Annotation of mandoc/read.c, Revision 1.191
1.191 ! schwarze 1: /* $Id: read.c,v 1.190 2017/07/08 14:51:04 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"
1.1 kristaps 45:
46: #define REPARSE_LIMIT 1000
47:
48: struct mparse {
1.160 schwarze 49: struct roff *roff; /* roff parser (!NULL) */
1.134 schwarze 50: struct roff_man *man; /* man parser */
1.45 schwarze 51: char *sodest; /* filename pointed to by .so */
1.83 schwarze 52: const char *file; /* filename of current input file */
53: struct buf *primary; /* buffer currently being parsed */
54: struct buf *secondary; /* preprocessed copy of input */
1.179 schwarze 55: const char *os_s; /* default operating system */
1.83 schwarze 56: mandocmsg mmsg; /* warning/error message handler */
57: enum mandoclevel file_status; /* status of current parse */
1.179 schwarze 58: enum mandocerr mmin; /* ignore messages below this */
1.83 schwarze 59: int options; /* parser options */
1.140 schwarze 60: int gzip; /* current input file is gzipped */
1.93 schwarze 61: int filenc; /* encoding of the current file */
1.1 kristaps 62: int reparse_count; /* finite interp. stack */
1.83 schwarze 63: int line; /* line number in the file */
1.1 kristaps 64: };
65:
1.84 schwarze 66: static void choose_parser(struct mparse *);
1.1 kristaps 67: static void resize_buf(struct buf *, size_t);
1.163 schwarze 68: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 69: static int read_whole_file(struct mparse *, const char *, int,
70: struct buf *, int *);
1.1 kristaps 71: static void mparse_end(struct mparse *);
1.37 schwarze 72: static void mparse_parse_buffer(struct mparse *, struct buf,
73: const char *);
1.1 kristaps 74:
1.3 kristaps 75: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
76: MANDOCERR_OK,
1.188 schwarze 77: MANDOCERR_OK,
1.3 kristaps 78: MANDOCERR_WARNING,
79: MANDOCERR_ERROR,
1.112 schwarze 80: MANDOCERR_UNSUPP,
1.3 kristaps 81: MANDOCERR_MAX,
82: MANDOCERR_MAX
83: };
84:
1.7 kristaps 85: static const char * const mandocerrs[MANDOCERR_MAX] = {
86: "ok",
87:
1.179 schwarze 88: "base system convention",
1.166 schwarze 89:
1.176 schwarze 90: "Mdocdate found",
91: "Mdocdate missing",
1.181 schwarze 92: "unknown architecture",
1.180 schwarze 93: "operating system explicitly specified",
1.179 schwarze 94: "RCS id missing",
1.184 schwarze 95: "referenced manual not found",
1.179 schwarze 96:
97: "generic style suggestion",
98:
1.176 schwarze 99: "legacy man(7) date format",
1.189 schwarze 100: "lower case character in document title",
1.178 schwarze 101: "duplicate RCS id",
1.182 schwarze 102: "typo in section name",
1.189 schwarze 103: "unterminated quoted argument",
1.166 schwarze 104: "useless macro",
1.167 schwarze 105: "consider using OS macro",
1.172 schwarze 106: "errnos out of order",
107: "duplicate errno",
1.187 schwarze 108: "trailing delimiter",
1.174 schwarze 109: "no blank before trailing delimiter",
1.189 schwarze 110: "fill mode already enabled, skipping",
111: "fill mode already disabled, skipping",
1.175 schwarze 112: "function name without markup",
1.189 schwarze 113: "whitespace at end of input line",
114: "bad comment style",
1.166 schwarze 115:
1.7 kristaps 116: "generic warning",
117:
118: /* related to the prologue */
1.79 schwarze 119: "missing manual title, using UNTITLED",
120: "missing manual title, using \"\"",
121: "missing manual section, using \"\"",
1.7 kristaps 122: "unknown manual section",
1.54 schwarze 123: "missing date, using today's date",
1.7 kristaps 124: "cannot parse date, using it verbatim",
