Annotation of src/usr.bin/mandoc/read.c, Revision 1.147
1.147 ! schwarze 1: /* $OpenBSD: read.c,v 1.146 2017/06/10 01:48:31 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.129 schwarze 4: * Copyright (c) 2010-2017 Ingo Schwarze <schwarze@openbsd.org>
1.20 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 schwarze 6: *
7: * Permission to use, copy, modify, and distribute this software for any
8: * purpose with or without fee is hereby granted, provided that the above
9: * copyright notice and this permission notice appear in all copies.
10: *
1.108 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.108 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18: */
1.58 schwarze 19: #include <sys/types.h>
20: #include <sys/mman.h>
1.1 schwarze 21: #include <sys/stat.h>
22:
23: #include <assert.h>
24: #include <ctype.h>
1.118 schwarze 25: #include <err.h>
1.18 schwarze 26: #include <errno.h>
1.1 schwarze 27: #include <fcntl.h>
28: #include <stdarg.h>
29: #include <stdio.h>
30: #include <stdlib.h>
31: #include <string.h>
32: #include <unistd.h>
1.115 schwarze 33: #include <zlib.h>
1.1 schwarze 34:
1.108 schwarze 35: #include "mandoc_aux.h"
1.1 schwarze 36: #include "mandoc.h"
1.108 schwarze 37: #include "roff.h"
1.1 schwarze 38: #include "mdoc.h"
39: #include "man.h"
1.108 schwarze 40: #include "libmandoc.h"
1.114 schwarze 41: #include "roff_int.h"
1.1 schwarze 42:
43: #define REPARSE_LIMIT 1000
44:
45: struct mparse {
1.133 schwarze 46: struct roff *roff; /* roff parser (!NULL) */
1.109 schwarze 47: struct roff_man *man; /* man parser */
1.23 schwarze 48: char *sodest; /* filename pointed to by .so */
1.59 schwarze 49: const char *file; /* filename of current input file */
50: struct buf *primary; /* buffer currently being parsed */
51: struct buf *secondary; /* preprocessed copy of input */
52: const char *defos; /* default operating system */
53: mandocmsg mmsg; /* warning/error message handler */
54: enum mandoclevel file_status; /* status of current parse */
55: enum mandoclevel wlevel; /* ignore messages below this */
56: int options; /* parser options */
1.115 schwarze 57: int gzip; /* current input file is gzipped */
1.70 schwarze 58: int filenc; /* encoding of the current file */
1.1 schwarze 59: int reparse_count; /* finite interp. stack */
1.59 schwarze 60: int line; /* line number in the file */
1.1 schwarze 61: };
62:
1.60 schwarze 63: static void choose_parser(struct mparse *);
1.1 schwarze 64: static void resize_buf(struct buf *, size_t);
1.135 schwarze 65: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.18 schwarze 66: static int read_whole_file(struct mparse *, const char *, int,
67: struct buf *, int *);
1.1 schwarze 68: static void mparse_end(struct mparse *);
1.15 schwarze 69: static void mparse_parse_buffer(struct mparse *, struct buf,
70: const char *);
1.1 schwarze 71:
72: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
73: MANDOCERR_OK,
1.137 schwarze 74: MANDOCERR_STYLE,
1.1 schwarze 75: MANDOCERR_WARNING,
76: MANDOCERR_ERROR,
1.87 schwarze 77: MANDOCERR_UNSUPP,
1.1 schwarze 78: MANDOCERR_MAX,
79: MANDOCERR_MAX
80: };
81:
82: static const char * const mandocerrs[MANDOCERR_MAX] = {
83: "ok",
84:
1.137 schwarze 85: "generic style suggestion",
1.138 schwarze 86:
87: "useless macro",
1.139 schwarze 88: "consider using OS macro",
1.144 schwarze 89: "errnos out of order",
90: "duplicate errno",
1.140 schwarze 91: "description line ends with a full stop",
1.146 schwarze 92: "no blank before trailing delimiter",
1.147 ! schwarze 93: "function name without markup",
1.138 schwarze 94:
1.1 schwarze 95: "generic warning",
96:
97: /* related to the prologue */
1.57 schwarze 98: "missing manual title, using UNTITLED",
