Annotation of mandoc/read.c, Revision 1.170
1.170 ! schwarze 1: /* $Id: read.c,v 1.169 2017/06/03 15:55:24 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",
1.168 schwarze 93: "description line ends with a full stop",
1.166 schwarze 94:
1.7 kristaps 95: "generic warning",
96:
97: /* related to the prologue */
1.79 schwarze 98: "missing manual title, using UNTITLED",
99: "missing manual title, using \"\"",
1.54 schwarze 100: "lower case character in document title",
1.79 schwarze 101: "missing manual section, using \"\"",
1.7 kristaps 102: "unknown manual section",
1.54 schwarze 103: "missing date, using today's date",
1.7 kristaps 104: "cannot parse date, using it verbatim",
1.79 schwarze 105: "missing Os macro, using \"\"",
106: "duplicate prologue macro",
107: "late prologue macro",
108: "skipping late title macro",
1.7 kristaps 109: "prologue macros out of order",
110:
111: /* related to document structure */
112: ".so is fragile, better use ln(1)",
1.50 schwarze 113: "no document body",
1.54 schwarze 114: "content before first section header",
115: "first section is not \"NAME\"",
1.156 schwarze 116: "NAME section without Nm before Nd",
1.128 schwarze 117: "NAME section without description",
118: "description not at the end of NAME",
119: "bad NAME section content",
1.156 schwarze 120: "missing comma before name",
1.121 schwarze 121: "missing description line, using \"\"",
1.162 schwarze 122: "description line outside NAME section",
1.7 kristaps 123: "sections out of conventional order",
1.54 schwarze 124: "duplicate section title",
125: "unexpected section",
1.87 schwarze 126: "unusual Xr order",
127: "unusual Xr punctuation",
1.86 schwarze 128: "AUTHORS section without An macro",
1.7 kristaps 129:
130: /* related to macros and nesting */
1.55 schwarze 131: "obsolete macro",
1.102 schwarze 132: "macro neither callable nor escaped",
1.7 kristaps 133: "skipping paragraph macro",
1.31 schwarze 134: "moving paragraph macro out of list",
1.7 kristaps 135: "skipping no-space macro",
136: "blocks badly nested",
137: "nested displays are not portable",
1.57 schwarze 138: "moving content out of list",
1.78 schwarze 139: "fill mode already enabled, skipping",
140: "fill mode already disabled, skipping",
1.7 kristaps 141: "line scope broken",
1.169 schwarze 142: "skipping blank line in line scope",
1.7 kristaps 143:
144: /* related to missing macro arguments */
1.58 schwarze 145: "skipping empty request",
146: "conditional request controls empty scope",
1.7 kristaps 147: "skipping empty macro",
1.124 schwarze 148: "empty block",
1.62 schwarze 149: "empty argument, using 0n",
1.60 schwarze 150: "missing display type, using -ragged",
151: "list type is not the first argument",
1.152 schwarze 152: "missing -width in -tag list, using 6n",
1.78 schwarze 153: "missing utility name, using \"\"",
1.123 schwarze 154: "missing function name, using \"\"",
1.60 schwarze 155: "empty head in list item",
156: "empty list item",
1.61 schwarze 157: "missing font type, using \\fR",
158: "unknown font type, using \\fR",
1.103 schwarze 159: "nothing follows prefix",
1.122 schwarze 160: "empty reference block",
1.155 schwarze 161: "missing section argument",
1.60 schwarze 162: "missing -std argument, adding it",
1.125 schwarze 163: "missing option string, using \"\"",
164: "missing resource identifier, using \"\"",
1.90 schwarze 165: "missing eqn box, using \"\"",
1.7 kristaps 166:
167: /* related to bad macro arguments */
1.64 schwarze 168: "unterminated quoted argument",
1.7 kristaps 169: "duplicate argument",
1.76 schwarze 170: "skipping duplicate argument",
1.63 schwarze 171: "skipping duplicate display type",
172: "skipping duplicate list type",
1.76 schwarze 173: "skipping -width argument",
1.126 schwarze 174: "wrong number of cells",
1.7 kristaps 175: "unknown AT&T UNIX version",
