Annotation of src/usr.bin/mandoc/read.c, Revision 1.135
1.135 ! schwarze 1: /* $OpenBSD: read.c,v 1.134 2017/03/06 17:25:24 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.129 schwarze 4: * Copyright (c) 2010-2017 Ingo Schwarze <schwarze@openbsd.org>
1.20 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 schwarze 6: *
7: * Permission to use, copy, modify, and distribute this software for any
8: * purpose with or without fee is hereby granted, provided that the above
9: * copyright notice and this permission notice appear in all copies.
10: *
1.108 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.108 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18: */
1.58 schwarze 19: #include <sys/types.h>
20: #include <sys/mman.h>
1.1 schwarze 21: #include <sys/stat.h>
22:
23: #include <assert.h>
24: #include <ctype.h>
1.118 schwarze 25: #include <err.h>
1.18 schwarze 26: #include <errno.h>
1.1 schwarze 27: #include <fcntl.h>
28: #include <stdarg.h>
29: #include <stdio.h>
30: #include <stdlib.h>
31: #include <string.h>
32: #include <unistd.h>
1.115 schwarze 33: #include <zlib.h>
1.1 schwarze 34:
1.108 schwarze 35: #include "mandoc_aux.h"
1.1 schwarze 36: #include "mandoc.h"
1.108 schwarze 37: #include "roff.h"
1.1 schwarze 38: #include "mdoc.h"
39: #include "man.h"
1.108 schwarze 40: #include "libmandoc.h"
1.114 schwarze 41: #include "roff_int.h"
1.1 schwarze 42:
43: #define REPARSE_LIMIT 1000
44:
45: struct mparse {
1.133 schwarze 46: struct roff *roff; /* roff parser (!NULL) */
1.109 schwarze 47: struct roff_man *man; /* man parser */
1.23 schwarze 48: char *sodest; /* filename pointed to by .so */
1.59 schwarze 49: const char *file; /* filename of current input file */
50: struct buf *primary; /* buffer currently being parsed */
51: struct buf *secondary; /* preprocessed copy of input */
52: const char *defos; /* default operating system */
53: mandocmsg mmsg; /* warning/error message handler */
54: enum mandoclevel file_status; /* status of current parse */
55: enum mandoclevel wlevel; /* ignore messages below this */
56: int options; /* parser options */
1.115 schwarze 57: int gzip; /* current input file is gzipped */
1.70 schwarze 58: int filenc; /* encoding of the current file */
1.1 schwarze 59: int reparse_count; /* finite interp. stack */
1.59 schwarze 60: int line; /* line number in the file */
1.1 schwarze 61: };
62:
1.60 schwarze 63: static void choose_parser(struct mparse *);
1.1 schwarze 64: static void resize_buf(struct buf *, size_t);
1.135 ! schwarze 65: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.18 schwarze 66: static int read_whole_file(struct mparse *, const char *, int,
67: struct buf *, int *);
1.1 schwarze 68: static void mparse_end(struct mparse *);
1.15 schwarze 69: static void mparse_parse_buffer(struct mparse *, struct buf,
70: const char *);
1.1 schwarze 71:
72: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
73: MANDOCERR_OK,
74: MANDOCERR_WARNING,
75: MANDOCERR_WARNING,
76: MANDOCERR_ERROR,
1.87 schwarze 77: MANDOCERR_UNSUPP,
1.1 schwarze 78: MANDOCERR_MAX,
79: MANDOCERR_MAX
80: };
81:
82: static const char * const mandocerrs[MANDOCERR_MAX] = {
83: "ok",
84:
85: "generic warning",
86:
87: /* related to the prologue */
1.57 schwarze 88: "missing manual title, using UNTITLED",
89: "missing manual title, using \"\"",
1.32 schwarze 90: "lower case character in document title",
1.57 schwarze 91: "missing manual section, using \"\"",
1.1 schwarze 92: "unknown manual section",
1.32 schwarze 93: "missing date, using today's date",
1.1 schwarze 94: "cannot parse date, using it verbatim",
1.57 schwarze 95: "missing Os macro, using \"\"",
96: "duplicate prologue macro",
97: "late prologue macro",
98: "skipping late title macro",
1.1 schwarze 99: "prologue macros out of order",
100:
101: /* related to document structure */
