Annotation of mandoc/read.c, Revision 1.175
1.175 ! schwarze 1: /* $Id: read.c,v 1.174 2017/06/10 01:48:53 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",
1.174 schwarze 96: "no blank before trailing delimiter",
1.175 ! schwarze 97: "function name without markup",
1.166 schwarze 98:
1.7 kristaps 99: "generic warning",
100:
101: /* related to the prologue */
1.79 schwarze 102: "missing manual title, using UNTITLED",
103: "missing manual title, using \"\"",
1.54 schwarze 104: "lower case character in document title",
1.79 schwarze 105: "missing manual section, using \"\"",
1.7 kristaps 106: "unknown manual section",
1.54 schwarze 107: "missing date, using today's date",
1.7 kristaps 108: "cannot parse date, using it verbatim",
1.79 schwarze 109: "missing Os macro, using \"\"",
110: "duplicate prologue macro",
111: "late prologue macro",
112: "skipping late title macro",
1.7 kristaps 113: "prologue macros out of order",
114:
115: /* related to document structure */
116: ".so is fragile, better use ln(1)",
1.50 schwarze 117: "no document body",
1.54 schwarze 118: "content before first section header",
119: "first section is not \"NAME\"",
1.156 schwarze 120: "NAME section without Nm before Nd",
1.128 schwarze 121: "NAME section without description",
122: "description not at the end of NAME",
123: "bad NAME section content",
1.156 schwarze 124: "missing comma before name",
1.121 schwarze 125: "missing description line, using \"\"",
1.162 schwarze 126: "description line outside NAME section",
1.7 kristaps 127: "sections out of conventional order",
1.54 schwarze 128: "duplicate section title",
129: "unexpected section",
1.87 schwarze 130: "unusual Xr order",
131: "unusual Xr punctuation",
1.86 schwarze 132: "AUTHORS section without An macro",
1.7 kristaps 133:
134: /* related to macros and nesting */
1.55 schwarze 135: "obsolete macro",
1.102 schwarze 136: "macro neither callable nor escaped",
1.7 kristaps 137: "skipping paragraph macro",
1.31 schwarze 138: "moving paragraph macro out of list",
1.7 kristaps 139: "skipping no-space macro",
140: "blocks badly nested",
141: "nested displays are not portable",
1.57 schwarze 142: "moving content out of list",
1.78 schwarze 143: "fill mode already enabled, skipping",
144: "fill mode already disabled, skipping",
1.7 kristaps 145: "line scope broken",
1.169 schwarze 146: "skipping blank line in line scope",
1.7 kristaps 147:
148: /* related to missing macro arguments */
1.58 schwarze 149: "skipping empty request",
150: "conditional request controls empty scope",
1.7 kristaps 151: "skipping empty macro",
1.124 schwarze 152: "empty block",
1.62 schwarze 153: "empty argument, using 0n",
1.60 schwarze 154: "missing display type, using -ragged",
155: "list type is not the first argument",
1.152 schwarze 156: "missing -width in -tag list, using 6n",
1.78 schwarze 157: "missing utility name, using \"\"",
1.123 schwarze 158: "missing function name, using \"\"",
1.60 schwarze 159: "empty head in list item",
160: "empty list item",
1.61 schwarze 161: "missing font type, using \\fR",
162: "unknown font type, using \\fR",
1.103 schwarze 163: "nothing follows prefix",
1.122 schwarze 164: "empty reference block",
1.155 schwarze 165: "missing section argument",
1.60 schwarze 166: "missing -std argument, adding it",
1.125 schwarze 167: "missing option string, using \"\"",
168: "missing resource identifier, using \"\"",
1.90 schwarze 169: "missing eqn box, using \"\"",
1.7 kristaps 170:
171: /* related to bad macro arguments */
1.64 schwarze 172: "unterminated quoted argument",
1.7 kristaps 173: "duplicate argument",
1.76 schwarze 174: "skipping duplicate argument",
1.63 schwarze 175: "skipping duplicate display type",
176: "skipping duplicate list type",
1.76 schwarze 177: "skipping -width argument",
1.126 schwarze 178: "wrong number of cells",
1.7 kristaps 179: "unknown AT&T UNIX version",
1.88 schwarze 180: "comma in function argument",
