Annotation of mandoc/read.c, Revision 1.96
1.96 ! schwarze 1: /* $Id: read.c,v 1.95 2014/11/01 04:08:43 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 */
1.1 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",
1.32 schwarze 99: "unknown manual volume or arch",
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",
1.71 schwarze 204: "missing list type, using -item",
1.70 schwarze 205: "missing manual name, using \"\"",
1.71 schwarze 206: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 207: "unknown standard specifier",
1.71 schwarze 208: "skipping request without numeric argument",
1.61 schwarze 209: "skipping all arguments",
210: "skipping excess arguments",
1.92 kristaps 211: "divide by zero",
1.7 kristaps 212:
213: "generic fatal error",
214:
1.40 schwarze 215: "input too large",
1.78 schwarze 216: "NOT IMPLEMENTED: Bd -file",
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;
1.95 schwarze 319: int of, rc;
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:
572: rc = -1;
573:
574: if (ROFF_TBL == rr)
575: while (NULL != (span = roff_span(curp->roff))) {
576: rc = curp->man ?
1.47 schwarze 577: man_addspan(curp->man, span) :
578: mdoc_addspan(curp->mdoc, span);
1.1 kristaps 579: if (0 == rc)
580: break;
581: }
582: else if (ROFF_EQN == rr)
1.47 schwarze 583: rc = curp->mdoc ?
584: mdoc_addeqn(curp->mdoc,
585: roff_eqn(curp->roff)) :
586: man_addeqn(curp->man,
587: roff_eqn(curp->roff));
1.1 kristaps 588: else if (curp->man || curp->mdoc)
589: rc = curp->man ?
1.47 schwarze 590: man_parseln(curp->man,
591: curp->line, ln.buf, of) :
592: mdoc_parseln(curp->mdoc,
593: curp->line, ln.buf, of);
1.1 kristaps 594:
595: if (0 == rc) {
596: assert(MANDOCLEVEL_FATAL <= curp->file_status);
597: break;
1.41 schwarze 598: } else if (2 == rc)
599: break;
1.1 kristaps 600:
601: /* Temporary buffers typically are not full. */
602:
603: if (0 == start && '\0' == blk.buf[i])
604: break;
605:
606: /* Start the next input line. */
607:
608: pos = 0;
609: }
610:
611: free(ln.buf);
612: }
613:
614: static int
1.40 schwarze 615: read_whole_file(struct mparse *curp, const char *file, int fd,
616: struct buf *fb, int *with_mmap)
1.1 kristaps 617: {
618: size_t off;
619: ssize_t ssz;
620:
1.81 schwarze 621: #if HAVE_MMAP
1.15 kristaps 622: struct stat st;
1.1 kristaps 623: if (-1 == fstat(fd, &st)) {
1.40 schwarze 624: curp->file_status = MANDOCLEVEL_SYSERR;
625: if (curp->mmsg)
626: (*curp->mmsg)(MANDOCERR_SYSSTAT, curp->file_status,
627: file, 0, 0, strerror(errno));
1.1 kristaps 628: return(0);
629: }
630:
631: /*
632: * If we're a regular file, try just reading in the whole entry
633: * via mmap(). This is faster than reading it into blocks, and
634: * since each file is only a few bytes to begin with, I'm not
635: * concerned that this is going to tank any machines.
636: */
637:
638: if (S_ISREG(st.st_mode)) {
639: if (st.st_size >= (1U << 31)) {
1.40 schwarze 640: curp->file_status = MANDOCLEVEL_FATAL;
641: if (curp->mmsg)
642: (*curp->mmsg)(MANDOCERR_TOOLARGE,
643: curp->file_status, file, 0, 0, NULL);
1.1 kristaps 644: return(0);
645: }
646: *with_mmap = 1;
647: fb->sz = (size_t)st.st_size;
1.37 schwarze 648: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 649: if (fb->buf != MAP_FAILED)
650: return(1);
651: }
1.15 kristaps 652: #endif
1.1 kristaps 653:
654: /*
655: * If this isn't a regular file (like, say, stdin), then we must
656: * go the old way and just read things in bit by bit.