1.88 schwarze 176: "comma in function argument",
1.89 schwarze 177: "parenthesis in function name",
1.67 schwarze 178: "invalid content in Rs block",
1.63 schwarze 179: "invalid Boolean argument",
180: "unknown font, skipping request",
1.126 schwarze 181: "odd number of characters in request",
1.7 kristaps 182:
183: /* related to plain text */
1.64 schwarze 184: "blank line in fill mode, using .sp",
185: "tab in filled text",
186: "whitespace at end of input line",
1.158 schwarze 187: "new sentence, new line",
1.7 kristaps 188: "bad comment style",
1.64 schwarze 189: "invalid escape sequence",
190: "undefined string, using \"\"",
1.16 kristaps 191:
1.118 schwarze 192: /* related to tables */
193: "tbl line starts with span",
194: "tbl column starts with span",
195: "skipping vertical bar in tbl layout",
196:
1.7 kristaps 197: "generic error",
198:
199: /* related to tables */
1.116 schwarze 200: "non-alphabetic character in tbl options",
201: "skipping unknown tbl option",
202: "missing tbl option argument",
203: "wrong tbl option argument size",
1.118 schwarze 204: "empty tbl layout",
205: "invalid character in tbl layout",
206: "unmatched parenthesis in tbl layout",
1.119 schwarze 207: "tbl without any data cells",
208: "ignoring data in spanned tbl cell",
209: "ignoring extra tbl data cells",
210: "data block open at end of tbl",
1.7 kristaps 211:
1.68 schwarze 212: /* related to document structure and macros */
1.108 schwarze 213: NULL,
1.7 kristaps 214: "input stack limit exceeded, infinite loop?",
215: "skipping bad character",
1.68 schwarze 216: "skipping unknown macro",
1.112 schwarze 217: "skipping insecure request",
1.73 schwarze 218: "skipping item outside list",
1.68 schwarze 219: "skipping column outside column list",
220: "skipping end of block that is not open",
1.115 schwarze 221: "fewer RS blocks open, skipping",
1.68 schwarze 222: "inserting missing end of block",
223: "appending missing end of block",
224:
225: /* related to request and macro arguments */
1.7 kristaps 226: "escaped character not allowed in a name",
1.99 schwarze 227: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 228: "skipping display without arguments",
1.71 schwarze 229: "missing list type, using -item",
1.70 schwarze 230: "missing manual name, using \"\"",
1.71 schwarze 231: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 232: "unknown standard specifier",
1.71 schwarze 233: "skipping request without numeric argument",
1.109 schwarze 234: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 235: ".so request failed",
1.61 schwarze 236: "skipping all arguments",
237: "skipping excess arguments",
1.92 kristaps 238: "divide by zero",
1.112 schwarze 239:
240: "unsupported feature",
241: "input too large",
1.114 schwarze 242: "unsupported control character",
1.112 schwarze 243: "unsupported roff request",
1.119 schwarze 244: "eqn delim option in tbl",
1.118 schwarze 245: "unsupported tbl layout modifier",
1.112 schwarze 246: "ignoring macro in table",
1.7 kristaps 247: };
248:
249: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
250: "SUCCESS",
1.165 schwarze 251: "STYLE",
1.7 kristaps 252: "WARNING",
253: "ERROR",
1.112 schwarze 254: "UNSUPP",
1.7 kristaps 255: "BADARG",
256: "SYSERR"
257: };
258:
1.47 schwarze 259:
1.1 kristaps 260: static void
261: resize_buf(struct buf *buf, size_t initial)
262: {
263:
264: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
265: buf->buf = mandoc_realloc(buf->buf, buf->sz);
266: }
267:
268: static void
1.84 schwarze 269: choose_parser(struct mparse *curp)
1.1 kristaps 270: {
1.83 schwarze 271: char *cp, *ep;
272: int format;
1.1 kristaps 273:
1.83 schwarze 274: /*
275: * If neither command line arguments -mdoc or -man select
276: * a parser nor the roff parser found a .Dd or .TH macro
277: * yet, look ahead in the main input buffer.
