Annotation of mandoc/read.c, Revision 1.167
1.167 ! schwarze 1: /* $Id: read.c,v 1.166 2017/05/30 19:30:39 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",
1.166 schwarze 93:
1.7 kristaps 94: "generic warning",
95:
96: /* related to the prologue */
1.79 schwarze 97: "missing manual title, using UNTITLED",
98: "missing manual title, using \"\"",
1.54 schwarze 99: "lower case character in document title",
1.79 schwarze 100: "missing manual section, using \"\"",
1.7 kristaps 101: "unknown manual section",
1.54 schwarze 102: "missing date, using today's date",
1.7 kristaps 103: "cannot parse date, using it verbatim",
1.79 schwarze 104: "missing Os macro, using \"\"",
105: "duplicate prologue macro",
106: "late prologue macro",
107: "skipping late title macro",
1.7 kristaps 108: "prologue macros out of order",
109:
110: /* related to document structure */
111: ".so is fragile, better use ln(1)",
1.50 schwarze 112: "no document body",
1.54 schwarze 113: "content before first section header",
114: "first section is not \"NAME\"",
1.156 schwarze 115: "NAME section without Nm before Nd",
1.128 schwarze 116: "NAME section without description",
117: "description not at the end of NAME",
118: "bad NAME section content",
1.156 schwarze 119: "missing comma before name",
1.121 schwarze 120: "missing description line, using \"\"",
1.162 schwarze 121: "description line outside NAME section",
1.7 kristaps 122: "sections out of conventional order",
1.54 schwarze 123: "duplicate section title",
124: "unexpected section",
1.87 schwarze 125: "unusual Xr order",
126: "unusual Xr punctuation",
1.86 schwarze 127: "AUTHORS section without An macro",
1.7 kristaps 128:
129: /* related to macros and nesting */
1.55 schwarze 130: "obsolete macro",
1.102 schwarze 131: "macro neither callable nor escaped",
1.7 kristaps 132: "skipping paragraph macro",
1.31 schwarze 133: "moving paragraph macro out of list",
1.7 kristaps 134: "skipping no-space macro",
135: "blocks badly nested",
136: "nested displays are not portable",
1.57 schwarze 137: "moving content out of list",
1.78 schwarze 138: "fill mode already enabled, skipping",
139: "fill mode already disabled, skipping",
1.7 kristaps 140: "line scope broken",
141:
142: /* related to missing macro arguments */
1.58 schwarze 143: "skipping empty request",
144: "conditional request controls empty scope",
1.7 kristaps 145: "skipping empty macro",
1.124 schwarze 146: "empty block",
1.62 schwarze 147: "empty argument, using 0n",
1.60 schwarze 148: "missing display type, using -ragged",
149: "list type is not the first argument",
1.152 schwarze 150: "missing -width in -tag list, using 6n",
1.78 schwarze 151: "missing utility name, using \"\"",
1.123 schwarze 152: "missing function name, using \"\"",
1.60 schwarze 153: "empty head in list item",
154: "empty list item",
1.61 schwarze 155: "missing font type, using \\fR",
156: "unknown font type, using \\fR",
1.103 schwarze 157: "nothing follows prefix",
1.122 schwarze 158: "empty reference block",
1.155 schwarze 159: "missing section argument",
1.60 schwarze 160: "missing -std argument, adding it",
1.125 schwarze 161: "missing option string, using \"\"",
162: "missing resource identifier, using \"\"",
1.90 schwarze 163: "missing eqn box, using \"\"",
1.7 kristaps 164:
165: /* related to bad macro arguments */
1.64 schwarze 166: "unterminated quoted argument",
1.7 kristaps 167: "duplicate argument",
1.76 schwarze 168: "skipping duplicate argument",
1.63 schwarze 169: "skipping duplicate display type",
170: "skipping duplicate list type",
1.76 schwarze 171: "skipping -width argument",
1.126 schwarze 172: "wrong number of cells",
1.7 kristaps 173: "unknown AT&T UNIX version",
1.88 schwarze 174: "comma in function argument",