102: ".so is fragile, better use ln(1)",
1.28 schwarze 103: "no document body",
1.32 schwarze 104: "content before first section header",
105: "first section is not \"NAME\"",
1.129 schwarze 106: "NAME section without Nm before Nd",
1.103 schwarze 107: "NAME section without description",
108: "description not at the end of NAME",
109: "bad NAME section content",
1.129 schwarze 110: "missing comma before name",
1.96 schwarze 111: "missing description line, using \"\"",
1.134 schwarze 112: "description line outside NAME section",
1.1 schwarze 113: "sections out of conventional order",
1.32 schwarze 114: "duplicate section title",
115: "unexpected section",
1.63 schwarze 116: "unusual Xr order",
117: "unusual Xr punctuation",
1.62 schwarze 118: "AUTHORS section without An macro",
1.1 schwarze 119:
120: /* related to macros and nesting */
1.33 schwarze 121: "obsolete macro",
1.78 schwarze 122: "macro neither callable nor escaped",
1.1 schwarze 123: "skipping paragraph macro",
1.10 schwarze 124: "moving paragraph macro out of list",
1.1 schwarze 125: "skipping no-space macro",
126: "blocks badly nested",
127: "nested displays are not portable",
1.35 schwarze 128: "moving content out of list",
1.56 schwarze 129: "fill mode already enabled, skipping",
130: "fill mode already disabled, skipping",
1.1 schwarze 131: "line scope broken",
132:
133: /* related to missing macro arguments */
1.36 schwarze 134: "skipping empty request",
135: "conditional request controls empty scope",
1.1 schwarze 136: "skipping empty macro",
1.99 schwarze 137: "empty block",
1.40 schwarze 138: "empty argument, using 0n",
1.38 schwarze 139: "missing display type, using -ragged",
140: "list type is not the first argument",
1.125 schwarze 141: "missing -width in -tag list, using 6n",
1.56 schwarze 142: "missing utility name, using \"\"",
1.98 schwarze 143: "missing function name, using \"\"",
1.38 schwarze 144: "empty head in list item",
145: "empty list item",
1.39 schwarze 146: "missing font type, using \\fR",
147: "unknown font type, using \\fR",
1.79 schwarze 148: "nothing follows prefix",
1.97 schwarze 149: "empty reference block",
1.128 schwarze 150: "missing section argument",
1.38 schwarze 151: "missing -std argument, adding it",
1.100 schwarze 152: "missing option string, using \"\"",
153: "missing resource identifier, using \"\"",
1.66 schwarze 154: "missing eqn box, using \"\"",
1.1 schwarze 155:
156: /* related to bad macro arguments */
1.42 schwarze 157: "unterminated quoted argument",
1.1 schwarze 158: "duplicate argument",
1.54 schwarze 159: "skipping duplicate argument",
1.41 schwarze 160: "skipping duplicate display type",
161: "skipping duplicate list type",
1.54 schwarze 162: "skipping -width argument",
1.101 schwarze 163: "wrong number of cells",
1.1 schwarze 164: "unknown AT&T UNIX version",
1.64 schwarze 165: "comma in function argument",
1.65 schwarze 166: "parenthesis in function name",
1.45 schwarze 167: "invalid content in Rs block",
1.41 schwarze 168: "invalid Boolean argument",
169: "unknown font, skipping request",
1.101 schwarze 170: "odd number of characters in request",
1.1 schwarze 171:
172: /* related to plain text */
1.42 schwarze 173: "blank line in fill mode, using .sp",
174: "tab in filled text",
175: "whitespace at end of input line",
1.131 schwarze 176: "new sentence, new line",
1.1 schwarze 177: "bad comment style",
1.42 schwarze 178: "invalid escape sequence",
179: "undefined string, using \"\"",
1.3 schwarze 180:
1.93 schwarze 181: /* related to tables */
182: "tbl line starts with span",
183: "tbl column starts with span",
184: "skipping vertical bar in tbl layout",
185:
1.1 schwarze 186: "generic error",
1.3 schwarze 187:
1.1 schwarze 188: /* related to tables */
1.91 schwarze 189: "non-alphabetic character in tbl options",