657: */
658:
659: *with_mmap = 0;
660: off = 0;
661: fb->sz = 0;
662: fb->buf = NULL;
663: for (;;) {
664: if (off == fb->sz) {
665: if (fb->sz == (1U << 31)) {
1.40 schwarze 666: curp->file_status = MANDOCLEVEL_FATAL;
667: if (curp->mmsg)
668: (*curp->mmsg)(MANDOCERR_TOOLARGE,
669: curp->file_status,
670: file, 0, 0, NULL);
1.1 kristaps 671: break;
672: }
673: resize_buf(fb, 65536);
674: }
675: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
676: if (ssz == 0) {
677: fb->sz = off;
678: return(1);
679: }
680: if (ssz == -1) {
1.40 schwarze 681: curp->file_status = MANDOCLEVEL_SYSERR;
682: if (curp->mmsg)
683: (*curp->mmsg)(MANDOCERR_SYSREAD,
684: curp->file_status, file, 0, 0,
685: strerror(errno));
1.1 kristaps 686: break;
687: }
688: off += (size_t)ssz;
689: }
690:
691: free(fb->buf);
692: fb->buf = NULL;
693: return(0);
694: }
695:
696: static void
697: mparse_end(struct mparse *curp)
698: {
699:
700: if (MANDOCLEVEL_FATAL <= curp->file_status)
701: return;
702:
1.72 schwarze 703: if (curp->mdoc == NULL &&
704: curp->man == NULL &&
705: curp->sodest == NULL) {
706: if (curp->options & MPARSE_MDOC)
707: curp->mdoc = curp->pmdoc;
708: else {
709: if (curp->pman == NULL)
710: curp->pman = man_alloc(curp->roff, curp,
711: curp->options & MPARSE_QUICK ? 1 : 0);
712: curp->man = curp->pman;
713: }
714: }
715:
1.1 kristaps 716: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
717: assert(MANDOCLEVEL_FATAL <= curp->file_status);
718: return;
719: }
720:
721: if (curp->man && ! man_endparse(curp->man)) {
722: assert(MANDOCLEVEL_FATAL <= curp->file_status);
723: return;
724: }
725:
726: roff_endparse(curp->roff);
727: }
728:
729: static void
1.36 schwarze 730: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 731: {
1.85 schwarze 732: struct buf *svprimary;
1.28 joerg 733: const char *svfile;
1.95 schwarze 734: size_t offset;
1.36 schwarze 735: static int recursion_depth;
736:
737: if (64 < recursion_depth) {
738: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
739: return;
740: }
1.28 joerg 741:
742: /* Line number is per-file. */
743: svfile = curp->file;
744: curp->file = file;
1.85 schwarze 745: svprimary = curp->primary;
1.83 schwarze 746: curp->primary = &blk;
1.28 joerg 747: curp->line = 1;
1.36 schwarze 748: recursion_depth++;
1.28 joerg 749:
1.93 schwarze 750: /* Skip an UTF-8 byte order mark. */
751: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
752: (unsigned char)blk.buf[0] == 0xef &&
753: (unsigned char)blk.buf[1] == 0xbb &&
754: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 755: offset = 3;
1.93 schwarze 756: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 757: } else
758: offset = 0;
1.93 schwarze 759:
1.95 schwarze 760: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 761:
1.36 schwarze 762: if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
1.28 joerg 763: mparse_end(curp);
764:
1.85 schwarze 765: curp->primary = svprimary;
1.28 joerg 766: curp->file = svfile;
767: }
768:
769: enum mandoclevel
770: mparse_readmem(struct mparse *curp, const void *buf, size_t len,
771: const char *file)
772: {
773: struct buf blk;
774:
775: blk.buf = UNCONST(buf);
776: blk.sz = len;
777:
1.36 schwarze 778: mparse_parse_buffer(curp, blk, file);
1.28 joerg 779: return(curp->file_status);
780: }
781:
1.36 schwarze 782: enum mandoclevel
783: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 784: {
1.28 joerg 785: struct buf blk;
786: int with_mmap;
1.93 schwarze 787: int save_filenc;
1.1 kristaps 788:
1.40 schwarze 789: if (-1 == fd && -1 == (fd = open(file, O_RDONLY, 0))) {
790: curp->file_status = MANDOCLEVEL_SYSERR;
791: if (curp->mmsg)
792: (*curp->mmsg)(MANDOCERR_SYSOPEN,
793: curp->file_status,
794: file, 0, 0, strerror(errno));
1.91 schwarze 795: return(curp->file_status);
1.40 schwarze 796: }
797:
1.28 joerg 798: /*
799: * Run for each opened file; may be called more than once for
800: * each full parse sequence if the opened file is nested (i.e.,
801: * from `so'). Simply sucks in the whole file and moves into
802: * the parse phase for the file.