1.89 schwarze 175: "parenthesis in function name",
1.67 schwarze 176: "invalid content in Rs block",
1.63 schwarze 177: "invalid Boolean argument",
178: "unknown font, skipping request",
1.126 schwarze 179: "odd number of characters in request",
1.7 kristaps 180:
181: /* related to plain text */
1.64 schwarze 182: "blank line in fill mode, using .sp",
183: "tab in filled text",
184: "whitespace at end of input line",
1.158 schwarze 185: "new sentence, new line",
1.7 kristaps 186: "bad comment style",
1.64 schwarze 187: "invalid escape sequence",
188: "undefined string, using \"\"",
1.16 kristaps 189:
1.118 schwarze 190: /* related to tables */
191: "tbl line starts with span",
192: "tbl column starts with span",
193: "skipping vertical bar in tbl layout",
194:
1.7 kristaps 195: "generic error",
196:
197: /* related to tables */
1.116 schwarze 198: "non-alphabetic character in tbl options",
199: "skipping unknown tbl option",
200: "missing tbl option argument",
201: "wrong tbl option argument size",
1.118 schwarze 202: "empty tbl layout",
203: "invalid character in tbl layout",
204: "unmatched parenthesis in tbl layout",
1.119 schwarze 205: "tbl without any data cells",
206: "ignoring data in spanned tbl cell",
207: "ignoring extra tbl data cells",
208: "data block open at end of tbl",
1.7 kristaps 209:
1.68 schwarze 210: /* related to document structure and macros */
1.108 schwarze 211: NULL,
1.7 kristaps 212: "input stack limit exceeded, infinite loop?",
213: "skipping bad character",
1.68 schwarze 214: "skipping unknown macro",
1.112 schwarze 215: "skipping insecure request",
1.73 schwarze 216: "skipping item outside list",
1.68 schwarze 217: "skipping column outside column list",
218: "skipping end of block that is not open",
1.115 schwarze 219: "fewer RS blocks open, skipping",
1.68 schwarze 220: "inserting missing end of block",
221: "appending missing end of block",
222:
223: /* related to request and macro arguments */
1.7 kristaps 224: "escaped character not allowed in a name",
1.99 schwarze 225: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 226: "skipping display without arguments",
1.71 schwarze 227: "missing list type, using -item",
1.70 schwarze 228: "missing manual name, using \"\"",
1.71 schwarze 229: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 230: "unknown standard specifier",
1.71 schwarze 231: "skipping request without numeric argument",
1.109 schwarze 232: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 233: ".so request failed",
1.61 schwarze 234: "skipping all arguments",
235: "skipping excess arguments",
1.92 kristaps 236: "divide by zero",
1.112 schwarze 237:
238: "unsupported feature",
239: "input too large",
1.114 schwarze 240: "unsupported control character",
1.112 schwarze 241: "unsupported roff request",
1.119 schwarze 242: "eqn delim option in tbl",
1.118 schwarze 243: "unsupported tbl layout modifier",
1.112 schwarze 244: "ignoring macro in table",
1.7 kristaps 245: };
246:
247: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
248: "SUCCESS",
1.165 schwarze 249: "STYLE",
1.7 kristaps 250: "WARNING",
251: "ERROR",
1.112 schwarze 252: "UNSUPP",
1.7 kristaps 253: "BADARG",
254: "SYSERR"
255: };
256:
1.47 schwarze 257:
1.1 kristaps 258: static void
259: resize_buf(struct buf *buf, size_t initial)
260: {
261:
262: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
263: buf->buf = mandoc_realloc(buf->buf, buf->sz);
264: }
265:
266: static void
1.84 schwarze 267: choose_parser(struct mparse *curp)
1.1 kristaps 268: {
1.83 schwarze 269: char *cp, *ep;
270: int format;
1.1 kristaps 271:
1.83 schwarze 272: /*
273: * If neither command line arguments -mdoc or -man select
274: * a parser nor the roff parser found a .Dd or .TH macro
275: * yet, look ahead in the main input buffer.
