Annotation of mandoc/read.c, Revision 1.166
1.166 ! schwarze 1: /* $Id: read.c,v 1.165 2017/05/16 19:06:30 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",
! 92:
1.7 kristaps 93: "generic warning",
94:
95: /* related to the prologue */
1.79 schwarze 96: "missing manual title, using UNTITLED",
97: "missing manual title, using \"\"",
1.54 schwarze 98: "lower case character in document title",
1.79 schwarze 99: "missing manual section, using \"\"",
1.7 kristaps 100: "unknown manual section",
1.54 schwarze 101: "missing date, using today's date",
1.7 kristaps 102: "cannot parse date, using it verbatim",
1.79 schwarze 103: "missing Os macro, using \"\"",
104: "duplicate prologue macro",
105: "late prologue macro",
106: "skipping late title macro",
1.7 kristaps 107: "prologue macros out of order",
108:
109: /* related to document structure */
110: ".so is fragile, better use ln(1)",
1.50 schwarze 111: "no document body",
1.54 schwarze 112: "content before first section header",
113: "first section is not \"NAME\"",
1.156 schwarze 114: "NAME section without Nm before Nd",
1.128 schwarze 115: "NAME section without description",
116: "description not at the end of NAME",
117: "bad NAME section content",
1.156 schwarze 118: "missing comma before name",
1.121 schwarze 119: "missing description line, using \"\"",
1.162 schwarze 120: "description line outside NAME section",
1.7 kristaps 121: "sections out of conventional order",
1.54 schwarze 122: "duplicate section title",
123: "unexpected section",
1.87 schwarze 124: "unusual Xr order",
125: "unusual Xr punctuation",
1.86 schwarze 126: "AUTHORS section without An macro",
1.7 kristaps 127:
128: /* related to macros and nesting */
1.55 schwarze 129: "obsolete macro",
1.102 schwarze 130: "macro neither callable nor escaped",
1.7 kristaps 131: "skipping paragraph macro",
1.31 schwarze 132: "moving paragraph macro out of list",
1.7 kristaps 133: "skipping no-space macro",
134: "blocks badly nested",
135: "nested displays are not portable",
1.57 schwarze 136: "moving content out of list",
1.78 schwarze 137: "fill mode already enabled, skipping",
138: "fill mode already disabled, skipping",
1.7 kristaps 139: "line scope broken",
140:
141: /* related to missing macro arguments */
1.58 schwarze 142: "skipping empty request",
143: "conditional request controls empty scope",
1.7 kristaps 144: "skipping empty macro",
1.124 schwarze 145: "empty block",
1.62 schwarze 146: "empty argument, using 0n",
1.60 schwarze 147: "missing display type, using -ragged",
148: "list type is not the first argument",
1.152 schwarze 149: "missing -width in -tag list, using 6n",
1.78 schwarze 150: "missing utility name, using \"\"",
1.123 schwarze 151: "missing function name, using \"\"",
1.60 schwarze 152: "empty head in list item",
153: "empty list item",
1.61 schwarze 154: "missing font type, using \\fR",
155: "unknown font type, using \\fR",
1.103 schwarze 156: "nothing follows prefix",
1.122 schwarze 157: "empty reference block",
1.155 schwarze 158: "missing section argument",
1.60 schwarze 159: "missing -std argument, adding it",
1.125 schwarze 160: "missing option string, using \"\"",
161: "missing resource identifier, using \"\"",
1.90 schwarze 162: "missing eqn box, using \"\"",
1.7 kristaps 163:
164: /* related to bad macro arguments */
1.64 schwarze 165: "unterminated quoted argument",
1.7 kristaps 166: "duplicate argument",
1.76 schwarze 167: "skipping duplicate argument",
1.63 schwarze 168: "skipping duplicate display type",
169: "skipping duplicate list type",
1.76 schwarze 170: "skipping -width argument",
1.126 schwarze 171: "wrong number of cells",
1.7 kristaps 172: "unknown AT&T UNIX version",
1.88 schwarze 173: "comma in function argument",
1.89 schwarze 174: "parenthesis in function name",