1.89 schwarze 181: "parenthesis in function name",
1.173 schwarze 182: "unknown library name",
1.67 schwarze 183: "invalid content in Rs block",
1.63 schwarze 184: "invalid Boolean argument",
185: "unknown font, skipping request",
1.126 schwarze 186: "odd number of characters in request",
1.7 kristaps 187:
188: /* related to plain text */
1.64 schwarze 189: "blank line in fill mode, using .sp",
190: "tab in filled text",
191: "whitespace at end of input line",
1.158 schwarze 192: "new sentence, new line",
1.7 kristaps 193: "bad comment style",
1.64 schwarze 194: "invalid escape sequence",
195: "undefined string, using \"\"",
1.16 kristaps 196:
1.118 schwarze 197: /* related to tables */
198: "tbl line starts with span",
199: "tbl column starts with span",
200: "skipping vertical bar in tbl layout",
201:
1.7 kristaps 202: "generic error",
203:
204: /* related to tables */
1.116 schwarze 205: "non-alphabetic character in tbl options",
206: "skipping unknown tbl option",
207: "missing tbl option argument",
208: "wrong tbl option argument size",
1.118 schwarze 209: "empty tbl layout",
210: "invalid character in tbl layout",
211: "unmatched parenthesis in tbl layout",
1.119 schwarze 212: "tbl without any data cells",
213: "ignoring data in spanned tbl cell",
214: "ignoring extra tbl data cells",
215: "data block open at end of tbl",
1.7 kristaps 216:
1.68 schwarze 217: /* related to document structure and macros */
1.108 schwarze 218: NULL,
1.7 kristaps 219: "input stack limit exceeded, infinite loop?",
220: "skipping bad character",
1.68 schwarze 221: "skipping unknown macro",
1.112 schwarze 222: "skipping insecure request",
1.73 schwarze 223: "skipping item outside list",
1.68 schwarze 224: "skipping column outside column list",
225: "skipping end of block that is not open",
1.115 schwarze 226: "fewer RS blocks open, skipping",
1.68 schwarze 227: "inserting missing end of block",
228: "appending missing end of block",
229:
230: /* related to request and macro arguments */
1.7 kristaps 231: "escaped character not allowed in a name",
1.99 schwarze 232: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 233: "skipping display without arguments",
1.71 schwarze 234: "missing list type, using -item",
1.171 schwarze 235: "argument is not numeric, using 1",
1.70 schwarze 236: "missing manual name, using \"\"",
1.71 schwarze 237: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 238: "unknown standard specifier",
1.71 schwarze 239: "skipping request without numeric argument",
1.109 schwarze 240: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 241: ".so request failed",
1.61 schwarze 242: "skipping all arguments",
243: "skipping excess arguments",
1.92 kristaps 244: "divide by zero",
1.112 schwarze 245:
246: "unsupported feature",
247: "input too large",
1.114 schwarze 248: "unsupported control character",
1.112 schwarze 249: "unsupported roff request",
1.119 schwarze 250: "eqn delim option in tbl",
1.118 schwarze 251: "unsupported tbl layout modifier",
1.112 schwarze 252: "ignoring macro in table",
1.7 kristaps 253: };
254:
255: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
256: "SUCCESS",
1.165 schwarze 257: "STYLE",
1.7 kristaps 258: "WARNING",
259: "ERROR",
1.112 schwarze 260: "UNSUPP",
1.7 kristaps 261: "BADARG",
262: "SYSERR"
263: };
264:
1.47 schwarze 265:
1.1 kristaps 266: static void
267: resize_buf(struct buf *buf, size_t initial)
268: {
269:
270: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
271: buf->buf = mandoc_realloc(buf->buf, buf->sz);
272: }
273:
274: static void
1.84 schwarze 275: choose_parser(struct mparse *curp)
1.1 kristaps 276: {
1.83 schwarze 277: char *cp, *ep;
278: int format;
1.1 kristaps 279:
1.83 schwarze 280: /*
281: * If neither command line arguments -mdoc or -man select
282: * a parser nor the roff parser found a .Dd or .TH macro
283: * yet, look ahead in the main input buffer.