190: "skipping unknown tbl option",
191: "missing tbl option argument",
192: "wrong tbl option argument size",
1.93 schwarze 193: "empty tbl layout",
194: "invalid character in tbl layout",
195: "unmatched parenthesis in tbl layout",
1.94 schwarze 196: "tbl without any data cells",
197: "ignoring data in spanned tbl cell",
198: "ignoring extra tbl data cells",
199: "data block open at end of tbl",
1.1 schwarze 200:
1.46 schwarze 201: /* related to document structure and macros */
1.83 schwarze 202: NULL,
1.1 schwarze 203: "input stack limit exceeded, infinite loop?",
204: "skipping bad character",
1.46 schwarze 205: "skipping unknown macro",
1.87 schwarze 206: "skipping insecure request",
1.51 schwarze 207: "skipping item outside list",
1.46 schwarze 208: "skipping column outside column list",
209: "skipping end of block that is not open",
1.90 schwarze 210: "fewer RS blocks open, skipping",
1.46 schwarze 211: "inserting missing end of block",
212: "appending missing end of block",
213:
214: /* related to request and macro arguments */
1.1 schwarze 215: "escaped character not allowed in a name",
1.75 schwarze 216: "NOT IMPLEMENTED: Bd -file",
1.120 schwarze 217: "skipping display without arguments",
1.49 schwarze 218: "missing list type, using -item",
1.48 schwarze 219: "missing manual name, using \"\"",
1.49 schwarze 220: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 221: "unknown standard specifier",
1.49 schwarze 222: "skipping request without numeric argument",
1.84 schwarze 223: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 224: ".so request failed",
1.39 schwarze 225: "skipping all arguments",
226: "skipping excess arguments",
1.68 schwarze 227: "divide by zero",
1.87 schwarze 228:
229: "unsupported feature",
230: "input too large",
1.89 schwarze 231: "unsupported control character",
1.87 schwarze 232: "unsupported roff request",
1.94 schwarze 233: "eqn delim option in tbl",
1.93 schwarze 234: "unsupported tbl layout modifier",
1.87 schwarze 235: "ignoring macro in table",
1.1 schwarze 236: };
237:
238: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
239: "SUCCESS",
240: "RESERVED",
241: "WARNING",
242: "ERROR",
1.87 schwarze 243: "UNSUPP",
1.1 schwarze 244: "BADARG",
245: "SYSERR"
246: };
247:
1.25 schwarze 248:
1.1 schwarze 249: static void
250: resize_buf(struct buf *buf, size_t initial)
251: {
252:
253: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
254: buf->buf = mandoc_realloc(buf->buf, buf->sz);
255: }
256:
257: static void
1.60 schwarze 258: choose_parser(struct mparse *curp)
1.1 schwarze 259: {
1.59 schwarze 260: char *cp, *ep;
261: int format;
1.1 schwarze 262:
1.59 schwarze 263: /*
264: * If neither command line arguments -mdoc or -man select
265: * a parser nor the roff parser found a .Dd or .TH macro
266: * yet, look ahead in the main input buffer.
267: */
268:
269: if ((format = roff_getformat(curp->roff)) == 0) {
270: cp = curp->primary->buf;
271: ep = cp + curp->primary->sz;
272: while (cp < ep) {
1.61 schwarze 273: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 274: cp++;
275: if (cp[0] == 'D' && cp[1] == 'd') {
276: format = MPARSE_MDOC;
277: break;
278: }
279: if (cp[0] == 'T' && cp[1] == 'H') {
280: format = MPARSE_MAN;
281: break;
282: }
283: }
284: cp = memchr(cp, '\n', ep - cp);
285: if (cp == NULL)
286: break;
287: cp++;
288: }
1.1 schwarze 289: }
290:
1.59 schwarze 291: if (format == MPARSE_MDOC) {
1.111 schwarze 292: mdoc_hash_init();
1.112 schwarze 293: curp->man->macroset = MACROSET_MDOC;
1.113 schwarze 294: curp->man->first->tok = TOKEN_NONE;
1.112 schwarze 295: } else {
296: man_hash_init();
297: curp->man->macroset = MACROSET_MAN;
1.113 schwarze 298: curp->man->first->tok = TOKEN_NONE;
1.25 schwarze 299: }
1.1 schwarze 300: }
301:
302: /*
1.71 schwarze 303: * Main parse routine for a buffer.