276: */
277:
278: if ((format = roff_getformat(curp->roff)) == 0) {
279: cp = curp->primary->buf;
280: ep = cp + curp->primary->sz;
281: while (cp < ep) {
1.85 schwarze 282: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 283: cp++;
284: if (cp[0] == 'D' && cp[1] == 'd') {
285: format = MPARSE_MDOC;
286: break;
287: }
288: if (cp[0] == 'T' && cp[1] == 'H') {
289: format = MPARSE_MAN;
290: break;
291: }
292: }
293: cp = memchr(cp, '\n', ep - cp);
294: if (cp == NULL)
295: break;
296: cp++;
297: }
1.1 kristaps 298: }
299:
1.83 schwarze 300: if (format == MPARSE_MDOC) {
1.137 schwarze 301: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 302: if (curp->man->mdocmac == NULL)
303: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 304: } else {
305: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 306: if (curp->man->manmac == NULL)
307: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 308: }
1.164 schwarze 309: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 310: }
311:
312: /*
1.95 schwarze 313: * Main parse routine for a buffer.
314: * It assumes encoding and line numbering are already set up.
315: * It can recurse directly (for invocations of user-defined
316: * macros, inline equations, and input line traps)
317: * and indirectly (for .so file inclusion).
1.1 kristaps 318: */
1.163 schwarze 319: static int
1.95 schwarze 320: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 321: {
322: const struct tbl_span *span;
323: struct buf ln;
1.113 schwarze 324: const char *save_file;
1.110 schwarze 325: char *cp;
1.95 schwarze 326: size_t pos; /* byte number in the ln buffer */
1.154 schwarze 327: size_t j; /* auxiliary byte number in the blk buffer */
1.1 kristaps 328: enum rofferr rr;
1.100 schwarze 329: int of;
1.1 kristaps 330: int lnn; /* line number in the real file */
1.110 schwarze 331: int fd;
1.1 kristaps 332: unsigned char c;
333:
1.95 schwarze 334: memset(&ln, 0, sizeof(ln));
1.1 kristaps 335:
1.47 schwarze 336: lnn = curp->line;
337: pos = 0;
1.1 kristaps 338:
1.95 schwarze 339: while (i < blk.sz) {
1.1 kristaps 340: if (0 == pos && '\0' == blk.buf[i])
341: break;
342:
343: if (start) {
344: curp->line = lnn;
345: curp->reparse_count = 0;
1.93 schwarze 346:
347: if (lnn < 3 &&
348: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 349: curp->filenc & MPARSE_LATIN1)
350: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 351: }
352:
1.95 schwarze 353: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 354:
355: /*
356: * When finding an unescaped newline character,
357: * leave the character loop to process the line.
358: * Skip a preceding carriage return, if any.
359: */
360:
1.95 schwarze 361: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 362: '\n' == blk.buf[i + 1])
363: ++i;
364: if ('\n' == blk.buf[i]) {
365: ++i;
366: ++lnn;
367: break;
368: }
369:
1.35 schwarze 370: /*
1.93 schwarze 371: * Make sure we have space for the worst
372: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 373: */
374:
1.95 schwarze 375: if (pos + 11 > ln.sz)
1.35 schwarze 376: resize_buf(&ln, 256);
377:
1.47 schwarze 378: /*
1.93 schwarze 379: * Encode 8-bit input.
1.1 kristaps 380: */
381:
1.93 schwarze 382: c = blk.buf[i];
383: if (c & 0x80) {
1.95 schwarze 384: if ( ! (curp->filenc && preconv_encode(
385: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 386: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
387: curp->line, pos, "0x%x", c);
1.93 schwarze 388: ln.buf[pos++] = '?';
389: i++;
390: }
391: continue;
392: }
393:
394: /*
395: * Exclude control characters.
396: */
1.1 kristaps 397:
1.93 schwarze 398: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 399: mandoc_vmsg(c == 0x00 || c == 0x04 ||
400: c > 0x0a ? MANDOCERR_CHAR_BAD :
401: MANDOCERR_CHAR_UNSUPP,
402: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 403: i++;
1.127 schwarze 404: if (c != '\r')
405: ln.buf[pos++] = '?';
1.1 kristaps 406: continue;
407: }
408:
409: /* Trailing backslash = a plain char. */
410:
1.95 schwarze 411: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 412: ln.buf[pos++] = blk.buf[i++];
413: continue;
414: }
415:
416: /*
417: * Found escape and at least one other character.
418: * When it's a newline character, skip it.
419: * When there is a carriage return in between,
420: * skip that one as well.