1.67 schwarze 175: "invalid content in Rs block",
1.63 schwarze 176: "invalid Boolean argument",
177: "unknown font, skipping request",
1.126 schwarze 178: "odd number of characters in request",
1.7 kristaps 179:
180: /* related to plain text */
1.64 schwarze 181: "blank line in fill mode, using .sp",
182: "tab in filled text",
183: "whitespace at end of input line",
1.158 schwarze 184: "new sentence, new line",
1.7 kristaps 185: "bad comment style",
1.64 schwarze 186: "invalid escape sequence",
187: "undefined string, using \"\"",
1.16 kristaps 188:
1.118 schwarze 189: /* related to tables */
190: "tbl line starts with span",
191: "tbl column starts with span",
192: "skipping vertical bar in tbl layout",
193:
1.7 kristaps 194: "generic error",
195:
196: /* related to tables */
1.116 schwarze 197: "non-alphabetic character in tbl options",
198: "skipping unknown tbl option",
199: "missing tbl option argument",
200: "wrong tbl option argument size",
1.118 schwarze 201: "empty tbl layout",
202: "invalid character in tbl layout",
203: "unmatched parenthesis in tbl layout",
1.119 schwarze 204: "tbl without any data cells",
205: "ignoring data in spanned tbl cell",
206: "ignoring extra tbl data cells",
207: "data block open at end of tbl",
1.7 kristaps 208:
1.68 schwarze 209: /* related to document structure and macros */
1.108 schwarze 210: NULL,
1.7 kristaps 211: "input stack limit exceeded, infinite loop?",
212: "skipping bad character",
1.68 schwarze 213: "skipping unknown macro",
1.112 schwarze 214: "skipping insecure request",
1.73 schwarze 215: "skipping item outside list",
1.68 schwarze 216: "skipping column outside column list",
217: "skipping end of block that is not open",
1.115 schwarze 218: "fewer RS blocks open, skipping",
1.68 schwarze 219: "inserting missing end of block",
220: "appending missing end of block",
221:
222: /* related to request and macro arguments */
1.7 kristaps 223: "escaped character not allowed in a name",
1.99 schwarze 224: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 225: "skipping display without arguments",
1.71 schwarze 226: "missing list type, using -item",
1.70 schwarze 227: "missing manual name, using \"\"",
1.71 schwarze 228: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 229: "unknown standard specifier",
1.71 schwarze 230: "skipping request without numeric argument",
1.109 schwarze 231: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 232: ".so request failed",
1.61 schwarze 233: "skipping all arguments",
234: "skipping excess arguments",
1.92 kristaps 235: "divide by zero",
1.112 schwarze 236:
237: "unsupported feature",
238: "input too large",
1.114 schwarze 239: "unsupported control character",
1.112 schwarze 240: "unsupported roff request",
1.119 schwarze 241: "eqn delim option in tbl",
1.118 schwarze 242: "unsupported tbl layout modifier",
1.112 schwarze 243: "ignoring macro in table",
1.7 kristaps 244: };
245:
246: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
247: "SUCCESS",
1.165 schwarze 248: "STYLE",
1.7 kristaps 249: "WARNING",
250: "ERROR",
1.112 schwarze 251: "UNSUPP",
1.7 kristaps 252: "BADARG",
253: "SYSERR"
254: };
255:
1.47 schwarze 256:
1.1 kristaps 257: static void
258: resize_buf(struct buf *buf, size_t initial)
259: {
260:
261: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
262: buf->buf = mandoc_realloc(buf->buf, buf->sz);
263: }
264:
265: static void
1.84 schwarze 266: choose_parser(struct mparse *curp)
1.1 kristaps 267: {
1.83 schwarze 268: char *cp, *ep;
269: int format;
1.1 kristaps 270:
1.83 schwarze 271: /*
272: * If neither command line arguments -mdoc or -man select
273: * a parser nor the roff parser found a .Dd or .TH macro
274: * yet, look ahead in the main input buffer.