99: "missing manual title, using \"\"",
1.32 schwarze 100: "lower case character in document title",
1.57 schwarze 101: "missing manual section, using \"\"",
1.1 schwarze 102: "unknown manual section",
1.32 schwarze 103: "missing date, using today's date",
1.1 schwarze 104: "cannot parse date, using it verbatim",
1.57 schwarze 105: "missing Os macro, using \"\"",
106: "duplicate prologue macro",
107: "late prologue macro",
108: "skipping late title macro",
1.1 schwarze 109: "prologue macros out of order",
110:
111: /* related to document structure */
112: ".so is fragile, better use ln(1)",
1.28 schwarze 113: "no document body",
1.32 schwarze 114: "content before first section header",
115: "first section is not \"NAME\"",
1.129 schwarze 116: "NAME section without Nm before Nd",
1.103 schwarze 117: "NAME section without description",
118: "description not at the end of NAME",
119: "bad NAME section content",
1.129 schwarze 120: "missing comma before name",
1.96 schwarze 121: "missing description line, using \"\"",
1.134 schwarze 122: "description line outside NAME section",
1.1 schwarze 123: "sections out of conventional order",
1.32 schwarze 124: "duplicate section title",
125: "unexpected section",
1.63 schwarze 126: "unusual Xr order",
127: "unusual Xr punctuation",
1.62 schwarze 128: "AUTHORS section without An macro",
1.1 schwarze 129:
130: /* related to macros and nesting */
1.33 schwarze 131: "obsolete macro",
1.78 schwarze 132: "macro neither callable nor escaped",
1.1 schwarze 133: "skipping paragraph macro",
1.10 schwarze 134: "moving paragraph macro out of list",
1.1 schwarze 135: "skipping no-space macro",
136: "blocks badly nested",
137: "nested displays are not portable",
1.35 schwarze 138: "moving content out of list",
1.56 schwarze 139: "fill mode already enabled, skipping",
140: "fill mode already disabled, skipping",
1.1 schwarze 141: "line scope broken",
1.141 schwarze 142: "skipping blank line in line scope",
1.1 schwarze 143:
144: /* related to missing macro arguments */
1.36 schwarze 145: "skipping empty request",
146: "conditional request controls empty scope",
1.1 schwarze 147: "skipping empty macro",
1.99 schwarze 148: "empty block",
1.40 schwarze 149: "empty argument, using 0n",
1.38 schwarze 150: "missing display type, using -ragged",
151: "list type is not the first argument",
1.125 schwarze 152: "missing -width in -tag list, using 6n",
1.56 schwarze 153: "missing utility name, using \"\"",
1.98 schwarze 154: "missing function name, using \"\"",
1.38 schwarze 155: "empty head in list item",
156: "empty list item",
1.39 schwarze 157: "missing font type, using \\fR",
158: "unknown font type, using \\fR",
1.79 schwarze 159: "nothing follows prefix",
1.97 schwarze 160: "empty reference block",
1.128 schwarze 161: "missing section argument",
1.38 schwarze 162: "missing -std argument, adding it",
1.100 schwarze 163: "missing option string, using \"\"",
164: "missing resource identifier, using \"\"",
1.66 schwarze 165: "missing eqn box, using \"\"",
1.1 schwarze 166:
167: /* related to bad macro arguments */
1.42 schwarze 168: "unterminated quoted argument",
1.1 schwarze 169: "duplicate argument",
1.54 schwarze 170: "skipping duplicate argument",
1.41 schwarze 171: "skipping duplicate display type",
172: "skipping duplicate list type",
1.54 schwarze 173: "skipping -width argument",
1.101 schwarze 174: "wrong number of cells",
1.1 schwarze 175: "unknown AT&T UNIX version",
1.64 schwarze 176: "comma in function argument",
1.65 schwarze 177: "parenthesis in function name",
1.145 schwarze 178: "unknown library name",
1.45 schwarze 179: "invalid content in Rs block",
1.41 schwarze 180: "invalid Boolean argument",
