Annotation of mandoc/read.c, Revision 1.171
1.171 ! schwarze 1: /* $Id: read.c,v 1.170 2017/06/04 00:13:15 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",
1.171 ! schwarze 230: "argument is not numeric, using 1",
1.70 schwarze 231: "missing manual name, using \"\"",
1.71 schwarze 232: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 233: "unknown standard specifier",
1.71 schwarze 234: "skipping request without numeric argument",
1.109 schwarze 235: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 236: ".so request failed",
1.61 schwarze 237: "skipping all arguments",
238: "skipping excess arguments",
1.92 kristaps 239: "divide by zero",
1.112 schwarze 240:
241: "unsupported feature",
242: "input too large",
1.114 schwarze 243: "unsupported control character",
1.112 schwarze 244: "unsupported roff request",
1.119 schwarze 245: "eqn delim option in tbl",
1.118 schwarze 246: "unsupported tbl layout modifier",
1.112 schwarze 247: "ignoring macro in table",
1.7 kristaps 248: };
249:
250: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
251: "SUCCESS",
1.165 schwarze 252: "STYLE",
1.7 kristaps 253: "WARNING",
254: "ERROR",
1.112 schwarze 255: "UNSUPP",
1.7 kristaps 256: "BADARG",
257: "SYSERR"
258: };
259:
1.47 schwarze 260:
1.1 kristaps 261: static void
262: resize_buf(struct buf *buf, size_t initial)
263: {
264:
265: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
266: buf->buf = mandoc_realloc(buf->buf, buf->sz);
267: }
268:
269: static void
1.84 schwarze 270: choose_parser(struct mparse *curp)
1.1 kristaps 271: {
1.83 schwarze 272: char *cp, *ep;
273: int format;
1.1 kristaps 274:
1.83 schwarze 275: /*
276: * If neither command line arguments -mdoc or -man select
277: * a parser nor the roff parser found a .Dd or .TH macro
278: * yet, look ahead in the main input buffer.
279: */
280:
281: if ((format = roff_getformat(curp->roff)) == 0) {
282: cp = curp->primary->buf;
283: ep = cp + curp->primary->sz;
284: while (cp < ep) {
1.85 schwarze 285: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 286: cp++;
287: if (cp[0] == 'D' && cp[1] == 'd') {
288: format = MPARSE_MDOC;
289: break;
290: }
291: if (cp[0] == 'T' && cp[1] == 'H') {
292: format = MPARSE_MAN;
293: break;
294: }
295: }
296: cp = memchr(cp, '\n', ep - cp);
297: if (cp == NULL)
298: break;
299: cp++;
300: }
1.1 kristaps 301: }
302:
1.83 schwarze 303: if (format == MPARSE_MDOC) {
1.137 schwarze 304: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 305: if (curp->man->mdocmac == NULL)
306: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 307: } else {
308: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 309: if (curp->man->manmac == NULL)
310: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 311: }
1.164 schwarze 312: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 313: }
314:
315: /*
1.95 schwarze 316: * Main parse routine for a buffer.
317: * It assumes encoding and line numbering are already set up.
318: * It can recurse directly (for invocations of user-defined
319: * macros, inline equations, and input line traps)
320: * and indirectly (for .so file inclusion).
1.1 kristaps 321: */
1.163 schwarze 322: static int
1.95 schwarze 323: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 324: {
325: const struct tbl_span *span;
326: struct buf ln;
1.113 schwarze 327: const char *save_file;
1.110 schwarze 328: char *cp;
1.95 schwarze 329: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 330: enum rofferr rr;
1.100 schwarze 331: int of;
1.1 kristaps 332: int lnn; /* line number in the real file */
1.110 schwarze 333: int fd;
1.1 kristaps 334: unsigned char c;
335:
1.95 schwarze 336: memset(&ln, 0, sizeof(ln));
1.1 kristaps 337:
1.47 schwarze 338: lnn = curp->line;
339: pos = 0;
1.1 kristaps 340:
1.95 schwarze 341: while (i < blk.sz) {
1.1 kristaps 342: if (0 == pos && '\0' == blk.buf[i])
343: break;
344:
345: if (start) {
346: curp->line = lnn;
347: curp->reparse_count = 0;
1.93 schwarze 348:
349: if (lnn < 3 &&
350: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 351: curp->filenc & MPARSE_LATIN1)
352: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 353: }
354:
1.95 schwarze 355: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 356:
357: /*
358: * When finding an unescaped newline character,
359: * leave the character loop to process the line.
