Annotation of mandoc/read.c, Revision 1.101
1.101 ! schwarze 1: /* $Id: read.c,v 1.100 2014/11/28 06:27:05 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.40 schwarze 4: * Copyright (c) 2010-2014 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"
1.28 joerg 44: #include "main.h"
1.1 kristaps 45:
46: #define REPARSE_LIMIT 1000
47:
48: struct mparse {
49: struct man *pman; /* persistent man parser */
50: struct mdoc *pmdoc; /* persistent mdoc parser */
51: struct man *man; /* man parser */
52: struct mdoc *mdoc; /* mdoc parser */
53: struct roff *roff; /* roff parser (!NULL) */
1.94 schwarze 54: const struct mchars *mchars; /* character table */
1.45 schwarze 55: char *sodest; /* filename pointed to by .so */
1.83 schwarze 56: const char *file; /* filename of current input file */
57: struct buf *primary; /* buffer currently being parsed */
58: struct buf *secondary; /* preprocessed copy of input */
59: const char *defos; /* default operating system */
60: mandocmsg mmsg; /* warning/error message handler */
61: enum mandoclevel file_status; /* status of current parse */
62: enum mandoclevel wlevel; /* ignore messages below this */
63: int options; /* parser options */
1.93 schwarze 64: int filenc; /* encoding of the current file */
1.1 kristaps 65: int reparse_count; /* finite interp. stack */
1.83 schwarze 66: int line; /* line number in the file */
1.97 schwarze 67: pid_t child; /* the gunzip(1) process */
1.1 kristaps 68: };
69:
1.84 schwarze 70: static void choose_parser(struct mparse *);
1.1 kristaps 71: static void resize_buf(struct buf *, size_t);
1.95 schwarze 72: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 73: static int read_whole_file(struct mparse *, const char *, int,
74: struct buf *, int *);
1.1 kristaps 75: static void mparse_end(struct mparse *);
1.37 schwarze 76: static void mparse_parse_buffer(struct mparse *, struct buf,
77: const char *);
1.1 kristaps 78:
1.3 kristaps 79: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
80: MANDOCERR_OK,
81: MANDOCERR_WARNING,
82: MANDOCERR_WARNING,
83: MANDOCERR_ERROR,
84: MANDOCERR_FATAL,
85: MANDOCERR_MAX,
86: MANDOCERR_MAX
87: };
88:
1.7 kristaps 89: static const char * const mandocerrs[MANDOCERR_MAX] = {
90: "ok",
91:
92: "generic warning",
93:
94: /* related to the prologue */
1.79 schwarze 95: "missing manual title, using UNTITLED",
96: "missing manual title, using \"\"",
1.54 schwarze 97: "lower case character in document title",
1.79 schwarze 98: "missing manual section, using \"\"",
1.7 kristaps 99: "unknown manual section",
1.54 schwarze 100: "missing date, using today's date",
1.7 kristaps 101: "cannot parse date, using it verbatim",
1.79 schwarze 102: "missing Os macro, using \"\"",
103: "duplicate prologue macro",
104: "late prologue macro",
105: "skipping late title macro",
1.7 kristaps 106: "prologue macros out of order",
107:
108: /* related to document structure */
109: ".so is fragile, better use ln(1)",
1.50 schwarze 110: "no document body",
1.54 schwarze 111: "content before first section header",
112: "first section is not \"NAME\"",
1.7 kristaps 113: "bad NAME section contents",
114: "sections out of conventional order",
1.54 schwarze 115: "duplicate section title",
116: "unexpected section",
1.87 schwarze 117: "unusual Xr order",
118: "unusual Xr punctuation",
1.86 schwarze 119: "AUTHORS section without An macro",
1.7 kristaps 120:
121: /* related to macros and nesting */
1.55 schwarze 122: "obsolete macro",
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",
1.60 schwarze 148: "missing -std argument, adding it",
1.90 schwarze 149: "missing eqn box, using \"\"",
1.7 kristaps 150:
151: /* related to bad macro arguments */
1.64 schwarze 152: "unterminated quoted argument",
