Annotation of mandoc/read.c, Revision 1.172
1.172 ! schwarze 1: /* $Id: read.c,v 1.171 2017/06/06 15:01:04 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.156 schwarze 4: * Copyright (c) 2010-2017 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: *
1.133 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 kristaps 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.133 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 kristaps 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>
1.82 schwarze 22: #include <sys/mman.h>
1.80 schwarze 23: #include <sys/stat.h>
1.1 kristaps 24:
25: #include <assert.h>
26: #include <ctype.h>
1.146 schwarze 27: #if HAVE_ERR
1.143 schwarze 28: #include <err.h>
1.146 schwarze 29: #endif
1.40 schwarze 30: #include <errno.h>
1.1 kristaps 31: #include <fcntl.h>
1.3 kristaps 32: #include <stdarg.h>
1.1 kristaps 33: #include <stdio.h>
34: #include <stdlib.h>
35: #include <string.h>
36: #include <unistd.h>
1.140 schwarze 37: #include <zlib.h>
1.1 kristaps 38:
1.133 schwarze 39: #include "mandoc_aux.h"
1.1 kristaps 40: #include "mandoc.h"
1.133 schwarze 41: #include "roff.h"
1.1 kristaps 42: #include "mdoc.h"
43: #include "man.h"
1.133 schwarze 44: #include "libmandoc.h"
1.139 schwarze 45: #include "roff_int.h"
1.1 kristaps 46:
47: #define REPARSE_LIMIT 1000
48:
49: struct mparse {
1.160 schwarze 50: struct roff *roff; /* roff parser (!NULL) */
1.134 schwarze 51: struct roff_man *man; /* man parser */
1.45 schwarze 52: char *sodest; /* filename pointed to by .so */
1.83 schwarze 53: const char *file; /* filename of current input file */
54: struct buf *primary; /* buffer currently being parsed */
55: struct buf *secondary; /* preprocessed copy of input */
56: const char *defos; /* default operating system */
57: mandocmsg mmsg; /* warning/error message handler */
58: enum mandoclevel file_status; /* status of current parse */
59: enum mandoclevel wlevel; /* ignore messages below this */
60: int options; /* parser options */
1.140 schwarze 61: int gzip; /* current input file is gzipped */
1.93 schwarze 62: int filenc; /* encoding of the current file */
1.1 kristaps 63: int reparse_count; /* finite interp. stack */
1.83 schwarze 64: int line; /* line number in the file */
1.1 kristaps 65: };
66:
1.84 schwarze 67: static void choose_parser(struct mparse *);
1.1 kristaps 68: static void resize_buf(struct buf *, size_t);
1.163 schwarze 69: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 70: static int read_whole_file(struct mparse *, const char *, int,
71: struct buf *, int *);
1.1 kristaps 72: static void mparse_end(struct mparse *);
1.37 schwarze 73: static void mparse_parse_buffer(struct mparse *, struct buf,
74: const char *);
1.1 kristaps 75:
1.3 kristaps 76: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
77: MANDOCERR_OK,
1.165 schwarze 78: MANDOCERR_STYLE,
1.3 kristaps 79: MANDOCERR_WARNING,
80: MANDOCERR_ERROR,
1.112 schwarze 81: MANDOCERR_UNSUPP,
1.3 kristaps 82: MANDOCERR_MAX,
83: MANDOCERR_MAX
84: };
85:
1.7 kristaps 86: static const char * const mandocerrs[MANDOCERR_MAX] = {
87: "ok",
88:
1.165 schwarze 89: "generic style suggestion",
1.166 schwarze 90:
91: "useless macro",
1.167 schwarze 92: "consider using OS macro",
1.172 ! schwarze 93: "errnos out of order",
! 94: "duplicate errno",
1.168 schwarze 95: "description line ends with a full stop",
1.166 schwarze 96:
1.7 kristaps 97: "generic warning",
98:
99: /* related to the prologue */
1.79 schwarze 100: "missing manual title, using UNTITLED",
101: "missing manual title, using \"\"",
1.54 schwarze 102: "lower case character in document title",