284: */
285:
286: if ((format = roff_getformat(curp->roff)) == 0) {
287: cp = curp->primary->buf;
288: ep = cp + curp->primary->sz;
289: while (cp < ep) {
1.85 schwarze 290: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 291: cp++;
292: if (cp[0] == 'D' && cp[1] == 'd') {
293: format = MPARSE_MDOC;
294: break;
295: }
296: if (cp[0] == 'T' && cp[1] == 'H') {
297: format = MPARSE_MAN;
298: break;
299: }
300: }
301: cp = memchr(cp, '\n', ep - cp);
302: if (cp == NULL)
303: break;
304: cp++;
305: }
1.1 kristaps 306: }
307:
1.83 schwarze 308: if (format == MPARSE_MDOC) {
1.137 schwarze 309: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 310: if (curp->man->mdocmac == NULL)
311: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 312: } else {
313: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 314: if (curp->man->manmac == NULL)
315: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 316: }
1.164 schwarze 317: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 318: }
319:
320: /*
1.95 schwarze 321: * Main parse routine for a buffer.
322: * It assumes encoding and line numbering are already set up.
323: * It can recurse directly (for invocations of user-defined
324: * macros, inline equations, and input line traps)
325: * and indirectly (for .so file inclusion).
1.1 kristaps 326: */
1.163 schwarze 327: static int
1.95 schwarze 328: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 329: {
330: const struct tbl_span *span;
331: struct buf ln;
1.113 schwarze 332: const char *save_file;
1.110 schwarze 333: char *cp;
1.95 schwarze 334: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 335: enum rofferr rr;
1.100 schwarze 336: int of;
1.1 kristaps 337: int lnn; /* line number in the real file */
1.110 schwarze 338: int fd;
1.1 kristaps 339: unsigned char c;
340:
1.95 schwarze 341: memset(&ln, 0, sizeof(ln));
1.1 kristaps 342:
1.47 schwarze 343: lnn = curp->line;
344: pos = 0;
1.1 kristaps 345:
1.95 schwarze 346: while (i < blk.sz) {
1.1 kristaps 347: if (0 == pos && '\0' == blk.buf[i])
348: break;
349:
350: if (start) {
351: curp->line = lnn;
352: curp->reparse_count = 0;
1.93 schwarze 353:
354: if (lnn < 3 &&
355: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 356: curp->filenc & MPARSE_LATIN1)
357: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 358: }
359:
1.95 schwarze 360: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 361:
362: /*
363: * When finding an unescaped newline character,
364: * leave the character loop to process the line.
365: * Skip a preceding carriage return, if any.
366: */
367:
1.95 schwarze 368: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 369: '\n' == blk.buf[i + 1])
370: ++i;
371: if ('\n' == blk.buf[i]) {
372: ++i;
373: ++lnn;
374: break;
375: }
376:
1.35 schwarze 377: /*
1.93 schwarze 378: * Make sure we have space for the worst
379: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 380: */
381:
1.95 schwarze 382: if (pos + 11 > ln.sz)
1.35 schwarze 383: resize_buf(&ln, 256);
384:
1.47 schwarze 385: /*
1.93 schwarze 386: * Encode 8-bit input.
1.1 kristaps 387: */
388:
1.93 schwarze 389: c = blk.buf[i];
390: if (c & 0x80) {
1.95 schwarze 391: if ( ! (curp->filenc && preconv_encode(
392: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 393: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
394: curp->line, pos, "0x%x", c);
1.93 schwarze 395: ln.buf[pos++] = '?';
396: i++;
397: }
398: continue;
399: }
400:
401: /*
402: * Exclude control characters.