1.186 schwarze 125: "date in the future, using it anyway",
1.79 schwarze 126: "missing Os macro, using \"\"",
127: "late prologue macro",
1.7 kristaps 128: "prologue macros out of order",
129:
130: /* related to document structure */
131: ".so is fragile, better use ln(1)",
1.50 schwarze 132: "no document body",
1.54 schwarze 133: "content before first section header",
134: "first section is not \"NAME\"",
1.156 schwarze 135: "NAME section without Nm before Nd",
1.128 schwarze 136: "NAME section without description",
137: "description not at the end of NAME",
138: "bad NAME section content",
1.156 schwarze 139: "missing comma before name",
1.121 schwarze 140: "missing description line, using \"\"",
1.162 schwarze 141: "description line outside NAME section",
1.7 kristaps 142: "sections out of conventional order",
1.54 schwarze 143: "duplicate section title",
144: "unexpected section",
1.185 schwarze 145: "cross reference to self",
1.87 schwarze 146: "unusual Xr order",
147: "unusual Xr punctuation",
1.86 schwarze 148: "AUTHORS section without An macro",
1.7 kristaps 149:
150: /* related to macros and nesting */
1.55 schwarze 151: "obsolete macro",
1.102 schwarze 152: "macro neither callable nor escaped",
1.7 kristaps 153: "skipping paragraph macro",
1.31 schwarze 154: "moving paragraph macro out of list",
1.7 kristaps 155: "skipping no-space macro",
156: "blocks badly nested",
157: "nested displays are not portable",
1.57 schwarze 158: "moving content out of list",
1.183 schwarze 159: "first macro on line",
1.7 kristaps 160: "line scope broken",
1.169 schwarze 161: "skipping blank line in line scope",
1.7 kristaps 162:
163: /* related to missing macro arguments */
1.58 schwarze 164: "skipping empty request",
165: "conditional request controls empty scope",
1.7 kristaps 166: "skipping empty macro",
1.124 schwarze 167: "empty block",
1.62 schwarze 168: "empty argument, using 0n",
1.60 schwarze 169: "missing display type, using -ragged",
170: "list type is not the first argument",
1.152 schwarze 171: "missing -width in -tag list, using 6n",
1.78 schwarze 172: "missing utility name, using \"\"",
1.123 schwarze 173: "missing function name, using \"\"",
1.60 schwarze 174: "empty head in list item",
175: "empty list item",
1.183 schwarze 176: "missing argument, using next line",
1.61 schwarze 177: "missing font type, using \\fR",
178: "unknown font type, using \\fR",
1.103 schwarze 179: "nothing follows prefix",
1.122 schwarze 180: "empty reference block",
1.155 schwarze 181: "missing section argument",
1.60 schwarze 182: "missing -std argument, adding it",
1.125 schwarze 183: "missing option string, using \"\"",
184: "missing resource identifier, using \"\"",
1.90 schwarze 185: "missing eqn box, using \"\"",
1.7 kristaps 186:
187: /* related to bad macro arguments */
188: "duplicate argument",
1.76 schwarze 189: "skipping duplicate argument",
1.63 schwarze 190: "skipping duplicate display type",
191: "skipping duplicate list type",
1.76 schwarze 192: "skipping -width argument",
1.126 schwarze 193: "wrong number of cells",
1.7 kristaps 194: "unknown AT&T UNIX version",
1.88 schwarze 195: "comma in function argument",
1.89 schwarze 196: "parenthesis in function name",
1.173 schwarze 197: "unknown library name",
1.67 schwarze 198: "invalid content in Rs block",
1.63 schwarze 199: "invalid Boolean argument",
200: "unknown font, skipping request",
1.126 schwarze 201: "odd number of characters in request",
1.7 kristaps 202:
203: /* related to plain text */
