad table syntax",
184: "bad table option",
185: "bad table layout",
186: "no table layout cells specified",
187: "no table data cells specified",
188: "ignore data in cell",
189: "data block still open",
190: "ignoring extra data cells",
1.105 schwarze 191: "ignoring macro in table",
1.7 kristaps 192:
1.68 schwarze 193: /* related to document structure and macros */
1.108 schwarze 194: NULL,
1.7 kristaps 195: "input stack limit exceeded, infinite loop?",
196: "skipping bad character",
1.68 schwarze 197: "skipping unknown macro",
1.73 schwarze 198: "skipping item outside list",
1.68 schwarze 199: "skipping column outside column list",
200: "skipping end of block that is not open",
201: "inserting missing end of block",
202: "appending missing end of block",
203:
204: /* related to request and macro arguments */
1.7 kristaps 205: "escaped character not allowed in a name",
206: "argument count wrong",
1.99 schwarze 207: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 208: "missing list type, using -item",
1.70 schwarze 209: "missing manual name, using \"\"",
1.71 schwarze 210: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 211: "unknown standard specifier",
1.71 schwarze 212: "skipping request without numeric argument",
1.109 ! schwarze 213: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.61 schwarze 214: "skipping all arguments",
215: "skipping excess arguments",
1.92 kristaps 216: "divide by zero",
1.7 kristaps 217:
218: "generic fatal error",
219:
1.40 schwarze 220: "input too large",
1.52 schwarze 221: ".so request failed",
1.7 kristaps 222: };
223:
224: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
225: "SUCCESS",
226: "RESERVED",
227: "WARNING",
228: "ERROR",
229: "FATAL",
230: "BADARG",
231: "SYSERR"
232: };
233:
1.47 schwarze 234:
1.1 kristaps 235: static void
236: resize_buf(struct buf *buf, size_t initial)
237: {
238:
239: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
240: buf->buf = mandoc_realloc(buf->buf, buf->sz);
241: }
242:
243: static void
1.84 schwarze 244: choose_parser(struct mparse *curp)
1.1 kristaps 245: {
1.83 schwarze 246: char *cp, *ep;
247: int format;
1.1 kristaps 248:
1.83 schwarze 249: /*
250: * If neither command line arguments -mdoc or -man select
251: * a parser nor the roff parser found a .Dd or .TH macro
252: * yet, look ahead in the main input buffer.
253: */
254:
255: if ((format = roff_getformat(curp->roff)) == 0) {
256: cp = curp->primary->buf;
257: ep = cp + curp->primary->sz;
258: while (cp < ep) {
1.85 schwarze 259: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 260: cp++;
261: if (cp[0] == 'D' && cp[1] == 'd') {
262: format = MPARSE_MDOC;
263: break;
264: }
265: if (cp[0] == 'T' && cp[1] == 'H') {
266: format = MPARSE_MAN;
267: break;
268: }
269: }
270: cp = memchr(cp, '\n', ep - cp);
271: if (cp == NULL)
272: break;
273: cp++;
274: }
1.1 kristaps 275: }
276:
1.83 schwarze 277: if (format == MPARSE_MDOC) {
1.47 schwarze 278: if (NULL == curp->pmdoc)
1.44 schwarze 279: curp->pmdoc = mdoc_alloc(
280: curp->roff, curp, curp->defos,
281: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 282: assert(curp->pmdoc);
283: curp->mdoc = curp->pmdoc;
284: return;
1.47 schwarze 285: }
1.1 kristaps 286:
1.83 schwarze 287: /* Fall back to man(7) as a last resort. */
288:
1.47 schwarze 289: if (NULL == curp->pman)
1.106 schwarze 290: curp->pman = man_alloc(
291: curp->roff, curp, curp->defos,
1.44 schwarze 292: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 293: assert(curp->pman);
294: curp->man = curp->pman;
295: }
296:
297: /*
1.95 schwarze 298: * Main parse routine for a buffer.
