Annotation of src/usr.bin/mandoc/read.c, Revision 1.112
1.112 ! schwarze 1: /* $OpenBSD: read.c,v 1.111 2015/04/18 17:01:28 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: *
1.108 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.108 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 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>
1.107 schwarze 28: #include <signal.h>
1.1 schwarze 29: #include <stdarg.h>
30: #include <stdio.h>
31: #include <stdlib.h>
32: #include <string.h>
33: #include <unistd.h>
34:
1.108 schwarze 35: #include "mandoc_aux.h"
1.1 schwarze 36: #include "mandoc.h"
1.108 schwarze 37: #include "roff.h"
1.1 schwarze 38: #include "mdoc.h"
39: #include "man.h"
1.108 schwarze 40: #include "libmandoc.h"
1.1 schwarze 41:
42: #define REPARSE_LIMIT 1000
43:
44: struct mparse {
1.109 schwarze 45: struct roff_man *man; /* man parser */
1.1 schwarze 46: struct roff *roff; /* roff parser (!NULL) */
1.69 schwarze 47: const struct mchars *mchars; /* character table */
1.23 schwarze 48: char *sodest; /* filename pointed to by .so */
1.59 schwarze 49: const char *file; /* filename of current input file */
50: struct buf *primary; /* buffer currently being parsed */
51: struct buf *secondary; /* preprocessed copy of input */
52: const char *defos; /* default operating system */
53: mandocmsg mmsg; /* warning/error message handler */
54: enum mandoclevel file_status; /* status of current parse */
55: enum mandoclevel wlevel; /* ignore messages below this */
56: int options; /* parser options */
1.70 schwarze 57: int filenc; /* encoding of the current file */
1.1 schwarze 58: int reparse_count; /* finite interp. stack */
1.59 schwarze 59: int line; /* line number in the file */
1.73 schwarze 60: pid_t child; /* the gunzip(1) process */
1.1 schwarze 61: };
62:
1.60 schwarze 63: static void choose_parser(struct mparse *);
1.1 schwarze 64: static void resize_buf(struct buf *, size_t);
1.71 schwarze 65: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.18 schwarze 66: static int read_whole_file(struct mparse *, const char *, int,
67: struct buf *, int *);
1.1 schwarze 68: static void mparse_end(struct mparse *);
1.15 schwarze 69: static void mparse_parse_buffer(struct mparse *, struct buf,
70: const char *);
1.1 schwarze 71:
72: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
73: MANDOCERR_OK,
74: MANDOCERR_WARNING,
75: MANDOCERR_WARNING,
76: MANDOCERR_ERROR,
1.87 schwarze 77: MANDOCERR_UNSUPP,
1.1 schwarze 78: MANDOCERR_MAX,
79: MANDOCERR_MAX
80: };
81:
82: static const char * const mandocerrs[MANDOCERR_MAX] = {
83: "ok",
84:
85: "generic warning",
86:
87: /* related to the prologue */
1.57 schwarze 88: "missing manual title, using UNTITLED",
89: "missing manual title, using \"\"",
1.32 schwarze 90: "lower case character in document title",
1.57 schwarze 91: "missing manual section, using \"\"",
1.1 schwarze 92: "unknown manual section",
1.32 schwarze 93: "missing date, using today's date",
1.1 schwarze 94: "cannot parse date, using it verbatim",
1.57 schwarze 95: "missing Os macro, using \"\"",
96: "duplicate prologue macro",
97: "late prologue macro",
98: "skipping late title macro",
1.1 schwarze 99: "prologue macros out of order",
100:
101: /* related to document structure */
102: ".so is fragile, better use ln(1)",
1.28 schwarze 103: "no document body",
1.32 schwarze 104: "content before first section header",
105: "first section is not \"NAME\"",
1.103 schwarze 106: "NAME section without name",
107: "NAME section without description",
108: "description not at the end of NAME",
109: "bad NAME section content",
1.96 schwarze 110: "missing description line, using \"\"",
1.1 schwarze 111: "sections out of conventional order",
1.32 schwarze 112: "duplicate section title",
113: "unexpected section",
1.63 schwarze 114: "unusual Xr order",
115: "unusual Xr punctuation",
1.62 schwarze 116: "AUTHORS section without An macro",
1.1 schwarze 117:
118: /* related to macros and nesting */
1.33 schwarze 119: "obsolete macro",