304: * It assumes encoding and line numbering are already set up.
305: * It can recurse directly (for invocations of user-defined
306: * macros, inline equations, and input line traps)
307: * and indirectly (for .so file inclusion).
1.1 schwarze 308: */
1.135 ! schwarze 309: static int
1.71 schwarze 310: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 311: {
312: const struct tbl_span *span;
313: struct buf ln;
1.88 schwarze 314: const char *save_file;
1.85 schwarze 315: char *cp;
1.71 schwarze 316: size_t pos; /* byte number in the ln buffer */
1.127 schwarze 317: size_t j; /* auxiliary byte number in the blk buffer */
1.1 schwarze 318: enum rofferr rr;
1.76 schwarze 319: int of;
1.1 schwarze 320: int lnn; /* line number in the real file */
1.85 schwarze 321: int fd;
1.1 schwarze 322: unsigned char c;
323:
1.71 schwarze 324: memset(&ln, 0, sizeof(ln));
1.1 schwarze 325:
1.25 schwarze 326: lnn = curp->line;
327: pos = 0;
1.1 schwarze 328:
1.71 schwarze 329: while (i < blk.sz) {
1.1 schwarze 330: if (0 == pos && '\0' == blk.buf[i])
331: break;
332:
333: if (start) {
334: curp->line = lnn;
335: curp->reparse_count = 0;
1.70 schwarze 336:
337: if (lnn < 3 &&
338: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 339: curp->filenc & MPARSE_LATIN1)
340: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 341: }
342:
1.71 schwarze 343: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 344:
345: /*
346: * When finding an unescaped newline character,
347: * leave the character loop to process the line.
348: * Skip a preceding carriage return, if any.
349: */
350:
1.71 schwarze 351: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 352: '\n' == blk.buf[i + 1])
353: ++i;
354: if ('\n' == blk.buf[i]) {
355: ++i;
356: ++lnn;
357: break;
358: }
359:
1.13 schwarze 360: /*
1.70 schwarze 361: * Make sure we have space for the worst
362: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 363: */
364:
1.71 schwarze 365: if (pos + 11 > ln.sz)
1.13 schwarze 366: resize_buf(&ln, 256);
367:
1.25 schwarze 368: /*
1.70 schwarze 369: * Encode 8-bit input.
1.1 schwarze 370: */
371:
1.70 schwarze 372: c = blk.buf[i];
373: if (c & 0x80) {
1.71 schwarze 374: if ( ! (curp->filenc && preconv_encode(
375: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 376: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
377: curp->line, pos, "0x%x", c);
1.70 schwarze 378: ln.buf[pos++] = '?';
379: i++;
380: }
381: continue;
382: }
383:
384: /*
385: * Exclude control characters.
386: */
1.1 schwarze 387:
1.70 schwarze 388: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 389: mandoc_vmsg(c == 0x00 || c == 0x04 ||
390: c > 0x0a ? MANDOCERR_CHAR_BAD :
391: MANDOCERR_CHAR_UNSUPP,
392: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 393: i++;
1.102 schwarze 394: if (c != '\r')
395: ln.buf[pos++] = '?';
1.1 schwarze 396: continue;
397: }
398:
399: /* Trailing backslash = a plain char. */
400:
1.71 schwarze 401: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 schwarze 402: ln.buf[pos++] = blk.buf[i++];
403: continue;
404: }
405:
406: /*
407: * Found escape and at least one other character.
408: * When it's a newline character, skip it.
409: * When there is a carriage return in between,
410: * skip that one as well.