1.64 schwarze 204: "blank line in fill mode, using .sp",
205: "tab in filled text",
1.158 schwarze 206: "new sentence, new line",
1.64 schwarze 207: "invalid escape sequence",
208: "undefined string, using \"\"",
1.16 kristaps 209:
1.118 schwarze 210: /* related to tables */
211: "tbl line starts with span",
212: "tbl column starts with span",
213: "skipping vertical bar in tbl layout",
214:
1.7 kristaps 215: "generic error",
216:
217: /* related to tables */
1.116 schwarze 218: "non-alphabetic character in tbl options",
219: "skipping unknown tbl option",
220: "missing tbl option argument",
221: "wrong tbl option argument size",
1.118 schwarze 222: "empty tbl layout",
223: "invalid character in tbl layout",
224: "unmatched parenthesis in tbl layout",
1.119 schwarze 225: "tbl without any data cells",
226: "ignoring data in spanned tbl cell",
227: "ignoring extra tbl data cells",
228: "data block open at end of tbl",
1.7 kristaps 229:
1.68 schwarze 230: /* related to document structure and macros */
1.108 schwarze 231: NULL,
1.189 schwarze 232: "duplicate prologue macro",
233: "skipping late title macro",
1.7 kristaps 234: "input stack limit exceeded, infinite loop?",
235: "skipping bad character",
1.68 schwarze 236: "skipping unknown macro",
1.112 schwarze 237: "skipping insecure request",
1.73 schwarze 238: "skipping item outside list",
1.68 schwarze 239: "skipping column outside column list",
240: "skipping end of block that is not open",
1.115 schwarze 241: "fewer RS blocks open, skipping",
1.68 schwarze 242: "inserting missing end of block",
243: "appending missing end of block",
244:
245: /* related to request and macro arguments */
1.7 kristaps 246: "escaped character not allowed in a name",
1.99 schwarze 247: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 248: "skipping display without arguments",
1.71 schwarze 249: "missing list type, using -item",
1.171 schwarze 250: "argument is not numeric, using 1",
1.70 schwarze 251: "missing manual name, using \"\"",
1.71 schwarze 252: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 253: "unknown standard specifier",
1.71 schwarze 254: "skipping request without numeric argument",
1.109 schwarze 255: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 256: ".so request failed",
1.61 schwarze 257: "skipping all arguments",
258: "skipping excess arguments",
1.92 kristaps 259: "divide by zero",
1.112 schwarze 260:
261: "unsupported feature",
262: "input too large",
1.114 schwarze 263: "unsupported control character",
1.112 schwarze 264: "unsupported roff request",
1.119 schwarze 265: "eqn delim option in tbl",
1.118 schwarze 266: "unsupported tbl layout modifier",
1.112 schwarze 267: "ignoring macro in table",
1.7 kristaps 268: };
269:
270: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
271: "SUCCESS",
1.165 schwarze 272: "STYLE",
1.7 kristaps 273: "WARNING",
274: "ERROR",
1.112 schwarze 275: "UNSUPP",
1.7 kristaps 276: "BADARG",
277: "SYSERR"
278: };
279:
1.47 schwarze 280:
1.1 kristaps 281: static void
282: resize_buf(struct buf *buf, size_t initial)
283: {
284:
285: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
286: buf->buf = mandoc_realloc(buf->buf, buf->sz);
287: }
288:
289: static void
1.84 schwarze 290: choose_parser(struct mparse *curp)
1.1 kristaps 291: {
1.83 schwarze 292: char *cp, *ep;
293: int format;
1.1 kristaps 294:
1.83 schwarze 295: /*
296: * If neither command line arguments -mdoc or -man select
297: * a parser nor the roff parser found a .Dd or .TH macro
298: * yet, look ahead in the main input buffer.