299: * It assumes encoding and line numbering are already set up.
300: * It can recurse directly (for invocations of user-defined
301: * macros, inline equations, and input line traps)
302: * and indirectly (for .so file inclusion).
1.1 kristaps 303: */
304: static void
1.95 schwarze 305: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 306: {
307: const struct tbl_span *span;
308: struct buf ln;
1.95 schwarze 309: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 310: enum rofferr rr;
1.100 schwarze 311: int of;
1.1 kristaps 312: int lnn; /* line number in the real file */
313: unsigned char c;
314:
1.95 schwarze 315: memset(&ln, 0, sizeof(ln));
1.1 kristaps 316:
1.47 schwarze 317: lnn = curp->line;
318: pos = 0;
1.1 kristaps 319:
1.95 schwarze 320: while (i < blk.sz) {
1.1 kristaps 321: if (0 == pos && '\0' == blk.buf[i])
322: break;
323:
324: if (start) {
325: curp->line = lnn;
326: curp->reparse_count = 0;
1.93 schwarze 327:
328: if (lnn < 3 &&
329: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 330: curp->filenc & MPARSE_LATIN1)
331: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 332: }
333:
1.95 schwarze 334: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 335:
336: /*
337: * When finding an unescaped newline character,
338: * leave the character loop to process the line.
339: * Skip a preceding carriage return, if any.
340: */
341:
1.95 schwarze 342: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 343: '\n' == blk.buf[i + 1])
344: ++i;
345: if ('\n' == blk.buf[i]) {
346: ++i;
347: ++lnn;
348: break;
349: }
350:
1.35 schwarze 351: /*
1.93 schwarze 352: * Make sure we have space for the worst
353: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 354: */
355:
1.95 schwarze 356: if (pos + 11 > ln.sz)
1.35 schwarze 357: resize_buf(&ln, 256);
358:
1.47 schwarze 359: /*
1.93 schwarze 360: * Encode 8-bit input.
1.1 kristaps 361: */
362:
1.93 schwarze 363: c = blk.buf[i];
364: if (c & 0x80) {
1.95 schwarze 365: if ( ! (curp->filenc && preconv_encode(
366: &blk, &i, &ln, &pos, &curp->filenc))) {
1.93 schwarze 367: mandoc_vmsg(MANDOCERR_BADCHAR,
368: curp, curp->line, pos,
369: "0x%x", c);
370: ln.buf[pos++] = '?';
371: i++;
372: }
373: continue;
374: }
375:
376: /*
377: * Exclude control characters.
378: */
1.1 kristaps 379:
1.93 schwarze 380: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.78 schwarze 381: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
382: curp->line, pos, "0x%x", c);
1.1 kristaps 383: i++;
1.27 joerg 384: ln.buf[pos++] = '?';
1.1 kristaps 385: continue;
386: }
387:
388: /* Trailing backslash = a plain char. */
389:
1.95 schwarze 390: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 391: ln.buf[pos++] = blk.buf[i++];
392: continue;
393: }
394:
395: /*
396: * Found escape and at least one other character.
397: * When it's a newline character, skip it.
398: * When there is a carriage return in between,
399: * skip that one as well.