1.79 schwarze 103: "missing manual section, using \"\"",
1.7 kristaps 104: "unknown manual section",
1.54 schwarze 105: "missing date, using today's date",
1.7 kristaps 106: "cannot parse date, using it verbatim",
1.79 schwarze 107: "missing Os macro, using \"\"",
108: "duplicate prologue macro",
109: "late prologue macro",
110: "skipping late title macro",
1.7 kristaps 111: "prologue macros out of order",
112:
113: /* related to document structure */
114: ".so is fragile, better use ln(1)",
1.50 schwarze 115: "no document body",
1.54 schwarze 116: "content before first section header",
117: "first section is not \"NAME\"",
1.156 schwarze 118: "NAME section without Nm before Nd",
1.128 schwarze 119: "NAME section without description",
120: "description not at the end of NAME",
121: "bad NAME section content",
1.156 schwarze 122: "missing comma before name",
1.121 schwarze 123: "missing description line, using \"\"",
1.162 schwarze 124: "description line outside NAME section",
1.7 kristaps 125: "sections out of conventional order",
1.54 schwarze 126: "duplicate section title",
127: "unexpected section",
1.87 schwarze 128: "unusual Xr order",
129: "unusual Xr punctuation",
1.86 schwarze 130: "AUTHORS section without An macro",
1.7 kristaps 131:
132: /* related to macros and nesting */
1.55 schwarze 133: "obsolete macro",
1.102 schwarze 134: "macro neither callable nor escaped",
1.7 kristaps 135: "skipping paragraph macro",
1.31 schwarze 136: "moving paragraph macro out of list",
1.7 kristaps 137: "skipping no-space macro",
138: "blocks badly nested",
139: "nested displays are not portable",
1.57 schwarze 140: "moving content out of list",
1.78 schwarze 141: "fill mode already enabled, skipping",
142: "fill mode already disabled, skipping",
1.7 kristaps 143: "line scope broken",
1.169 schwarze 144: "skipping blank line in line scope",
1.7 kristaps 145:
146: /* related to missing macro arguments */
1.58 schwarze 147: "skipping empty request",
148: "conditional request controls empty scope",
1.7 kristaps 149: "skipping empty macro",
1.124 schwarze 150: "empty block",
1.62 schwarze 151: "empty argument, using 0n",
1.60 schwarze 152: "missing display type, using -ragged",
153: "list type is not the first argument",
1.152 schwarze 154: "missing -width in -tag list, using 6n",
1.78 schwarze 155: "missing utility name, using \"\"",
1.123 schwarze 156: "missing function name, using \"\"",
1.60 schwarze 157: "empty head in list item",
158: "empty list item",
1.61 schwarze 159: "missing font type, using \\fR",
160: "unknown font type, using \\fR",
1.103 schwarze 161: "nothing follows prefix",
1.122 schwarze 162: "empty reference block",
1.155 schwarze 163: "missing section argument",
1.60 schwarze 164: "missing -std argument, adding it",
1.125 schwarze 165: "missing option string, using \"\"",
166: "missing resource identifier, using \"\"",
1.90 schwarze 167: "missing eqn box, using \"\"",
1.7 kristaps 168:
169: /* related to bad macro arguments */
1.64 schwarze 170: "unterminated quoted argument",
1.7 kristaps 171: "duplicate argument",
1.76 schwarze 172: "skipping duplicate argument",
1.63 schwarze 173: "skipping duplicate display type",
174: "skipping duplicate list type",
1.76 schwarze 175: "skipping -width argument",
1.126 schwarze 176: "wrong number of cells",
1.7 kristaps 177: "unknown AT&T UNIX version",
1.88 schwarze 178: "comma in function argument",
1.89 schwarze 179: "parenthesis in function name",
1.67 schwarze 180: "invalid content in Rs block",
1.63 schwarze 181: "invalid Boolean argument",
182: "unknown font, skipping request",
1.126 schwarze 183: "odd number of characters in request",
