Annotation of src/usr.bin/mandoc/read.c, Revision 1.140
1.140 ! schwarze 1: /* $OpenBSD: read.c,v 1.139 2017/05/31 15:30:12 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.129 schwarze 4: * Copyright (c) 2010-2017 Ingo Schwarze <schwarze@openbsd.org>
1.20 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 schwarze 6: *
7: * Permission to use, copy, modify, and distribute this software for any
8: * purpose with or without fee is hereby granted, provided that the above
9: * copyright notice and this permission notice appear in all copies.
10: *
1.108 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.108 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18: */
1.58 schwarze 19: #include <sys/types.h>
20: #include <sys/mman.h>
1.1 schwarze 21: #include <sys/stat.h>
22:
23: #include <assert.h>
24: #include <ctype.h>
1.118 schwarze 25: #include <err.h>
1.18 schwarze 26: #include <errno.h>
1.1 schwarze 27: #include <fcntl.h>
28: #include <stdarg.h>
29: #include <stdio.h>
30: #include <stdlib.h>
31: #include <string.h>
32: #include <unistd.h>
1.115 schwarze 33: #include <zlib.h>
1.1 schwarze 34:
1.108 schwarze 35: #include "mandoc_aux.h"
1.1 schwarze 36: #include "mandoc.h"
1.108 schwarze 37: #include "roff.h"
1.1 schwarze 38: #include "mdoc.h"
39: #include "man.h"
1.108 schwarze 40: #include "libmandoc.h"
1.114 schwarze 41: #include "roff_int.h"
1.1 schwarze 42:
43: #define REPARSE_LIMIT 1000
44:
45: struct mparse {
1.133 schwarze 46: struct roff *roff; /* roff parser (!NULL) */
1.109 schwarze 47: struct roff_man *man; /* man parser */
1.23 schwarze 48: char *sodest; /* filename pointed to by .so */
1.59 schwarze 49: const char *file; /* filename of current input file */
50: struct buf *primary; /* buffer currently being parsed */
51: struct buf *secondary; /* preprocessed copy of input */
52: const char *defos; /* default operating system */
53: mandocmsg mmsg; /* warning/error message handler */
54: enum mandoclevel file_status; /* status of current parse */
55: enum mandoclevel wlevel; /* ignore messages below this */
56: int options; /* parser options */
1.115 schwarze 57: int gzip; /* current input file is gzipped */
1.70 schwarze 58: int filenc; /* encoding of the current file */
1.1 schwarze 59: int reparse_count; /* finite interp. stack */
1.59 schwarze 60: int line; /* line number in the file */
1.1 schwarze 61: };
62:
1.60 schwarze 63: static void choose_parser(struct mparse *);
1.1 schwarze 64: static void resize_buf(struct buf *, size_t);
1.135 schwarze 65: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.18 schwarze 66: static int read_whole_file(struct mparse *, const char *, int,
67: struct buf *, int *);
1.1 schwarze 68: static void mparse_end(struct mparse *);
1.15 schwarze 69: static void mparse_parse_buffer(struct mparse *, struct buf,
70: const char *);
1.1 schwarze 71:
72: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
73: MANDOCERR_OK,
1.137 schwarze 74: MANDOCERR_STYLE,
1.1 schwarze 75: MANDOCERR_WARNING,
76: MANDOCERR_ERROR,
1.87 schwarze 77: MANDOCERR_UNSUPP,
1.1 schwarze 78: MANDOCERR_MAX,
79: MANDOCERR_MAX
80: };
81:
82: static const char * const mandocerrs[MANDOCERR_MAX] = {
83: "ok",
84:
1.137 schwarze 85: "generic style suggestion",
1.138 schwarze 86:
87: "useless macro",
1.139 schwarze 88: "consider using OS macro",
1.140 ! schwarze 89: "description line ends with a full stop",
1.138 schwarze 90:
1.1 schwarze 91: "generic warning",
92:
93: /* related to the prologue */
1.57 schwarze 94: "missing manual title, using UNTITLED",
95: "missing manual title, using \"\"",
1.32 schwarze 96: "lower case character in document title",
1.57 schwarze 97: "missing manual section, using \"\"",
1.1 schwarze 98: "unknown manual section",
1.32 schwarze 99: "missing date, using today's date",
1.1 schwarze 100: "cannot parse date, using it verbatim",
1.57 schwarze 101: "missing Os macro, using \"\"",