1.78 schwarze 120: "macro neither callable nor escaped",
1.1 schwarze 121: "skipping paragraph macro",
1.10 schwarze 122: "moving paragraph macro out of list",
1.1 schwarze 123: "skipping no-space macro",
124: "blocks badly nested",
125: "nested displays are not portable",
1.35 schwarze 126: "moving content out of list",
127: ".Vt block has child macro",
1.56 schwarze 128: "fill mode already enabled, skipping",
129: "fill mode already disabled, skipping",
1.1 schwarze 130: "line scope broken",
131:
132: /* related to missing macro arguments */
1.36 schwarze 133: "skipping empty request",
134: "conditional request controls empty scope",
1.1 schwarze 135: "skipping empty macro",
1.99 schwarze 136: "empty block",
1.40 schwarze 137: "empty argument, using 0n",
1.38 schwarze 138: "missing display type, using -ragged",
139: "list type is not the first argument",
140: "missing -width in -tag list, using 8n",
1.56 schwarze 141: "missing utility name, using \"\"",
1.98 schwarze 142: "missing function name, using \"\"",
1.38 schwarze 143: "empty head in list item",
144: "empty list item",
1.39 schwarze 145: "missing font type, using \\fR",
146: "unknown font type, using \\fR",
1.79 schwarze 147: "nothing follows prefix",
1.97 schwarze 148: "empty reference block",
1.38 schwarze 149: "missing -std argument, adding it",
1.100 schwarze 150: "missing option string, using \"\"",
151: "missing resource identifier, using \"\"",
1.66 schwarze 152: "missing eqn box, using \"\"",
1.1 schwarze 153:
154: /* related to bad macro arguments */
1.42 schwarze 155: "unterminated quoted argument",
1.1 schwarze 156: "duplicate argument",
1.54 schwarze 157: "skipping duplicate argument",
1.41 schwarze 158: "skipping duplicate display type",
159: "skipping duplicate list type",
1.54 schwarze 160: "skipping -width argument",
1.101 schwarze 161: "wrong number of cells",
1.1 schwarze 162: "unknown AT&T UNIX version",
1.64 schwarze 163: "comma in function argument",
1.65 schwarze 164: "parenthesis in function name",
1.45 schwarze 165: "invalid content in Rs block",
1.41 schwarze 166: "invalid Boolean argument",
167: "unknown font, skipping request",
1.101 schwarze 168: "odd number of characters in request",
1.1 schwarze 169:
170: /* related to plain text */
1.42 schwarze 171: "blank line in fill mode, using .sp",
172: "tab in filled text",
173: "whitespace at end of input line",
1.1 schwarze 174: "bad comment style",
1.42 schwarze 175: "invalid escape sequence",
176: "undefined string, using \"\"",
1.3 schwarze 177:
1.93 schwarze 178: /* related to tables */
179: "tbl line starts with span",
180: "tbl column starts with span",
181: "skipping vertical bar in tbl layout",
182:
1.1 schwarze 183: "generic error",
1.3 schwarze 184:
1.1 schwarze 185: /* related to tables */
1.91 schwarze 186: "non-alphabetic character in tbl options",
187: "skipping unknown tbl option",
188: "missing tbl option argument",
189: "wrong tbl option argument size",
1.93 schwarze 190: "empty tbl layout",
191: "invalid character in tbl layout",
192: "unmatched parenthesis in tbl layout",
1.94 schwarze 193: "tbl without any data cells",
194: "ignoring data in spanned tbl cell",
195: "ignoring extra tbl data cells",
196: "data block open at end of tbl",
1.1 schwarze 197:
1.46 schwarze 198: /* related to document structure and macros */
1.83 schwarze 199: NULL,
1.1 schwarze 200: "input stack limit exceeded, infinite loop?",
201: "skipping bad character",
1.46 schwarze 202: "skipping unknown macro",
1.87 schwarze 203: "skipping insecure request",
1.51 schwarze 204: "skipping item outside list",
1.46 schwarze 205: "skipping column outside column list",
206: "skipping end of block that is not open",
1.90 schwarze 207: "fewer RS blocks open, skipping",
1.46 schwarze 208: "inserting missing end of block",
209: "appending missing end of block",
210:
211: /* related to request and macro arguments */
1.1 schwarze 212: "escaped character not allowed in a name",
1.75 schwarze 213: "NOT IMPLEMENTED: Bd -file",
1.49 schwarze 214: "missing list type, using -item",