181: "unknown font, skipping request",
1.101 schwarze 182: "odd number of characters in request",
1.1 schwarze 183:
184: /* related to plain text */
1.42 schwarze 185: "blank line in fill mode, using .sp",
186: "tab in filled text",
187: "whitespace at end of input line",
1.131 schwarze 188: "new sentence, new line",
1.1 schwarze 189: "bad comment style",
1.42 schwarze 190: "invalid escape sequence",
191: "undefined string, using \"\"",
1.3 schwarze 192:
1.93 schwarze 193: /* related to tables */
194: "tbl line starts with span",
195: "tbl column starts with span",
196: "skipping vertical bar in tbl layout",
197:
1.1 schwarze 198: "generic error",
1.3 schwarze 199:
1.1 schwarze 200: /* related to tables */
1.91 schwarze 201: "non-alphabetic character in tbl options",
202: "skipping unknown tbl option",
203: "missing tbl option argument",
204: "wrong tbl option argument size",
1.93 schwarze 205: "empty tbl layout",
206: "invalid character in tbl layout",
207: "unmatched parenthesis in tbl layout",
1.94 schwarze 208: "tbl without any data cells",
209: "ignoring data in spanned tbl cell",
210: "ignoring extra tbl data cells",
211: "data block open at end of tbl",
1.1 schwarze 212:
1.46 schwarze 213: /* related to document structure and macros */
1.83 schwarze 214: NULL,
1.1 schwarze 215: "input stack limit exceeded, infinite loop?",
216: "skipping bad character",
1.46 schwarze 217: "skipping unknown macro",
1.87 schwarze 218: "skipping insecure request",
1.51 schwarze 219: "skipping item outside list",
1.46 schwarze 220: "skipping column outside column list",
221: "skipping end of block that is not open",
1.90 schwarze 222: "fewer RS blocks open, skipping",
1.46 schwarze 223: "inserting missing end of block",
224: "appending missing end of block",
225:
226: /* related to request and macro arguments */
1.1 schwarze 227: "escaped character not allowed in a name",
1.75 schwarze 228: "NOT IMPLEMENTED: Bd -file",
1.120 schwarze 229: "skipping display without arguments",
1.49 schwarze 230: "missing list type, using -item",
1.143 schwarze 231: "argument is not numeric, using 1",
1.48 schwarze 232: "missing manual name, using \"\"",
1.49 schwarze 233: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 234: "unknown standard specifier",
1.49 schwarze 235: "skipping request without numeric argument",
1.84 schwarze 236: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 237: ".so request failed",
1.39 schwarze 238: "skipping all arguments",
239: "skipping excess arguments",
1.68 schwarze 240: "divide by zero",
1.87 schwarze 241:
242: "unsupported feature",
243: "input too large",
1.89 schwarze 244: "unsupported control character",
1.87 schwarze 245: "unsupported roff request",
1.94 schwarze 246: "eqn delim option in tbl",
1.93 schwarze 247: "unsupported tbl layout modifier",
1.87 schwarze 248: "ignoring macro in table",
1.1 schwarze 249: };
250:
251: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
252: "SUCCESS",
1.137 schwarze 253: "STYLE",
1.1 schwarze 254: "WARNING",
255: "ERROR",
1.87 schwarze 256: "UNSUPP",
1.1 schwarze 257: "BADARG",
258: "SYSERR"
259: };
260:
1.25 schwarze 261:
1.1 schwarze 262: static void
263: resize_buf(struct buf *buf, size_t initial)
264: {
265:
266: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
267: buf->buf = mandoc_realloc(buf->buf, buf->sz);
268: }
269:
270: static void
1.60 schwarze 271: choose_parser(struct mparse *curp)
1.1 schwarze 272: {
1.59 schwarze 273: char *cp, *ep;
274: int format;
1.1 schwarze 275:
1.59 schwarze 276: /*
277: * If neither command line arguments -mdoc or -man select
278: * a parser nor the roff parser found a .Dd or .TH macro
279: * yet, look ahead in the main input buffer.