1.7 kristaps 153: "duplicate argument",
1.76 schwarze 154: "skipping duplicate argument",
1.63 schwarze 155: "skipping duplicate display type",
156: "skipping duplicate list type",
1.76 schwarze 157: "skipping -width argument",
1.7 kristaps 158: "unknown AT&T UNIX version",
1.88 schwarze 159: "comma in function argument",
1.89 schwarze 160: "parenthesis in function name",
1.67 schwarze 161: "invalid content in Rs block",
1.63 schwarze 162: "invalid Boolean argument",
163: "unknown font, skipping request",
1.7 kristaps 164:
165: /* related to plain text */
1.64 schwarze 166: "blank line in fill mode, using .sp",
167: "tab in filled text",
168: "whitespace at end of input line",
1.7 kristaps 169: "bad comment style",
1.64 schwarze 170: "invalid escape sequence",
171: "undefined string, using \"\"",
1.16 kristaps 172:
1.7 kristaps 173: "generic error",
1.17 kristaps 174:
175: /* related to equations */
1.20 kristaps 176: "unexpected equation scope closure",
177: "equation scope open on exit",
1.21 kristaps 178: "overlapping equation scopes",
179: "unexpected end of equation",
1.7 kristaps 180:
181: /* related to tables */
182: "bad table syntax",
183: "bad table option",
184: "bad table layout",
185: "no table layout cells specified",
186: "no table data cells specified",
187: "ignore data in cell",
188: "data block still open",
189: "ignoring extra data cells",
190:
1.68 schwarze 191: /* related to document structure and macros */
1.7 kristaps 192: "input stack limit exceeded, infinite loop?",
193: "skipping bad character",
1.68 schwarze 194: "skipping unknown macro",
1.73 schwarze 195: "skipping item outside list",
1.68 schwarze 196: "skipping column outside column list",
197: "skipping end of block that is not open",
198: "inserting missing end of block",
199: "appending missing end of block",
200:
201: /* related to request and macro arguments */
1.7 kristaps 202: "escaped character not allowed in a name",
203: "argument count wrong",
1.99 schwarze 204: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 205: "missing list type, using -item",
1.70 schwarze 206: "missing manual name, using \"\"",
1.71 schwarze 207: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 208: "unknown standard specifier",
1.71 schwarze 209: "skipping request without numeric argument",
1.61 schwarze 210: "skipping all arguments",
211: "skipping excess arguments",
1.92 kristaps 212: "divide by zero",
1.7 kristaps 213:
214: "generic fatal error",
215:
1.40 schwarze 216: "input too large",
1.7 kristaps 217: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.52 schwarze 218: ".so request failed",
1.40 schwarze 219:
220: /* system errors */
1.82 schwarze 221: "cannot dup file descriptor",
222: "cannot exec",
223: "gunzip failed with code",
224: "cannot fork",
1.51 schwarze 225: NULL,
1.82 schwarze 226: "cannot open pipe",
227: "cannot read file",
228: "gunzip died from signal",
1.40 schwarze 229: "cannot stat file",
1.82 schwarze 230: "wait failed",
1.7 kristaps 231: };
232:
233: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
234: "SUCCESS",
235: "RESERVED",
236: "WARNING",
237: "ERROR",
238: "FATAL",
239: "BADARG",
240: "SYSERR"
241: };
242:
1.47 schwarze 243:
1.1 kristaps 244: static void
245: resize_buf(struct buf *buf, size_t initial)
246: {
247:
248: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
249: buf->buf = mandoc_realloc(buf->buf, buf->sz);
250: }
251:
252: static void
1.84 schwarze 253: choose_parser(struct mparse *curp)
1.1 kristaps 254: {
1.83 schwarze 255: char *cp, *ep;
256: int format;
1.1 kristaps 257:
1.83 schwarze 258: /*
259: * If neither command line arguments -mdoc or -man select
260: * a parser nor the roff parser found a .Dd or .TH macro
261: * yet, look ahead in the main input buffer.