299: */
300:
301: if ((format = roff_getformat(curp->roff)) == 0) {
302: cp = curp->primary->buf;
303: ep = cp + curp->primary->sz;
304: while (cp < ep) {
1.85 schwarze 305: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 306: cp++;
307: if (cp[0] == 'D' && cp[1] == 'd') {
308: format = MPARSE_MDOC;
309: break;
310: }
311: if (cp[0] == 'T' && cp[1] == 'H') {
312: format = MPARSE_MAN;
313: break;
314: }
315: }
316: cp = memchr(cp, '\n', ep - cp);
317: if (cp == NULL)
318: break;
319: cp++;
320: }
1.1 kristaps 321: }
322:
1.83 schwarze 323: if (format == MPARSE_MDOC) {
1.137 schwarze 324: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 325: if (curp->man->mdocmac == NULL)
326: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 327: } else {
328: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 329: if (curp->man->manmac == NULL)
330: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 331: }
1.164 schwarze 332: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 333: }
334:
335: /*
1.95 schwarze 336: * Main parse routine for a buffer.
337: * It assumes encoding and line numbering are already set up.
338: * It can recurse directly (for invocations of user-defined
339: * macros, inline equations, and input line traps)
340: * and indirectly (for .so file inclusion).
1.1 kristaps 341: */
1.163 schwarze 342: static int
1.95 schwarze 343: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 344: {
345: struct buf ln;
1.113 schwarze 346: const char *save_file;
1.110 schwarze 347: char *cp;
1.95 schwarze 348: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 349: enum rofferr rr;
1.100 schwarze 350: int of;
1.1 kristaps 351: int lnn; /* line number in the real file */
1.110 schwarze 352: int fd;
1.1 kristaps 353: unsigned char c;
354:
1.95 schwarze 355: memset(&ln, 0, sizeof(ln));
1.1 kristaps 356:
1.47 schwarze 357: lnn = curp->line;
358: pos = 0;
1.1 kristaps 359:
1.95 schwarze 360: while (i < blk.sz) {
1.1 kristaps 361: if (0 == pos && '\0' == blk.buf[i])
362: break;
363:
364: if (start) {
365: curp->line = lnn;
366: curp->reparse_count = 0;
1.93 schwarze 367:
368: if (lnn < 3 &&
369: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 370: curp->filenc & MPARSE_LATIN1)
371: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 372: }
373:
1.95 schwarze 374: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 375:
376: /*
377: * When finding an unescaped newline character,
378: * leave the character loop to process the line.
379: * Skip a preceding carriage return, if any.
380: */
381:
1.95 schwarze 382: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 383: '\n' == blk.buf[i + 1])
384: ++i;
385: if ('\n' == blk.buf[i]) {
386: ++i;
387: ++lnn;
388: break;
389: }
390:
1.35 schwarze 391: /*
1.93 schwarze 392: * Make sure we have space for the worst
393: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 394: */
395:
1.95 schwarze 396: if (pos + 11 > ln.sz)
1.35 schwarze 397: resize_buf(&ln, 256);
398:
1.47 schwarze 399: /*
1.93 schwarze 400: * Encode 8-bit input.
1.1 kristaps 401: */
402:
1.93 schwarze 403: c = blk.buf[i];
404: if (c & 0x80) {
1.95 schwarze 405: if ( ! (curp->filenc && preconv_encode(
406: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 407: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
408: curp->line, pos, "0x%x", c);
1.93 schwarze 409: ln.buf[pos++] = '?';
410: i++;
411: }
412: continue;
413: }
414:
415: /*
416: * Exclude control characters.