280: */
281:
282: if ((format = roff_getformat(curp->roff)) == 0) {
283: cp = curp->primary->buf;
284: ep = cp + curp->primary->sz;
285: while (cp < ep) {
1.61 schwarze 286: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 287: cp++;
288: if (cp[0] == 'D' && cp[1] == 'd') {
289: format = MPARSE_MDOC;
290: break;
291: }
292: if (cp[0] == 'T' && cp[1] == 'H') {
293: format = MPARSE_MAN;
294: break;
295: }
296: }
297: cp = memchr(cp, '\n', ep - cp);
298: if (cp == NULL)
299: break;
300: cp++;
301: }
1.1 schwarze 302: }
303:
1.59 schwarze 304: if (format == MPARSE_MDOC) {
1.112 schwarze 305: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 306: if (curp->man->mdocmac == NULL)
307: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 308: } else {
309: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 310: if (curp->man->manmac == NULL)
311: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 312: }
1.136 schwarze 313: curp->man->first->tok = TOKEN_NONE;
1.1 schwarze 314: }
315:
316: /*
1.71 schwarze 317: * Main parse routine for a buffer.
318: * It assumes encoding and line numbering are already set up.
319: * It can recurse directly (for invocations of user-defined
320: * macros, inline equations, and input line traps)
321: * and indirectly (for .so file inclusion).
1.1 schwarze 322: */
1.135 schwarze 323: static int
1.71 schwarze 324: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 325: {
326: const struct tbl_span *span;
327: struct buf ln;
1.88 schwarze 328: const char *save_file;
1.85 schwarze 329: char *cp;
1.71 schwarze 330: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 331: enum rofferr rr;
1.76 schwarze 332: int of;
1.1 schwarze 333: int lnn; /* line number in the real file */
1.85 schwarze 334: int fd;
1.1 schwarze 335: unsigned char c;
336:
1.71 schwarze 337: memset(&ln, 0, sizeof(ln));
1.1 schwarze 338:
1.25 schwarze 339: lnn = curp->line;
340: pos = 0;
1.1 schwarze 341:
1.71 schwarze 342: while (i < blk.sz) {
1.1 schwarze 343: if (0 == pos && '\0' == blk.buf[i])
344: break;
345:
346: if (start) {
347: curp->line = lnn;
348: curp->reparse_count = 0;
1.70 schwarze 349:
350: if (lnn < 3 &&
351: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 352: curp->filenc & MPARSE_LATIN1)
353: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 354: }
355:
1.71 schwarze 356: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 357:
358: /*
359: * When finding an unescaped newline character,
360: * leave the character loop to process the line.
361: * Skip a preceding carriage return, if any.
362: */
363:
1.71 schwarze 364: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 365: '\n' == blk.buf[i + 1])
366: ++i;
367: if ('\n' == blk.buf[i]) {
368: ++i;
369: ++lnn;
370: break;
371: }
372:
1.13 schwarze 373: /*
1.70 schwarze 374: * Make sure we have space for the worst
375: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 376: */
377:
1.71 schwarze 378: if (pos + 11 > ln.sz)
1.13 schwarze 379: resize_buf(&ln, 256);
380:
1.25 schwarze 381: /*
1.70 schwarze 382: * Encode 8-bit input.
1.1 schwarze 383: */
384:
1.70 schwarze 385: c = blk.buf[i];
386: if (c & 0x80) {
1.71 schwarze 387: if ( ! (curp->filenc && preconv_encode(
388: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 389: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
390: curp->line, pos, "0x%x", c);
1.70 schwarze 391: ln.buf[pos++] = '?';
392: i++;
393: }
394: continue;
395: }
396:
397: /*
398: * Exclude control characters.