403: */
1.1 kristaps 404:
1.93 schwarze 405: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 406: mandoc_vmsg(c == 0x00 || c == 0x04 ||
407: c > 0x0a ? MANDOCERR_CHAR_BAD :
408: MANDOCERR_CHAR_UNSUPP,
409: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 410: i++;
1.127 schwarze 411: if (c != '\r')
412: ln.buf[pos++] = '?';
1.1 kristaps 413: continue;
414: }
415:
416: ln.buf[pos++] = blk.buf[i++];
417: }
418:
1.170 schwarze 419: if (pos + 1 >= ln.sz)
1.1 kristaps 420: resize_buf(&ln, 256);
421:
1.170 schwarze 422: if (i == blk.sz || blk.buf[i] == '\0')
423: ln.buf[pos++] = '\n';
1.1 kristaps 424: ln.buf[pos] = '\0';
425:
426: /*
427: * A significant amount of complexity is contained by
428: * the roff preprocessor. It's line-oriented but can be
429: * expressed on one line, so we need at times to
430: * readjust our starting point and re-run it. The roff
431: * preprocessor can also readjust the buffers with new
432: * data, so we pass them in wholesale.
433: */
434:
435: of = 0;
436:
1.24 kristaps 437: /*
438: * Maintain a lookaside buffer of all parsed lines. We
439: * only do this if mparse_keep() has been invoked (the
440: * buffer may be accessed with mparse_getkeep()).
441: */
442:
443: if (curp->secondary) {
1.47 schwarze 444: curp->secondary->buf = mandoc_realloc(
445: curp->secondary->buf,
446: curp->secondary->sz + pos + 2);
447: memcpy(curp->secondary->buf +
448: curp->secondary->sz,
449: ln.buf, pos);
1.24 kristaps 450: curp->secondary->sz += pos;
451: curp->secondary->buf
452: [curp->secondary->sz] = '\n';
453: curp->secondary->sz++;
454: curp->secondary->buf
455: [curp->secondary->sz] = '\0';
456: }
1.1 kristaps 457: rerun:
1.96 schwarze 458: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 459:
460: switch (rr) {
1.47 schwarze 461: case ROFF_REPARSE:
1.163 schwarze 462: if (++curp->reparse_count > REPARSE_LIMIT)
1.3 kristaps 463: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 464: curp->line, pos, NULL);
1.163 schwarze 465: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
466: start == 1) {
467: pos = 0;
468: continue;
469: }
470: free(ln.buf);
471: return 0;
1.47 schwarze 472: case ROFF_APPEND:
1.95 schwarze 473: pos = strlen(ln.buf);
1.1 kristaps 474: continue;
1.47 schwarze 475: case ROFF_RERUN:
1.1 kristaps 476: goto rerun;
1.47 schwarze 477: case ROFF_IGN:
1.1 kristaps 478: pos = 0;
479: continue;
1.47 schwarze 480: case ROFF_SO:
1.95 schwarze 481: if ( ! (curp->options & MPARSE_SO) &&
482: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 483: curp->sodest = mandoc_strdup(ln.buf + of);
484: free(ln.buf);
1.163 schwarze 485: return 1;
1.45 schwarze 486: }
1.24 kristaps 487: /*
488: * We remove `so' clauses from our lookaside
489: * buffer because we're going to descend into
490: * the file recursively.
491: */
1.47 schwarze 492: if (curp->secondary)
1.25 kristaps 493: curp->secondary->sz -= pos + 1;
1.113 schwarze 494: save_file = curp->file;
1.148 schwarze 495: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 496: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 497: close(fd);
1.113 schwarze 498: curp->file = save_file;
499: } else {
500: curp->file = save_file;
1.52 schwarze 501: mandoc_vmsg(MANDOCERR_SO_FAIL,
502: curp, curp->line, pos,
503: ".so %s", ln.buf + of);
1.110 schwarze 504: ln.sz = mandoc_asprintf(&cp,
505: ".sp\nSee the file %s.\n.sp",
506: ln.buf + of);
507: free(ln.buf);
508: ln.buf = cp;
509: of = 0;
510: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 511: }
1.1 kristaps 512: pos = 0;
513: continue;
514: default:
515: break;
516: }
517:
1.150 schwarze 518: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 519: choose_parser(curp);
1.1 kristaps 520:
1.47 schwarze 521: /*
1.84 schwarze 522: * Lastly, push down into the parsers themselves.