262: */
263:
264: if ((format = roff_getformat(curp->roff)) == 0) {
265: cp = curp->primary->buf;
266: ep = cp + curp->primary->sz;
267: while (cp < ep) {
1.85 schwarze 268: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 269: cp++;
270: if (cp[0] == 'D' && cp[1] == 'd') {
271: format = MPARSE_MDOC;
272: break;
273: }
274: if (cp[0] == 'T' && cp[1] == 'H') {
275: format = MPARSE_MAN;
276: break;
277: }
278: }
279: cp = memchr(cp, '\n', ep - cp);
280: if (cp == NULL)
281: break;
282: cp++;
283: }
1.1 kristaps 284: }
285:
1.83 schwarze 286: if (format == MPARSE_MDOC) {
1.47 schwarze 287: if (NULL == curp->pmdoc)
1.44 schwarze 288: curp->pmdoc = mdoc_alloc(
289: curp->roff, curp, curp->defos,
290: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 291: assert(curp->pmdoc);
292: curp->mdoc = curp->pmdoc;
293: return;
1.47 schwarze 294: }
1.1 kristaps 295:
1.83 schwarze 296: /* Fall back to man(7) as a last resort. */
297:
1.47 schwarze 298: if (NULL == curp->pman)
1.44 schwarze 299: curp->pman = man_alloc(curp->roff, curp,
300: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 301: assert(curp->pman);
302: curp->man = curp->pman;
303: }
304:
305: /*
1.95 schwarze 306: * Main parse routine for a buffer.
307: * It assumes encoding and line numbering are already set up.
308: * It can recurse directly (for invocations of user-defined
309: * macros, inline equations, and input line traps)
310: * and indirectly (for .so file inclusion).
1.1 kristaps 311: */
312: static void
1.95 schwarze 313: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 314: {
315: const struct tbl_span *span;
316: struct buf ln;
1.95 schwarze 317: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 318: enum rofferr rr;
1.100 schwarze 319: int of;
1.1 kristaps 320: int lnn; /* line number in the real file */
321: unsigned char c;
322:
1.95 schwarze 323: memset(&ln, 0, sizeof(ln));
1.1 kristaps 324:
1.47 schwarze 325: lnn = curp->line;
326: pos = 0;
1.1 kristaps 327:
1.95 schwarze 328: while (i < blk.sz) {
1.1 kristaps 329: if (0 == pos && '\0' == blk.buf[i])
330: break;
331:
332: if (start) {
333: curp->line = lnn;
334: curp->reparse_count = 0;
1.93 schwarze 335:
336: if (lnn < 3 &&
337: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 338: curp->filenc & MPARSE_LATIN1)
339: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 340: }
341:
1.95 schwarze 342: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 343:
344: /*
345: * When finding an unescaped newline character,
346: * leave the character loop to process the line.
347: * Skip a preceding carriage return, if any.
348: */
349:
1.95 schwarze 350: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 351: '\n' == blk.buf[i + 1])
352: ++i;
353: if ('\n' == blk.buf[i]) {
354: ++i;
355: ++lnn;
356: break;
357: }
358:
1.35 schwarze 359: /*
1.93 schwarze 360: * Make sure we have space for the worst
361: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 362: */
363:
1.95 schwarze 364: if (pos + 11 > ln.sz)
1.35 schwarze 365: resize_buf(&ln, 256);
366:
1.47 schwarze 367: /*
1.93 schwarze 368: * Encode 8-bit input.
1.1 kristaps 369: */
370:
1.93 schwarze 371: c = blk.buf[i];
372: if (c & 0x80) {
1.95 schwarze 373: if ( ! (curp->filenc && preconv_encode(
374: &blk, &i, &ln, &pos, &curp->filenc))) {
1.93 schwarze 375: mandoc_vmsg(MANDOCERR_BADCHAR,
376: curp, curp->line, pos,
377: "0x%x", c);
378: ln.buf[pos++] = '?';
379: i++;
380: }
381: continue;
382: }
383:
384: /*
385: * Exclude control characters.
386: */
1.1 kristaps 387:
1.93 schwarze 388: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.78 schwarze 389: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
390: curp->line, pos, "0x%x", c);
1.1 kristaps 391: i++;
1.27 joerg 392: ln.buf[pos++] = '?';
1.1 kristaps 393: continue;
394: }
395:
396: /* Trailing backslash = a plain char. */
397:
1.95 schwarze 398: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 399: ln.buf[pos++] = blk.buf[i++];
400: continue;
401: }
402:
403: /*
404: * Found escape and at least one other character.
405: * When it's a newline character, skip it.
406: * When there is a carriage return in between,
407: * skip that one as well.