1.7 kristaps 184:
185: /* related to plain text */
1.64 schwarze 186: "blank line in fill mode, using .sp",
187: "tab in filled text",
188: "whitespace at end of input line",
1.158 schwarze 189: "new sentence, new line",
1.7 kristaps 190: "bad comment style",
1.64 schwarze 191: "invalid escape sequence",
192: "undefined string, using \"\"",
1.16 kristaps 193:
1.118 schwarze 194: /* related to tables */
195: "tbl line starts with span",
196: "tbl column starts with span",
197: "skipping vertical bar in tbl layout",
198:
1.7 kristaps 199: "generic error",
200:
201: /* related to tables */
1.116 schwarze 202: "non-alphabetic character in tbl options",
203: "skipping unknown tbl option",
204: "missing tbl option argument",
205: "wrong tbl option argument size",
1.118 schwarze 206: "empty tbl layout",
207: "invalid character in tbl layout",
208: "unmatched parenthesis in tbl layout",
1.119 schwarze 209: "tbl without any data cells",
210: "ignoring data in spanned tbl cell",
211: "ignoring extra tbl data cells",
212: "data block open at end of tbl",
1.7 kristaps 213:
1.68 schwarze 214: /* related to document structure and macros */
1.108 schwarze 215: NULL,
1.7 kristaps 216: "input stack limit exceeded, infinite loop?",
217: "skipping bad character",
1.68 schwarze 218: "skipping unknown macro",
1.112 schwarze 219: "skipping insecure request",
1.73 schwarze 220: "skipping item outside list",
1.68 schwarze 221: "skipping column outside column list",
222: "skipping end of block that is not open",
1.115 schwarze 223: "fewer RS blocks open, skipping",
1.68 schwarze 224: "inserting missing end of block",
225: "appending missing end of block",
226:
227: /* related to request and macro arguments */
1.7 kristaps 228: "escaped character not allowed in a name",
1.99 schwarze 229: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 230: "skipping display without arguments",
1.71 schwarze 231: "missing list type, using -item",
1.171 schwarze 232: "argument is not numeric, using 1",
1.70 schwarze 233: "missing manual name, using \"\"",
1.71 schwarze 234: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 235: "unknown standard specifier",
1.71 schwarze 236: "skipping request without numeric argument",
1.109 schwarze 237: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 238: ".so request failed",
1.61 schwarze 239: "skipping all arguments",
240: "skipping excess arguments",
1.92 kristaps 241: "divide by zero",
1.112 schwarze 242:
243: "unsupported feature",
244: "input too large",
1.114 schwarze 245: "unsupported control character",
1.112 schwarze 246: "unsupported roff request",
1.119 schwarze 247: "eqn delim option in tbl",
1.118 schwarze 248: "unsupported tbl layout modifier",
1.112 schwarze 249: "ignoring macro in table",
1.7 kristaps 250: };
251:
252: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
253: "SUCCESS",
1.165 schwarze 254: "STYLE",
1.7 kristaps 255: "WARNING",
256: "ERROR",
1.112 schwarze 257: "UNSUPP",
1.7 kristaps 258: "BADARG",
259: "SYSERR"
260: };
261:
1.47 schwarze 262:
1.1 kristaps 263: static void
264: resize_buf(struct buf *buf, size_t initial)
265: {
266:
267: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
268: buf->buf = mandoc_realloc(buf->buf, buf->sz);
269: }
270:
271: static void
1.84 schwarze 272: choose_parser(struct mparse *curp)
1.1 kristaps 273: {
1.83 schwarze 274: char *cp, *ep;
275: int format;
1.1 kristaps 276:
1.83 schwarze 277: /*
278: * If neither command line arguments -mdoc or -man select
279: * a parser nor the roff parser found a .Dd or .TH macro
280: * yet, look ahead in the main input buffer.