421: */
422:
1.95 schwarze 423: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 424: '\n' == blk.buf[i + 2])
425: ++i;
426: if ('\n' == blk.buf[i + 1]) {
427: i += 2;
428: ++lnn;
429: continue;
430: }
431:
1.13 kristaps 432: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.154 schwarze 433: j = i;
1.1 kristaps 434: i += 2;
435: /* Comment, skip to end of line */
1.95 schwarze 436: for (; i < blk.sz; ++i) {
1.153 schwarze 437: if (blk.buf[i] != '\n')
438: continue;
439: if (blk.buf[i - 1] == ' ' ||
440: blk.buf[i - 1] == '\t')
441: mandoc_msg(
442: MANDOCERR_SPACE_EOL,
443: curp, curp->line,
1.154 schwarze 444: pos + i-1 - j, NULL);
1.153 schwarze 445: ++i;
446: ++lnn;
447: break;
1.1 kristaps 448: }
449:
450: /* Backout trailing whitespaces */
451: for (; pos > 0; --pos) {
452: if (ln.buf[pos - 1] != ' ')
453: break;
454: if (pos > 2 && ln.buf[pos - 2] == '\\')
455: break;
456: }
457: break;
458: }
459:
1.35 schwarze 460: /* Catch escaped bogus characters. */
461:
462: c = (unsigned char) blk.buf[i+1];
463:
1.47 schwarze 464: if ( ! (isascii(c) &&
465: (isgraph(c) || isblank(c)))) {
1.114 schwarze 466: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 467: curp->line, pos, "0x%x", c);
1.35 schwarze 468: i += 2;
469: ln.buf[pos++] = '?';
470: continue;
471: }
472:
1.1 kristaps 473: /* Some other escape sequence, copy & cont. */
474:
475: ln.buf[pos++] = blk.buf[i++];
476: ln.buf[pos++] = blk.buf[i++];
477: }
478:
1.95 schwarze 479: if (pos >= ln.sz)
1.1 kristaps 480: resize_buf(&ln, 256);
481:
482: ln.buf[pos] = '\0';
483:
484: /*
485: * A significant amount of complexity is contained by
486: * the roff preprocessor. It's line-oriented but can be
487: * expressed on one line, so we need at times to
488: * readjust our starting point and re-run it. The roff
489: * preprocessor can also readjust the buffers with new
490: * data, so we pass them in wholesale.
491: */
492:
493: of = 0;
494:
1.24 kristaps 495: /*
496: * Maintain a lookaside buffer of all parsed lines. We
497: * only do this if mparse_keep() has been invoked (the
498: * buffer may be accessed with mparse_getkeep()).
499: */
500:
501: if (curp->secondary) {
1.47 schwarze 502: curp->secondary->buf = mandoc_realloc(
503: curp->secondary->buf,
504: curp->secondary->sz + pos + 2);
505: memcpy(curp->secondary->buf +
506: curp->secondary->sz,
507: ln.buf, pos);
1.24 kristaps 508: curp->secondary->sz += pos;
509: curp->secondary->buf
510: [curp->secondary->sz] = '\n';
511: curp->secondary->sz++;
512: curp->secondary->buf
513: [curp->secondary->sz] = '\0';
514: }
1.1 kristaps 515: rerun:
1.96 schwarze 516: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 517:
518: switch (rr) {
1.47 schwarze 519: case ROFF_REPARSE:
1.163 schwarze 520: if (++curp->reparse_count > REPARSE_LIMIT)
1.3 kristaps 521: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 522: curp->line, pos, NULL);
1.163 schwarze 523: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
524: start == 1) {
525: pos = 0;
526: continue;
527: }
528: free(ln.buf);
529: return 0;
1.47 schwarze 530: case ROFF_APPEND:
1.95 schwarze 531: pos = strlen(ln.buf);
1.1 kristaps 532: continue;
1.47 schwarze 533: case ROFF_RERUN:
1.1 kristaps 534: goto rerun;
1.47 schwarze 535: case ROFF_IGN:
1.1 kristaps 536: pos = 0;
537: continue;
1.47 schwarze 538: case ROFF_SO:
1.95 schwarze 539: if ( ! (curp->options & MPARSE_SO) &&
540: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 541: curp->sodest = mandoc_strdup(ln.buf + of);
542: free(ln.buf);
1.163 schwarze 543: return 1;
1.45 schwarze 544: }
1.24 kristaps 545: /*
546: * We remove `so' clauses from our lookaside
547: * buffer because we're going to descend into
548: * the file recursively.