408: */
409:
1.95 schwarze 410: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 411: '\n' == blk.buf[i + 2])
412: ++i;
413: if ('\n' == blk.buf[i + 1]) {
414: i += 2;
415: ++lnn;
416: continue;
417: }
418:
1.13 kristaps 419: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 420: i += 2;
421: /* Comment, skip to end of line */
1.95 schwarze 422: for (; i < blk.sz; ++i) {
1.1 kristaps 423: if ('\n' == blk.buf[i]) {
424: ++i;
425: ++lnn;
426: break;
427: }
428: }
429:
430: /* Backout trailing whitespaces */
431: for (; pos > 0; --pos) {
432: if (ln.buf[pos - 1] != ' ')
433: break;
434: if (pos > 2 && ln.buf[pos - 2] == '\\')
435: break;
436: }
437: break;
438: }
439:
1.35 schwarze 440: /* Catch escaped bogus characters. */
441:
442: c = (unsigned char) blk.buf[i+1];
443:
1.47 schwarze 444: if ( ! (isascii(c) &&
445: (isgraph(c) || isblank(c)))) {
1.78 schwarze 446: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
447: curp->line, pos, "0x%x", c);
1.35 schwarze 448: i += 2;
449: ln.buf[pos++] = '?';
450: continue;
451: }
452:
1.1 kristaps 453: /* Some other escape sequence, copy & cont. */
454:
455: ln.buf[pos++] = blk.buf[i++];
456: ln.buf[pos++] = blk.buf[i++];
457: }
458:
1.95 schwarze 459: if (pos >= ln.sz)
1.1 kristaps 460: resize_buf(&ln, 256);
461:
462: ln.buf[pos] = '\0';
463:
464: /*
465: * A significant amount of complexity is contained by
466: * the roff preprocessor. It's line-oriented but can be
467: * expressed on one line, so we need at times to
468: * readjust our starting point and re-run it. The roff
469: * preprocessor can also readjust the buffers with new
470: * data, so we pass them in wholesale.
471: */
472:
473: of = 0;
474:
1.24 kristaps 475: /*
476: * Maintain a lookaside buffer of all parsed lines. We
477: * only do this if mparse_keep() has been invoked (the
478: * buffer may be accessed with mparse_getkeep()).
479: */
480:
481: if (curp->secondary) {
1.47 schwarze 482: curp->secondary->buf = mandoc_realloc(
483: curp->secondary->buf,
484: curp->secondary->sz + pos + 2);
485: memcpy(curp->secondary->buf +
486: curp->secondary->sz,
487: ln.buf, pos);
1.24 kristaps 488: curp->secondary->sz += pos;
489: curp->secondary->buf
490: [curp->secondary->sz] = '\n';
491: curp->secondary->sz++;
492: curp->secondary->buf
493: [curp->secondary->sz] = '\0';
494: }
1.1 kristaps 495: rerun:
1.96 schwarze 496: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 497:
498: switch (rr) {
1.47 schwarze 499: case ROFF_REPARSE:
1.1 kristaps 500: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 501: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 502: else
1.3 kristaps 503: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 504: curp->line, pos, NULL);
1.1 kristaps 505: pos = 0;
506: continue;
1.47 schwarze 507: case ROFF_APPEND:
1.95 schwarze 508: pos = strlen(ln.buf);
1.1 kristaps 509: continue;
1.47 schwarze 510: case ROFF_RERUN:
1.1 kristaps 511: goto rerun;
1.47 schwarze 512: case ROFF_IGN:
1.1 kristaps 513: pos = 0;
514: continue;
1.47 schwarze 515: case ROFF_ERR:
1.1 kristaps 516: assert(MANDOCLEVEL_FATAL <= curp->file_status);
517: break;
1.47 schwarze 518: case ROFF_SO:
1.95 schwarze 519: if ( ! (curp->options & MPARSE_SO) &&
520: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 521: curp->sodest = mandoc_strdup(ln.buf + of);
522: free(ln.buf);
523: return;
524: }
1.24 kristaps 525: /*
526: * We remove `so' clauses from our lookaside
527: * buffer because we're going to descend into
528: * the file recursively.
529: */
1.47 schwarze 530: if (curp->secondary)
1.25 kristaps 531: curp->secondary->sz -= pos + 1;
1.36 schwarze 532: mparse_readfd(curp, -1, ln.buf + of);
1.52 schwarze 533: if (MANDOCLEVEL_FATAL <= curp->file_status) {
534: mandoc_vmsg(MANDOCERR_SO_FAIL,
535: curp, curp->line, pos,
536: ".so %s", ln.buf + of);
1.1 kristaps 537: break;
1.52 schwarze 538: }
1.1 kristaps 539: pos = 0;
540: continue;
541: default:
542: break;
543: }
544:
545: /*
546: * If we encounter errors in the recursive parse, make
547: * sure we don't continue parsing.