281: */
282:
283: if ((format = roff_getformat(curp->roff)) == 0) {
284: cp = curp->primary->buf;
285: ep = cp + curp->primary->sz;
286: while (cp < ep) {
1.85 schwarze 287: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 288: cp++;
289: if (cp[0] == 'D' && cp[1] == 'd') {
290: format = MPARSE_MDOC;
291: break;
292: }
293: if (cp[0] == 'T' && cp[1] == 'H') {
294: format = MPARSE_MAN;
295: break;
296: }
297: }
298: cp = memchr(cp, '\n', ep - cp);
299: if (cp == NULL)
300: break;
301: cp++;
302: }
1.1 kristaps 303: }
304:
1.83 schwarze 305: if (format == MPARSE_MDOC) {
1.137 schwarze 306: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 307: if (curp->man->mdocmac == NULL)
308: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 309: } else {
310: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 311: if (curp->man->manmac == NULL)
312: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 313: }
1.164 schwarze 314: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 315: }
316:
317: /*
1.95 schwarze 318: * Main parse routine for a buffer.
319: * It assumes encoding and line numbering are already set up.
320: * It can recurse directly (for invocations of user-defined
321: * macros, inline equations, and input line traps)
322: * and indirectly (for .so file inclusion).
1.1 kristaps 323: */
1.163 schwarze 324: static int
1.95 schwarze 325: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 326: {
327: const struct tbl_span *span;
328: struct buf ln;
1.113 schwarze 329: const char *save_file;
1.110 schwarze 330: char *cp;
1.95 schwarze 331: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 332: enum rofferr rr;
1.100 schwarze 333: int of;
1.1 kristaps 334: int lnn; /* line number in the real file */
1.110 schwarze 335: int fd;
1.1 kristaps 336: unsigned char c;
337:
1.95 schwarze 338: memset(&ln, 0, sizeof(ln));
1.1 kristaps 339:
1.47 schwarze 340: lnn = curp->line;
341: pos = 0;
1.1 kristaps 342:
1.95 schwarze 343: while (i < blk.sz) {
1.1 kristaps 344: if (0 == pos && '\0' == blk.buf[i])
345: break;
346:
347: if (start) {
348: curp->line = lnn;
349: curp->reparse_count = 0;
1.93 schwarze 350:
351: if (lnn < 3 &&
352: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 353: curp->filenc & MPARSE_LATIN1)
354: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 355: }
356:
1.95 schwarze 357: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 358:
359: /*
360: * When finding an unescaped newline character,
361: * leave the character loop to process the line.
362: * Skip a preceding carriage return, if any.
363: */
364:
1.95 schwarze 365: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 366: '\n' == blk.buf[i + 1])
367: ++i;
368: if ('\n' == blk.buf[i]) {
369: ++i;
370: ++lnn;
371: break;
372: }
373:
1.35 schwarze 374: /*
1.93 schwarze 375: * Make sure we have space for the worst
376: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 377: */
378:
1.95 schwarze 379: if (pos + 11 > ln.sz)
1.35 schwarze 380: resize_buf(&ln, 256);
381:
1.47 schwarze 382: /*
1.93 schwarze 383: * Encode 8-bit input.
1.1 kristaps 384: */
385:
1.93 schwarze 386: c = blk.buf[i];
387: if (c & 0x80) {
1.95 schwarze 388: if ( ! (curp->filenc && preconv_encode(
389: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 390: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
391: curp->line, pos, "0x%x", c);
1.93 schwarze 392: ln.buf[pos++] = '?';
393: i++;
394: }
395: continue;
396: }
397:
398: /*
399: * Exclude control characters.