1.1 kristaps 523: * If libroff returns ROFF_TBL, then add it to the
524: * currently open parse. Since we only get here if
525: * there does exist data (see tbl_data.c), we're
526: * guaranteed that something's been allocated.
527: * Do the same for ROFF_EQN.
528: */
529:
1.139 schwarze 530: if (rr == ROFF_TBL)
1.100 schwarze 531: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 532: roff_addtbl(curp->man, span);
533: else if (rr == ROFF_EQN)
534: roff_addeqn(curp->man, roff_eqn(curp->roff));
535: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 536: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 537: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
538: break;
1.1 kristaps 539:
540: /* Temporary buffers typically are not full. */
541:
542: if (0 == start && '\0' == blk.buf[i])
543: break;
544:
545: /* Start the next input line. */
546:
547: pos = 0;
548: }
549:
550: free(ln.buf);
1.163 schwarze 551: return 1;
1.1 kristaps 552: }
553:
554: static int
1.40 schwarze 555: read_whole_file(struct mparse *curp, const char *file, int fd,
556: struct buf *fb, int *with_mmap)
1.1 kristaps 557: {
1.161 schwarze 558: struct stat st;
1.140 schwarze 559: gzFile gz;
1.1 kristaps 560: size_t off;
561: ssize_t ssz;
1.143 schwarze 562:
563: if (fstat(fd, &st) == -1)
564: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 565:
566: /*
567: * If we're a regular file, try just reading in the whole entry
568: * via mmap(). This is faster than reading it into blocks, and
569: * since each file is only a few bytes to begin with, I'm not
570: * concerned that this is going to tank any machines.
571: */
572:
1.140 schwarze 573: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 574: if (st.st_size > 0x7fffffff) {
1.111 schwarze 575: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 576: return 0;
1.1 kristaps 577: }
578: *with_mmap = 1;
579: fb->sz = (size_t)st.st_size;
1.37 schwarze 580: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 581: if (fb->buf != MAP_FAILED)
1.142 schwarze 582: return 1;
1.1 kristaps 583: }
584:
1.140 schwarze 585: if (curp->gzip) {
1.143 schwarze 586: if ((gz = gzdopen(fd, "rb")) == NULL)
587: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 588: } else
589: gz = NULL;
590:
1.1 kristaps 591: /*
592: * If this isn't a regular file (like, say, stdin), then we must
593: * go the old way and just read things in bit by bit.
594: */
595:
596: *with_mmap = 0;
597: off = 0;
598: fb->sz = 0;
599: fb->buf = NULL;
600: for (;;) {
601: if (off == fb->sz) {
602: if (fb->sz == (1U << 31)) {
1.111 schwarze 603: mandoc_msg(MANDOCERR_TOOLARGE, curp,
604: 0, 0, NULL);
1.1 kristaps 605: break;
606: }
607: resize_buf(fb, 65536);
608: }
1.140 schwarze 609: ssz = curp->gzip ?