548: */
549:
550: if (MANDOCLEVEL_FATAL <= curp->file_status)
551: break;
552:
553: /*
554: * If input parsers have not been allocated, do so now.
1.14 kristaps 555: * We keep these instanced between parsers, but set them
1.1 kristaps 556: * locally per parse routine since we can use different
557: * parsers with each one.
558: */
559:
560: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 561: choose_parser(curp);
1.1 kristaps 562:
1.47 schwarze 563: /*
1.84 schwarze 564: * Lastly, push down into the parsers themselves.
1.1 kristaps 565: * If libroff returns ROFF_TBL, then add it to the
566: * currently open parse. Since we only get here if
567: * there does exist data (see tbl_data.c), we're
568: * guaranteed that something's been allocated.
569: * Do the same for ROFF_EQN.
570: */
571:
1.100 schwarze 572: if (rr == ROFF_TBL) {
573: while ((span = roff_span(curp->roff)) != NULL)
574: if (curp->man == NULL)
575: mdoc_addspan(curp->mdoc, span);
576: else
577: man_addspan(curp->man, span);
578: } else if (rr == ROFF_EQN) {
579: if (curp->man == NULL)
580: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
581: else
582: man_addeqn(curp->man, roff_eqn(curp->roff));
583: } else if ((curp->man == NULL ?
584: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
585: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
586: break;
1.1 kristaps 587:
588: /* Temporary buffers typically are not full. */
589:
590: if (0 == start && '\0' == blk.buf[i])
591: break;
592:
593: /* Start the next input line. */
594:
595: pos = 0;
596: }
597:
598: free(ln.buf);
599: }
600:
601: static int
1.40 schwarze 602: read_whole_file(struct mparse *curp, const char *file, int fd,
603: struct buf *fb, int *with_mmap)
1.1 kristaps 604: {
605: size_t off;
606: ssize_t ssz;
607:
1.81 schwarze 608: #if HAVE_MMAP
1.15 kristaps 609: struct stat st;
1.1 kristaps 610: if (-1 == fstat(fd, &st)) {
1.40 schwarze 611: curp->file_status = MANDOCLEVEL_SYSERR;
612: if (curp->mmsg)
613: (*curp->mmsg)(MANDOCERR_SYSSTAT, curp->file_status,
614: file, 0, 0, strerror(errno));
1.1 kristaps 615: return(0);
616: }
617:
618: /*
619: * If we're a regular file, try just reading in the whole entry
620: * via mmap(). This is faster than reading it into blocks, and
621: * since each file is only a few bytes to begin with, I'm not
622: * concerned that this is going to tank any machines.
623: */
624:
625: if (S_ISREG(st.st_mode)) {
626: if (st.st_size >= (1U << 31)) {
1.40 schwarze 627: curp->file_status = MANDOCLEVEL_FATAL;
628: if (curp->mmsg)
629: (*curp->mmsg)(MANDOCERR_TOOLARGE,
630: curp->file_status, file, 0, 0, NULL);
1.1 kristaps 631: return(0);
632: }
633: *with_mmap = 1;
634: fb->sz = (size_t)st.st_size;
1.37 schwarze 635: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 636: if (fb->buf != MAP_FAILED)
637: return(1);
638: }
1.15 kristaps 639: #endif
1.1 kristaps 640:
641: /*
642: * If this isn't a regular file (like, say, stdin), then we must
643: * go the old way and just read things in bit by bit.