400: */
1.1 kristaps 401:
1.93 schwarze 402: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 403: mandoc_vmsg(c == 0x00 || c == 0x04 ||
404: c > 0x0a ? MANDOCERR_CHAR_BAD :
405: MANDOCERR_CHAR_UNSUPP,
406: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 407: i++;
1.127 schwarze 408: if (c != '\r')
409: ln.buf[pos++] = '?';
1.1 kristaps 410: continue;
411: }
412:
413: ln.buf[pos++] = blk.buf[i++];
414: }
415:
1.170 schwarze 416: if (pos + 1 >= ln.sz)
1.1 kristaps 417: resize_buf(&ln, 256);
418:
1.170 schwarze 419: if (i == blk.sz || blk.buf[i] == '\0')
420: ln.buf[pos++] = '\n';
1.1 kristaps 421: ln.buf[pos] = '\0';
422:
423: /*
424: * A significant amount of complexity is contained by
425: * the roff preprocessor. It's line-oriented but can be
426: * expressed on one line, so we need at times to
427: * readjust our starting point and re-run it. The roff
428: * preprocessor can also readjust the buffers with new
429: * data, so we pass them in wholesale.
430: */
431:
432: of = 0;
433:
1.24 kristaps 434: /*
435: * Maintain a lookaside buffer of all parsed lines. We
436: * only do this if mparse_keep() has been invoked (the
437: * buffer may be accessed with mparse_getkeep()).
438: */
439:
440: if (curp->secondary) {
1.47 schwarze 441: curp->secondary->buf = mandoc_realloc(
442: curp->secondary->buf,
443: curp->secondary->sz + pos + 2);
444: memcpy(curp->secondary->buf +
445: curp->secondary->sz,
446: ln.buf, pos);
1.24 kristaps 447: curp->secondary->sz += pos;
448: curp->secondary->buf
449: [curp->secondary->sz] = '\n';
450: curp->secondary->sz++;
451: curp->secondary->buf
452: [curp->secondary->sz] = '\0';
453: }
1.1 kristaps 454: rerun:
1.96 schwarze 455: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 456:
457: switch (rr) {
1.47 schwarze 458: case ROFF_REPARSE:
1.163 schwarze 459: if (++curp->reparse_count > REPARSE_LIMIT)
1.3 kristaps 460: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 461: curp->line, pos, NULL);
1.163 schwarze 462: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
463: start == 1) {
464: pos = 0;
465: continue;
466: }
467: free(ln.buf);
468: return 0;
1.47 schwarze 469: case ROFF_APPEND:
1.95 schwarze 470: pos = strlen(ln.buf);
1.1 kristaps 471: continue;
1.47 schwarze 472: case ROFF_RERUN:
1.1 kristaps 473: goto rerun;
1.47 schwarze 474: case ROFF_IGN:
1.1 kristaps 475: pos = 0;
476: continue;
1.47 schwarze 477: case ROFF_SO:
1.95 schwarze 478: if ( ! (curp->options & MPARSE_SO) &&
479: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 480: curp->sodest = mandoc_strdup(ln.buf + of);
481: free(ln.buf);
1.163 schwarze 482: return 1;
1.45 schwarze 483: }
1.24 kristaps 484: /*
485: * We remove `so' clauses from our lookaside
486: * buffer because we're going to descend into
487: * the file recursively.
488: */
1.47 schwarze 489: if (curp->secondary)
1.25 kristaps 490: curp->secondary->sz -= pos + 1;
1.113 schwarze 491: save_file = curp->file;
1.148 schwarze 492: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 493: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 494: close(fd);
1.113 schwarze 495: curp->file = save_file;
496: } else {
497: curp->file = save_file;
1.52 schwarze 498: mandoc_vmsg(MANDOCERR_SO_FAIL,
499: curp, curp->line, pos,
500: ".so %s", ln.buf + of);
1.110 schwarze 501: ln.sz = mandoc_asprintf(&cp,
502: ".sp\nSee the file %s.\n.sp",
503: ln.buf + of);
504: free(ln.buf);
505: ln.buf = cp;
506: of = 0;
507: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 508: }
1.1 kristaps 509: pos = 0;
510: continue;
511: default:
512: break;
513: }
514:
1.150 schwarze 515: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 516: choose_parser(curp);
1.1 kristaps 517:
1.47 schwarze 518: /*
1.84 schwarze 519: * Lastly, push down into the parsers themselves.