549: */
1.47 schwarze 550: if (curp->secondary)
1.25 kristaps 551: curp->secondary->sz -= pos + 1;
1.113 schwarze 552: save_file = curp->file;
1.148 schwarze 553: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 554: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 555: close(fd);
1.113 schwarze 556: curp->file = save_file;
557: } else {
558: curp->file = save_file;
1.52 schwarze 559: mandoc_vmsg(MANDOCERR_SO_FAIL,
560: curp, curp->line, pos,
561: ".so %s", ln.buf + of);
1.110 schwarze 562: ln.sz = mandoc_asprintf(&cp,
563: ".sp\nSee the file %s.\n.sp",
564: ln.buf + of);
565: free(ln.buf);
566: ln.buf = cp;
567: of = 0;
568: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 569: }
1.1 kristaps 570: pos = 0;
571: continue;
572: default:
573: break;
574: }
575:
1.150 schwarze 576: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 577: choose_parser(curp);
1.1 kristaps 578:
1.47 schwarze 579: /*
1.84 schwarze 580: * Lastly, push down into the parsers themselves.
1.1 kristaps 581: * If libroff returns ROFF_TBL, then add it to the
582: * currently open parse. Since we only get here if
583: * there does exist data (see tbl_data.c), we're
584: * guaranteed that something's been allocated.
585: * Do the same for ROFF_EQN.
586: */
587:
1.139 schwarze 588: if (rr == ROFF_TBL)
1.100 schwarze 589: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 590: roff_addtbl(curp->man, span);
591: else if (rr == ROFF_EQN)
592: roff_addeqn(curp->man, roff_eqn(curp->roff));
593: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 594: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 595: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
596: break;
1.1 kristaps 597:
598: /* Temporary buffers typically are not full. */
599:
600: if (0 == start && '\0' == blk.buf[i])
601: break;
602:
603: /* Start the next input line. */
604:
605: pos = 0;
606: }
607:
608: free(ln.buf);
1.163 schwarze 609: return 1;
1.1 kristaps 610: }
611:
612: static int
1.40 schwarze 613: read_whole_file(struct mparse *curp, const char *file, int fd,
614: struct buf *fb, int *with_mmap)
1.1 kristaps 615: {
1.161 schwarze 616: struct stat st;
1.140 schwarze 617: gzFile gz;
1.1 kristaps 618: size_t off;
619: ssize_t ssz;
1.143 schwarze 620:
621: if (fstat(fd, &st) == -1)
622: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 623:
624: /*
625: * If we're a regular file, try just reading in the whole entry
626: * via mmap(). This is faster than reading it into blocks, and
627: * since each file is only a few bytes to begin with, I'm not
628: * concerned that this is going to tank any machines.
629: */
630:
1.140 schwarze 631: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 632: if (st.st_size > 0x7fffffff) {
1.111 schwarze 633: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 634: return 0;
1.1 kristaps 635: }
636: *with_mmap = 1;
637: fb->sz = (size_t)st.st_size;
1.37 schwarze 638: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 639: if (fb->buf != MAP_FAILED)
1.142 schwarze 640: return 1;
1.1 kristaps 641: }
642:
1.140 schwarze 643: if (curp->gzip) {
1.143 schwarze 644: if ((gz = gzdopen(fd, "rb")) == NULL)
645: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 646: } else
647: gz = NULL;
648:
1.1 kristaps 649: /*
650: * If this isn't a regular file (like, say, stdin), then we must
651: * go the old way and just read things in bit by bit.
652: */
653:
654: *with_mmap = 0;
655: off = 0;
656: fb->sz = 0;
657: fb->buf = NULL;
658: for (;;) {
659: if (off == fb->sz) {
660: if (fb->sz == (1U << 31)) {
1.111 schwarze 661: mandoc_msg(MANDOCERR_TOOLARGE, curp,
662: 0, 0, NULL);
1.1 kristaps 663: break;
664: }
665: resize_buf(fb, 65536);
666: }
1.140 schwarze 667: ssz = curp->gzip ?