644: */
645:
646: *with_mmap = 0;
647: off = 0;
648: fb->sz = 0;
649: fb->buf = NULL;
650: for (;;) {
651: if (off == fb->sz) {
652: if (fb->sz == (1U << 31)) {
1.40 schwarze 653: curp->file_status = MANDOCLEVEL_FATAL;
654: if (curp->mmsg)
655: (*curp->mmsg)(MANDOCERR_TOOLARGE,
656: curp->file_status,
657: file, 0, 0, NULL);
1.1 kristaps 658: break;
659: }
660: resize_buf(fb, 65536);
661: }
662: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
663: if (ssz == 0) {
664: fb->sz = off;
665: return(1);
666: }
667: if (ssz == -1) {
1.40 schwarze 668: curp->file_status = MANDOCLEVEL_SYSERR;
669: if (curp->mmsg)
670: (*curp->mmsg)(MANDOCERR_SYSREAD,
671: curp->file_status, file, 0, 0,
672: strerror(errno));
1.1 kristaps 673: break;
674: }
675: off += (size_t)ssz;
676: }
677:
678: free(fb->buf);
679: fb->buf = NULL;
680: return(0);
681: }
682:
683: static void
684: mparse_end(struct mparse *curp)
685: {
686:
687: if (MANDOCLEVEL_FATAL <= curp->file_status)
688: return;
689:
1.72 schwarze 690: if (curp->mdoc == NULL &&
691: curp->man == NULL &&
692: curp->sodest == NULL) {
693: if (curp->options & MPARSE_MDOC)
694: curp->mdoc = curp->pmdoc;
695: else {
696: if (curp->pman == NULL)
697: curp->pman = man_alloc(curp->roff, curp,
698: curp->options & MPARSE_QUICK ? 1 : 0);
699: curp->man = curp->pman;
700: }
701: }
702:
1.1 kristaps 703: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
704: assert(MANDOCLEVEL_FATAL <= curp->file_status);
705: return;
706: }
707:
708: if (curp->man && ! man_endparse(curp->man)) {
709: assert(MANDOCLEVEL_FATAL <= curp->file_status);
710: return;
711: }
712:
713: roff_endparse(curp->roff);
714: }
715:
716: static void
1.36 schwarze 717: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 718: {
1.85 schwarze 719: struct buf *svprimary;
1.28 joerg 720: const char *svfile;
1.95 schwarze 721: size_t offset;
1.36 schwarze 722: static int recursion_depth;
723:
724: if (64 < recursion_depth) {
725: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
726: return;
727: }
1.28 joerg 728:
729: /* Line number is per-file. */
730: svfile = curp->file;
731: curp->file = file;
1.85 schwarze 732: svprimary = curp->primary;
1.83 schwarze 733: curp->primary = &blk;
1.28 joerg 734: curp->line = 1;
1.36 schwarze 735: recursion_depth++;
1.28 joerg 736:
1.93 schwarze 737: /* Skip an UTF-8 byte order mark. */
738: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
739: (unsigned char)blk.buf[0] == 0xef &&
740: (unsigned char)blk.buf[1] == 0xbb &&
741: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 742: offset = 3;
1.93 schwarze 743: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 744: } else
745: offset = 0;
1.93 schwarze 746:
1.95 schwarze 747: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 748:
1.36 schwarze 749: if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
1.28 joerg 750: mparse_end(curp);
751:
1.85 schwarze 752: curp->primary = svprimary;
1.28 joerg 753: curp->file = svfile;
754: }
755:
756: enum mandoclevel
757: mparse_readmem(struct mparse *curp, const void *buf, size_t len,
758: const char *file)
759: {
760: struct buf blk;
761:
762: blk.buf = UNCONST(buf);
763: blk.sz = len;
764:
1.36 schwarze 765: mparse_parse_buffer(curp, blk, file);
1.28 joerg 766: return(curp->file_status);
767: }
768:
1.98 schwarze 769: /*
770: * If a file descriptor is given, use it and assume it points
771: * to the named file. Otherwise, open the named file.
772: * Read the whole file into memory and call the parsers.
773: * Called recursively when an .so request is encountered.