400: */
401:
1.95 schwarze 402: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 403: '\n' == blk.buf[i + 2])
404: ++i;
405: if ('\n' == blk.buf[i + 1]) {
406: i += 2;
407: ++lnn;
408: continue;
409: }
410:
1.13 kristaps 411: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 412: i += 2;
413: /* Comment, skip to end of line */
1.95 schwarze 414: for (; i < blk.sz; ++i) {
1.1 kristaps 415: if ('\n' == blk.buf[i]) {
416: ++i;
417: ++lnn;
418: break;
419: }
420: }
421:
422: /* Backout trailing whitespaces */
423: for (; pos > 0; --pos) {
424: if (ln.buf[pos - 1] != ' ')
425: break;
426: if (pos > 2 && ln.buf[pos - 2] == '\\')
427: break;
428: }
429: break;
430: }
431:
1.35 schwarze 432: /* Catch escaped bogus characters. */
433:
434: c = (unsigned char) blk.buf[i+1];
435:
1.47 schwarze 436: if ( ! (isascii(c) &&
437: (isgraph(c) || isblank(c)))) {
1.78 schwarze 438: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
439: curp->line, pos, "0x%x", c);
1.35 schwarze 440: i += 2;
441: ln.buf[pos++] = '?';
442: continue;
443: }
444:
1.1 kristaps 445: /* Some other escape sequence, copy & cont. */
446:
447: ln.buf[pos++] = blk.buf[i++];
448: ln.buf[pos++] = blk.buf[i++];
449: }
450:
1.95 schwarze 451: if (pos >= ln.sz)
1.1 kristaps 452: resize_buf(&ln, 256);
453:
454: ln.buf[pos] = '\0';
455:
456: /*
457: * A significant amount of complexity is contained by
458: * the roff preprocessor. It's line-oriented but can be
459: * expressed on one line, so we need at times to
460: * readjust our starting point and re-run it. The roff
461: * preprocessor can also readjust the buffers with new
462: * data, so we pass them in wholesale.
463: */
464:
465: of = 0;
466:
1.24 kristaps 467: /*
468: * Maintain a lookaside buffer of all parsed lines. We
469: * only do this if mparse_keep() has been invoked (the
470: * buffer may be accessed with mparse_getkeep()).
471: */
472:
473: if (curp->secondary) {
1.47 schwarze 474: curp->secondary->buf = mandoc_realloc(
475: curp->secondary->buf,
476: curp->secondary->sz + pos + 2);
477: memcpy(curp->secondary->buf +
478: curp->secondary->sz,
479: ln.buf, pos);
1.24 kristaps 480: curp->secondary->sz += pos;
481: curp->secondary->buf
482: [curp->secondary->sz] = '\n';
483: curp->secondary->sz++;
484: curp->secondary->buf
485: [curp->secondary->sz] = '\0';
486: }
1.1 kristaps 487: rerun:
1.96 schwarze 488: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 489:
490: switch (rr) {
1.47 schwarze 491: case ROFF_REPARSE:
1.1 kristaps 492: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 493: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 494: else
1.3 kristaps 495: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 496: curp->line, pos, NULL);
1.1 kristaps 497: pos = 0;
498: continue;
1.47 schwarze 499: case ROFF_APPEND:
1.95 schwarze 500: pos = strlen(ln.buf);
1.1 kristaps 501: continue;
1.47 schwarze 502: case ROFF_RERUN:
1.1 kristaps 503: goto rerun;
1.47 schwarze 504: case ROFF_IGN:
1.1 kristaps 505: pos = 0;
506: continue;
1.47 schwarze 507: case ROFF_SO:
1.95 schwarze 508: if ( ! (curp->options & MPARSE_SO) &&
509: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 510: curp->sodest = mandoc_strdup(ln.buf + of);
511: free(ln.buf);
512: return;
513: }
1.24 kristaps 514: /*
515: * We remove `so' clauses from our lookaside
516: * buffer because we're going to descend into
517: * the file recursively.
518: */
1.47 schwarze 519: if (curp->secondary)
1.25 kristaps 520: curp->secondary->sz -= pos + 1;
1.36 schwarze 521: mparse_readfd(curp, -1, ln.buf + of);
1.52 schwarze 522: if (MANDOCLEVEL_FATAL <= curp->file_status) {
523: mandoc_vmsg(MANDOCERR_SO_FAIL,
524: curp, curp->line, pos,
525: ".so %s", ln.buf + of);
1.1 kristaps 526: break;
1.52 schwarze 527: }
1.1 kristaps 528: pos = 0;
529: continue;
530: default:
531: break;
532: }
533:
534: /*
535: * If we encounter errors in the recursive parse, make
536: * sure we don't continue parsing.