399: */
1.1 schwarze 400:
1.70 schwarze 401: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 402: mandoc_vmsg(c == 0x00 || c == 0x04 ||
403: c > 0x0a ? MANDOCERR_CHAR_BAD :
404: MANDOCERR_CHAR_UNSUPP,
405: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 406: i++;
1.102 schwarze 407: if (c != '\r')
408: ln.buf[pos++] = '?';
1.1 schwarze 409: continue;
410: }
411:
412: ln.buf[pos++] = blk.buf[i++];
413: }
414:
1.142 schwarze 415: if (pos + 1 >= ln.sz)
1.1 schwarze 416: resize_buf(&ln, 256);
417:
1.142 schwarze 418: if (i == blk.sz || blk.buf[i] == '\0')
419: ln.buf[pos++] = '\n';
1.1 schwarze 420: ln.buf[pos] = '\0';
421:
422: /*
423: * A significant amount of complexity is contained by
424: * the roff preprocessor. It's line-oriented but can be
425: * expressed on one line, so we need at times to
426: * readjust our starting point and re-run it. The roff
427: * preprocessor can also readjust the buffers with new
428: * data, so we pass them in wholesale.
429: */
430:
431: of = 0;
432:
1.4 schwarze 433: /*
434: * Maintain a lookaside buffer of all parsed lines. We
435: * only do this if mparse_keep() has been invoked (the
436: * buffer may be accessed with mparse_getkeep()).
437: */
438:
439: if (curp->secondary) {
1.25 schwarze 440: curp->secondary->buf = mandoc_realloc(
441: curp->secondary->buf,
442: curp->secondary->sz + pos + 2);
443: memcpy(curp->secondary->buf +
444: curp->secondary->sz,
445: ln.buf, pos);
1.4 schwarze 446: curp->secondary->sz += pos;
447: curp->secondary->buf
448: [curp->secondary->sz] = '\n';
449: curp->secondary->sz++;
450: curp->secondary->buf
451: [curp->secondary->sz] = '\0';
452: }
1.1 schwarze 453: rerun:
1.72 schwarze 454: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 455:
456: switch (rr) {
1.25 schwarze 457: case ROFF_REPARSE:
1.135 schwarze 458: if (++curp->reparse_count > REPARSE_LIMIT)
1.1 schwarze 459: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 460: curp->line, pos, NULL);
1.135 schwarze 461: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
462: start == 1) {
463: pos = 0;
464: continue;
465: }
466: free(ln.buf);
467: return 0;
1.25 schwarze 468: case ROFF_APPEND:
1.71 schwarze 469: pos = strlen(ln.buf);
1.1 schwarze 470: continue;
1.25 schwarze 471: case ROFF_RERUN:
1.1 schwarze 472: goto rerun;
1.25 schwarze 473: case ROFF_IGN:
1.1 schwarze 474: pos = 0;
475: continue;
1.25 schwarze 476: case ROFF_SO:
1.71 schwarze 477: if ( ! (curp->options & MPARSE_SO) &&
478: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 479: curp->sodest = mandoc_strdup(ln.buf + of);
480: free(ln.buf);
1.135 schwarze 481: return 1;
1.23 schwarze 482: }
1.4 schwarze 483: /*
484: * We remove `so' clauses from our lookaside
485: * buffer because we're going to descend into
486: * the file recursively.
487: */
1.25 schwarze 488: if (curp->secondary)
1.4 schwarze 489: curp->secondary->sz -= pos + 1;
1.88 schwarze 490: save_file = curp->file;
1.122 schwarze 491: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 492: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 493: close(fd);
1.88 schwarze 494: curp->file = save_file;
495: } else {
496: curp->file = save_file;
1.30 schwarze 497: mandoc_vmsg(MANDOCERR_SO_FAIL,
498: curp, curp->line, pos,
499: ".so %s", ln.buf + of);
1.85 schwarze 500: ln.sz = mandoc_asprintf(&cp,
501: ".sp\nSee the file %s.\n.sp",
502: ln.buf + of);
503: free(ln.buf);
504: ln.buf = cp;
505: of = 0;
506: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 507: }
1.1 schwarze 508: pos = 0;
509: continue;
510: default:
511: break;
512: }
513:
1.124 schwarze 514: if (curp->man->macroset == MACROSET_NONE)
1.60 schwarze 515: choose_parser(curp);
1.1 schwarze 516:
1.25 schwarze 517: /*
1.60 schwarze 518: * Lastly, push down into the parsers themselves.