278: */
279:
280: if ((format = roff_getformat(curp->roff)) == 0) {
281: cp = curp->primary->buf;
282: ep = cp + curp->primary->sz;
283: while (cp < ep) {
1.85 schwarze 284: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 285: cp++;
286: if (cp[0] == 'D' && cp[1] == 'd') {
287: format = MPARSE_MDOC;
288: break;
289: }
290: if (cp[0] == 'T' && cp[1] == 'H') {
291: format = MPARSE_MAN;
292: break;
293: }
294: }
295: cp = memchr(cp, '\n', ep - cp);
296: if (cp == NULL)
297: break;
298: cp++;
299: }
1.1 kristaps 300: }
301:
1.83 schwarze 302: if (format == MPARSE_MDOC) {
1.137 schwarze 303: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 304: if (curp->man->mdocmac == NULL)
305: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 306: } else {
307: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 308: if (curp->man->manmac == NULL)
309: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 310: }
1.164 schwarze 311: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 312: }
313:
314: /*
1.95 schwarze 315: * Main parse routine for a buffer.
316: * It assumes encoding and line numbering are already set up.
317: * It can recurse directly (for invocations of user-defined
318: * macros, inline equations, and input line traps)
319: * and indirectly (for .so file inclusion).
1.1 kristaps 320: */
1.163 schwarze 321: static int
1.95 schwarze 322: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 323: {
324: const struct tbl_span *span;
325: struct buf ln;
1.113 schwarze 326: const char *save_file;
1.110 schwarze 327: char *cp;
1.95 schwarze 328: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 329: enum rofferr rr;
1.100 schwarze 330: int of;
1.1 kristaps 331: int lnn; /* line number in the real file */
1.110 schwarze 332: int fd;
1.1 kristaps 333: unsigned char c;
334:
1.95 schwarze 335: memset(&ln, 0, sizeof(ln));
1.1 kristaps 336:
1.47 schwarze 337: lnn = curp->line;
338: pos = 0;
1.1 kristaps 339:
1.95 schwarze 340: while (i < blk.sz) {
1.1 kristaps 341: if (0 == pos && '\0' == blk.buf[i])
342: break;
343:
344: if (start) {
345: curp->line = lnn;
346: curp->reparse_count = 0;
1.93 schwarze 347:
348: if (lnn < 3 &&
349: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 350: curp->filenc & MPARSE_LATIN1)
351: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 352: }
353:
1.95 schwarze 354: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 355:
356: /*
357: * When finding an unescaped newline character,
358: * leave the character loop to process the line.
359: * Skip a preceding carriage return, if any.
360: */
361:
1.95 schwarze 362: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 363: '\n' == blk.buf[i + 1])
364: ++i;
365: if ('\n' == blk.buf[i]) {
366: ++i;
367: ++lnn;
368: break;
369: }
370:
1.35 schwarze 371: /*
1.93 schwarze 372: * Make sure we have space for the worst
373: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 374: */
375:
1.95 schwarze 376: if (pos + 11 > ln.sz)
1.35 schwarze 377: resize_buf(&ln, 256);
378:
1.47 schwarze 379: /*
1.93 schwarze 380: * Encode 8-bit input.
1.1 kristaps 381: */
382:
1.93 schwarze 383: c = blk.buf[i];
384: if (c & 0x80) {
1.95 schwarze 385: if ( ! (curp->filenc && preconv_encode(
386: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 387: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
388: curp->line, pos, "0x%x", c);
1.93 schwarze 389: ln.buf[pos++] = '?';
390: i++;
391: }
392: continue;
393: }
394:
395: /*
396: * Exclude control characters.