1.48 schwarze 215: "missing manual name, using \"\"",
1.49 schwarze 216: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 217: "unknown standard specifier",
1.49 schwarze 218: "skipping request without numeric argument",
1.84 schwarze 219: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 220: ".so request failed",
1.39 schwarze 221: "skipping all arguments",
222: "skipping excess arguments",
1.68 schwarze 223: "divide by zero",
1.87 schwarze 224:
225: "unsupported feature",
226: "input too large",
1.89 schwarze 227: "unsupported control character",
1.87 schwarze 228: "unsupported roff request",
1.94 schwarze 229: "eqn delim option in tbl",
1.93 schwarze 230: "unsupported tbl layout modifier",
1.87 schwarze 231: "ignoring macro in table",
1.1 schwarze 232: };
233:
234: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
235: "SUCCESS",
236: "RESERVED",
237: "WARNING",
238: "ERROR",
1.87 schwarze 239: "UNSUPP",
1.1 schwarze 240: "BADARG",
241: "SYSERR"
242: };
243:
1.25 schwarze 244:
1.1 schwarze 245: static void
246: resize_buf(struct buf *buf, size_t initial)
247: {
248:
249: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
250: buf->buf = mandoc_realloc(buf->buf, buf->sz);
251: }
252:
253: static void
1.60 schwarze 254: choose_parser(struct mparse *curp)
1.1 schwarze 255: {
1.59 schwarze 256: char *cp, *ep;
257: int format;
1.1 schwarze 258:
1.59 schwarze 259: /*
260: * If neither command line arguments -mdoc or -man select
261: * a parser nor the roff parser found a .Dd or .TH macro
262: * yet, look ahead in the main input buffer.
263: */
264:
265: if ((format = roff_getformat(curp->roff)) == 0) {
266: cp = curp->primary->buf;
267: ep = cp + curp->primary->sz;
268: while (cp < ep) {
1.61 schwarze 269: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 270: cp++;
271: if (cp[0] == 'D' && cp[1] == 'd') {
272: format = MPARSE_MDOC;
273: break;
274: }
275: if (cp[0] == 'T' && cp[1] == 'H') {
276: format = MPARSE_MAN;
277: break;
278: }
279: }
280: cp = memchr(cp, '\n', ep - cp);
281: if (cp == NULL)
282: break;
283: cp++;
284: }
1.1 schwarze 285: }
286:
1.112 ! schwarze 287: if (curp->man == NULL) {
! 288: curp->man = roff_man_alloc(curp->roff, curp, curp->defos,
! 289: curp->options & MPARSE_QUICK ? 1 : 0);
! 290: curp->man->macroset = MACROSET_MAN;
! 291: curp->man->first->tok = MDOC_MAX;
! 292: }
! 293:
1.59 schwarze 294: if (format == MPARSE_MDOC) {
1.111 schwarze 295: mdoc_hash_init();
1.112 ! schwarze 296: curp->man->macroset = MACROSET_MDOC;
! 297: curp->man->first->tok = MDOC_MAX;
! 298: } else {
! 299: man_hash_init();
! 300: curp->man->macroset = MACROSET_MAN;
! 301: curp->man->first->tok = MAN_MAX;
1.25 schwarze 302: }
1.1 schwarze 303: }
304:
305: /*
1.71 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 schwarze 311: */
312: static void
1.71 schwarze 313: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 314: {
315: const struct tbl_span *span;
316: struct buf ln;
1.88 schwarze 317: const char *save_file;
1.85 schwarze 318: char *cp;
1.71 schwarze 319: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 320: enum rofferr rr;
1.76 schwarze 321: int of;
1.1 schwarze 322: int lnn; /* line number in the real file */
1.85 schwarze 323: int fd;
324: pid_t save_child;
1.1 schwarze 325: unsigned char c;
326:
1.71 schwarze 327: memset(&ln, 0, sizeof(ln));
1.1 schwarze 328:
1.25 schwarze 329: lnn = curp->line;
330: pos = 0;
1.1 schwarze 331:
1.71 schwarze 332: while (i < blk.sz) {
1.1 schwarze 333: if (0 == pos && '\0' == blk.buf[i])
334: break;
335:
336: if (start) {
337: curp->line = lnn;
338: curp->reparse_count = 0;
1.70 schwarze 339:
340: if (lnn < 3 &&
341: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 342: curp->filenc & MPARSE_LATIN1)
343: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 344: }
345:
1.71 schwarze 346: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 347:
348: /*
349: * When finding an unescaped newline character,
350: * leave the character loop to process the line.