774: */
1.36 schwarze 775: enum mandoclevel
776: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 777: {
1.28 joerg 778: struct buf blk;
779: int with_mmap;
1.93 schwarze 780: int save_filenc;
1.98 schwarze 781: pid_t save_child;
1.1 kristaps 782:
1.98 schwarze 783: save_child = curp->child;
784: if (fd != -1)
785: curp->child = 0;
786: else if (mparse_open(curp, &fd, file) >= MANDOCLEVEL_SYSERR)
787: goto out;
1.1 kristaps 788:
1.91 schwarze 789: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 790: save_filenc = curp->filenc;
791: curp->filenc = curp->options &
792: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 793: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 794: curp->filenc = save_filenc;
1.81 schwarze 795: #if HAVE_MMAP
1.91 schwarze 796: if (with_mmap)
797: munmap(blk.buf, blk.sz);
798: else
1.28 joerg 799: #endif
1.91 schwarze 800: free(blk.buf);
801: }
1.1 kristaps 802:
1.98 schwarze 803: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 804: perror(file);
1.91 schwarze 805:
1.98 schwarze 806: mparse_wait(curp);
807: out:
808: curp->child = save_child;
1.1 kristaps 809: return(curp->file_status);
1.82 schwarze 810: }
811:
812: enum mandoclevel
1.97 schwarze 813: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 814: {
815: int pfd[2];
1.98 schwarze 816: int save_errno;
1.82 schwarze 817: char *cp;
818: enum mandocerr err;
819:
820: pfd[1] = -1;
821: curp->file = file;
1.98 schwarze 822:
823: /* Unless zipped, try to just open the file. */
824:
1.82 schwarze 825: if ((cp = strrchr(file, '.')) == NULL ||
826: strcmp(cp + 1, "gz")) {
1.97 schwarze 827: curp->child = 0;
1.98 schwarze 828: if ((*fd = open(file, O_RDONLY)) != -1)
829: return(MANDOCLEVEL_OK);
830:
831: /* Open failed; try to append ".gz". */
832:
833: mandoc_asprintf(&cp, "%s.gz", file);
834: file = cp;
835: } else
836: cp = NULL;
837:
838: /* Before forking, make sure the file can be read. */
839:
840: save_errno = errno;
841: if (access(file, R_OK) == -1) {
842: if (cp != NULL)
843: errno = save_errno;
844: err = MANDOCERR_SYSOPEN;
845: goto out;
1.82 schwarze 846: }
847:
1.98 schwarze 848: /* Run gunzip(1). */
849:
1.82 schwarze 850: if (pipe(pfd) == -1) {
851: err = MANDOCERR_SYSPIPE;
852: goto out;
853: }
854:
1.97 schwarze 855: switch (curp->child = fork()) {
1.82 schwarze 856: case -1:
857: err = MANDOCERR_SYSFORK;
858: close(pfd[0]);
859: close(pfd[1]);
860: pfd[1] = -1;
861: break;
862: case 0:
863: close(pfd[0]);
864: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
865: err = MANDOCERR_SYSDUP;
866: break;
867: }
868: execlp("gunzip", "gunzip", "-c", file, NULL);
869: err = MANDOCERR_SYSEXEC;
870: break;
871: default:
872: close(pfd[1]);
873: *fd = pfd[0];
874: return(MANDOCLEVEL_OK);
875: }
876:
877: out:
1.98 schwarze 878: free(cp);
1.82 schwarze 879: *fd = -1;
1.97 schwarze 880: curp->child = 0;
1.82 schwarze 881: curp->file_status = MANDOCLEVEL_SYSERR;
882: if (curp->mmsg)
1.98 schwarze 883: (*curp->mmsg)(err, curp->file_status, curp->file,
1.82 schwarze 884: 0, 0, strerror(errno));
885: if (pfd[1] != -1)
886: exit(1);
887: return(curp->file_status);
888: }
889:
890: enum mandoclevel
1.97 schwarze 891: mparse_wait(struct mparse *curp)
1.82 schwarze 892: {
893: int status;
894:
1.97 schwarze 895: if (curp->child == 0)
896: return(MANDOCLEVEL_OK);
897:
898: if (waitpid(curp->child, &status, 0) == -1) {
1.82 schwarze 899: mandoc_msg(MANDOCERR_SYSWAIT, curp, 0, 0,
900: strerror(errno));
901: curp->file_status = MANDOCLEVEL_SYSERR;
902: return(curp->file_status);
903: }
904: if (WIFSIGNALED(status)) {
905: mandoc_vmsg(MANDOCERR_SYSSIG, curp, 0, 0,
906: "%d", WTERMSIG(status));
907: curp->file_status = MANDOCLEVEL_SYSERR;
908: return(curp->file_status);
909: }
910: if (WEXITSTATUS(status)) {
911: mandoc_vmsg(MANDOCERR_SYSEXIT, curp, 0, 0,