411: */
412:
1.71 schwarze 413: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 schwarze 414: '\n' == blk.buf[i + 2])
415: ++i;
416: if ('\n' == blk.buf[i + 1]) {
417: i += 2;
418: ++lnn;
419: continue;
420: }
421:
422: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.127 schwarze 423: j = i;
1.1 schwarze 424: i += 2;
425: /* Comment, skip to end of line */
1.71 schwarze 426: for (; i < blk.sz; ++i) {
1.126 schwarze 427: if (blk.buf[i] != '\n')
428: continue;
429: if (blk.buf[i - 1] == ' ' ||
430: blk.buf[i - 1] == '\t')
431: mandoc_msg(
432: MANDOCERR_SPACE_EOL,
433: curp, curp->line,
1.127 schwarze 434: pos + i-1 - j, NULL);
1.126 schwarze 435: ++i;
436: ++lnn;
437: break;
1.1 schwarze 438: }
439:
440: /* Backout trailing whitespaces */
441: for (; pos > 0; --pos) {
442: if (ln.buf[pos - 1] != ' ')
443: break;
444: if (pos > 2 && ln.buf[pos - 2] == '\\')
445: break;
446: }
447: break;
448: }
449:
1.13 schwarze 450: /* Catch escaped bogus characters. */
451:
452: c = (unsigned char) blk.buf[i+1];
453:
1.25 schwarze 454: if ( ! (isascii(c) &&
455: (isgraph(c) || isblank(c)))) {
1.89 schwarze 456: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.56 schwarze 457: curp->line, pos, "0x%x", c);
1.13 schwarze 458: i += 2;
459: ln.buf[pos++] = '?';
460: continue;
461: }
462:
1.1 schwarze 463: /* Some other escape sequence, copy & cont. */
464:
465: ln.buf[pos++] = blk.buf[i++];
466: ln.buf[pos++] = blk.buf[i++];
467: }
468:
1.71 schwarze 469: if (pos >= ln.sz)
1.1 schwarze 470: resize_buf(&ln, 256);
471:
472: ln.buf[pos] = '\0';
473:
474: /*
475: * A significant amount of complexity is contained by
476: * the roff preprocessor. It's line-oriented but can be
477: * expressed on one line, so we need at times to
478: * readjust our starting point and re-run it. The roff
479: * preprocessor can also readjust the buffers with new
480: * data, so we pass them in wholesale.
481: */
482:
483: of = 0;
484:
1.4 schwarze 485: /*
486: * Maintain a lookaside buffer of all parsed lines. We
487: * only do this if mparse_keep() has been invoked (the
488: * buffer may be accessed with mparse_getkeep()).
489: */
490:
491: if (curp->secondary) {
1.25 schwarze 492: curp->secondary->buf = mandoc_realloc(
493: curp->secondary->buf,
494: curp->secondary->sz + pos + 2);
495: memcpy(curp->secondary->buf +
496: curp->secondary->sz,
497: ln.buf, pos);
1.4 schwarze 498: curp->secondary->sz += pos;
499: curp->secondary->buf
500: [curp->secondary->sz] = '\n';
501: curp->secondary->sz++;
502: curp->secondary->buf
503: [curp->secondary->sz] = '\0';
504: }
1.1 schwarze 505: rerun:
1.72 schwarze 506: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 507:
508: switch (rr) {
1.25 schwarze 509: case ROFF_REPARSE:
1.135 ! schwarze 510: if (++curp->reparse_count > REPARSE_LIMIT)
1.1 schwarze 511: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 512: curp->line, pos, NULL);
1.135 ! schwarze 513: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
! 514: start == 1) {
! 515: pos = 0;
! 516: continue;
! 517: }
! 518: free(ln.buf);
! 519: return 0;
1.25 schwarze 520: case ROFF_APPEND:
1.71 schwarze 521: pos = strlen(ln.buf);
1.1 schwarze 522: continue;
1.25 schwarze 523: case ROFF_RERUN:
1.1 schwarze 524: goto rerun;
1.25 schwarze 525: case ROFF_IGN:
1.1 schwarze 526: pos = 0;
527: continue;
1.25 schwarze 528: case ROFF_SO:
1.71 schwarze 529: if ( ! (curp->options & MPARSE_SO) &&
530: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 531: curp->sodest = mandoc_strdup(ln.buf + of);
532: free(ln.buf);
1.135 ! schwarze 533: return 1;
1.23 schwarze 534: }
1.4 schwarze 535: /*
536: * We remove `so' clauses from our lookaside
537: * buffer because we're going to descend into
538: * the file recursively.