1.1 kristaps 520: * If libroff returns ROFF_TBL, then add it to the
521: * currently open parse. Since we only get here if
522: * there does exist data (see tbl_data.c), we're
523: * guaranteed that something's been allocated.
524: * Do the same for ROFF_EQN.
525: */
526:
1.139 schwarze 527: if (rr == ROFF_TBL)
1.100 schwarze 528: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 529: roff_addtbl(curp->man, span);
530: else if (rr == ROFF_EQN)
531: roff_addeqn(curp->man, roff_eqn(curp->roff));
532: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 533: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 534: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
535: break;
1.1 kristaps 536:
537: /* Temporary buffers typically are not full. */
538:
539: if (0 == start && '\0' == blk.buf[i])
540: break;
541:
542: /* Start the next input line. */
543:
544: pos = 0;
545: }
546:
547: free(ln.buf);
1.163 schwarze 548: return 1;
1.1 kristaps 549: }
550:
551: static int
1.40 schwarze 552: read_whole_file(struct mparse *curp, const char *file, int fd,
553: struct buf *fb, int *with_mmap)
1.1 kristaps 554: {
1.161 schwarze 555: struct stat st;
1.140 schwarze 556: gzFile gz;
1.1 kristaps 557: size_t off;
558: ssize_t ssz;
1.143 schwarze 559:
560: if (fstat(fd, &st) == -1)
561: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 562:
563: /*
564: * If we're a regular file, try just reading in the whole entry
565: * via mmap(). This is faster than reading it into blocks, and
566: * since each file is only a few bytes to begin with, I'm not
567: * concerned that this is going to tank any machines.
568: */
569:
1.140 schwarze 570: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 571: if (st.st_size > 0x7fffffff) {
1.111 schwarze 572: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 573: return 0;
1.1 kristaps 574: }
575: *with_mmap = 1;
576: fb->sz = (size_t)st.st_size;
1.37 schwarze 577: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 578: if (fb->buf != MAP_FAILED)
1.142 schwarze 579: return 1;
1.1 kristaps 580: }
581:
1.140 schwarze 582: if (curp->gzip) {
1.143 schwarze 583: if ((gz = gzdopen(fd, "rb")) == NULL)
584: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 585: } else
586: gz = NULL;
587:
1.1 kristaps 588: /*
589: * If this isn't a regular file (like, say, stdin), then we must
590: * go the old way and just read things in bit by bit.
591: */
592:
593: *with_mmap = 0;
594: off = 0;
595: fb->sz = 0;
596: fb->buf = NULL;
597: for (;;) {
598: if (off == fb->sz) {
599: if (fb->sz == (1U << 31)) {
1.111 schwarze 600: mandoc_msg(MANDOCERR_TOOLARGE, curp,
601: 0, 0, NULL);
1.1 kristaps 602: break;
603: }
604: resize_buf(fb, 65536);
605: }
1.140 schwarze 606: ssz = curp->gzip ?