537: */
538:
539: if (MANDOCLEVEL_FATAL <= curp->file_status)
540: break;
541:
542: /*
543: * If input parsers have not been allocated, do so now.
1.14 kristaps 544: * We keep these instanced between parsers, but set them
1.1 kristaps 545: * locally per parse routine since we can use different
546: * parsers with each one.
547: */
548:
549: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 550: choose_parser(curp);
1.1 kristaps 551:
1.47 schwarze 552: /*
1.84 schwarze 553: * Lastly, push down into the parsers themselves.
1.1 kristaps 554: * If libroff returns ROFF_TBL, then add it to the
555: * currently open parse. Since we only get here if
556: * there does exist data (see tbl_data.c), we're
557: * guaranteed that something's been allocated.
558: * Do the same for ROFF_EQN.
559: */
560:
1.100 schwarze 561: if (rr == ROFF_TBL) {
562: while ((span = roff_span(curp->roff)) != NULL)
563: if (curp->man == NULL)
564: mdoc_addspan(curp->mdoc, span);
565: else
566: man_addspan(curp->man, span);
567: } else if (rr == ROFF_EQN) {
568: if (curp->man == NULL)
569: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
570: else
571: man_addeqn(curp->man, roff_eqn(curp->roff));
572: } else if ((curp->man == NULL ?
573: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
574: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
575: break;
1.1 kristaps 576:
577: /* Temporary buffers typically are not full. */
578:
579: if (0 == start && '\0' == blk.buf[i])
580: break;
581:
582: /* Start the next input line. */
583:
584: pos = 0;
585: }
586:
587: free(ln.buf);
588: }
589:
590: static int
1.40 schwarze 591: read_whole_file(struct mparse *curp, const char *file, int fd,
592: struct buf *fb, int *with_mmap)
1.1 kristaps 593: {
594: size_t off;
595: ssize_t ssz;
596:
1.81 schwarze 597: #if HAVE_MMAP
1.15 kristaps 598: struct stat st;
1.1 kristaps 599: if (-1 == fstat(fd, &st)) {
1.107 schwarze 600: perror(file);
601: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 602: }
603:
604: /*
605: * If we're a regular file, try just reading in the whole entry
606: * via mmap(). This is faster than reading it into blocks, and
607: * since each file is only a few bytes to begin with, I'm not
608: * concerned that this is going to tank any machines.
609: */
610:
611: if (S_ISREG(st.st_mode)) {
612: if (st.st_size >= (1U << 31)) {
1.40 schwarze 613: curp->file_status = MANDOCLEVEL_FATAL;
614: if (curp->mmsg)
615: (*curp->mmsg)(MANDOCERR_TOOLARGE,
616: curp->file_status, file, 0, 0, NULL);
1.1 kristaps 617: return(0);
618: }
619: *with_mmap = 1;
620: fb->sz = (size_t)st.st_size;
1.37 schwarze 621: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 622: if (fb->buf != MAP_FAILED)
623: return(1);
624: }
1.15 kristaps 625: #endif
1.1 kristaps 626:
627: /*
628: * If this isn't a regular file (like, say, stdin), then we must
629: * go the old way and just read things in bit by bit.