397: */
1.1 kristaps 398:
1.93 schwarze 399: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 400: mandoc_vmsg(c == 0x00 || c == 0x04 ||
401: c > 0x0a ? MANDOCERR_CHAR_BAD :
402: MANDOCERR_CHAR_UNSUPP,
403: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 404: i++;
1.127 schwarze 405: if (c != '\r')
406: ln.buf[pos++] = '?';
1.1 kristaps 407: continue;
408: }
409:
410: ln.buf[pos++] = blk.buf[i++];
411: }
412:
1.170 ! schwarze 413: if (pos + 1 >= ln.sz)
1.1 kristaps 414: resize_buf(&ln, 256);
415:
1.170 ! schwarze 416: if (i == blk.sz || blk.buf[i] == '\0')
! 417: ln.buf[pos++] = '\n';
1.1 kristaps 418: ln.buf[pos] = '\0';
419:
420: /*
421: * A significant amount of complexity is contained by
422: * the roff preprocessor. It's line-oriented but can be
423: * expressed on one line, so we need at times to
424: * readjust our starting point and re-run it. The roff
425: * preprocessor can also readjust the buffers with new
426: * data, so we pass them in wholesale.
427: */
428:
429: of = 0;
430:
1.24 kristaps 431: /*
432: * Maintain a lookaside buffer of all parsed lines. We
433: * only do this if mparse_keep() has been invoked (the
434: * buffer may be accessed with mparse_getkeep()).
435: */
436:
437: if (curp->secondary) {
1.47 schwarze 438: curp->secondary->buf = mandoc_realloc(
439: curp->secondary->buf,
440: curp->secondary->sz + pos + 2);
441: memcpy(curp->secondary->buf +
442: curp->secondary->sz,
443: ln.buf, pos);
1.24 kristaps 444: curp->secondary->sz += pos;
445: curp->secondary->buf
446: [curp->secondary->sz] = '\n';
447: curp->secondary->sz++;
448: curp->secondary->buf
449: [curp->secondary->sz] = '\0';
450: }
1.1 kristaps 451: rerun:
1.96 schwarze 452: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 453:
454: switch (rr) {
1.47 schwarze 455: case ROFF_REPARSE:
1.163 schwarze 456: if (++curp->reparse_count > REPARSE_LIMIT)
1.3 kristaps 457: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 458: curp->line, pos, NULL);
1.163 schwarze 459: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
460: start == 1) {
461: pos = 0;
462: continue;
463: }
464: free(ln.buf);
465: return 0;
1.47 schwarze 466: case ROFF_APPEND:
1.95 schwarze 467: pos = strlen(ln.buf);
1.1 kristaps 468: continue;
1.47 schwarze 469: case ROFF_RERUN:
1.1 kristaps 470: goto rerun;
1.47 schwarze 471: case ROFF_IGN:
1.1 kristaps 472: pos = 0;
473: continue;
1.47 schwarze 474: case ROFF_SO:
1.95 schwarze 475: if ( ! (curp->options & MPARSE_SO) &&
476: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 477: curp->sodest = mandoc_strdup(ln.buf + of);
478: free(ln.buf);
1.163 schwarze 479: return 1;
1.45 schwarze 480: }
1.24 kristaps 481: /*
482: * We remove `so' clauses from our lookaside
483: * buffer because we're going to descend into
484: * the file recursively.
485: */
1.47 schwarze 486: if (curp->secondary)
1.25 kristaps 487: curp->secondary->sz -= pos + 1;
1.113 schwarze 488: save_file = curp->file;
1.148 schwarze 489: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 490: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 491: close(fd);
1.113 schwarze 492: curp->file = save_file;
493: } else {
494: curp->file = save_file;
1.52 schwarze 495: mandoc_vmsg(MANDOCERR_SO_FAIL,
496: curp, curp->line, pos,
497: ".so %s", ln.buf + of);
1.110 schwarze 498: ln.sz = mandoc_asprintf(&cp,
499: ".sp\nSee the file %s.\n.sp",
500: ln.buf + of);
501: free(ln.buf);
502: ln.buf = cp;
503: of = 0;
504: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 505: }
1.1 kristaps 506: pos = 0;
507: continue;
508: default:
509: break;
510: }
511:
1.150 schwarze 512: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 513: choose_parser(curp);
1.1 kristaps 514:
1.47 schwarze 515: /*
1.84 schwarze 516: * Lastly, push down into the parsers themselves.