102: "duplicate prologue macro",
103: "late prologue macro",
104: "skipping late title macro",
1.1 schwarze 105: "prologue macros out of order",
106:
107: /* related to document structure */
108: ".so is fragile, better use ln(1)",
1.28 schwarze 109: "no document body",
1.32 schwarze 110: "content before first section header",
111: "first section is not \"NAME\"",
1.129 schwarze 112: "NAME section without Nm before Nd",
1.103 schwarze 113: "NAME section without description",
114: "description not at the end of NAME",
115: "bad NAME section content",
1.129 schwarze 116: "missing comma before name",
1.96 schwarze 117: "missing description line, using \"\"",
1.134 schwarze 118: "description line outside NAME section",
1.1 schwarze 119: "sections out of conventional order",
1.32 schwarze 120: "duplicate section title",
121: "unexpected section",
1.63 schwarze 122: "unusual Xr order",
123: "unusual Xr punctuation",
1.62 schwarze 124: "AUTHORS section without An macro",
1.1 schwarze 125:
126: /* related to macros and nesting */
1.33 schwarze 127: "obsolete macro",
1.78 schwarze 128: "macro neither callable nor escaped",
1.1 schwarze 129: "skipping paragraph macro",
1.10 schwarze 130: "moving paragraph macro out of list",
1.1 schwarze 131: "skipping no-space macro",
132: "blocks badly nested",
133: "nested displays are not portable",
1.35 schwarze 134: "moving content out of list",
1.56 schwarze 135: "fill mode already enabled, skipping",
136: "fill mode already disabled, skipping",
1.1 schwarze 137: "line scope broken",
138:
139: /* related to missing macro arguments */
1.36 schwarze 140: "skipping empty request",
141: "conditional request controls empty scope",
1.1 schwarze 142: "skipping empty macro",
1.99 schwarze 143: "empty block",
1.40 schwarze 144: "empty argument, using 0n",
1.38 schwarze 145: "missing display type, using -ragged",
146: "list type is not the first argument",
1.125 schwarze 147: "missing -width in -tag list, using 6n",
1.56 schwarze 148: "missing utility name, using \"\"",
1.98 schwarze 149: "missing function name, using \"\"",
1.38 schwarze 150: "empty head in list item",
151: "empty list item",
1.39 schwarze 152: "missing font type, using \\fR",
153: "unknown font type, using \\fR",
1.79 schwarze 154: "nothing follows prefix",
1.97 schwarze 155: "empty reference block",
1.128 schwarze 156: "missing section argument",
1.38 schwarze 157: "missing -std argument, adding it",
1.100 schwarze 158: "missing option string, using \"\"",
159: "missing resource identifier, using \"\"",
1.66 schwarze 160: "missing eqn box, using \"\"",
1.1 schwarze 161:
162: /* related to bad macro arguments */
1.42 schwarze 163: "unterminated quoted argument",
1.1 schwarze 164: "duplicate argument",
1.54 schwarze 165: "skipping duplicate argument",
1.41 schwarze 166: "skipping duplicate display type",
167: "skipping duplicate list type",
1.54 schwarze 168: "skipping -width argument",
1.101 schwarze 169: "wrong number of cells",
1.1 schwarze 170: "unknown AT&T UNIX version",
1.64 schwarze 171: "comma in function argument",
1.65 schwarze 172: "parenthesis in function name",
1.45 schwarze 173: "invalid content in Rs block",
1.41 schwarze 174: "invalid Boolean argument",
175: "unknown font, skipping request",
1.101 schwarze 176: "odd number of characters in request",
1.1 schwarze 177:
178: /* related to plain text */
1.42 schwarze 179: "blank line in fill mode, using .sp",
180: "tab in filled text",
181: "whitespace at end of input line",
1.131 schwarze 182: "new sentence, new line",
1.1 schwarze 183: "bad comment style",
1.42 schwarze 184: "invalid escape sequence",
185: "undefined string, using \"\"",
1.3 schwarze 186:
1.93 schwarze 187: /* related to tables */
188: "tbl line starts with span",
189: "tbl column starts with span",
190: "skipping vertical bar in tbl layout",
191:
1.1 schwarze 192: "generic error",
1.3 schwarze 193:
1.1 schwarze 194: /* related to tables */