668: gzread(gz, fb->buf + (int)off, fb->sz - off) :
669: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 670: if (ssz == 0) {
671: fb->sz = off;
1.142 schwarze 672: return 1;
1.1 kristaps 673: }
1.143 schwarze 674: if (ssz == -1)
675: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 676: off += (size_t)ssz;
677: }
678:
679: free(fb->buf);
680: fb->buf = NULL;
1.142 schwarze 681: return 0;
1.1 kristaps 682: }
683:
684: static void
685: mparse_end(struct mparse *curp)
686: {
1.135 schwarze 687: if (curp->man->macroset == MACROSET_NONE)
688: curp->man->macroset = MACROSET_MAN;
689: if (curp->man->macroset == MACROSET_MDOC)
690: mdoc_endparse(curp->man);
691: else
1.111 schwarze 692: man_endparse(curp->man);
1.1 kristaps 693: roff_endparse(curp->roff);
694: }
695:
696: static void
1.36 schwarze 697: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 698: {
1.85 schwarze 699: struct buf *svprimary;
1.28 joerg 700: const char *svfile;
1.95 schwarze 701: size_t offset;
1.36 schwarze 702: static int recursion_depth;
703:
704: if (64 < recursion_depth) {
705: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
706: return;
707: }
1.28 joerg 708:
709: /* Line number is per-file. */
710: svfile = curp->file;
711: curp->file = file;
1.85 schwarze 712: svprimary = curp->primary;
1.83 schwarze 713: curp->primary = &blk;
1.28 joerg 714: curp->line = 1;
1.36 schwarze 715: recursion_depth++;
1.28 joerg 716:
1.93 schwarze 717: /* Skip an UTF-8 byte order mark. */
718: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
719: (unsigned char)blk.buf[0] == 0xef &&
720: (unsigned char)blk.buf[1] == 0xbb &&
721: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 722: offset = 3;
1.93 schwarze 723: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 724: } else
725: offset = 0;
1.93 schwarze 726:
1.95 schwarze 727: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 728:
1.111 schwarze 729: if (--recursion_depth == 0)
1.28 joerg 730: mparse_end(curp);
731:
1.85 schwarze 732: curp->primary = svprimary;
1.28 joerg 733: curp->file = svfile;
734: }
735:
736: enum mandoclevel
1.104 schwarze 737: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 738: const char *file)
739: {
740: struct buf blk;
741:
1.104 schwarze 742: blk.buf = buf;
1.28 joerg 743: blk.sz = len;
744:
1.36 schwarze 745: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 746: return curp->file_status;
1.28 joerg 747: }
748:
1.98 schwarze 749: /*
750: * Read the whole file into memory and call the parsers.
751: * Called recursively when an .so request is encountered.
752: */
1.36 schwarze 753: enum mandoclevel
754: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 755: {
1.28 joerg 756: struct buf blk;
757: int with_mmap;
1.93 schwarze 758: int save_filenc;
1.1 kristaps 759:
1.91 schwarze 760: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 761: save_filenc = curp->filenc;
762: curp->filenc = curp->options &
763: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 764: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 765: curp->filenc = save_filenc;
1.91 schwarze 766: if (with_mmap)
767: munmap(blk.buf, blk.sz);
768: else
769: free(blk.buf);
770: }
1.142 schwarze 771: return curp->file_status;
1.82 schwarze 772: }
773:
1.148 schwarze 774: int
775: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 776: {
777: char *cp;
1.148 schwarze 778: int fd;
1.82 schwarze 779:
780: curp->file = file;
1.140 schwarze 781: cp = strrchr(file, '.');
782: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 783:
1.140 schwarze 784: /* First try to use the filename as it is. */
1.98 schwarze 785:
1.148 schwarze 786: if ((fd = open(file, O_RDONLY)) != -1)
787: return fd;
1.98 schwarze 788:
1.140 schwarze 789: /*
790: * If that doesn't work and the filename doesn't
791: * already end in .gz, try appending .gz.