607: gzread(gz, fb->buf + (int)off, fb->sz - off) :
608: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 609: if (ssz == 0) {
610: fb->sz = off;
1.142 schwarze 611: return 1;
1.1 kristaps 612: }
1.143 schwarze 613: if (ssz == -1)
614: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 615: off += (size_t)ssz;
616: }
617:
618: free(fb->buf);
619: fb->buf = NULL;
1.142 schwarze 620: return 0;
1.1 kristaps 621: }
622:
623: static void
624: mparse_end(struct mparse *curp)
625: {
1.135 schwarze 626: if (curp->man->macroset == MACROSET_NONE)
627: curp->man->macroset = MACROSET_MAN;
628: if (curp->man->macroset == MACROSET_MDOC)
629: mdoc_endparse(curp->man);
630: else
1.111 schwarze 631: man_endparse(curp->man);
1.1 kristaps 632: roff_endparse(curp->roff);
633: }
634:
635: static void
1.36 schwarze 636: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 637: {
1.85 schwarze 638: struct buf *svprimary;
1.28 joerg 639: const char *svfile;
1.95 schwarze 640: size_t offset;
1.36 schwarze 641: static int recursion_depth;
642:
643: if (64 < recursion_depth) {
644: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
645: return;
646: }
1.28 joerg 647:
648: /* Line number is per-file. */
649: svfile = curp->file;
650: curp->file = file;
1.85 schwarze 651: svprimary = curp->primary;
1.83 schwarze 652: curp->primary = &blk;
1.28 joerg 653: curp->line = 1;
1.36 schwarze 654: recursion_depth++;
1.28 joerg 655:
1.93 schwarze 656: /* Skip an UTF-8 byte order mark. */
657: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
658: (unsigned char)blk.buf[0] == 0xef &&
659: (unsigned char)blk.buf[1] == 0xbb &&
660: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 661: offset = 3;
1.93 schwarze 662: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 663: } else
664: offset = 0;
1.93 schwarze 665:
1.95 schwarze 666: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 667:
1.111 schwarze 668: if (--recursion_depth == 0)
1.28 joerg 669: mparse_end(curp);
670:
1.85 schwarze 671: curp->primary = svprimary;
1.28 joerg 672: curp->file = svfile;
673: }
674:
675: enum mandoclevel
1.104 schwarze 676: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 677: const char *file)
678: {
679: struct buf blk;
680:
1.104 schwarze 681: blk.buf = buf;
1.28 joerg 682: blk.sz = len;
683:
1.36 schwarze 684: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 685: return curp->file_status;
1.28 joerg 686: }
687:
1.98 schwarze 688: /*
689: * Read the whole file into memory and call the parsers.
690: * Called recursively when an .so request is encountered.
691: */
1.36 schwarze 692: enum mandoclevel
693: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 694: {
1.28 joerg 695: struct buf blk;
696: int with_mmap;
1.93 schwarze 697: int save_filenc;
1.1 kristaps 698:
1.91 schwarze 699: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 700: save_filenc = curp->filenc;
701: curp->filenc = curp->options &
702: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 703: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 704: curp->filenc = save_filenc;
1.91 schwarze 705: if (with_mmap)
706: munmap(blk.buf, blk.sz);
707: else
708: free(blk.buf);
709: }
1.142 schwarze 710: return curp->file_status;
1.82 schwarze 711: }
712:
1.148 schwarze 713: int
714: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 715: {
716: char *cp;
1.148 schwarze 717: int fd;
1.82 schwarze 718:
719: curp->file = file;
1.140 schwarze 720: cp = strrchr(file, '.');
721: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 722:
1.140 schwarze 723: /* First try to use the filename as it is. */
1.98 schwarze 724:
1.148 schwarze 725: if ((fd = open(file, O_RDONLY)) != -1)
726: return fd;
1.98 schwarze 727:
1.140 schwarze 728: /*
729: * If that doesn't work and the filename doesn't
730: * already end in .gz, try appending .gz.