417: */
1.1 kristaps 418:
1.93 schwarze 419: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 420: mandoc_vmsg(c == 0x00 || c == 0x04 ||
421: c > 0x0a ? MANDOCERR_CHAR_BAD :
422: MANDOCERR_CHAR_UNSUPP,
423: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 424: i++;
1.127 schwarze 425: if (c != '\r')
426: ln.buf[pos++] = '?';
1.1 kristaps 427: continue;
428: }
429:
430: ln.buf[pos++] = blk.buf[i++];
431: }
432:
1.170 schwarze 433: if (pos + 1 >= ln.sz)
1.1 kristaps 434: resize_buf(&ln, 256);
435:
1.170 schwarze 436: if (i == blk.sz || blk.buf[i] == '\0')
437: ln.buf[pos++] = '\n';
1.1 kristaps 438: ln.buf[pos] = '\0';
439:
440: /*
441: * A significant amount of complexity is contained by
442: * the roff preprocessor. It's line-oriented but can be
443: * expressed on one line, so we need at times to
444: * readjust our starting point and re-run it. The roff
445: * preprocessor can also readjust the buffers with new
446: * data, so we pass them in wholesale.
447: */
448:
449: of = 0;
450:
1.24 kristaps 451: /*
452: * Maintain a lookaside buffer of all parsed lines. We
453: * only do this if mparse_keep() has been invoked (the
454: * buffer may be accessed with mparse_getkeep()).
455: */
456:
457: if (curp->secondary) {
1.47 schwarze 458: curp->secondary->buf = mandoc_realloc(
459: curp->secondary->buf,
460: curp->secondary->sz + pos + 2);
461: memcpy(curp->secondary->buf +
462: curp->secondary->sz,
463: ln.buf, pos);
1.24 kristaps 464: curp->secondary->sz += pos;
465: curp->secondary->buf
466: [curp->secondary->sz] = '\n';
467: curp->secondary->sz++;
468: curp->secondary->buf
469: [curp->secondary->sz] = '\0';
470: }
1.1 kristaps 471: rerun:
1.96 schwarze 472: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 473:
474: switch (rr) {
1.47 schwarze 475: case ROFF_REPARSE:
1.163 schwarze 476: if (++curp->reparse_count > REPARSE_LIMIT)
1.3 kristaps 477: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 478: curp->line, pos, NULL);
1.163 schwarze 479: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
480: start == 1) {
481: pos = 0;
482: continue;
483: }
484: free(ln.buf);
485: return 0;
1.47 schwarze 486: case ROFF_APPEND:
1.95 schwarze 487: pos = strlen(ln.buf);
1.1 kristaps 488: continue;
1.47 schwarze 489: case ROFF_RERUN:
1.1 kristaps 490: goto rerun;
1.47 schwarze 491: case ROFF_IGN:
1.1 kristaps 492: pos = 0;
493: continue;
1.47 schwarze 494: case ROFF_SO:
1.95 schwarze 495: if ( ! (curp->options & MPARSE_SO) &&
496: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 497: curp->sodest = mandoc_strdup(ln.buf + of);
498: free(ln.buf);
1.163 schwarze 499: return 1;
1.45 schwarze 500: }
1.24 kristaps 501: /*
502: * We remove `so' clauses from our lookaside
503: * buffer because we're going to descend into
504: * the file recursively.