1.1 schwarze 519: * If libroff returns ROFF_TBL, then add it to the
520: * currently open parse. Since we only get here if
521: * there does exist data (see tbl_data.c), we're
522: * guaranteed that something's been allocated.
523: * Do the same for ROFF_EQN.
524: */
525:
1.114 schwarze 526: if (rr == ROFF_TBL)
1.76 schwarze 527: while ((span = roff_span(curp->roff)) != NULL)
1.114 schwarze 528: roff_addtbl(curp->man, span);
529: else if (rr == ROFF_EQN)
530: roff_addeqn(curp->man, roff_eqn(curp->roff));
531: else if ((curp->man->macroset == MACROSET_MDOC ?
1.110 schwarze 532: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 533: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
534: break;
1.1 schwarze 535:
536: /* Temporary buffers typically are not full. */
537:
538: if (0 == start && '\0' == blk.buf[i])
539: break;
540:
541: /* Start the next input line. */
542:
543: pos = 0;
544: }
545:
546: free(ln.buf);
1.135 schwarze 547: return 1;
1.1 schwarze 548: }
549:
550: static int
1.18 schwarze 551: read_whole_file(struct mparse *curp, const char *file, int fd,
552: struct buf *fb, int *with_mmap)
1.1 schwarze 553: {
554: struct stat st;
1.115 schwarze 555: gzFile gz;
1.1 schwarze 556: size_t off;
557: ssize_t ssz;
558:
1.118 schwarze 559: if (fstat(fd, &st) == -1)
560: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 561:
562: /*
563: * If we're a regular file, try just reading in the whole entry
564: * via mmap(). This is faster than reading it into blocks, and
565: * since each file is only a few bytes to begin with, I'm not
566: * concerned that this is going to tank any machines.
567: */
568:
1.115 schwarze 569: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 570: if (st.st_size > 0x7fffffff) {
1.86 schwarze 571: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.117 schwarze 572: return 0;
1.1 schwarze 573: }
574: *with_mmap = 1;
575: fb->sz = (size_t)st.st_size;
1.15 schwarze 576: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 577: if (fb->buf != MAP_FAILED)
1.117 schwarze 578: return 1;
1.1 schwarze 579: }
580:
1.115 schwarze 581: if (curp->gzip) {
1.118 schwarze 582: if ((gz = gzdopen(fd, "rb")) == NULL)
583: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.115 schwarze 584: } else
585: gz = NULL;
586:
1.1 schwarze 587: /*
588: * If this isn't a regular file (like, say, stdin), then we must
589: * go the old way and just read things in bit by bit.
590: */
591:
592: *with_mmap = 0;
593: off = 0;
594: fb->sz = 0;
595: fb->buf = NULL;
596: for (;;) {
597: if (off == fb->sz) {
598: if (fb->sz == (1U << 31)) {
1.86 schwarze 599: mandoc_msg(MANDOCERR_TOOLARGE, curp,
600: 0, 0, NULL);
1.1 schwarze 601: break;
602: }
603: resize_buf(fb, 65536);
604: }
1.115 schwarze 605: ssz = curp->gzip ?