792: */
1.98 schwarze 793:
1.140 schwarze 794: if ( ! curp->gzip) {
1.98 schwarze 795: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 796: fd = open(cp, O_RDONLY);
1.108 schwarze 797: free(cp);
1.148 schwarze 798: if (fd != -1) {
1.140 schwarze 799: curp->gzip = 1;
1.148 schwarze 800: return fd;
1.82 schwarze 801: }
802: }
803:
1.140 schwarze 804: /* Neither worked, give up. */
1.97 schwarze 805:
1.140 schwarze 806: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 807: return -1;
1.1 kristaps 808: }
809:
810: struct mparse *
1.94 schwarze 811: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.144 schwarze 812: const char *defos)
1.1 kristaps 813: {
814: struct mparse *curp;
1.10 kristaps 815:
1.1 kristaps 816: curp = mandoc_calloc(1, sizeof(struct mparse));
817:
1.44 schwarze 818: curp->options = options;
1.3 kristaps 819: curp->wlevel = wlevel;
1.1 kristaps 820: curp->mmsg = mmsg;
1.29 schwarze 821: curp->defos = defos;
1.1 kristaps 822:
1.144 schwarze 823: curp->roff = roff_alloc(curp, options);
1.137 schwarze 824: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
825: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 826: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 827: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 828: if (curp->man->mdocmac == NULL)
829: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 830: } else if (curp->options & MPARSE_MAN) {
831: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 832: if (curp->man->manmac == NULL)
833: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 834: }
1.138 schwarze 835: curp->man->first->tok = TOKEN_NONE;
1.142 schwarze 836: return curp;
1.1 kristaps 837: }
838:
839: void
840: mparse_reset(struct mparse *curp)
841: {
842: roff_reset(curp->roff);
1.150 schwarze 843: roff_man_reset(curp->man);
1.160 schwarze 844:
845: free(curp->sodest);
846: curp->sodest = NULL;
847:
1.24 kristaps 848: if (curp->secondary)
849: curp->secondary->sz = 0;
1.1 kristaps 850:
851: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 852: curp->gzip = 0;
1.1 kristaps 853: }
854:
855: void
856: mparse_free(struct mparse *curp)
857: {
858:
1.164 schwarze 859: roffhash_free(curp->man->mdocmac);
860: roffhash_free(curp->man->manmac);
1.137 schwarze 861: roff_man_free(curp->man);
1.160 schwarze 862: roff_free(curp->roff);
1.24 kristaps 863: if (curp->secondary)
864: free(curp->secondary->buf);
1.1 kristaps 865:
1.24 kristaps 866: free(curp->secondary);
1.45 schwarze 867: free(curp->sodest);
1.1 kristaps 868: free(curp);
869: }
870:
871: void
1.135 schwarze 872: mparse_result(struct mparse *curp, struct roff_man **man,
873: char **sodest)
1.1 kristaps 874: {
875:
1.45 schwarze 876: if (sodest && NULL != (*sodest = curp->sodest)) {
877: *man = NULL;
878: return;
879: }
1.9 kristaps 880: if (man)
881: *man = curp->man;
1.157 schwarze 882: }
883:
884: void
885: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
886: {
887: if (curp->file_status > *rc)
888: *rc = curp->file_status;
1.3 kristaps 889: }
890:
891: void
892: mandoc_vmsg(enum mandocerr t, struct mparse *m,
893: int ln, int pos, const char *fmt, ...)
894: {
895: char buf[256];
896: va_list ap;
897:
898: va_start(ap, fmt);
1.48 schwarze 899: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 900: va_end(ap);
901:
902: mandoc_msg(t, m, ln, pos, buf);
903: }
904:
905: void
1.47 schwarze 906: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 907: int ln, int col, const char *msg)
908: {
909: enum mandoclevel level;
910:
1.112 schwarze 911: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 912: while (er < mandoclimits[level])
913: level--;
914:
1.108 schwarze 915: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 916: return;
917:
1.8 kristaps 918: if (m->mmsg)
919: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 920:
921: if (m->file_status < level)
922: m->file_status = level;
1.7 kristaps 923: }
924:
925: const char *
926: mparse_strerror(enum mandocerr er)
927: {
928:
1.142 schwarze 929: return mandocerrs[er];
1.7 kristaps 930: }
931:
932: const char *
933: mparse_strlevel(enum mandoclevel lvl)
934: {
1.142 schwarze 935: return mandoclevels[lvl];
1.24 kristaps 936: }
937:
938: void
939: mparse_keep(struct mparse *p)
940: {
941:
942: assert(NULL == p->secondary);
943: p->secondary = mandoc_calloc(1, sizeof(struct buf));
944: }
945:
946: const char *
947: mparse_getkeep(const struct mparse *p)
948: {
949:
950: assert(p->secondary);
1.142 schwarze 951: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 952: }
CVSweb