539: */
1.25 schwarze 540: if (curp->secondary)
1.4 schwarze 541: curp->secondary->sz -= pos + 1;
1.88 schwarze 542: save_file = curp->file;
1.122 schwarze 543: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 544: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 545: close(fd);
1.88 schwarze 546: curp->file = save_file;
547: } else {
548: curp->file = save_file;
1.30 schwarze 549: mandoc_vmsg(MANDOCERR_SO_FAIL,
550: curp, curp->line, pos,
551: ".so %s", ln.buf + of);
1.85 schwarze 552: ln.sz = mandoc_asprintf(&cp,
553: ".sp\nSee the file %s.\n.sp",
554: ln.buf + of);
555: free(ln.buf);
556: ln.buf = cp;
557: of = 0;
558: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 559: }
1.1 schwarze 560: pos = 0;
561: continue;
562: default:
563: break;
564: }
565:
1.124 schwarze 566: if (curp->man->macroset == MACROSET_NONE)
1.60 schwarze 567: choose_parser(curp);
1.1 schwarze 568:
1.25 schwarze 569: /*
1.60 schwarze 570: * Lastly, push down into the parsers themselves.
1.1 schwarze 571: * If libroff returns ROFF_TBL, then add it to the
572: * currently open parse. Since we only get here if
573: * there does exist data (see tbl_data.c), we're
574: * guaranteed that something's been allocated.
575: * Do the same for ROFF_EQN.
576: */
577:
1.114 schwarze 578: if (rr == ROFF_TBL)
1.76 schwarze 579: while ((span = roff_span(curp->roff)) != NULL)
1.114 schwarze 580: roff_addtbl(curp->man, span);
581: else if (rr == ROFF_EQN)
582: roff_addeqn(curp->man, roff_eqn(curp->roff));
583: else if ((curp->man->macroset == MACROSET_MDOC ?
1.110 schwarze 584: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 585: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
586: break;
1.1 schwarze 587:
588: /* Temporary buffers typically are not full. */
589:
590: if (0 == start && '\0' == blk.buf[i])
591: break;
592:
593: /* Start the next input line. */
594:
595: pos = 0;
596: }
597:
598: free(ln.buf);
1.135 ! schwarze 599: return 1;
1.1 schwarze 600: }
601:
602: static int
1.18 schwarze 603: read_whole_file(struct mparse *curp, const char *file, int fd,
604: struct buf *fb, int *with_mmap)
1.1 schwarze 605: {
606: struct stat st;
1.115 schwarze 607: gzFile gz;
1.1 schwarze 608: size_t off;
609: ssize_t ssz;
610:
1.118 schwarze 611: if (fstat(fd, &st) == -1)
612: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 613:
614: /*
615: * If we're a regular file, try just reading in the whole entry
616: * via mmap(). This is faster than reading it into blocks, and
617: * since each file is only a few bytes to begin with, I'm not
618: * concerned that this is going to tank any machines.
619: */
620:
1.115 schwarze 621: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 622: if (st.st_size > 0x7fffffff) {
1.86 schwarze 623: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.117 schwarze 624: return 0;
1.1 schwarze 625: }
626: *with_mmap = 1;
627: fb->sz = (size_t)st.st_size;
1.15 schwarze 628: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 629: if (fb->buf != MAP_FAILED)
1.117 schwarze 630: return 1;
1.1 schwarze 631: }
632:
1.115 schwarze 633: if (curp->gzip) {
1.118 schwarze 634: if ((gz = gzdopen(fd, "rb")) == NULL)
635: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.115 schwarze 636: } else
637: gz = NULL;
638:
1.1 schwarze 639: /*
640: * If this isn't a regular file (like, say, stdin), then we must
641: * go the old way and just read things in bit by bit.
642: */
643:
644: *with_mmap = 0;
645: off = 0;
646: fb->sz = 0;
647: fb->buf = NULL;
648: for (;;) {
649: if (off == fb->sz) {
650: if (fb->sz == (1U << 31)) {
1.86 schwarze 651: mandoc_msg(MANDOCERR_TOOLARGE, curp,
652: 0, 0, NULL);
1.1 schwarze 653: break;
654: }
655: resize_buf(fb, 65536);
656: }
1.115 schwarze 657: ssz = curp->gzip ?