606: gzread(gz, fb->buf + (int)off, fb->sz - off) :
607: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 608: if (ssz == 0) {
609: fb->sz = off;
1.117 schwarze 610: return 1;
1.1 schwarze 611: }
1.118 schwarze 612: if (ssz == -1)
613: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 614: off += (size_t)ssz;
615: }
616:
617: free(fb->buf);
618: fb->buf = NULL;
1.117 schwarze 619: return 0;
1.1 schwarze 620: }
621:
622: static void
623: mparse_end(struct mparse *curp)
624: {
1.110 schwarze 625: if (curp->man->macroset == MACROSET_NONE)
626: curp->man->macroset = MACROSET_MAN;
627: if (curp->man->macroset == MACROSET_MDOC)
628: mdoc_endparse(curp->man);
629: else
1.86 schwarze 630: man_endparse(curp->man);
1.1 schwarze 631: roff_endparse(curp->roff);
632: }
633:
1.15 schwarze 634: static void
635: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 636: {
1.61 schwarze 637: struct buf *svprimary;
1.1 schwarze 638: const char *svfile;
1.71 schwarze 639: size_t offset;
1.14 schwarze 640: static int recursion_depth;
641:
642: if (64 < recursion_depth) {
643: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 644: return;
1.14 schwarze 645: }
1.1 schwarze 646:
1.15 schwarze 647: /* Line number is per-file. */
648: svfile = curp->file;
649: curp->file = file;
1.61 schwarze 650: svprimary = curp->primary;
1.59 schwarze 651: curp->primary = &blk;
1.15 schwarze 652: curp->line = 1;
653: recursion_depth++;
654:
1.70 schwarze 655: /* Skip an UTF-8 byte order mark. */
656: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
657: (unsigned char)blk.buf[0] == 0xef &&
658: (unsigned char)blk.buf[1] == 0xbb &&
659: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 660: offset = 3;
1.70 schwarze 661: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 662: } else
663: offset = 0;
1.70 schwarze 664:
1.71 schwarze 665: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 666:
1.86 schwarze 667: if (--recursion_depth == 0)
1.15 schwarze 668: mparse_end(curp);
669:
1.61 schwarze 670: curp->primary = svprimary;
1.15 schwarze 671: curp->file = svfile;
672: }
673:
1.74 schwarze 674: /*
675: * Read the whole file into memory and call the parsers.
676: * Called recursively when an .so request is encountered.
677: */
1.15 schwarze 678: enum mandoclevel
679: mparse_readfd(struct mparse *curp, int fd, const char *file)
680: {
681: struct buf blk;
682: int with_mmap;
1.70 schwarze 683: int save_filenc;
1.1 schwarze 684:
1.67 schwarze 685: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 686: save_filenc = curp->filenc;
687: curp->filenc = curp->options &
688: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 689: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 690: curp->filenc = save_filenc;
1.67 schwarze 691: if (with_mmap)
692: munmap(blk.buf, blk.sz);
693: else
694: free(blk.buf);
695: }
1.117 schwarze 696: return curp->file_status;
1.58 schwarze 697: }
698:
1.122 schwarze 699: int
700: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 701: {
702: char *cp;
1.122 schwarze 703: int fd;
1.58 schwarze 704:
705: curp->file = file;
1.115 schwarze 706: cp = strrchr(file, '.');
707: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 708:
1.115 schwarze 709: /* First try to use the filename as it is. */
1.74 schwarze 710:
1.122 schwarze 711: if ((fd = open(file, O_RDONLY)) != -1)
712: return fd;
1.74 schwarze 713:
1.115 schwarze 714: /*
715: * If that doesn't work and the filename doesn't
716: * already end in .gz, try appending .gz.