360: * Skip a preceding carriage return, if any.
361: */
362:
1.95 schwarze 363: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 364: '\n' == blk.buf[i + 1])
365: ++i;
366: if ('\n' == blk.buf[i]) {
367: ++i;
368: ++lnn;
369: break;
370: }
371:
1.35 schwarze 372: /*
1.93 schwarze 373: * Make sure we have space for the worst
374: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 375: */
376:
1.95 schwarze 377: if (pos + 11 > ln.sz)
1.35 schwarze 378: resize_buf(&ln, 256);
379:
1.47 schwarze 380: /*
1.93 schwarze 381: * Encode 8-bit input.
1.1 kristaps 382: */
383:
1.93 schwarze 384: c = blk.buf[i];
385: if (c & 0x80) {
1.95 schwarze 386: if ( ! (curp->filenc && preconv_encode(
387: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 388: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
389: curp->line, pos, "0x%x", c);
1.93 schwarze 390: ln.buf[pos++] = '?';
391: i++;
392: }
393: continue;
394: }
395:
396: /*
397: * Exclude control characters.
398: */
1.1 kristaps 399:
1.93 schwarze 400: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 401: mandoc_vmsg(c == 0x00 || c == 0x04 ||
402: c > 0x0a ? MANDOCERR_CHAR_BAD :
403: MANDOCERR_CHAR_UNSUPP,
404: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 405: i++;
1.127 schwarze 406: if (c != '\r')
407: ln.buf[pos++] = '?';
1.1 kristaps 408: continue;
409: }
410:
411: ln.buf[pos++] = blk.buf[i++];
412: }
413:
1.170 schwarze 414: if (pos + 1 >= ln.sz)
1.1 kristaps 415: resize_buf(&ln, 256);
416:
1.170 schwarze 417: if (i == blk.sz || blk.buf[i] == '\0')
418: ln.buf[pos++] = '\n';
1.1 kristaps 419: ln.buf[pos] = '\0';
420:
421: /*
422: * A significant amount of complexity is contained by
423: * the roff preprocessor. It's line-oriented but can be
424: * expressed on one line, so we need at times to
425: * readjust our starting point and re-run it. The roff
426: * preprocessor can also readjust the buffers with new
427: * data, so we pass them in wholesale.
428: */
429:
430: of = 0;
431:
1.24 kristaps 432: /*
433: * Maintain a lookaside buffer of all parsed lines. We
434: * only do this if mparse_keep() has been invoked (the
435: * buffer may be accessed with mparse_getkeep()).
436: */
437:
438: if (curp->secondary) {
1.47 schwarze 439: curp->secondary->buf = mandoc_realloc(
440: curp->secondary->buf,
441: curp->secondary->sz + pos + 2);
442: memcpy(curp->secondary->buf +
443: curp->secondary->sz,
444: ln.buf, pos);
1.24 kristaps 445: curp->secondary->sz += pos;
446: curp->secondary->buf
447: [curp->secondary->sz] = '\n';
448: curp->secondary->sz++;
449: curp->secondary->buf
450: [curp->secondary->sz] = '\0';
451: }
1.1 kristaps 452: rerun:
1.96 schwarze 453: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 454:
455: switch (rr) {
1.47 schwarze 456: case ROFF_REPARSE:
1.163 schwarze 457: if (++curp->reparse_count > REPARSE_LIMIT)
1.3 kristaps 458: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 459: curp->line, pos, NULL);
1.163 schwarze 460: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
461: start == 1) {
462: pos = 0;
463: continue;
464: }
465: free(ln.buf);
466: return 0;
1.47 schwarze 467: case ROFF_APPEND:
1.95 schwarze 468: pos = strlen(ln.buf);
1.1 kristaps 469: continue;
1.47 schwarze 470: case ROFF_RERUN:
1.1 kristaps 471: goto rerun;
1.47 schwarze 472: case ROFF_IGN:
1.1 kristaps 473: pos = 0;
474: continue;
1.47 schwarze 475: case ROFF_SO:
1.95 schwarze 476: if ( ! (curp->options & MPARSE_SO) &&
477: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 478: curp->sodest = mandoc_strdup(ln.buf + of);
479: free(ln.buf);
1.163 schwarze 480: return 1;
1.45 schwarze 481: }
1.24 kristaps 482: /*
483: * We remove `so' clauses from our lookaside
484: * buffer because we're going to descend into
485: * the file recursively.