803: */
1.1 kristaps 804:
1.91 schwarze 805: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 806: save_filenc = curp->filenc;
807: curp->filenc = curp->options &
808: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 809: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 810: curp->filenc = save_filenc;
1.81 schwarze 811: #if HAVE_MMAP
1.91 schwarze 812: if (with_mmap)
813: munmap(blk.buf, blk.sz);
814: else
1.28 joerg 815: #endif
1.91 schwarze 816: free(blk.buf);
817: }
1.1 kristaps 818:
819: if (STDIN_FILENO != fd && -1 == close(fd))
820: perror(file);
1.91 schwarze 821:
1.1 kristaps 822: return(curp->file_status);
1.82 schwarze 823: }
824:
825: enum mandoclevel
826: mparse_open(struct mparse *curp, int *fd, const char *file,
827: pid_t *child_pid)
828: {
829: int pfd[2];
830: char *cp;
831: enum mandocerr err;
832:
833: pfd[1] = -1;
834: curp->file = file;
835: if ((cp = strrchr(file, '.')) == NULL ||
836: strcmp(cp + 1, "gz")) {
837: *child_pid = 0;
838: if ((*fd = open(file, O_RDONLY)) == -1) {
839: err = MANDOCERR_SYSOPEN;
840: goto out;
841: }
842: return(MANDOCLEVEL_OK);
843: }
844:
845: if (pipe(pfd) == -1) {
846: err = MANDOCERR_SYSPIPE;
847: goto out;
848: }
849:
850: switch (*child_pid = fork()) {
851: case -1:
852: err = MANDOCERR_SYSFORK;
853: close(pfd[0]);
854: close(pfd[1]);
855: pfd[1] = -1;
856: break;
857: case 0:
858: close(pfd[0]);
859: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
860: err = MANDOCERR_SYSDUP;
861: break;
862: }
863: execlp("gunzip", "gunzip", "-c", file, NULL);
864: err = MANDOCERR_SYSEXEC;
865: break;
866: default:
867: close(pfd[1]);
868: *fd = pfd[0];
869: return(MANDOCLEVEL_OK);
870: }
871:
872: out:
873: *fd = -1;
874: *child_pid = 0;
875: curp->file_status = MANDOCLEVEL_SYSERR;
876: if (curp->mmsg)
877: (*curp->mmsg)(err, curp->file_status, file,
878: 0, 0, strerror(errno));
879: if (pfd[1] != -1)
880: exit(1);
881: return(curp->file_status);
882: }
883:
884: enum mandoclevel
885: mparse_wait(struct mparse *curp, pid_t child_pid)
886: {
887: int status;
888:
889: if (waitpid(child_pid, &status, 0) == -1) {
890: mandoc_msg(MANDOCERR_SYSWAIT, curp, 0, 0,
891: strerror(errno));
892: curp->file_status = MANDOCLEVEL_SYSERR;
893: return(curp->file_status);
894: }
895: if (WIFSIGNALED(status)) {
896: mandoc_vmsg(MANDOCERR_SYSSIG, curp, 0, 0,
897: "%d", WTERMSIG(status));
898: curp->file_status = MANDOCLEVEL_SYSERR;
899: return(curp->file_status);
900: }
901: if (WEXITSTATUS(status)) {
902: mandoc_vmsg(MANDOCERR_SYSEXIT, curp, 0, 0,
903: "%d", WEXITSTATUS(status));
904: curp->file_status = MANDOCLEVEL_SYSERR;
905: return(curp->file_status);
906: }
907: return(MANDOCLEVEL_OK);
1.1 kristaps 908: }
909:
910: struct mparse *
1.94 schwarze 911: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
912: const struct mchars *mchars, const char *defos)
1.1 kristaps 913: {
914: struct mparse *curp;
1.10 kristaps 915:
916: assert(wlevel <= MANDOCLEVEL_FATAL);
1.1 kristaps 917:
918: curp = mandoc_calloc(1, sizeof(struct mparse));
919:
1.44 schwarze 920: curp->options = options;
1.3 kristaps 921: curp->wlevel = wlevel;
1.1 kristaps 922: curp->mmsg = mmsg;
1.29 schwarze 923: curp->defos = defos;
1.1 kristaps 924:
1.94 schwarze 925: curp->mchars = mchars;
926: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 927: if (curp->options & MPARSE_MDOC)
928: curp->pmdoc = mdoc_alloc(
929: curp->roff, curp, curp->defos,
930: curp->options & MPARSE_QUICK ? 1 : 0);
931: if (curp->options & MPARSE_MAN)
932: curp->pman = man_alloc(curp->roff, curp,
933: curp->options & MPARSE_QUICK ? 1 : 0);
934:
1.1 kristaps 935: return(curp);
936: }
937:
938: void
939: mparse_reset(struct mparse *curp)
940: {
941:
942: roff_reset(curp->roff);
943:
944: if (curp->mdoc)
945: mdoc_reset(curp->mdoc);
946: if (curp->man)
947: man_reset(curp->man);
1.24 kristaps 948: if (curp->secondary)
949: curp->secondary->sz = 0;
1.1 kristaps 950:
951: curp->file_status = MANDOCLEVEL_OK;
952: curp->mdoc = NULL;
953: curp->man = NULL;
1.45 schwarze 954:
955: free(curp->sodest);
956: curp->sodest = NULL;
1.1 kristaps 957: }
958:
959: void
960: mparse_free(struct mparse *curp)
961: {
962:
963: if (curp->pmdoc)
964: mdoc_free(curp->pmdoc);
965: if (curp->pman)
966: man_free(curp->pman);
967: if (curp->roff)
968: roff_free(curp->roff);
1.24 kristaps 969: if (curp->secondary)
970: free(curp->secondary->buf);
1.1 kristaps 971:
1.24 kristaps 972: free(curp->secondary);
1.45 schwarze 973: free(curp->sodest);
1.1 kristaps 974: free(curp);
975: }
976:
977: void
1.45 schwarze 978: mparse_result(struct mparse *curp,
979: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 980: {
981:
1.45 schwarze 982: if (sodest && NULL != (*sodest = curp->sodest)) {
983: *mdoc = NULL;
984: *man = NULL;
985: return;
986: }
1.9 kristaps 987: if (mdoc)
988: *mdoc = curp->mdoc;
989: if (man)
990: *man = curp->man;
1.3 kristaps 991: }
992:
993: void
994: mandoc_vmsg(enum mandocerr t, struct mparse *m,
995: int ln, int pos, const char *fmt, ...)
996: {
997: char buf[256];
998: va_list ap;
999:
1000: va_start(ap, fmt);
1.48 schwarze 1001: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 1002: va_end(ap);
1003:
1004: mandoc_msg(t, m, ln, pos, buf);
1005: }
1006:
1007: void
1.47 schwarze 1008: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 1009: int ln, int col, const char *msg)
1010: {
1011: enum mandoclevel level;
1012:
1013: level = MANDOCLEVEL_FATAL;
1014: while (er < mandoclimits[level])
1015: level--;
1016:
1017: if (level < m->wlevel)
1018: return;
1019:
1.8 kristaps 1020: if (m->mmsg)
1021: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 1022:
1023: if (m->file_status < level)
1024: m->file_status = level;
1.7 kristaps 1025: }
1026:
1027: const char *
1028: mparse_strerror(enum mandocerr er)
1029: {
1030:
1031: return(mandocerrs[er]);
1032: }
1033:
1034: const char *
1035: mparse_strlevel(enum mandoclevel lvl)
1036: {
1037: return(mandoclevels[lvl]);
1.24 kristaps 1038: }
1039:
1040: void
1041: mparse_keep(struct mparse *p)
1042: {
1043:
1044: assert(NULL == p->secondary);
1045: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1046: }
1047:
1048: const char *
1049: mparse_getkeep(const struct mparse *p)
1050: {
1051:
1052: assert(p->secondary);
1053: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1054: }
CVSweb