275: */
276:
277: if ((format = roff_getformat(curp->roff)) == 0) {
278: cp = curp->primary->buf;
279: ep = cp + curp->primary->sz;
280: while (cp < ep) {
1.85 schwarze 281: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 282: cp++;
283: if (cp[0] == 'D' && cp[1] == 'd') {
284: format = MPARSE_MDOC;
285: break;
286: }
287: if (cp[0] == 'T' && cp[1] == 'H') {
288: format = MPARSE_MAN;
289: break;
290: }
291: }
292: cp = memchr(cp, '\n', ep - cp);
293: if (cp == NULL)
294: break;
295: cp++;
296: }
1.1 kristaps 297: }
298:
1.83 schwarze 299: if (format == MPARSE_MDOC) {
1.137 schwarze 300: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 301: if (curp->man->mdocmac == NULL)
302: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 303: } else {
304: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 305: if (curp->man->manmac == NULL)
306: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 307: }
1.164 schwarze 308: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 309: }
310:
311: /*
1.95 schwarze 312: * Main parse routine for a buffer.
313: * It assumes encoding and line numbering are already set up.
314: * It can recurse directly (for invocations of user-defined
315: * macros, inline equations, and input line traps)
316: * and indirectly (for .so file inclusion).
1.1 kristaps 317: */
1.163 schwarze 318: static int
1.95 schwarze 319: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 320: {
321: const struct tbl_span *span;
322: struct buf ln;
1.113 schwarze 323: const char *save_file;
1.110 schwarze 324: char *cp;
1.95 schwarze 325: size_t pos; /* byte number in the ln buffer */
1.154 schwarze 326: size_t j; /* auxiliary byte number in the blk buffer */
1.1 kristaps 327: enum rofferr rr;
1.100 schwarze 328: int of;
1.1 kristaps 329: int lnn; /* line number in the real file */
1.110 schwarze 330: int fd;
1.1 kristaps 331: unsigned char c;
332:
1.95 schwarze 333: memset(&ln, 0, sizeof(ln));
1.1 kristaps 334:
1.47 schwarze 335: lnn = curp->line;
336: pos = 0;
1.1 kristaps 337:
1.95 schwarze 338: while (i < blk.sz) {
1.1 kristaps 339: if (0 == pos && '\0' == blk.buf[i])
340: break;
341:
342: if (start) {
343: curp->line = lnn;
344: curp->reparse_count = 0;
1.93 schwarze 345:
346: if (lnn < 3 &&
347: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 348: curp->filenc & MPARSE_LATIN1)
349: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 350: }
351:
1.95 schwarze 352: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 353:
354: /*
355: * When finding an unescaped newline character,
356: * leave the character loop to process the line.
357: * Skip a preceding carriage return, if any.
358: */
359:
1.95 schwarze 360: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 361: '\n' == blk.buf[i + 1])
362: ++i;
363: if ('\n' == blk.buf[i]) {
364: ++i;
365: ++lnn;
366: break;
367: }
368:
1.35 schwarze 369: /*
1.93 schwarze 370: * Make sure we have space for the worst
371: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 372: */
373:
1.95 schwarze 374: if (pos + 11 > ln.sz)
1.35 schwarze 375: resize_buf(&ln, 256);
376:
1.47 schwarze 377: /*
1.93 schwarze 378: * Encode 8-bit input.
1.1 kristaps 379: */
380:
1.93 schwarze 381: c = blk.buf[i];
382: if (c & 0x80) {
1.95 schwarze 383: if ( ! (curp->filenc && preconv_encode(
384: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 385: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
386: curp->line, pos, "0x%x", c);
1.93 schwarze 387: ln.buf[pos++] = '?';
388: i++;
389: }
390: continue;
391: }
392:
393: /*
394: * Exclude control characters.
395: */
1.1 kristaps 396:
1.93 schwarze 397: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 398: mandoc_vmsg(c == 0x00 || c == 0x04 ||
399: c > 0x0a ? MANDOCERR_CHAR_BAD :
400: MANDOCERR_CHAR_UNSUPP,
401: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 402: i++;
1.127 schwarze 403: if (c != '\r')
404: ln.buf[pos++] = '?';
1.1 kristaps 405: continue;
406: }
407:
408: /* Trailing backslash = a plain char. */
409:
1.95 schwarze 410: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 411: ln.buf[pos++] = blk.buf[i++];
412: continue;
413: }
414:
415: /*
416: * Found escape and at least one other character.
417: * When it's a newline character, skip it.
418: * When there is a carriage return in between,
419: * skip that one as well.