486: */
1.47 schwarze 487: if (curp->secondary)
1.25 kristaps 488: curp->secondary->sz -= pos + 1;
1.113 schwarze 489: save_file = curp->file;
1.148 schwarze 490: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 491: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 492: close(fd);
1.113 schwarze 493: curp->file = save_file;
494: } else {
495: curp->file = save_file;
1.52 schwarze 496: mandoc_vmsg(MANDOCERR_SO_FAIL,
497: curp, curp->line, pos,
498: ".so %s", ln.buf + of);
1.110 schwarze 499: ln.sz = mandoc_asprintf(&cp,
500: ".sp\nSee the file %s.\n.sp",
501: ln.buf + of);
502: free(ln.buf);
503: ln.buf = cp;
504: of = 0;
505: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 506: }
1.1 kristaps 507: pos = 0;
508: continue;
509: default:
510: break;
511: }
512:
1.150 schwarze 513: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 514: choose_parser(curp);
1.1 kristaps 515:
1.47 schwarze 516: /*
1.84 schwarze 517: * Lastly, push down into the parsers themselves.
1.1 kristaps 518: * If libroff returns ROFF_TBL, then add it to the
519: * currently open parse. Since we only get here if
520: * there does exist data (see tbl_data.c), we're
521: * guaranteed that something's been allocated.
522: * Do the same for ROFF_EQN.
523: */
524:
1.139 schwarze 525: if (rr == ROFF_TBL)
1.100 schwarze 526: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 527: roff_addtbl(curp->man, span);
528: else if (rr == ROFF_EQN)
529: roff_addeqn(curp->man, roff_eqn(curp->roff));
530: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 531: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 532: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
533: break;
1.1 kristaps 534:
535: /* Temporary buffers typically are not full. */
536:
537: if (0 == start && '\0' == blk.buf[i])
538: break;
539:
540: /* Start the next input line. */
541:
542: pos = 0;
543: }
544:
545: free(ln.buf);
1.163 schwarze 546: return 1;
1.1 kristaps 547: }
548:
549: static int
1.40 schwarze 550: read_whole_file(struct mparse *curp, const char *file, int fd,
551: struct buf *fb, int *with_mmap)
1.1 kristaps 552: {
1.161 schwarze 553: struct stat st;
1.140 schwarze 554: gzFile gz;
1.1 kristaps 555: size_t off;
556: ssize_t ssz;
1.143 schwarze 557:
558: if (fstat(fd, &st) == -1)
559: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 560:
561: /*
562: * If we're a regular file, try just reading in the whole entry
563: * via mmap(). This is faster than reading it into blocks, and
564: * since each file is only a few bytes to begin with, I'm not
565: * concerned that this is going to tank any machines.
566: */
567:
1.140 schwarze 568: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 569: if (st.st_size > 0x7fffffff) {
1.111 schwarze 570: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 571: return 0;
1.1 kristaps 572: }
573: *with_mmap = 1;
574: fb->sz = (size_t)st.st_size;
1.37 schwarze 575: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 576: if (fb->buf != MAP_FAILED)
1.142 schwarze 577: return 1;
1.1 kristaps 578: }
579:
1.140 schwarze 580: if (curp->gzip) {
1.143 schwarze 581: if ((gz = gzdopen(fd, "rb")) == NULL)
582: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 583: } else
584: gz = NULL;
585:
1.1 kristaps 586: /*
587: * If this isn't a regular file (like, say, stdin), then we must
588: * go the old way and just read things in bit by bit.
589: */
590:
591: *with_mmap = 0;
592: off = 0;
593: fb->sz = 0;
594: fb->buf = NULL;
595: for (;;) {
596: if (off == fb->sz) {
597: if (fb->sz == (1U << 31)) {
1.111 schwarze 598: mandoc_msg(MANDOCERR_TOOLARGE, curp,
599: 0, 0, NULL);
1.1 kristaps 600: break;
601: }
602: resize_buf(fb, 65536);
603: }
1.140 schwarze 604: ssz = curp->gzip ?