351: * Skip a preceding carriage return, if any.
352: */
353:
1.71 schwarze 354: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 355: '\n' == blk.buf[i + 1])
356: ++i;
357: if ('\n' == blk.buf[i]) {
358: ++i;
359: ++lnn;
360: break;
361: }
362:
1.13 schwarze 363: /*
1.70 schwarze 364: * Make sure we have space for the worst
365: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 366: */
367:
1.71 schwarze 368: if (pos + 11 > ln.sz)
1.13 schwarze 369: resize_buf(&ln, 256);
370:
1.25 schwarze 371: /*
1.70 schwarze 372: * Encode 8-bit input.
1.1 schwarze 373: */
374:
1.70 schwarze 375: c = blk.buf[i];
376: if (c & 0x80) {
1.71 schwarze 377: if ( ! (curp->filenc && preconv_encode(
378: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 379: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
380: curp->line, pos, "0x%x", c);
1.70 schwarze 381: ln.buf[pos++] = '?';
382: i++;
383: }
384: continue;
385: }
386:
387: /*
388: * Exclude control characters.
389: */
1.1 schwarze 390:
1.70 schwarze 391: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 392: mandoc_vmsg(c == 0x00 || c == 0x04 ||
393: c > 0x0a ? MANDOCERR_CHAR_BAD :
394: MANDOCERR_CHAR_UNSUPP,
395: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 396: i++;
1.102 schwarze 397: if (c != '\r')
398: ln.buf[pos++] = '?';
1.1 schwarze 399: continue;
400: }
401:
402: /* Trailing backslash = a plain char. */
403:
1.71 schwarze 404: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 schwarze 405: ln.buf[pos++] = blk.buf[i++];
406: continue;
407: }
408:
409: /*
410: * Found escape and at least one other character.
411: * When it's a newline character, skip it.
412: * When there is a carriage return in between,
413: * skip that one as well.
414: */
415:
1.71 schwarze 416: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 schwarze 417: '\n' == blk.buf[i + 2])
418: ++i;
419: if ('\n' == blk.buf[i + 1]) {
420: i += 2;
421: ++lnn;
422: continue;
423: }
424:
425: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
426: i += 2;
427: /* Comment, skip to end of line */
1.71 schwarze 428: for (; i < blk.sz; ++i) {
1.1 schwarze 429: if ('\n' == blk.buf[i]) {
430: ++i;
431: ++lnn;
432: break;
433: }
434: }
435:
436: /* Backout trailing whitespaces */
437: for (; pos > 0; --pos) {
438: if (ln.buf[pos - 1] != ' ')
439: break;
440: if (pos > 2 && ln.buf[pos - 2] == '\\')
441: break;
442: }
443: break;
444: }
445:
1.13 schwarze 446: /* Catch escaped bogus characters. */
447:
448: c = (unsigned char) blk.buf[i+1];
449:
1.25 schwarze 450: if ( ! (isascii(c) &&
451: (isgraph(c) || isblank(c)))) {
1.89 schwarze 452: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.56 schwarze 453: curp->line, pos, "0x%x", c);
1.13 schwarze 454: i += 2;
455: ln.buf[pos++] = '?';
456: continue;
457: }
458:
1.1 schwarze 459: /* Some other escape sequence, copy & cont. */
460:
461: ln.buf[pos++] = blk.buf[i++];
462: ln.buf[pos++] = blk.buf[i++];
463: }
464:
1.71 schwarze 465: if (pos >= ln.sz)
1.1 schwarze 466: resize_buf(&ln, 256);
467:
468: ln.buf[pos] = '\0';
469:
470: /*
471: * A significant amount of complexity is contained by
472: * the roff preprocessor. It's line-oriented but can be
473: * expressed on one line, so we need at times to
474: * readjust our starting point and re-run it. The roff
475: * preprocessor can also readjust the buffers with new
476: * data, so we pass them in wholesale.
477: */
478:
479: of = 0;
480:
1.4 schwarze 481: /*
482: * Maintain a lookaside buffer of all parsed lines. We
483: * only do this if mparse_keep() has been invoked (the
484: * buffer may be accessed with mparse_getkeep()).