630: */
631:
632: *with_mmap = 0;
633: off = 0;
634: fb->sz = 0;
635: fb->buf = NULL;
636: for (;;) {
637: if (off == fb->sz) {
638: if (fb->sz == (1U << 31)) {
1.40 schwarze 639: curp->file_status = MANDOCLEVEL_FATAL;
640: if (curp->mmsg)
641: (*curp->mmsg)(MANDOCERR_TOOLARGE,
642: curp->file_status,
643: file, 0, 0, NULL);
1.1 kristaps 644: break;
645: }
646: resize_buf(fb, 65536);
647: }
648: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
649: if (ssz == 0) {
650: fb->sz = off;
651: return(1);
652: }
653: if (ssz == -1) {
1.107 schwarze 654: perror(file);
655: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 656: }
657: off += (size_t)ssz;
658: }
659:
660: free(fb->buf);
661: fb->buf = NULL;
662: return(0);
663: }
664:
665: static void
666: mparse_end(struct mparse *curp)
667: {
668:
669: if (MANDOCLEVEL_FATAL <= curp->file_status)
670: return;
671:
1.72 schwarze 672: if (curp->mdoc == NULL &&
673: curp->man == NULL &&
674: curp->sodest == NULL) {
675: if (curp->options & MPARSE_MDOC)
676: curp->mdoc = curp->pmdoc;
677: else {
678: if (curp->pman == NULL)
1.106 schwarze 679: curp->pman = man_alloc(
680: curp->roff, curp, curp->defos,
1.72 schwarze 681: curp->options & MPARSE_QUICK ? 1 : 0);
682: curp->man = curp->pman;
683: }
684: }
685:
1.1 kristaps 686: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
687: assert(MANDOCLEVEL_FATAL <= curp->file_status);
688: return;
689: }
690:
691: if (curp->man && ! man_endparse(curp->man)) {
692: assert(MANDOCLEVEL_FATAL <= curp->file_status);
693: return;
694: }
695:
696: roff_endparse(curp->roff);
697: }
698:
699: static void
1.36 schwarze 700: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 701: {
1.85 schwarze 702: struct buf *svprimary;
1.28 joerg 703: const char *svfile;
1.95 schwarze 704: size_t offset;
1.36 schwarze 705: static int recursion_depth;
706:
707: if (64 < recursion_depth) {
708: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
709: return;
710: }
1.28 joerg 711:
712: /* Line number is per-file. */
713: svfile = curp->file;
714: curp->file = file;
1.85 schwarze 715: svprimary = curp->primary;
1.83 schwarze 716: curp->primary = &blk;
1.28 joerg 717: curp->line = 1;
1.36 schwarze 718: recursion_depth++;
1.28 joerg 719:
1.93 schwarze 720: /* Skip an UTF-8 byte order mark. */
721: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
722: (unsigned char)blk.buf[0] == 0xef &&
723: (unsigned char)blk.buf[1] == 0xbb &&
724: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 725: offset = 3;
1.93 schwarze 726: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 727: } else
728: offset = 0;
1.93 schwarze 729:
1.95 schwarze 730: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 731:
1.36 schwarze 732: if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
1.28 joerg 733: mparse_end(curp);
734:
1.85 schwarze 735: curp->primary = svprimary;
1.28 joerg 736: curp->file = svfile;
737: }
738:
739: enum mandoclevel
1.104 schwarze 740: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 741: const char *file)
742: {
743: struct buf blk;
744:
1.104 schwarze 745: blk.buf = buf;
1.28 joerg 746: blk.sz = len;
747:
1.36 schwarze 748: mparse_parse_buffer(curp, blk, file);
1.28 joerg 749: return(curp->file_status);
750: }
751:
1.98 schwarze 752: /*
753: * If a file descriptor is given, use it and assume it points
754: * to the named file. Otherwise, open the named file.
755: * Read the whole file into memory and call the parsers.
756: * Called recursively when an .so request is encountered.
757: */
1.36 schwarze 758: enum mandoclevel
759: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 760: {
1.28 joerg 761: struct buf blk;
762: int with_mmap;
1.93 schwarze 763: int save_filenc;
1.98 schwarze 764: pid_t save_child;
1.1 kristaps 765:
1.98 schwarze 766: save_child = curp->child;
767: if (fd != -1)
768: curp->child = 0;
1.108 schwarze 769: else if (mparse_open(curp, &fd, file) != MANDOCLEVEL_OK)
1.98 schwarze 770: goto out;
1.1 kristaps 771:
1.91 schwarze 772: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 773: save_filenc = curp->filenc;
774: curp->filenc = curp->options &
775: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 776: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 777: curp->filenc = save_filenc;
1.81 schwarze 778: #if HAVE_MMAP
1.91 schwarze 779: if (with_mmap)
780: munmap(blk.buf, blk.sz);
781: else
1.28 joerg 782: #endif
1.91 schwarze 783: free(blk.buf);
784: }
1.1 kristaps 785:
1.98 schwarze 786: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 787: perror(file);
1.91 schwarze 788:
1.98 schwarze 789: mparse_wait(curp);
790: out:
791: curp->child = save_child;
1.1 kristaps 792: return(curp->file_status);
1.82 schwarze 793: }
794:
795: enum mandoclevel
1.97 schwarze 796: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 797: {
798: int pfd[2];
1.98 schwarze 799: int save_errno;
1.82 schwarze 800: char *cp;
801:
802: curp->file = file;
1.98 schwarze 803:
804: /* Unless zipped, try to just open the file. */
805:
1.82 schwarze 806: if ((cp = strrchr(file, '.')) == NULL ||
807: strcmp(cp + 1, "gz")) {
1.97 schwarze 808: curp->child = 0;
1.98 schwarze 809: if ((*fd = open(file, O_RDONLY)) != -1)
810: return(MANDOCLEVEL_OK);
811:
812: /* Open failed; try to append ".gz". */
813:
814: mandoc_asprintf(&cp, "%s.gz", file);
815: file = cp;
816: } else
817: cp = NULL;
818:
819: /* Before forking, make sure the file can be read. */
820:
821: save_errno = errno;
822: if (access(file, R_OK) == -1) {
823: if (cp != NULL)
824: errno = save_errno;
1.108 schwarze 825: free(cp);
826: *fd = -1;
827: curp->child = 0;
828: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
829: return(MANDOCLEVEL_ERROR);
1.82 schwarze 830: }
831:
1.98 schwarze 832: /* Run gunzip(1). */
833:
1.82 schwarze 834: if (pipe(pfd) == -1) {
1.107 schwarze 835: perror("pipe");
836: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 837: }
838:
1.97 schwarze 839: switch (curp->child = fork()) {
1.82 schwarze 840: case -1:
1.107 schwarze 841: perror("fork");
842: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 843: case 0:
844: close(pfd[0]);
845: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 846: perror("dup");
847: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 848: }
849: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 850: perror("exec");
851: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 852: default:
853: close(pfd[1]);
854: *fd = pfd[0];
855: return(MANDOCLEVEL_OK);
856: }
857: }
858:
859: enum mandoclevel
1.97 schwarze 860: mparse_wait(struct mparse *curp)
1.82 schwarze 861: {
862: int status;
863:
1.97 schwarze 864: if (curp->child == 0)
865: return(MANDOCLEVEL_OK);
866:
867: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 868: perror("wait");
869: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 870: }
871: if (WIFSIGNALED(status)) {
1.108 schwarze 872: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
873: "gunzip died from signal %d", WTERMSIG(status));
874: return(MANDOCLEVEL_ERROR);
1.82 schwarze 875: }
876: if (WEXITSTATUS(status)) {
1.108 schwarze 877: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
878: "gunzip failed with code %d", WEXITSTATUS(status));
879: return(MANDOCLEVEL_ERROR);
1.82 schwarze 880: }