420: */
421:
1.95 schwarze 422: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 423: '\n' == blk.buf[i + 2])
424: ++i;
425: if ('\n' == blk.buf[i + 1]) {
426: i += 2;
427: ++lnn;
428: continue;
429: }
430:
1.13 kristaps 431: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.154 schwarze 432: j = i;
1.1 kristaps 433: i += 2;
434: /* Comment, skip to end of line */
1.95 schwarze 435: for (; i < blk.sz; ++i) {
1.153 schwarze 436: if (blk.buf[i] != '\n')
437: continue;
438: if (blk.buf[i - 1] == ' ' ||
439: blk.buf[i - 1] == '\t')
440: mandoc_msg(
441: MANDOCERR_SPACE_EOL,
442: curp, curp->line,
1.154 schwarze 443: pos + i-1 - j, NULL);
1.153 schwarze 444: ++i;
445: ++lnn;
446: break;
1.1 kristaps 447: }
448:
449: /* Backout trailing whitespaces */
450: for (; pos > 0; --pos) {
451: if (ln.buf[pos - 1] != ' ')
452: break;
453: if (pos > 2 && ln.buf[pos - 2] == '\\')
454: break;
455: }
456: break;
457: }
458:
1.35 schwarze 459: /* Catch escaped bogus characters. */
460:
461: c = (unsigned char) blk.buf[i+1];
462:
1.47 schwarze 463: if ( ! (isascii(c) &&
464: (isgraph(c) || isblank(c)))) {
1.114 schwarze 465: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 466: curp->line, pos, "0x%x", c);
1.35 schwarze 467: i += 2;
468: ln.buf[pos++] = '?';
469: continue;
470: }
471:
1.1 kristaps 472: /* Some other escape sequence, copy & cont. */
473:
474: ln.buf[pos++] = blk.buf[i++];
475: ln.buf[pos++] = blk.buf[i++];
476: }
477:
1.95 schwarze 478: if (pos >= ln.sz)
1.1 kristaps 479: resize_buf(&ln, 256);
480:
481: ln.buf[pos] = '\0';
482:
483: /*
484: * A significant amount of complexity is contained by
485: * the roff preprocessor. It's line-oriented but can be
486: * expressed on one line, so we need at times to
487: * readjust our starting point and re-run it. The roff
488: * preprocessor can also readjust the buffers with new
489: * data, so we pass them in wholesale.
490: */
491:
492: of = 0;
493:
1.24 kristaps 494: /*
495: * Maintain a lookaside buffer of all parsed lines. We
496: * only do this if mparse_keep() has been invoked (the
497: * buffer may be accessed with mparse_getkeep()).
498: */
499:
500: if (curp->secondary) {
1.47 schwarze 501: curp->secondary->buf = mandoc_realloc(
502: curp->secondary->buf,
503: curp->secondary->sz + pos + 2);
504: memcpy(curp->secondary->buf +
505: curp->secondary->sz,
506: ln.buf, pos);
1.24 kristaps 507: curp->secondary->sz += pos;
508: curp->secondary->buf
509: [curp->secondary->sz] = '\n';
510: curp->secondary->sz++;
511: curp->secondary->buf
512: [curp->secondary->sz] = '\0';
513: }
1.1 kristaps 514: rerun:
1.96 schwarze 515: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 516:
517: switch (rr) {
1.47 schwarze 518: case ROFF_REPARSE:
1.163 schwarze 519: if (++curp->reparse_count > REPARSE_LIMIT)
1.3 kristaps 520: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 521: curp->line, pos, NULL);
1.163 schwarze 522: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
523: start == 1) {
524: pos = 0;
525: continue;
526: }
527: free(ln.buf);
528: return 0;
1.47 schwarze 529: case ROFF_APPEND:
1.95 schwarze 530: pos = strlen(ln.buf);
1.1 kristaps 531: continue;
1.47 schwarze 532: case ROFF_RERUN:
1.1 kristaps 533: goto rerun;
1.47 schwarze 534: case ROFF_IGN:
1.1 kristaps 535: pos = 0;
536: continue;
1.47 schwarze 537: case ROFF_SO:
1.95 schwarze 538: if ( ! (curp->options & MPARSE_SO) &&
539: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 540: curp->sodest = mandoc_strdup(ln.buf + of);
541: free(ln.buf);
1.163 schwarze 542: return 1;
1.45 schwarze 543: }
1.24 kristaps 544: /*
545: * We remove `so' clauses from our lookaside
546: * buffer because we're going to descend into
547: * the file recursively.