1.1 kristaps 517: * If libroff returns ROFF_TBL, then add it to the
518: * currently open parse. Since we only get here if
519: * there does exist data (see tbl_data.c), we're
520: * guaranteed that something's been allocated.
521: * Do the same for ROFF_EQN.
522: */
523:
1.139 schwarze 524: if (rr == ROFF_TBL)
1.100 schwarze 525: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 526: roff_addtbl(curp->man, span);
527: else if (rr == ROFF_EQN)
528: roff_addeqn(curp->man, roff_eqn(curp->roff));
529: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 530: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 531: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
532: break;
1.1 kristaps 533:
534: /* Temporary buffers typically are not full. */
535:
536: if (0 == start && '\0' == blk.buf[i])
537: break;
538:
539: /* Start the next input line. */
540:
541: pos = 0;
542: }
543:
544: free(ln.buf);
1.163 schwarze 545: return 1;
1.1 kristaps 546: }
547:
548: static int
1.40 schwarze 549: read_whole_file(struct mparse *curp, const char *file, int fd,
550: struct buf *fb, int *with_mmap)
1.1 kristaps 551: {
1.161 schwarze 552: struct stat st;
1.140 schwarze 553: gzFile gz;
1.1 kristaps 554: size_t off;
555: ssize_t ssz;
1.143 schwarze 556:
557: if (fstat(fd, &st) == -1)
558: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 559:
560: /*
561: * If we're a regular file, try just reading in the whole entry
562: * via mmap(). This is faster than reading it into blocks, and
563: * since each file is only a few bytes to begin with, I'm not
564: * concerned that this is going to tank any machines.
565: */
566:
1.140 schwarze 567: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 568: if (st.st_size > 0x7fffffff) {
1.111 schwarze 569: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 570: return 0;
1.1 kristaps 571: }
572: *with_mmap = 1;
573: fb->sz = (size_t)st.st_size;
1.37 schwarze 574: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 575: if (fb->buf != MAP_FAILED)
1.142 schwarze 576: return 1;
1.1 kristaps 577: }
578:
1.140 schwarze 579: if (curp->gzip) {
1.143 schwarze 580: if ((gz = gzdopen(fd, "rb")) == NULL)
581: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 582: } else
583: gz = NULL;
584:
1.1 kristaps 585: /*
586: * If this isn't a regular file (like, say, stdin), then we must
587: * go the old way and just read things in bit by bit.
588: */
589:
590: *with_mmap = 0;
591: off = 0;
592: fb->sz = 0;
593: fb->buf = NULL;
594: for (;;) {
595: if (off == fb->sz) {
596: if (fb->sz == (1U << 31)) {
1.111 schwarze 597: mandoc_msg(MANDOCERR_TOOLARGE, curp,
598: 0, 0, NULL);
1.1 kristaps 599: break;
600: }
601: resize_buf(fb, 65536);
602: }
1.140 schwarze 603: ssz = curp->gzip ?