658: gzread(gz, fb->buf + (int)off, fb->sz - off) :
659: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 660: if (ssz == 0) {
661: fb->sz = off;
1.117 schwarze 662: return 1;
1.1 schwarze 663: }
1.118 schwarze 664: if (ssz == -1)
665: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 666: off += (size_t)ssz;
667: }
668:
669: free(fb->buf);
670: fb->buf = NULL;
1.117 schwarze 671: return 0;
1.1 schwarze 672: }
673:
674: static void
675: mparse_end(struct mparse *curp)
676: {
1.110 schwarze 677: if (curp->man->macroset == MACROSET_NONE)
678: curp->man->macroset = MACROSET_MAN;
679: if (curp->man->macroset == MACROSET_MDOC)
680: mdoc_endparse(curp->man);
681: else
1.86 schwarze 682: man_endparse(curp->man);
1.1 schwarze 683: roff_endparse(curp->roff);
684: }
685:
1.15 schwarze 686: static void
687: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 688: {
1.61 schwarze 689: struct buf *svprimary;
1.1 schwarze 690: const char *svfile;
1.71 schwarze 691: size_t offset;
1.14 schwarze 692: static int recursion_depth;
693:
694: if (64 < recursion_depth) {
695: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 696: return;
1.14 schwarze 697: }
1.1 schwarze 698:
1.15 schwarze 699: /* Line number is per-file. */
700: svfile = curp->file;
701: curp->file = file;
1.61 schwarze 702: svprimary = curp->primary;
1.59 schwarze 703: curp->primary = &blk;
1.15 schwarze 704: curp->line = 1;
705: recursion_depth++;
706:
1.70 schwarze 707: /* Skip an UTF-8 byte order mark. */
708: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
709: (unsigned char)blk.buf[0] == 0xef &&
710: (unsigned char)blk.buf[1] == 0xbb &&
711: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 712: offset = 3;
1.70 schwarze 713: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 714: } else
715: offset = 0;
1.70 schwarze 716:
1.71 schwarze 717: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 718:
1.86 schwarze 719: if (--recursion_depth == 0)
1.15 schwarze 720: mparse_end(curp);
721:
1.61 schwarze 722: curp->primary = svprimary;
1.15 schwarze 723: curp->file = svfile;
724: }
725:
1.74 schwarze 726: /*
727: * Read the whole file into memory and call the parsers.
728: * Called recursively when an .so request is encountered.
729: */
1.15 schwarze 730: enum mandoclevel
731: mparse_readfd(struct mparse *curp, int fd, const char *file)
732: {
733: struct buf blk;
734: int with_mmap;
1.70 schwarze 735: int save_filenc;
1.1 schwarze 736:
1.67 schwarze 737: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 738: save_filenc = curp->filenc;
739: curp->filenc = curp->options &
740: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 741: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 742: curp->filenc = save_filenc;
1.67 schwarze 743: if (with_mmap)
744: munmap(blk.buf, blk.sz);
745: else
746: free(blk.buf);
747: }
1.117 schwarze 748: return curp->file_status;
1.58 schwarze 749: }
750:
1.122 schwarze 751: int
752: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 753: {
754: char *cp;
1.122 schwarze 755: int fd;
1.58 schwarze 756:
757: curp->file = file;
1.115 schwarze 758: cp = strrchr(file, '.');
759: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 760:
1.115 schwarze 761: /* First try to use the filename as it is. */
1.74 schwarze 762:
1.122 schwarze 763: if ((fd = open(file, O_RDONLY)) != -1)
764: return fd;
1.74 schwarze 765:
1.115 schwarze 766: /*
767: * If that doesn't work and the filename doesn't
768: * already end in .gz, try appending .gz.