485: */
486:
487: if (curp->secondary) {
1.25 schwarze 488: curp->secondary->buf = mandoc_realloc(
489: curp->secondary->buf,
490: curp->secondary->sz + pos + 2);
491: memcpy(curp->secondary->buf +
492: curp->secondary->sz,
493: ln.buf, pos);
1.4 schwarze 494: curp->secondary->sz += pos;
495: curp->secondary->buf
496: [curp->secondary->sz] = '\n';
497: curp->secondary->sz++;
498: curp->secondary->buf
499: [curp->secondary->sz] = '\0';
500: }
1.1 schwarze 501: rerun:
1.72 schwarze 502: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 503:
504: switch (rr) {
1.25 schwarze 505: case ROFF_REPARSE:
1.1 schwarze 506: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.71 schwarze 507: mparse_buf_r(curp, ln, of, 0);
1.1 schwarze 508: else
509: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 510: curp->line, pos, NULL);
1.1 schwarze 511: pos = 0;
512: continue;
1.25 schwarze 513: case ROFF_APPEND:
1.71 schwarze 514: pos = strlen(ln.buf);
1.1 schwarze 515: continue;
1.25 schwarze 516: case ROFF_RERUN:
1.1 schwarze 517: goto rerun;
1.25 schwarze 518: case ROFF_IGN:
1.1 schwarze 519: pos = 0;
520: continue;
1.25 schwarze 521: case ROFF_SO:
1.71 schwarze 522: if ( ! (curp->options & MPARSE_SO) &&
523: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 524: curp->sodest = mandoc_strdup(ln.buf + of);
525: free(ln.buf);
526: return;
527: }
1.4 schwarze 528: /*
529: * We remove `so' clauses from our lookaside
530: * buffer because we're going to descend into
531: * the file recursively.
532: */
1.25 schwarze 533: if (curp->secondary)
1.4 schwarze 534: curp->secondary->sz -= pos + 1;
1.88 schwarze 535: save_file = curp->file;
1.85 schwarze 536: save_child = curp->child;
537: if (mparse_open(curp, &fd, ln.buf + of) ==
1.88 schwarze 538: MANDOCLEVEL_OK) {
1.85 schwarze 539: mparse_readfd(curp, fd, ln.buf + of);
1.88 schwarze 540: curp->file = save_file;
541: } else {
542: curp->file = save_file;
1.30 schwarze 543: mandoc_vmsg(MANDOCERR_SO_FAIL,
544: curp, curp->line, pos,
545: ".so %s", ln.buf + of);
1.85 schwarze 546: ln.sz = mandoc_asprintf(&cp,
547: ".sp\nSee the file %s.\n.sp",
548: ln.buf + of);
549: free(ln.buf);
550: ln.buf = cp;
551: of = 0;
552: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 553: }
1.85 schwarze 554: curp->child = save_child;
1.1 schwarze 555: pos = 0;
556: continue;
557: default:
558: break;
559: }
560:
561: /*
562: * If input parsers have not been allocated, do so now.
1.2 schwarze 563: * We keep these instanced between parsers, but set them
1.1 schwarze 564: * locally per parse routine since we can use different
565: * parsers with each one.
566: */
567:
1.110 schwarze 568: if (curp->man == NULL ||
569: curp->man->macroset == MACROSET_NONE)
1.60 schwarze 570: choose_parser(curp);
1.1 schwarze 571:
1.25 schwarze 572: /*
1.60 schwarze 573: * Lastly, push down into the parsers themselves.
1.1 schwarze 574: * If libroff returns ROFF_TBL, then add it to the
575: * currently open parse. Since we only get here if
576: * there does exist data (see tbl_data.c), we're
577: * guaranteed that something's been allocated.
578: * Do the same for ROFF_EQN.
579: */
580:
1.76 schwarze 581: if (rr == ROFF_TBL) {
582: while ((span = roff_span(curp->roff)) != NULL)
1.110 schwarze 583: if (curp->man->macroset == MACROSET_MDOC)
584: mdoc_addspan(curp->man, span);
1.76 schwarze 585: else
586: man_addspan(curp->man, span);
587: } else if (rr == ROFF_EQN) {
1.110 schwarze 588: if (curp->man->macroset == MACROSET_MDOC)
589: mdoc_addeqn(curp->man, roff_eqn(curp->roff));
1.76 schwarze 590: else
591: man_addeqn(curp->man, roff_eqn(curp->roff));
1.110 schwarze 592: } else if ((curp->man->macroset == MACROSET_MDOC ?