604: gzread(gz, fb->buf + (int)off, fb->sz - off) :
605: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 606: if (ssz == 0) {
607: fb->sz = off;
1.142 schwarze 608: return 1;
1.1 kristaps 609: }
1.143 schwarze 610: if (ssz == -1)
611: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 612: off += (size_t)ssz;
613: }
614:
615: free(fb->buf);
616: fb->buf = NULL;
1.142 schwarze 617: return 0;
1.1 kristaps 618: }
619:
620: static void
621: mparse_end(struct mparse *curp)
622: {
1.135 schwarze 623: if (curp->man->macroset == MACROSET_NONE)
624: curp->man->macroset = MACROSET_MAN;
625: if (curp->man->macroset == MACROSET_MDOC)
626: mdoc_endparse(curp->man);
627: else
1.111 schwarze 628: man_endparse(curp->man);
1.1 kristaps 629: roff_endparse(curp->roff);
630: }
631:
632: static void
1.36 schwarze 633: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 634: {
1.85 schwarze 635: struct buf *svprimary;
1.28 joerg 636: const char *svfile;
1.95 schwarze 637: size_t offset;
1.36 schwarze 638: static int recursion_depth;
639:
640: if (64 < recursion_depth) {
641: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
642: return;
643: }
1.28 joerg 644:
645: /* Line number is per-file. */
646: svfile = curp->file;
647: curp->file = file;
1.85 schwarze 648: svprimary = curp->primary;
1.83 schwarze 649: curp->primary = &blk;
1.28 joerg 650: curp->line = 1;
1.36 schwarze 651: recursion_depth++;
1.28 joerg 652:
1.93 schwarze 653: /* Skip an UTF-8 byte order mark. */
654: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
655: (unsigned char)blk.buf[0] == 0xef &&
656: (unsigned char)blk.buf[1] == 0xbb &&
657: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 658: offset = 3;
1.93 schwarze 659: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 660: } else
661: offset = 0;
1.93 schwarze 662:
1.95 schwarze 663: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 664:
1.111 schwarze 665: if (--recursion_depth == 0)
1.28 joerg 666: mparse_end(curp);
667:
1.85 schwarze 668: curp->primary = svprimary;
1.28 joerg 669: curp->file = svfile;
670: }
671:
672: enum mandoclevel
1.104 schwarze 673: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 674: const char *file)
675: {
676: struct buf blk;
677:
1.104 schwarze 678: blk.buf = buf;
1.28 joerg 679: blk.sz = len;
680:
1.36 schwarze 681: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 682: return curp->file_status;
1.28 joerg 683: }
684:
1.98 schwarze 685: /*
686: * Read the whole file into memory and call the parsers.
687: * Called recursively when an .so request is encountered.
688: */
1.36 schwarze 689: enum mandoclevel
690: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 691: {
1.28 joerg 692: struct buf blk;
693: int with_mmap;
1.93 schwarze 694: int save_filenc;
1.1 kristaps 695:
1.91 schwarze 696: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 697: save_filenc = curp->filenc;
698: curp->filenc = curp->options &
699: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 700: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 701: curp->filenc = save_filenc;
1.91 schwarze 702: if (with_mmap)
703: munmap(blk.buf, blk.sz);
704: else
705: free(blk.buf);
706: }
1.142 schwarze 707: return curp->file_status;
1.82 schwarze 708: }
709:
1.148 schwarze 710: int
711: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 712: {
713: char *cp;
1.148 schwarze 714: int fd;
1.82 schwarze 715:
716: curp->file = file;
1.140 schwarze 717: cp = strrchr(file, '.');
718: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 719:
1.140 schwarze 720: /* First try to use the filename as it is. */
1.98 schwarze 721:
1.148 schwarze 722: if ((fd = open(file, O_RDONLY)) != -1)
723: return fd;
1.98 schwarze 724:
1.140 schwarze 725: /*
726: * If that doesn't work and the filename doesn't
727: * already end in .gz, try appending .gz.