610: gzread(gz, fb->buf + (int)off, fb->sz - off) :
611: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 612: if (ssz == 0) {
613: fb->sz = off;
1.142 schwarze 614: return 1;
1.1 kristaps 615: }
1.143 schwarze 616: if (ssz == -1)
617: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 618: off += (size_t)ssz;
619: }
620:
621: free(fb->buf);
622: fb->buf = NULL;
1.142 schwarze 623: return 0;
1.1 kristaps 624: }
625:
626: static void
627: mparse_end(struct mparse *curp)
628: {
1.135 schwarze 629: if (curp->man->macroset == MACROSET_NONE)
630: curp->man->macroset = MACROSET_MAN;
631: if (curp->man->macroset == MACROSET_MDOC)
632: mdoc_endparse(curp->man);
633: else
1.111 schwarze 634: man_endparse(curp->man);
1.1 kristaps 635: roff_endparse(curp->roff);
636: }
637:
638: static void
1.36 schwarze 639: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 640: {
1.85 schwarze 641: struct buf *svprimary;
1.28 joerg 642: const char *svfile;
1.95 schwarze 643: size_t offset;
1.36 schwarze 644: static int recursion_depth;
645:
646: if (64 < recursion_depth) {
647: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
648: return;
649: }
1.28 joerg 650:
651: /* Line number is per-file. */
652: svfile = curp->file;
653: curp->file = file;
1.85 schwarze 654: svprimary = curp->primary;
1.83 schwarze 655: curp->primary = &blk;
1.28 joerg 656: curp->line = 1;
1.36 schwarze 657: recursion_depth++;
1.28 joerg 658:
1.93 schwarze 659: /* Skip an UTF-8 byte order mark. */
660: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
661: (unsigned char)blk.buf[0] == 0xef &&
662: (unsigned char)blk.buf[1] == 0xbb &&
663: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 664: offset = 3;
1.93 schwarze 665: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 666: } else
667: offset = 0;
1.93 schwarze 668:
1.95 schwarze 669: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 670:
1.111 schwarze 671: if (--recursion_depth == 0)
1.28 joerg 672: mparse_end(curp);
673:
1.85 schwarze 674: curp->primary = svprimary;
1.28 joerg 675: curp->file = svfile;
676: }
677:
678: enum mandoclevel
1.104 schwarze 679: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 680: const char *file)
681: {
682: struct buf blk;
683:
1.104 schwarze 684: blk.buf = buf;
1.28 joerg 685: blk.sz = len;
686:
1.36 schwarze 687: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 688: return curp->file_status;
1.28 joerg 689: }
690:
1.98 schwarze 691: /*
692: * Read the whole file into memory and call the parsers.
693: * Called recursively when an .so request is encountered.
694: */
1.36 schwarze 695: enum mandoclevel
696: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 697: {
1.28 joerg 698: struct buf blk;
699: int with_mmap;
1.93 schwarze 700: int save_filenc;
1.1 kristaps 701:
1.91 schwarze 702: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 703: save_filenc = curp->filenc;
704: curp->filenc = curp->options &
705: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 706: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 707: curp->filenc = save_filenc;
1.91 schwarze 708: if (with_mmap)
709: munmap(blk.buf, blk.sz);
710: else
711: free(blk.buf);
712: }
1.142 schwarze 713: return curp->file_status;
1.82 schwarze 714: }
715:
1.148 schwarze 716: int
717: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 718: {
719: char *cp;
1.148 schwarze 720: int fd;
1.82 schwarze 721:
722: curp->file = file;
1.140 schwarze 723: cp = strrchr(file, '.');
724: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 725:
1.140 schwarze 726: /* First try to use the filename as it is. */
1.98 schwarze 727:
1.148 schwarze 728: if ((fd = open(file, O_RDONLY)) != -1)
729: return fd;
1.98 schwarze 730:
1.140 schwarze 731: /*
732: * If that doesn't work and the filename doesn't
733: * already end in .gz, try appending .gz.