593: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 594: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
595: break;
1.1 schwarze 596:
597: /* Temporary buffers typically are not full. */
598:
599: if (0 == start && '\0' == blk.buf[i])
600: break;
601:
602: /* Start the next input line. */
603:
604: pos = 0;
605: }
606:
607: free(ln.buf);
608: }
609:
610: static int
1.18 schwarze 611: read_whole_file(struct mparse *curp, const char *file, int fd,
612: struct buf *fb, int *with_mmap)
1.1 schwarze 613: {
614: struct stat st;
615: size_t off;
616: ssize_t ssz;
617:
618: if (-1 == fstat(fd, &st)) {
1.82 schwarze 619: perror(file);
620: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 621: }
622:
623: /*
624: * If we're a regular file, try just reading in the whole entry
625: * via mmap(). This is faster than reading it into blocks, and
626: * since each file is only a few bytes to begin with, I'm not
627: * concerned that this is going to tank any machines.
628: */
629:
630: if (S_ISREG(st.st_mode)) {
1.106 schwarze 631: if (st.st_size > 0x7fffffff) {
1.86 schwarze 632: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 schwarze 633: return(0);
634: }
635: *with_mmap = 1;
636: fb->sz = (size_t)st.st_size;
1.15 schwarze 637: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 638: if (fb->buf != MAP_FAILED)
639: return(1);
640: }
641:
642: /*
643: * If this isn't a regular file (like, say, stdin), then we must
644: * go the old way and just read things in bit by bit.
645: */
646:
647: *with_mmap = 0;
648: off = 0;
649: fb->sz = 0;
650: fb->buf = NULL;
651: for (;;) {
652: if (off == fb->sz) {
653: if (fb->sz == (1U << 31)) {
1.86 schwarze 654: mandoc_msg(MANDOCERR_TOOLARGE, curp,
655: 0, 0, NULL);
1.1 schwarze 656: break;
657: }
658: resize_buf(fb, 65536);
659: }
660: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
661: if (ssz == 0) {
662: fb->sz = off;
663: return(1);
664: }
665: if (ssz == -1) {
1.82 schwarze 666: perror(file);
667: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 668: }
669: off += (size_t)ssz;
670: }
671:
672: free(fb->buf);
673: fb->buf = NULL;
674: return(0);
675: }
676:
677: static void
678: mparse_end(struct mparse *curp)
679: {
680:
1.110 schwarze 681: if (curp->man == NULL && curp->sodest == NULL)
1.112 ! schwarze 682: curp->man = roff_man_alloc(curp->roff, curp, curp->defos,
1.110 schwarze 683: curp->options & MPARSE_QUICK ? 1 : 0);
684: if (curp->man->macroset == MACROSET_NONE)
685: curp->man->macroset = MACROSET_MAN;
686: if (curp->man->macroset == MACROSET_MDOC)
687: mdoc_endparse(curp->man);
688: else
1.86 schwarze 689: man_endparse(curp->man);
1.1 schwarze 690: roff_endparse(curp->roff);
691: }
692:
1.15 schwarze 693: static void
694: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 695: {
1.61 schwarze 696: struct buf *svprimary;
1.1 schwarze 697: const char *svfile;
1.71 schwarze 698: size_t offset;
1.14 schwarze 699: static int recursion_depth;
700:
701: if (64 < recursion_depth) {
702: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 703: return;
1.14 schwarze 704: }
1.1 schwarze 705:
1.15 schwarze 706: /* Line number is per-file. */
707: svfile = curp->file;
708: curp->file = file;
1.61 schwarze 709: svprimary = curp->primary;
1.59 schwarze 710: curp->primary = &blk;
1.15 schwarze 711: curp->line = 1;
712: recursion_depth++;
713:
1.70 schwarze 714: /* Skip an UTF-8 byte order mark. */
715: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
716: (unsigned char)blk.buf[0] == 0xef &&
717: (unsigned char)blk.buf[1] == 0xbb &&
718: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 719: offset = 3;
1.70 schwarze 720: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 721: } else
722: offset = 0;
1.70 schwarze 723:
1.71 schwarze 724: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 725:
1.86 schwarze 726: if (--recursion_depth == 0)
1.15 schwarze 727: mparse_end(curp);
728:
1.61 schwarze 729: curp->primary = svprimary;
1.15 schwarze 730: curp->file = svfile;
731: }
732:
1.74 schwarze 733: /*
734: * Read the whole file into memory and call the parsers.