548: */
1.47 schwarze 549: if (curp->secondary)
1.25 kristaps 550: curp->secondary->sz -= pos + 1;
1.113 schwarze 551: save_file = curp->file;
1.148 schwarze 552: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 553: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 554: close(fd);
1.113 schwarze 555: curp->file = save_file;
556: } else {
557: curp->file = save_file;
1.52 schwarze 558: mandoc_vmsg(MANDOCERR_SO_FAIL,
559: curp, curp->line, pos,
560: ".so %s", ln.buf + of);
1.110 schwarze 561: ln.sz = mandoc_asprintf(&cp,
562: ".sp\nSee the file %s.\n.sp",
563: ln.buf + of);
564: free(ln.buf);
565: ln.buf = cp;
566: of = 0;
567: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 568: }
1.1 kristaps 569: pos = 0;
570: continue;
571: default:
572: break;
573: }
574:
1.150 schwarze 575: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 576: choose_parser(curp);
1.1 kristaps 577:
1.47 schwarze 578: /*
1.84 schwarze 579: * Lastly, push down into the parsers themselves.
1.1 kristaps 580: * If libroff returns ROFF_TBL, then add it to the
581: * currently open parse. Since we only get here if
582: * there does exist data (see tbl_data.c), we're
583: * guaranteed that something's been allocated.
584: * Do the same for ROFF_EQN.
585: */
586:
1.139 schwarze 587: if (rr == ROFF_TBL)
1.100 schwarze 588: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 589: roff_addtbl(curp->man, span);
590: else if (rr == ROFF_EQN)
591: roff_addeqn(curp->man, roff_eqn(curp->roff));
592: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 593: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 594: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
595: break;
1.1 kristaps 596:
597: /* Temporary buffers typically are not full. */
598:
599: if (0 == start && '\0' == blk.buf[i])
600: break;
601:
602: /* Start the next input line. */
603:
604: pos = 0;
605: }
606:
607: free(ln.buf);
1.163 schwarze 608: return 1;
1.1 kristaps 609: }
610:
611: static int
1.40 schwarze 612: read_whole_file(struct mparse *curp, const char *file, int fd,
613: struct buf *fb, int *with_mmap)
1.1 kristaps 614: {
1.161 schwarze 615: struct stat st;
1.140 schwarze 616: gzFile gz;
1.1 kristaps 617: size_t off;
618: ssize_t ssz;
1.143 schwarze 619:
620: if (fstat(fd, &st) == -1)
621: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 622:
623: /*
624: * If we're a regular file, try just reading in the whole entry
625: * via mmap(). This is faster than reading it into blocks, and
626: * since each file is only a few bytes to begin with, I'm not
627: * concerned that this is going to tank any machines.
628: */
629:
1.140 schwarze 630: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 631: if (st.st_size > 0x7fffffff) {
1.111 schwarze 632: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 633: return 0;
1.1 kristaps 634: }
635: *with_mmap = 1;
636: fb->sz = (size_t)st.st_size;
1.37 schwarze 637: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 638: if (fb->buf != MAP_FAILED)
1.142 schwarze 639: return 1;
1.1 kristaps 640: }
641:
1.140 schwarze 642: if (curp->gzip) {
1.143 schwarze 643: if ((gz = gzdopen(fd, "rb")) == NULL)
644: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 645: } else
646: gz = NULL;
647:
1.1 kristaps 648: /*
649: * If this isn't a regular file (like, say, stdin), then we must
650: * go the old way and just read things in bit by bit.
651: */
652:
653: *with_mmap = 0;
654: off = 0;
655: fb->sz = 0;
656: fb->buf = NULL;
657: for (;;) {
658: if (off == fb->sz) {
659: if (fb->sz == (1U << 31)) {
1.111 schwarze 660: mandoc_msg(MANDOCERR_TOOLARGE, curp,
661: 0, 0, NULL);
1.1 kristaps 662: break;
663: }
664: resize_buf(fb, 65536);
665: }
1.140 schwarze 666: ssz = curp->gzip ?