731: */
1.98 schwarze 732:
1.140 schwarze 733: if ( ! curp->gzip) {
1.98 schwarze 734: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 735: fd = open(cp, O_RDONLY);
1.108 schwarze 736: free(cp);
1.148 schwarze 737: if (fd != -1) {
1.140 schwarze 738: curp->gzip = 1;
1.148 schwarze 739: return fd;
1.82 schwarze 740: }
741: }
742:
1.140 schwarze 743: /* Neither worked, give up. */
1.97 schwarze 744:
1.140 schwarze 745: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 746: return -1;
1.1 kristaps 747: }
748:
749: struct mparse *
1.94 schwarze 750: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.144 schwarze 751: const char *defos)
1.1 kristaps 752: {
753: struct mparse *curp;
1.10 kristaps 754:
1.1 kristaps 755: curp = mandoc_calloc(1, sizeof(struct mparse));
756:
1.44 schwarze 757: curp->options = options;
1.3 kristaps 758: curp->wlevel = wlevel;
1.1 kristaps 759: curp->mmsg = mmsg;
1.29 schwarze 760: curp->defos = defos;
1.1 kristaps 761:
1.144 schwarze 762: curp->roff = roff_alloc(curp, options);
1.137 schwarze 763: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
764: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 765: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 766: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 767: if (curp->man->mdocmac == NULL)
768: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 769: } else if (curp->options & MPARSE_MAN) {
770: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 771: if (curp->man->manmac == NULL)
772: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 773: }
1.138 schwarze 774: curp->man->first->tok = TOKEN_NONE;
1.142 schwarze 775: return curp;
1.1 kristaps 776: }
777:
778: void
779: mparse_reset(struct mparse *curp)
780: {
781: roff_reset(curp->roff);
1.150 schwarze 782: roff_man_reset(curp->man);
1.160 schwarze 783:
784: free(curp->sodest);
785: curp->sodest = NULL;
786:
1.24 kristaps 787: if (curp->secondary)
788: curp->secondary->sz = 0;
1.1 kristaps 789:
790: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 791: curp->gzip = 0;
1.1 kristaps 792: }
793:
794: void
795: mparse_free(struct mparse *curp)
796: {
797:
1.164 schwarze 798: roffhash_free(curp->man->mdocmac);
799: roffhash_free(curp->man->manmac);
1.137 schwarze 800: roff_man_free(curp->man);
1.160 schwarze 801: roff_free(curp->roff);
1.24 kristaps 802: if (curp->secondary)
803: free(curp->secondary->buf);
1.1 kristaps 804:
1.24 kristaps 805: free(curp->secondary);
1.45 schwarze 806: free(curp->sodest);
1.1 kristaps 807: free(curp);
808: }
809:
810: void
1.135 schwarze 811: mparse_result(struct mparse *curp, struct roff_man **man,
812: char **sodest)
1.1 kristaps 813: {
814:
1.45 schwarze 815: if (sodest && NULL != (*sodest = curp->sodest)) {
816: *man = NULL;
817: return;
818: }
1.9 kristaps 819: if (man)
820: *man = curp->man;
1.157 schwarze 821: }
822:
823: void
824: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
825: {
826: if (curp->file_status > *rc)
827: *rc = curp->file_status;
1.3 kristaps 828: }
829:
830: void
831: mandoc_vmsg(enum mandocerr t, struct mparse *m,
832: int ln, int pos, const char *fmt, ...)
833: {
834: char buf[256];
835: va_list ap;
836:
837: va_start(ap, fmt);
1.48 schwarze 838: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 839: va_end(ap);
840:
841: mandoc_msg(t, m, ln, pos, buf);
842: }
843:
844: void
1.47 schwarze 845: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 846: int ln, int col, const char *msg)
847: {
848: enum mandoclevel level;
849:
1.112 schwarze 850: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 851: while (er < mandoclimits[level])
852: level--;
853:
1.108 schwarze 854: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 855: return;
856:
1.8 kristaps 857: if (m->mmsg)
858: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 859:
860: if (m->file_status < level)
861: m->file_status = level;
1.7 kristaps 862: }
863:
864: const char *
865: mparse_strerror(enum mandocerr er)
866: {
867:
1.142 schwarze 868: return mandocerrs[er];
1.7 kristaps 869: }
870:
871: const char *
872: mparse_strlevel(enum mandoclevel lvl)
873: {
1.142 schwarze 874: return mandoclevels[lvl];
1.24 kristaps 875: }
876:
877: void
878: mparse_keep(struct mparse *p)
879: {
880:
881: assert(NULL == p->secondary);
882: p->secondary = mandoc_calloc(1, sizeof(struct buf));
883: }
884:
885: const char *
886: mparse_getkeep(const struct mparse *p)
887: {
888:
889: assert(p->secondary);
1.142 schwarze 890: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 891: }
CVSweb