717: */
1.74 schwarze 718:
1.115 schwarze 719: if ( ! curp->gzip) {
1.74 schwarze 720: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 721: fd = open(cp, O_RDONLY);
1.83 schwarze 722: free(cp);
1.122 schwarze 723: if (fd != -1) {
1.115 schwarze 724: curp->gzip = 1;
1.122 schwarze 725: return fd;
1.58 schwarze 726: }
727: }
728:
1.115 schwarze 729: /* Neither worked, give up. */
1.73 schwarze 730:
1.115 schwarze 731: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.122 schwarze 732: return -1;
1.1 schwarze 733: }
734:
735: struct mparse *
1.69 schwarze 736: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.119 schwarze 737: const char *defos)
1.1 schwarze 738: {
739: struct mparse *curp;
740:
741: curp = mandoc_calloc(1, sizeof(struct mparse));
742:
1.22 schwarze 743: curp->options = options;
1.1 schwarze 744: curp->wlevel = wlevel;
745: curp->mmsg = mmsg;
1.7 schwarze 746: curp->defos = defos;
1.1 schwarze 747:
1.119 schwarze 748: curp->roff = roff_alloc(curp, options);
1.112 schwarze 749: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
750: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 751: if (curp->options & MPARSE_MDOC) {
1.112 schwarze 752: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 753: if (curp->man->mdocmac == NULL)
754: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 755: } else if (curp->options & MPARSE_MAN) {
756: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 757: if (curp->man->manmac == NULL)
758: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 759: }
1.113 schwarze 760: curp->man->first->tok = TOKEN_NONE;
1.117 schwarze 761: return curp;
1.1 schwarze 762: }
763:
764: void
765: mparse_reset(struct mparse *curp)
766: {
767: roff_reset(curp->roff);
1.124 schwarze 768: roff_man_reset(curp->man);
1.133 schwarze 769:
770: free(curp->sodest);
771: curp->sodest = NULL;
772:
1.4 schwarze 773: if (curp->secondary)
774: curp->secondary->sz = 0;
1.1 schwarze 775:
776: curp->file_status = MANDOCLEVEL_OK;
1.132 schwarze 777: curp->gzip = 0;
1.1 schwarze 778: }
779:
780: void
781: mparse_free(struct mparse *curp)
782: {
783:
1.136 schwarze 784: roffhash_free(curp->man->mdocmac);
785: roffhash_free(curp->man->manmac);
1.112 schwarze 786: roff_man_free(curp->man);
1.133 schwarze 787: roff_free(curp->roff);
1.4 schwarze 788: if (curp->secondary)
789: free(curp->secondary->buf);
1.1 schwarze 790:
1.4 schwarze 791: free(curp->secondary);
1.23 schwarze 792: free(curp->sodest);
1.1 schwarze 793: free(curp);
794: }
795:
796: void
1.110 schwarze 797: mparse_result(struct mparse *curp, struct roff_man **man,
798: char **sodest)
1.1 schwarze 799: {
800:
1.23 schwarze 801: if (sodest && NULL != (*sodest = curp->sodest)) {
802: *man = NULL;
803: return;
804: }
1.1 schwarze 805: if (man)
806: *man = curp->man;
1.130 schwarze 807: }
808:
809: void
810: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
811: {
812: if (curp->file_status > *rc)
813: *rc = curp->file_status;
1.1 schwarze 814: }
815:
816: void
817: mandoc_vmsg(enum mandocerr t, struct mparse *m,
818: int ln, int pos, const char *fmt, ...)
819: {
820: char buf[256];
821: va_list ap;
822:
823: va_start(ap, fmt);
1.26 schwarze 824: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 825: va_end(ap);
826:
827: mandoc_msg(t, m, ln, pos, buf);
828: }
829:
830: void
1.25 schwarze 831: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 832: int ln, int col, const char *msg)
833: {
834: enum mandoclevel level;
835:
1.87 schwarze 836: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 837: while (er < mandoclimits[level])
838: level--;
839:
1.83 schwarze 840: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 841: return;
842:
843: if (m->mmsg)
844: (*m->mmsg)(er, level, m->file, ln, col, msg);
845:
846: if (m->file_status < level)
847: m->file_status = level;
848: }
849:
850: const char *
851: mparse_strerror(enum mandocerr er)
852: {
853:
1.117 schwarze 854: return mandocerrs[er];
1.1 schwarze 855: }
856:
857: const char *
858: mparse_strlevel(enum mandoclevel lvl)
859: {
1.117 schwarze 860: return mandoclevels[lvl];
1.4 schwarze 861: }
862:
863: void
864: mparse_keep(struct mparse *p)
865: {
866:
867: assert(NULL == p->secondary);
868: p->secondary = mandoc_calloc(1, sizeof(struct buf));
869: }
870:
871: const char *
872: mparse_getkeep(const struct mparse *p)
873: {
874:
875: assert(p->secondary);
1.117 schwarze 876: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 schwarze 877: }