734: */
1.98 schwarze 735:
1.140 schwarze 736: if ( ! curp->gzip) {
1.98 schwarze 737: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 738: fd = open(cp, O_RDONLY);
1.108 schwarze 739: free(cp);
1.148 schwarze 740: if (fd != -1) {
1.140 schwarze 741: curp->gzip = 1;
1.148 schwarze 742: return fd;
1.82 schwarze 743: }
744: }
745:
1.140 schwarze 746: /* Neither worked, give up. */
1.97 schwarze 747:
1.140 schwarze 748: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 749: return -1;
1.1 kristaps 750: }
751:
752: struct mparse *
1.94 schwarze 753: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.144 schwarze 754: const char *defos)
1.1 kristaps 755: {
756: struct mparse *curp;
1.10 kristaps 757:
1.1 kristaps 758: curp = mandoc_calloc(1, sizeof(struct mparse));
759:
1.44 schwarze 760: curp->options = options;
1.3 kristaps 761: curp->wlevel = wlevel;
1.1 kristaps 762: curp->mmsg = mmsg;
1.29 schwarze 763: curp->defos = defos;
1.1 kristaps 764:
1.144 schwarze 765: curp->roff = roff_alloc(curp, options);
1.137 schwarze 766: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
767: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 768: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 769: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 770: if (curp->man->mdocmac == NULL)
771: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 772: } else if (curp->options & MPARSE_MAN) {
773: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 774: if (curp->man->manmac == NULL)
775: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 776: }
1.138 schwarze 777: curp->man->first->tok = TOKEN_NONE;
1.142 schwarze 778: return curp;
1.1 kristaps 779: }
780:
781: void
782: mparse_reset(struct mparse *curp)
783: {
784: roff_reset(curp->roff);
1.150 schwarze 785: roff_man_reset(curp->man);
1.160 schwarze 786:
787: free(curp->sodest);
788: curp->sodest = NULL;
789:
1.24 kristaps 790: if (curp->secondary)
791: curp->secondary->sz = 0;
1.1 kristaps 792:
793: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 794: curp->gzip = 0;
1.1 kristaps 795: }
796:
797: void
798: mparse_free(struct mparse *curp)
799: {
800:
1.164 schwarze 801: roffhash_free(curp->man->mdocmac);
802: roffhash_free(curp->man->manmac);
1.137 schwarze 803: roff_man_free(curp->man);
1.160 schwarze 804: roff_free(curp->roff);
1.24 kristaps 805: if (curp->secondary)
806: free(curp->secondary->buf);
1.1 kristaps 807:
1.24 kristaps 808: free(curp->secondary);
1.45 schwarze 809: free(curp->sodest);
1.1 kristaps 810: free(curp);
811: }
812:
813: void
1.135 schwarze 814: mparse_result(struct mparse *curp, struct roff_man **man,
815: char **sodest)
1.1 kristaps 816: {
817:
1.45 schwarze 818: if (sodest && NULL != (*sodest = curp->sodest)) {
819: *man = NULL;
820: return;
821: }
1.9 kristaps 822: if (man)
823: *man = curp->man;
1.157 schwarze 824: }
825:
826: void
827: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
828: {
829: if (curp->file_status > *rc)
830: *rc = curp->file_status;
1.3 kristaps 831: }
832:
833: void
834: mandoc_vmsg(enum mandocerr t, struct mparse *m,
835: int ln, int pos, const char *fmt, ...)
836: {
837: char buf[256];
838: va_list ap;
839:
840: va_start(ap, fmt);
1.48 schwarze 841: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 842: va_end(ap);
843:
844: mandoc_msg(t, m, ln, pos, buf);
845: }
846:
847: void
1.47 schwarze 848: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 849: int ln, int col, const char *msg)
850: {
851: enum mandoclevel level;
852:
1.112 schwarze 853: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 854: while (er < mandoclimits[level])
855: level--;
856:
1.108 schwarze 857: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 858: return;
859:
1.8 kristaps 860: if (m->mmsg)
861: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 862:
863: if (m->file_status < level)
864: m->file_status = level;
1.7 kristaps 865: }
866:
867: const char *
868: mparse_strerror(enum mandocerr er)
869: {
870:
1.142 schwarze 871: return mandocerrs[er];
1.7 kristaps 872: }
873:
874: const char *
875: mparse_strlevel(enum mandoclevel lvl)
876: {
1.142 schwarze 877: return mandoclevels[lvl];
1.24 kristaps 878: }
879:
880: void
881: mparse_keep(struct mparse *p)
882: {
883:
884: assert(NULL == p->secondary);
885: p->secondary = mandoc_calloc(1, sizeof(struct buf));
886: }
887:
888: const char *
889: mparse_getkeep(const struct mparse *p)
890: {
891:
892: assert(p->secondary);
1.142 schwarze 893: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 894: }
CVSweb