505: */
1.47 schwarze 506: if (curp->secondary)
1.25 kristaps 507: curp->secondary->sz -= pos + 1;
1.113 schwarze 508: save_file = curp->file;
1.148 schwarze 509: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 510: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 511: close(fd);
1.113 schwarze 512: curp->file = save_file;
513: } else {
514: curp->file = save_file;
1.52 schwarze 515: mandoc_vmsg(MANDOCERR_SO_FAIL,
516: curp, curp->line, pos,
517: ".so %s", ln.buf + of);
1.110 schwarze 518: ln.sz = mandoc_asprintf(&cp,
519: ".sp\nSee the file %s.\n.sp",
520: ln.buf + of);
521: free(ln.buf);
522: ln.buf = cp;
523: of = 0;
524: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 525: }
1.1 kristaps 526: pos = 0;
527: continue;
528: default:
529: break;
530: }
531:
1.150 schwarze 532: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 533: choose_parser(curp);
1.1 kristaps 534:
1.191 ! schwarze 535: 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 *
1.179 schwarze 753: mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
754: enum mandoc_os os_e, const char *os_s)
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;
1.179 schwarze 761: curp->mmin = mmin;
1.1 kristaps 762: curp->mmsg = mmsg;
1.179 schwarze 763: curp->os_s = os_s;
1.1 kristaps 764:
1.144 schwarze 765: curp->roff = roff_alloc(curp, options);
1.179 schwarze 766: curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
1.137 schwarze 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;
1.179 schwarze 778: curp->man->meta.os_e = os_e;
1.142 schwarze 779: return curp;
1.1 kristaps 780: }
781:
782: void
783: mparse_reset(struct mparse *curp)
784: {
785: roff_reset(curp->roff);
1.150 schwarze 786: roff_man_reset(curp->man);
1.160 schwarze 787:
788: free(curp->sodest);
789: curp->sodest = NULL;
790:
1.24 kristaps 791: if (curp->secondary)
792: curp->secondary->sz = 0;
1.1 kristaps 793:
794: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 795: curp->gzip = 0;
1.1 kristaps 796: }
797:
798: void
799: mparse_free(struct mparse *curp)
800: {
801:
1.164 schwarze 802: roffhash_free(curp->man->mdocmac);
803: roffhash_free(curp->man->manmac);
1.137 schwarze 804: roff_man_free(curp->man);
1.160 schwarze 805: roff_free(curp->roff);
1.24 kristaps 806: if (curp->secondary)
807: free(curp->secondary->buf);
1.1 kristaps 808:
1.24 kristaps 809: free(curp->secondary);
1.45 schwarze 810: free(curp->sodest);
1.1 kristaps 811: free(curp);
812: }
813:
814: void
1.135 schwarze 815: mparse_result(struct mparse *curp, struct roff_man **man,
816: char **sodest)
1.1 kristaps 817: {
818:
1.45 schwarze 819: if (sodest && NULL != (*sodest = curp->sodest)) {
820: *man = NULL;
821: return;
822: }
1.9 kristaps 823: if (man)
824: *man = curp->man;
1.157 schwarze 825: }
826:
827: void
828: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
829: {
830: if (curp->file_status > *rc)
831: *rc = curp->file_status;
1.3 kristaps 832: }
833:
834: void
835: mandoc_vmsg(enum mandocerr t, struct mparse *m,
836: int ln, int pos, const char *fmt, ...)
837: {
838: char buf[256];
839: va_list ap;
840:
841: va_start(ap, fmt);
1.48 schwarze 842: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 843: va_end(ap);
844:
845: mandoc_msg(t, m, ln, pos, buf);
846: }
847:
848: void
1.47 schwarze 849: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 850: int ln, int col, const char *msg)
851: {
852: enum mandoclevel level;
853:
1.179 schwarze 854: if (er < m->mmin && er != MANDOCERR_FILE)
855: return;
856:
1.112 schwarze 857: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 858: while (er < mandoclimits[level])
859: level--;
860:
1.8 kristaps 861: if (m->mmsg)
862: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 863:
864: if (m->file_status < level)
865: m->file_status = level;
1.7 kristaps 866: }
867:
868: const char *
869: mparse_strerror(enum mandocerr er)
870: {
871:
1.142 schwarze 872: return mandocerrs[er];
1.7 kristaps 873: }
874:
875: const char *
876: mparse_strlevel(enum mandoclevel lvl)
877: {
1.142 schwarze 878: return mandoclevels[lvl];
1.24 kristaps 879: }
880:
881: void
882: mparse_keep(struct mparse *p)
883: {
884:
885: assert(NULL == p->secondary);
886: p->secondary = mandoc_calloc(1, sizeof(struct buf));
887: }
888:
889: const char *
890: mparse_getkeep(const struct mparse *p)
891: {
892:
893: assert(p->secondary);
1.142 schwarze 894: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 895: }
CVSweb