728: */
1.98 schwarze 729:
1.140 schwarze 730: if ( ! curp->gzip) {
1.98 schwarze 731: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 732: fd = open(cp, O_RDONLY);
1.108 schwarze 733: free(cp);
1.148 schwarze 734: if (fd != -1) {
1.140 schwarze 735: curp->gzip = 1;
1.148 schwarze 736: return fd;
1.82 schwarze 737: }
738: }
739:
1.140 schwarze 740: /* Neither worked, give up. */
1.97 schwarze 741:
1.140 schwarze 742: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 743: return -1;
1.1 kristaps 744: }
745:
746: struct mparse *
1.94 schwarze 747: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.144 schwarze 748: const char *defos)
1.1 kristaps 749: {
750: struct mparse *curp;
1.10 kristaps 751:
1.1 kristaps 752: curp = mandoc_calloc(1, sizeof(struct mparse));
753:
1.44 schwarze 754: curp->options = options;
1.3 kristaps 755: curp->wlevel = wlevel;
1.1 kristaps 756: curp->mmsg = mmsg;
1.29 schwarze 757: curp->defos = defos;
1.1 kristaps 758:
1.144 schwarze 759: curp->roff = roff_alloc(curp, options);
1.137 schwarze 760: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
761: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 762: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 763: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 764: if (curp->man->mdocmac == NULL)
765: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 766: } else if (curp->options & MPARSE_MAN) {
767: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 768: if (curp->man->manmac == NULL)
769: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 770: }
1.138 schwarze 771: curp->man->first->tok = TOKEN_NONE;
1.142 schwarze 772: return curp;
1.1 kristaps 773: }
774:
775: void
776: mparse_reset(struct mparse *curp)
777: {
778: roff_reset(curp->roff);
1.150 schwarze 779: roff_man_reset(curp->man);
1.160 schwarze 780:
781: free(curp->sodest);
782: curp->sodest = NULL;
783:
1.24 kristaps 784: if (curp->secondary)
785: curp->secondary->sz = 0;
1.1 kristaps 786:
787: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 788: curp->gzip = 0;
1.1 kristaps 789: }
790:
791: void
792: mparse_free(struct mparse *curp)
793: {
794:
1.164 schwarze 795: roffhash_free(curp->man->mdocmac);
796: roffhash_free(curp->man->manmac);
1.137 schwarze 797: roff_man_free(curp->man);
1.160 schwarze 798: roff_free(curp->roff);
1.24 kristaps 799: if (curp->secondary)
800: free(curp->secondary->buf);
1.1 kristaps 801:
1.24 kristaps 802: free(curp->secondary);
1.45 schwarze 803: free(curp->sodest);
1.1 kristaps 804: free(curp);
805: }
806:
807: void
1.135 schwarze 808: mparse_result(struct mparse *curp, struct roff_man **man,
809: char **sodest)
1.1 kristaps 810: {
811:
1.45 schwarze 812: if (sodest && NULL != (*sodest = curp->sodest)) {
813: *man = NULL;
814: return;
815: }
1.9 kristaps 816: if (man)
817: *man = curp->man;
1.157 schwarze 818: }
819:
820: void
821: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
822: {
823: if (curp->file_status > *rc)
824: *rc = curp->file_status;
1.3 kristaps 825: }
826:
827: void
828: mandoc_vmsg(enum mandocerr t, struct mparse *m,
829: int ln, int pos, const char *fmt, ...)
830: {
831: char buf[256];
832: va_list ap;
833:
834: va_start(ap, fmt);
1.48 schwarze 835: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 836: va_end(ap);
837:
838: mandoc_msg(t, m, ln, pos, buf);
839: }
840:
841: void
1.47 schwarze 842: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 843: int ln, int col, const char *msg)
844: {
845: enum mandoclevel level;
846:
1.112 schwarze 847: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 848: while (er < mandoclimits[level])
849: level--;
850:
1.108 schwarze 851: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 852: return;
853:
1.8 kristaps 854: if (m->mmsg)
855: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 856:
857: if (m->file_status < level)
858: m->file_status = level;
1.7 kristaps 859: }
860:
861: const char *
862: mparse_strerror(enum mandocerr er)
863: {
864:
1.142 schwarze 865: return mandocerrs[er];
1.7 kristaps 866: }
867:
868: const char *
869: mparse_strlevel(enum mandoclevel lvl)
870: {
1.142 schwarze 871: return mandoclevels[lvl];
1.24 kristaps 872: }
873:
874: void
875: mparse_keep(struct mparse *p)
876: {
877:
878: assert(NULL == p->secondary);
879: p->secondary = mandoc_calloc(1, sizeof(struct buf));
880: }
881:
882: const char *
883: mparse_getkeep(const struct mparse *p)
884: {
885:
886: assert(p->secondary);
1.142 schwarze 887: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 888: }
CVSweb