605: gzread(gz, fb->buf + (int)off, fb->sz - off) :
606: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 607: if (ssz == 0) {
608: fb->sz = off;
1.142 schwarze 609: return 1;
1.1 kristaps 610: }
1.143 schwarze 611: if (ssz == -1)
612: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 613: off += (size_t)ssz;
614: }
615:
616: free(fb->buf);
617: fb->buf = NULL;
1.142 schwarze 618: return 0;
1.1 kristaps 619: }
620:
621: static void
622: mparse_end(struct mparse *curp)
623: {
1.135 schwarze 624: if (curp->man->macroset == MACROSET_NONE)
625: curp->man->macroset = MACROSET_MAN;
626: if (curp->man->macroset == MACROSET_MDOC)
627: mdoc_endparse(curp->man);
628: else
1.111 schwarze 629: man_endparse(curp->man);
1.1 kristaps 630: roff_endparse(curp->roff);
631: }
632:
633: static void
1.36 schwarze 634: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 635: {
1.85 schwarze 636: struct buf *svprimary;
1.28 joerg 637: const char *svfile;
1.95 schwarze 638: size_t offset;
1.36 schwarze 639: static int recursion_depth;
640:
641: if (64 < recursion_depth) {
642: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
643: return;
644: }
1.28 joerg 645:
646: /* Line number is per-file. */
647: svfile = curp->file;
648: curp->file = file;
1.85 schwarze 649: svprimary = curp->primary;
1.83 schwarze 650: curp->primary = &blk;
1.28 joerg 651: curp->line = 1;
1.36 schwarze 652: recursion_depth++;
1.28 joerg 653:
1.93 schwarze 654: /* Skip an UTF-8 byte order mark. */
655: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
656: (unsigned char)blk.buf[0] == 0xef &&
657: (unsigned char)blk.buf[1] == 0xbb &&
658: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 659: offset = 3;
1.93 schwarze 660: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 661: } else
662: offset = 0;
1.93 schwarze 663:
1.95 schwarze 664: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 665:
1.111 schwarze 666: if (--recursion_depth == 0)
1.28 joerg 667: mparse_end(curp);
668:
1.85 schwarze 669: curp->primary = svprimary;
1.28 joerg 670: curp->file = svfile;
671: }
672:
673: enum mandoclevel
1.104 schwarze 674: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 675: const char *file)
676: {
677: struct buf blk;
678:
1.104 schwarze 679: blk.buf = buf;
1.28 joerg 680: blk.sz = len;
681:
1.36 schwarze 682: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 683: return curp->file_status;
1.28 joerg 684: }
685:
1.98 schwarze 686: /*
687: * Read the whole file into memory and call the parsers.
688: * Called recursively when an .so request is encountered.
689: */
1.36 schwarze 690: enum mandoclevel
691: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 692: {
1.28 joerg 693: struct buf blk;
694: int with_mmap;
1.93 schwarze 695: int save_filenc;
1.1 kristaps 696:
1.91 schwarze 697: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 698: save_filenc = curp->filenc;
699: curp->filenc = curp->options &
700: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 701: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 702: curp->filenc = save_filenc;
1.91 schwarze 703: if (with_mmap)
704: munmap(blk.buf, blk.sz);
705: else
706: free(blk.buf);
707: }
1.142 schwarze 708: return curp->file_status;
1.82 schwarze 709: }
710:
1.148 schwarze 711: int
712: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 713: {
714: char *cp;
1.148 schwarze 715: int fd;
1.82 schwarze 716:
717: curp->file = file;
1.140 schwarze 718: cp = strrchr(file, '.');
719: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 720:
1.140 schwarze 721: /* First try to use the filename as it is. */
1.98 schwarze 722:
1.148 schwarze 723: if ((fd = open(file, O_RDONLY)) != -1)
724: return fd;
1.98 schwarze 725:
1.140 schwarze 726: /*
727: * If that doesn't work and the filename doesn't
728: * already end in .gz, try appending .gz.