667: gzread(gz, fb->buf + (int)off, fb->sz - off) :
668: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 669: if (ssz == 0) {
670: fb->sz = off;
1.142 schwarze 671: return 1;
1.1 kristaps 672: }
1.143 schwarze 673: if (ssz == -1)
674: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 675: off += (size_t)ssz;
676: }
677:
678: free(fb->buf);
679: fb->buf = NULL;
1.142 schwarze 680: return 0;
1.1 kristaps 681: }
682:
683: static void
684: mparse_end(struct mparse *curp)
685: {
1.135 schwarze 686: if (curp->man->macroset == MACROSET_NONE)
687: curp->man->macroset = MACROSET_MAN;
688: if (curp->man->macroset == MACROSET_MDOC)
689: mdoc_endparse(curp->man);
690: else
1.111 schwarze 691: man_endparse(curp->man);
1.1 kristaps 692: roff_endparse(curp->roff);
693: }
694:
695: static void
1.36 schwarze 696: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 697: {
1.85 schwarze 698: struct buf *svprimary;
1.28 joerg 699: const char *svfile;
1.95 schwarze 700: size_t offset;
1.36 schwarze 701: static int recursion_depth;
702:
703: if (64 < recursion_depth) {
704: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
705: return;
706: }
1.28 joerg 707:
708: /* Line number is per-file. */
709: svfile = curp->file;
710: curp->file = file;
1.85 schwarze 711: svprimary = curp->primary;
1.83 schwarze 712: curp->primary = &blk;
1.28 joerg 713: curp->line = 1;
1.36 schwarze 714: recursion_depth++;
1.28 joerg 715:
1.93 schwarze 716: /* Skip an UTF-8 byte order mark. */
717: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
718: (unsigned char)blk.buf[0] == 0xef &&
719: (unsigned char)blk.buf[1] == 0xbb &&
720: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 721: offset = 3;
1.93 schwarze 722: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 723: } else
724: offset = 0;
1.93 schwarze 725:
1.95 schwarze 726: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 727:
1.111 schwarze 728: if (--recursion_depth == 0)
1.28 joerg 729: mparse_end(curp);
730:
1.85 schwarze 731: curp->primary = svprimary;
1.28 joerg 732: curp->file = svfile;
733: }
734:
735: enum mandoclevel
1.104 schwarze 736: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 737: const char *file)
738: {
739: struct buf blk;
740:
1.104 schwarze 741: blk.buf = buf;
1.28 joerg 742: blk.sz = len;
743:
1.36 schwarze 744: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 745: return curp->file_status;
1.28 joerg 746: }
747:
1.98 schwarze 748: /*
749: * Read the whole file into memory and call the parsers.
750: * Called recursively when an .so request is encountered.
751: */
1.36 schwarze 752: enum mandoclevel
753: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 754: {
1.28 joerg 755: struct buf blk;
756: int with_mmap;
1.93 schwarze 757: int save_filenc;
1.1 kristaps 758:
1.91 schwarze 759: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 760: save_filenc = curp->filenc;
761: curp->filenc = curp->options &
762: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 763: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 764: curp->filenc = save_filenc;
1.91 schwarze 765: if (with_mmap)
766: munmap(blk.buf, blk.sz);
767: else
768: free(blk.buf);
769: }
1.142 schwarze 770: return curp->file_status;
1.82 schwarze 771: }
772:
1.148 schwarze 773: int
774: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 775: {
776: char *cp;
1.148 schwarze 777: int fd;
1.82 schwarze 778:
779: curp->file = file;
1.140 schwarze 780: cp = strrchr(file, '.');
781: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 782:
1.140 schwarze 783: /* First try to use the filename as it is. */
1.98 schwarze 784:
1.148 schwarze 785: if ((fd = open(file, O_RDONLY)) != -1)
786: return fd;
1.98 schwarze 787:
1.140 schwarze 788: /*
789: * If that doesn't work and the filename doesn't
790: * already end in .gz, try appending .gz.