769: */
1.74 schwarze 770:
1.115 schwarze 771: if ( ! curp->gzip) {
1.74 schwarze 772: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 773: fd = open(cp, O_RDONLY);
1.83 schwarze 774: free(cp);
1.122 schwarze 775: if (fd != -1) {
1.115 schwarze 776: curp->gzip = 1;
1.122 schwarze 777: return fd;
1.58 schwarze 778: }
779: }
780:
1.115 schwarze 781: /* Neither worked, give up. */
1.73 schwarze 782:
1.115 schwarze 783: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.122 schwarze 784: return -1;
1.1 schwarze 785: }
786:
787: struct mparse *
1.69 schwarze 788: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.119 schwarze 789: const char *defos)
1.1 schwarze 790: {
791: struct mparse *curp;
792:
793: curp = mandoc_calloc(1, sizeof(struct mparse));
794:
1.22 schwarze 795: curp->options = options;
1.1 schwarze 796: curp->wlevel = wlevel;
797: curp->mmsg = mmsg;
1.7 schwarze 798: curp->defos = defos;
1.1 schwarze 799:
1.119 schwarze 800: curp->roff = roff_alloc(curp, options);
1.112 schwarze 801: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
802: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 803: if (curp->options & MPARSE_MDOC) {
804: mdoc_hash_init();
1.112 schwarze 805: curp->man->macroset = MACROSET_MDOC;
806: } else if (curp->options & MPARSE_MAN) {
1.111 schwarze 807: man_hash_init();
1.112 schwarze 808: curp->man->macroset = MACROSET_MAN;
1.111 schwarze 809: }
1.113 schwarze 810: curp->man->first->tok = TOKEN_NONE;
1.117 schwarze 811: return curp;
1.1 schwarze 812: }
813:
814: void
815: mparse_reset(struct mparse *curp)
816: {
817: roff_reset(curp->roff);
1.124 schwarze 818: roff_man_reset(curp->man);
1.133 schwarze 819:
820: free(curp->sodest);
821: curp->sodest = NULL;
822:
1.4 schwarze 823: if (curp->secondary)
824: curp->secondary->sz = 0;
1.1 schwarze 825:
826: curp->file_status = MANDOCLEVEL_OK;
1.132 schwarze 827: curp->gzip = 0;
1.1 schwarze 828: }
829:
830: void
831: mparse_free(struct mparse *curp)
832: {
833:
1.112 schwarze 834: roff_man_free(curp->man);
1.133 schwarze 835: roff_free(curp->roff);
1.4 schwarze 836: if (curp->secondary)
837: free(curp->secondary->buf);
1.1 schwarze 838:
1.4 schwarze 839: free(curp->secondary);
1.23 schwarze 840: free(curp->sodest);
1.1 schwarze 841: free(curp);
842: }
843:
844: void
1.110 schwarze 845: mparse_result(struct mparse *curp, struct roff_man **man,
846: char **sodest)
1.1 schwarze 847: {
848:
1.23 schwarze 849: if (sodest && NULL != (*sodest = curp->sodest)) {
850: *man = NULL;
851: return;
852: }
1.1 schwarze 853: if (man)
854: *man = curp->man;
1.130 schwarze 855: }
856:
857: void
858: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
859: {
860: if (curp->file_status > *rc)
861: *rc = curp->file_status;
1.1 schwarze 862: }
863:
864: void
865: mandoc_vmsg(enum mandocerr t, struct mparse *m,
866: int ln, int pos, const char *fmt, ...)
867: {
868: char buf[256];
869: va_list ap;
870:
871: va_start(ap, fmt);
1.26 schwarze 872: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 873: va_end(ap);
874:
875: mandoc_msg(t, m, ln, pos, buf);
876: }
877:
878: void
1.25 schwarze 879: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 880: int ln, int col, const char *msg)
881: {
882: enum mandoclevel level;
883:
1.87 schwarze 884: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 885: while (er < mandoclimits[level])
886: level--;
887:
1.83 schwarze 888: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 889: return;
890:
891: if (m->mmsg)
892: (*m->mmsg)(er, level, m->file, ln, col, msg);
893:
894: if (m->file_status < level)
895: m->file_status = level;
896: }
897:
898: const char *
899: mparse_strerror(enum mandocerr er)
900: {
901:
1.117 schwarze 902: return mandocerrs[er];
1.1 schwarze 903: }
904:
905: const char *
906: mparse_strlevel(enum mandoclevel lvl)
907: {
1.117 schwarze 908: return mandoclevels[lvl];
1.4 schwarze 909: }
910:
911: void
912: mparse_keep(struct mparse *p)
913: {
914:
915: assert(NULL == p->secondary);
916: p->secondary = mandoc_calloc(1, sizeof(struct buf));
917: }
918:
919: const char *
920: mparse_getkeep(const struct mparse *p)
921: {
922:
923: assert(p->secondary);
1.117 schwarze 924: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 schwarze 925: }
![[BACK]](/icons/back.gif){kind=icon}
Return to read.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc