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