791: */
1.98 schwarze 792:
1.140 schwarze 793: if ( ! curp->gzip) {
1.98 schwarze 794: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 795: fd = open(cp, O_RDONLY);
1.108 schwarze 796: free(cp);
1.148 schwarze 797: if (fd != -1) {
1.140 schwarze 798: curp->gzip = 1;
1.148 schwarze 799: return fd;
1.82 schwarze 800: }
801: }
802:
1.140 schwarze 803: /* Neither worked, give up. */
1.97 schwarze 804:
1.140 schwarze 805: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 806: return -1;
1.1 kristaps 807: }
808:
809: struct mparse *
1.94 schwarze 810: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.144 schwarze 811: const char *defos)
1.1 kristaps 812: {
813: struct mparse *curp;
1.10 kristaps 814:
1.1 kristaps 815: curp = mandoc_calloc(1, sizeof(struct mparse));
816:
1.44 schwarze 817: curp->options = options;
1.3 kristaps 818: curp->wlevel = wlevel;
1.1 kristaps 819: curp->mmsg = mmsg;
1.29 schwarze 820: curp->defos = defos;
1.1 kristaps 821:
1.144 schwarze 822: curp->roff = roff_alloc(curp, options);
1.137 schwarze 823: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
824: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 825: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 826: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 827: if (curp->man->mdocmac == NULL)
828: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 829: } else if (curp->options & MPARSE_MAN) {
830: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 831: if (curp->man->manmac == NULL)
832: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 833: }
1.138 schwarze 834: curp->man->first->tok = TOKEN_NONE;
1.142 schwarze 835: return curp;
1.1 kristaps 836: }
837:
838: void
839: mparse_reset(struct mparse *curp)
840: {
841: roff_reset(curp->roff);
1.150 schwarze 842: roff_man_reset(curp->man);
1.160 schwarze 843:
844: free(curp->sodest);
845: curp->sodest = NULL;
846:
1.24 kristaps 847: if (curp->secondary)
848: curp->secondary->sz = 0;
1.1 kristaps 849:
850: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 851: curp->gzip = 0;
1.1 kristaps 852: }
853:
854: void
855: mparse_free(struct mparse *curp)
856: {
857:
1.164 schwarze 858: roffhash_free(curp->man->mdocmac);
859: roffhash_free(curp->man->manmac);
1.137 schwarze 860: roff_man_free(curp->man);
1.160 schwarze 861: roff_free(curp->roff);
1.24 kristaps 862: if (curp->secondary)
863: free(curp->secondary->buf);
1.1 kristaps 864:
1.24 kristaps 865: free(curp->secondary);
1.45 schwarze 866: free(curp->sodest);
1.1 kristaps 867: free(curp);
868: }
869:
870: void
1.135 schwarze 871: mparse_result(struct mparse *curp, struct roff_man **man,
872: char **sodest)
1.1 kristaps 873: {
874:
1.45 schwarze 875: if (sodest && NULL != (*sodest = curp->sodest)) {
876: *man = NULL;
877: return;
878: }
1.9 kristaps 879: if (man)
880: *man = curp->man;
1.157 schwarze 881: }
882:
883: void
884: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
885: {
886: if (curp->file_status > *rc)
887: *rc = curp->file_status;
1.3 kristaps 888: }
889:
890: void
891: mandoc_vmsg(enum mandocerr t, struct mparse *m,
892: int ln, int pos, const char *fmt, ...)
893: {
894: char buf[256];
895: va_list ap;
896:
897: va_start(ap, fmt);
1.48 schwarze 898: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 899: va_end(ap);
900:
901: mandoc_msg(t, m, ln, pos, buf);
902: }
903:
904: void
1.47 schwarze 905: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 906: int ln, int col, const char *msg)
907: {
908: enum mandoclevel level;
909:
1.112 schwarze 910: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 911: while (er < mandoclimits[level])
912: level--;
913:
1.108 schwarze 914: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 915: return;
916:
1.8 kristaps 917: if (m->mmsg)
918: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 919:
920: if (m->file_status < level)
921: m->file_status = level;
1.7 kristaps 922: }
923:
924: const char *
925: mparse_strerror(enum mandocerr er)
926: {
927:
1.142 schwarze 928: return mandocerrs[er];
1.7 kristaps 929: }
930:
931: const char *
932: mparse_strlevel(enum mandoclevel lvl)
933: {
1.142 schwarze 934: return mandoclevels[lvl];
1.24 kristaps 935: }
936:
937: void
938: mparse_keep(struct mparse *p)
939: {
940:
941: assert(NULL == p->secondary);
942: p->secondary = mandoc_calloc(1, sizeof(struct buf));
943: }
944:
945: const char *
946: mparse_getkeep(const struct mparse *p)
947: {
948:
949: assert(p->secondary);
1.142 schwarze 950: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 951: }
CVSweb