735: * Called recursively when an .so request is encountered.
736: */
1.15 schwarze 737: enum mandoclevel
738: mparse_readfd(struct mparse *curp, int fd, const char *file)
739: {
740: struct buf blk;
741: int with_mmap;
1.70 schwarze 742: int save_filenc;
1.1 schwarze 743:
1.67 schwarze 744: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 745: save_filenc = curp->filenc;
746: curp->filenc = curp->options &
747: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 748: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 749: curp->filenc = save_filenc;
1.67 schwarze 750: if (with_mmap)
751: munmap(blk.buf, blk.sz);
752: else
753: free(blk.buf);
754: }
1.1 schwarze 755:
1.74 schwarze 756: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 schwarze 757: perror(file);
1.67 schwarze 758:
1.74 schwarze 759: mparse_wait(curp);
1.1 schwarze 760: return(curp->file_status);
1.58 schwarze 761: }
762:
763: enum mandoclevel
1.73 schwarze 764: mparse_open(struct mparse *curp, int *fd, const char *file)
1.58 schwarze 765: {
766: int pfd[2];
1.74 schwarze 767: int save_errno;
1.58 schwarze 768: char *cp;
769:
770: curp->file = file;
1.74 schwarze 771:
772: /* Unless zipped, try to just open the file. */
773:
1.58 schwarze 774: if ((cp = strrchr(file, '.')) == NULL ||
775: strcmp(cp + 1, "gz")) {
1.73 schwarze 776: curp->child = 0;
1.74 schwarze 777: if ((*fd = open(file, O_RDONLY)) != -1)
778: return(MANDOCLEVEL_OK);
779:
780: /* Open failed; try to append ".gz". */
781:
782: mandoc_asprintf(&cp, "%s.gz", file);
783: file = cp;
784: } else
785: cp = NULL;
786:
787: /* Before forking, make sure the file can be read. */
788:
789: save_errno = errno;
790: if (access(file, R_OK) == -1) {
791: if (cp != NULL)
792: errno = save_errno;
1.83 schwarze 793: free(cp);
794: *fd = -1;
795: curp->child = 0;
796: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
797: return(MANDOCLEVEL_ERROR);
1.58 schwarze 798: }
799:
1.74 schwarze 800: /* Run gunzip(1). */
801:
1.58 schwarze 802: if (pipe(pfd) == -1) {
1.82 schwarze 803: perror("pipe");
804: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 805: }
806:
1.73 schwarze 807: switch (curp->child = fork()) {
1.58 schwarze 808: case -1:
1.82 schwarze 809: perror("fork");
810: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 811: case 0:
812: close(pfd[0]);
813: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.82 schwarze 814: perror("dup");
815: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 816: }
1.107 schwarze 817: signal(SIGPIPE, SIG_DFL);
1.58 schwarze 818: execlp("gunzip", "gunzip", "-c", file, NULL);
1.82 schwarze 819: perror("exec");
820: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 821: default:
822: close(pfd[1]);
823: *fd = pfd[0];
824: return(MANDOCLEVEL_OK);
825: }
826: }
827:
828: enum mandoclevel
1.73 schwarze 829: mparse_wait(struct mparse *curp)
1.58 schwarze 830: {
831: int status;
832:
1.73 schwarze 833: if (curp->child == 0)
834: return(MANDOCLEVEL_OK);
835:
836: if (waitpid(curp->child, &status, 0) == -1) {
1.82 schwarze 837: perror("wait");
838: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 839: }
1.104 schwarze 840: curp->child = 0;
1.58 schwarze 841: if (WIFSIGNALED(status)) {
1.83 schwarze 842: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
843: "gunzip died from signal %d", WTERMSIG(status));
844: return(MANDOCLEVEL_ERROR);
1.58 schwarze 845: }
846: if (WEXITSTATUS(status)) {
1.83 schwarze 847: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
848: "gunzip failed with code %d", WEXITSTATUS(status));
849: return(MANDOCLEVEL_ERROR);
1.58 schwarze 850: }
851: return(MANDOCLEVEL_OK);
1.1 schwarze 852: }
853:
854: struct mparse *
1.69 schwarze 855: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
856: const struct mchars *mchars, const char *defos)
1.1 schwarze 857: {
858: struct mparse *curp;
859:
860: curp = mandoc_calloc(1, sizeof(struct mparse));
861:
1.22 schwarze 862: curp->options = options;
1.1 schwarze 863: curp->wlevel = wlevel;
864: curp->mmsg = mmsg;
1.7 schwarze 865: curp->defos = defos;
1.1 schwarze 866:
1.69 schwarze 867: curp->mchars = mchars;
868: curp->roff = roff_alloc(curp, curp->mchars, options);
1.112 ! schwarze 869: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
! 870: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 871: if (curp->options & MPARSE_MDOC) {
872: mdoc_hash_init();
1.112 ! schwarze 873: curp->man->macroset = MACROSET_MDOC;
! 874: curp->man->first->tok = MDOC_MAX;
! 875: } else if (curp->options & MPARSE_MAN) {
1.111 schwarze 876: man_hash_init();
1.112 ! schwarze 877: curp->man->macroset = MACROSET_MAN;
! 878: curp->man->first->tok = MAN_MAX;
1.111 schwarze 879: }
1.1 schwarze 880: return(curp);
881: }
882:
883: void
884: mparse_reset(struct mparse *curp)
885: {
886:
887: roff_reset(curp->roff);
888:
1.112 ! schwarze 889: if (curp->man != NULL)
! 890: roff_man_reset(curp->man);
1.4 schwarze 891: if (curp->secondary)
892: curp->secondary->sz = 0;
1.1 schwarze 893:
894: curp->file_status = MANDOCLEVEL_OK;
1.23 schwarze 895:
896: free(curp->sodest);
897: curp->sodest = NULL;
1.1 schwarze 898: }
899:
900: void
901: mparse_free(struct mparse *curp)
902: {
903:
1.112 ! schwarze 904: roff_man_free(curp->man);
1.1 schwarze 905: if (curp->roff)
906: roff_free(curp->roff);
1.4 schwarze 907: if (curp->secondary)
908: free(curp->secondary->buf);
1.1 schwarze 909:
1.4 schwarze 910: free(curp->secondary);
1.23 schwarze 911: free(curp->sodest);
1.1 schwarze 912: free(curp);
913: }
914:
915: void
1.110 schwarze 916: mparse_result(struct mparse *curp, struct roff_man **man,
917: char **sodest)
1.1 schwarze 918: {
919:
1.23 schwarze 920: if (sodest && NULL != (*sodest = curp->sodest)) {
921: *man = NULL;
922: return;
923: }
1.1 schwarze 924: if (man)
925: *man = curp->man;
926: }
927:
928: void
929: mandoc_vmsg(enum mandocerr t, struct mparse *m,
930: int ln, int pos, const char *fmt, ...)
931: {
932: char buf[256];
933: va_list ap;
934:
935: va_start(ap, fmt);
1.26 schwarze 936: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 937: va_end(ap);
938:
939: mandoc_msg(t, m, ln, pos, buf);
940: }
941:
942: void
1.25 schwarze 943: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 944: int ln, int col, const char *msg)
945: {
946: enum mandoclevel level;
947:
1.87 schwarze 948: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 949: while (er < mandoclimits[level])
950: level--;
951:
1.83 schwarze 952: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 953: return;
954:
955: if (m->mmsg)
956: (*m->mmsg)(er, level, m->file, ln, col, msg);
957:
958: if (m->file_status < level)
959: m->file_status = level;
960: }
961:
962: const char *
963: mparse_strerror(enum mandocerr er)
964: {
965:
966: return(mandocerrs[er]);
967: }
968:
969: const char *
970: mparse_strlevel(enum mandoclevel lvl)
971: {
972: return(mandoclevels[lvl]);
1.4 schwarze 973: }
974:
975: void
976: mparse_keep(struct mparse *p)
977: {
978:
979: assert(NULL == p->secondary);
980: p->secondary = mandoc_calloc(1, sizeof(struct buf));
981: }
982:
983: const char *
984: mparse_getkeep(const struct mparse *p)
985: {
986:
987: assert(p->secondary);
988: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 schwarze 989: }
![[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