912: "%d", WEXITSTATUS(status));
913: curp->file_status = MANDOCLEVEL_SYSERR;
914: return(curp->file_status);
915: }
916: return(MANDOCLEVEL_OK);
1.1 kristaps 917: }
918:
919: struct mparse *
1.94 schwarze 920: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
921: const struct mchars *mchars, const char *defos)
1.1 kristaps 922: {
923: struct mparse *curp;
1.10 kristaps 924:
925: assert(wlevel <= MANDOCLEVEL_FATAL);
1.1 kristaps 926:
927: curp = mandoc_calloc(1, sizeof(struct mparse));
928:
1.44 schwarze 929: curp->options = options;
1.3 kristaps 930: curp->wlevel = wlevel;
1.1 kristaps 931: curp->mmsg = mmsg;
1.29 schwarze 932: curp->defos = defos;
1.1 kristaps 933:
1.94 schwarze 934: curp->mchars = mchars;
935: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 936: if (curp->options & MPARSE_MDOC)
937: curp->pmdoc = mdoc_alloc(
938: curp->roff, curp, curp->defos,
939: curp->options & MPARSE_QUICK ? 1 : 0);
940: if (curp->options & MPARSE_MAN)
941: curp->pman = man_alloc(curp->roff, curp,
942: curp->options & MPARSE_QUICK ? 1 : 0);
943:
1.1 kristaps 944: return(curp);
945: }
946:
947: void
948: mparse_reset(struct mparse *curp)
949: {
950:
951: roff_reset(curp->roff);
952:
953: if (curp->mdoc)
954: mdoc_reset(curp->mdoc);
955: if (curp->man)
956: man_reset(curp->man);
1.24 kristaps 957: if (curp->secondary)
958: curp->secondary->sz = 0;
1.1 kristaps 959:
960: curp->file_status = MANDOCLEVEL_OK;
961: curp->mdoc = NULL;
962: curp->man = NULL;
1.45 schwarze 963:
964: free(curp->sodest);
965: curp->sodest = NULL;
1.1 kristaps 966: }
967:
968: void
969: mparse_free(struct mparse *curp)
970: {
971:
972: if (curp->pmdoc)
973: mdoc_free(curp->pmdoc);
974: if (curp->pman)
975: man_free(curp->pman);
976: if (curp->roff)
977: roff_free(curp->roff);
1.24 kristaps 978: if (curp->secondary)
979: free(curp->secondary->buf);
1.1 kristaps 980:
1.24 kristaps 981: free(curp->secondary);
1.45 schwarze 982: free(curp->sodest);
1.1 kristaps 983: free(curp);
984: }
985:
986: void
1.45 schwarze 987: mparse_result(struct mparse *curp,
988: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 989: {
990:
1.45 schwarze 991: if (sodest && NULL != (*sodest = curp->sodest)) {
992: *mdoc = NULL;
993: *man = NULL;
994: return;
995: }
1.9 kristaps 996: if (mdoc)
997: *mdoc = curp->mdoc;
998: if (man)
999: *man = curp->man;
1.3 kristaps 1000: }
1001:
1002: void
1003: mandoc_vmsg(enum mandocerr t, struct mparse *m,
1004: int ln, int pos, const char *fmt, ...)
1005: {
1006: char buf[256];
1007: va_list ap;
1008:
1009: va_start(ap, fmt);
1.48 schwarze 1010: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 1011: va_end(ap);
1012:
1013: mandoc_msg(t, m, ln, pos, buf);
1014: }
1015:
1016: void
1.47 schwarze 1017: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 1018: int ln, int col, const char *msg)
1019: {
1020: enum mandoclevel level;
1021:
1022: level = MANDOCLEVEL_FATAL;
1023: while (er < mandoclimits[level])
1024: level--;
1025:
1026: if (level < m->wlevel)
1027: return;
1028:
1.8 kristaps 1029: if (m->mmsg)
1030: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 1031:
1032: if (m->file_status < level)
1033: m->file_status = level;
1.7 kristaps 1034: }
1035:
1036: const char *
1037: mparse_strerror(enum mandocerr er)
1038: {
1039:
1040: return(mandocerrs[er]);
1041: }
1042:
1043: const char *
1044: mparse_strlevel(enum mandoclevel lvl)
1045: {
1046: return(mandoclevels[lvl]);
1.24 kristaps 1047: }
1048:
1049: void
1050: mparse_keep(struct mparse *p)
1051: {
1052:
1053: assert(NULL == p->secondary);
1054: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1055: }
1056:
1057: const char *
1058: mparse_getkeep(const struct mparse *p)
1059: {
1060:
1061: assert(p->secondary);
1062: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1063: }
CVSweb