881: return(MANDOCLEVEL_OK);
1.1 kristaps 882: }
883:
884: struct mparse *
1.94 schwarze 885: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
886: const struct mchars *mchars, const char *defos)
1.1 kristaps 887: {
888: struct mparse *curp;
1.10 kristaps 889:
890: assert(wlevel <= MANDOCLEVEL_FATAL);
1.1 kristaps 891:
892: curp = mandoc_calloc(1, sizeof(struct mparse));
893:
1.44 schwarze 894: curp->options = options;
1.3 kristaps 895: curp->wlevel = wlevel;
1.1 kristaps 896: curp->mmsg = mmsg;
1.29 schwarze 897: curp->defos = defos;
1.1 kristaps 898:
1.94 schwarze 899: curp->mchars = mchars;
900: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 901: if (curp->options & MPARSE_MDOC)
902: curp->pmdoc = mdoc_alloc(
903: curp->roff, curp, curp->defos,
904: curp->options & MPARSE_QUICK ? 1 : 0);
905: if (curp->options & MPARSE_MAN)
1.106 schwarze 906: curp->pman = man_alloc(
907: curp->roff, curp, curp->defos,
1.72 schwarze 908: curp->options & MPARSE_QUICK ? 1 : 0);
909:
1.1 kristaps 910: return(curp);
911: }
912:
913: void
914: mparse_reset(struct mparse *curp)
915: {
916:
917: roff_reset(curp->roff);
918:
919: if (curp->mdoc)
920: mdoc_reset(curp->mdoc);
921: if (curp->man)
922: man_reset(curp->man);
1.24 kristaps 923: if (curp->secondary)
924: curp->secondary->sz = 0;
1.1 kristaps 925:
926: curp->file_status = MANDOCLEVEL_OK;
927: curp->mdoc = NULL;
928: curp->man = NULL;
1.45 schwarze 929:
930: free(curp->sodest);
931: curp->sodest = NULL;
1.1 kristaps 932: }
933:
934: void
935: mparse_free(struct mparse *curp)
936: {
937:
938: if (curp->pmdoc)
939: mdoc_free(curp->pmdoc);
940: if (curp->pman)
941: man_free(curp->pman);
942: if (curp->roff)
943: roff_free(curp->roff);
1.24 kristaps 944: if (curp->secondary)
945: free(curp->secondary->buf);
1.1 kristaps 946:
1.24 kristaps 947: free(curp->secondary);
1.45 schwarze 948: free(curp->sodest);
1.1 kristaps 949: free(curp);
950: }
951:
952: void
1.45 schwarze 953: mparse_result(struct mparse *curp,
954: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 955: {
956:
1.45 schwarze 957: if (sodest && NULL != (*sodest = curp->sodest)) {
958: *mdoc = NULL;
959: *man = NULL;
960: return;
961: }
1.9 kristaps 962: if (mdoc)
963: *mdoc = curp->mdoc;
964: if (man)
965: *man = curp->man;
1.3 kristaps 966: }
967:
968: void
969: mandoc_vmsg(enum mandocerr t, struct mparse *m,
970: int ln, int pos, const char *fmt, ...)
971: {
972: char buf[256];
973: va_list ap;
974:
975: va_start(ap, fmt);
1.48 schwarze 976: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 977: va_end(ap);
978:
979: mandoc_msg(t, m, ln, pos, buf);
980: }
981:
982: void
1.47 schwarze 983: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 984: int ln, int col, const char *msg)
985: {
986: enum mandoclevel level;
987:
988: level = MANDOCLEVEL_FATAL;
989: while (er < mandoclimits[level])
990: level--;
991:
1.108 schwarze 992: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 993: return;
994:
1.8 kristaps 995: if (m->mmsg)
996: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 997:
998: if (m->file_status < level)
999: m->file_status = level;
1.7 kristaps 1000: }
1001:
1002: const char *
1003: mparse_strerror(enum mandocerr er)
1004: {
1005:
1006: return(mandocerrs[er]);
1007: }
1008:
1009: const char *
1010: mparse_strlevel(enum mandoclevel lvl)
1011: {
1012: return(mandoclevels[lvl]);
1.24 kristaps 1013: }
1014:
1015: void
1016: mparse_keep(struct mparse *p)
1017: {
1018:
1019: assert(NULL == p->secondary);
1020: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1021: }
1022:
1023: const char *
1024: mparse_getkeep(const struct mparse *p)
1025: {
1026:
1027: assert(p->secondary);
1028: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1029: }
CVSweb