729: */
1.98 schwarze 730:
1.140 schwarze 731: if ( ! curp->gzip) {
1.98 schwarze 732: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 733: fd = open(cp, O_RDONLY);
1.108 schwarze 734: free(cp);
1.148 schwarze 735: if (fd != -1) {
1.140 schwarze 736: curp->gzip = 1;
1.148 schwarze 737: return fd;
1.82 schwarze 738: }
739: }
740:
1.140 schwarze 741: /* Neither worked, give up. */
1.97 schwarze 742:
1.140 schwarze 743: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 744: return -1;
1.1 kristaps 745: }
746:
747: struct mparse *
1.94 schwarze 748: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.144 schwarze 749: const char *defos)
1.1 kristaps 750: {
751: struct mparse *curp;
1.10 kristaps 752:
1.1 kristaps 753: curp = mandoc_calloc(1, sizeof(struct mparse));
754:
1.44 schwarze 755: curp->options = options;
1.3 kristaps 756: curp->wlevel = wlevel;
1.1 kristaps 757: curp->mmsg = mmsg;
1.29 schwarze 758: curp->defos = defos;
1.1 kristaps 759:
1.144 schwarze 760: curp->roff = roff_alloc(curp, options);
1.137 schwarze 761: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
762: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 763: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 764: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 765: if (curp->man->mdocmac == NULL)
766: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 767: } else if (curp->options & MPARSE_MAN) {
768: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 769: if (curp->man->manmac == NULL)
770: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 771: }
1.138 schwarze 772: curp->man->first->tok = TOKEN_NONE;
1.142 schwarze 773: return curp;
1.1 kristaps 774: }
775:
776: void
777: mparse_reset(struct mparse *curp)
778: {
779: roff_reset(curp->roff);
1.150 schwarze 780: roff_man_reset(curp->man);
1.160 schwarze 781:
782: free(curp->sodest);
783: curp->sodest = NULL;
784:
1.24 kristaps 785: if (curp->secondary)
786: curp->secondary->sz = 0;
1.1 kristaps 787:
788: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 789: curp->gzip = 0;
1.1 kristaps 790: }
791:
792: void
793: mparse_free(struct mparse *curp)
794: {
795:
1.164 schwarze 796: roffhash_free(curp->man->mdocmac);
797: roffhash_free(curp->man->manmac);
1.137 schwarze 798: roff_man_free(curp->man);
1.160 schwarze 799: roff_free(curp->roff);
1.24 kristaps 800: if (curp->secondary)
801: free(curp->secondary->buf);
1.1 kristaps 802:
1.24 kristaps 803: free(curp->secondary);
1.45 schwarze 804: free(curp->sodest);
1.1 kristaps 805: free(curp);
806: }
807:
808: void
1.135 schwarze 809: mparse_result(struct mparse *curp, struct roff_man **man,
810: char **sodest)
1.1 kristaps 811: {
812:
1.45 schwarze 813: if (sodest && NULL != (*sodest = curp->sodest)) {
814: *man = NULL;
815: return;
816: }
1.9 kristaps 817: if (man)
818: *man = curp->man;
1.157 schwarze 819: }
820:
821: void
822: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
823: {
824: if (curp->file_status > *rc)
825: *rc = curp->file_status;
1.3 kristaps 826: }
827:
828: void
829: mandoc_vmsg(enum mandocerr t, struct mparse *m,
830: int ln, int pos, const char *fmt, ...)
831: {
832: char buf[256];
833: va_list ap;
834:
835: va_start(ap, fmt);
1.48 schwarze 836: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 837: va_end(ap);
838:
839: mandoc_msg(t, m, ln, pos, buf);
840: }
841:
842: void
1.47 schwarze 843: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 844: int ln, int col, const char *msg)
845: {
846: enum mandoclevel level;
847:
1.112 schwarze 848: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 849: while (er < mandoclimits[level])
850: level--;
851:
1.108 schwarze 852: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 853: return;
854:
1.8 kristaps 855: if (m->mmsg)
856: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 857:
858: if (m->file_status < level)
859: m->file_status = level;
1.7 kristaps 860: }
861:
862: const char *
863: mparse_strerror(enum mandocerr er)
864: {
865:
1.142 schwarze 866: return mandocerrs[er];
1.7 kristaps 867: }
868:
869: const char *
870: mparse_strlevel(enum mandoclevel lvl)
871: {
1.142 schwarze 872: return mandoclevels[lvl];
1.24 kristaps 873: }
874:
875: void
876: mparse_keep(struct mparse *p)
877: {
878:
879: assert(NULL == p->secondary);
880: p->secondary = mandoc_calloc(1, sizeof(struct buf));
881: }
882:
883: const char *
884: mparse_getkeep(const struct mparse *p)
885: {
886:
887: assert(p->secondary);
1.142 schwarze 888: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 889: }
CVSweb