1.91 schwarze 195: "non-alphabetic character in tbl options",
196: "skipping unknown tbl option",
197: "missing tbl option argument",
198: "wrong tbl option argument size",
1.93 schwarze 199: "empty tbl layout",
200: "invalid character in tbl layout",
201: "unmatched parenthesis in tbl layout",
1.94 schwarze 202: "tbl without any data cells",
203: "ignoring data in spanned tbl cell",
204: "ignoring extra tbl data cells",
205: "data block open at end of tbl",
1.1 schwarze 206:
1.46 schwarze 207: /* related to document structure and macros */
1.83 schwarze 208: NULL,
1.1 schwarze 209: "input stack limit exceeded, infinite loop?",
210: "skipping bad character",
1.46 schwarze 211: "skipping unknown macro",
1.87 schwarze 212: "skipping insecure request",
1.51 schwarze 213: "skipping item outside list",
1.46 schwarze 214: "skipping column outside column list",
215: "skipping end of block that is not open",
1.90 schwarze 216: "fewer RS blocks open, skipping",
1.46 schwarze 217: "inserting missing end of block",
218: "appending missing end of block",
219:
220: /* related to request and macro arguments */
1.1 schwarze 221: "escaped character not allowed in a name",
1.75 schwarze 222: "NOT IMPLEMENTED: Bd -file",
1.120 schwarze 223: "skipping display without arguments",
1.49 schwarze 224: "missing list type, using -item",
1.48 schwarze 225: "missing manual name, using \"\"",
1.49 schwarze 226: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 227: "unknown standard specifier",
1.49 schwarze 228: "skipping request without numeric argument",
1.84 schwarze 229: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 230: ".so request failed",
1.39 schwarze 231: "skipping all arguments",
232: "skipping excess arguments",
1.68 schwarze 233: "divide by zero",
1.87 schwarze 234:
235: "unsupported feature",
236: "input too large",
1.89 schwarze 237: "unsupported control character",
1.87 schwarze 238: "unsupported roff request",
1.94 schwarze 239: "eqn delim option in tbl",
1.93 schwarze 240: "unsupported tbl layout modifier",
1.87 schwarze 241: "ignoring macro in table",
1.1 schwarze 242: };
243:
244: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
245: "SUCCESS",
1.137 schwarze 246: "STYLE",
1.1 schwarze 247: "WARNING",
248: "ERROR",
1.87 schwarze 249: "UNSUPP",
1.1 schwarze 250: "BADARG",
251: "SYSERR"
252: };
253:
1.25 schwarze 254:
1.1 schwarze 255: static void
256: resize_buf(struct buf *buf, size_t initial)
257: {
258:
259: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
260: buf->buf = mandoc_realloc(buf->buf, buf->sz);
261: }
262:
263: static void
1.60 schwarze 264: choose_parser(struct mparse *curp)
1.1 schwarze 265: {
1.59 schwarze 266: char *cp, *ep;
267: int format;
1.1 schwarze 268:
1.59 schwarze 269: /*
270: * If neither command line arguments -mdoc or -man select
271: * a parser nor the roff parser found a .Dd or .TH macro
272: * yet, look ahead in the main input buffer.
273: */
274:
275: if ((format = roff_getformat(curp->roff)) == 0) {
276: cp = curp->primary->buf;
277: ep = cp + curp->primary->sz;
278: while (cp < ep) {
1.61 schwarze 279: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 280: cp++;
281: if (cp[0] == 'D' && cp[1] == 'd') {
282: format = MPARSE_MDOC;
283: break;
284: }
285: if (cp[0] == 'T' && cp[1] == 'H') {
286: format = MPARSE_MAN;
287: break;
288: }
289: }
290: cp = memchr(cp, '\n', ep - cp);
291: if (cp == NULL)
292: break;
293: cp++;
294: }
1.1 schwarze 295: }
296:
1.59 schwarze 297: if (format == MPARSE_MDOC) {
1.112 schwarze 298: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 299: if (curp->man->mdocmac == NULL)
300: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 301: } else {
302: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 303: if (curp->man->manmac == NULL)
304: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 305: }
1.136 schwarze 306: curp->man->first->tok = TOKEN_NONE;
1.1 schwarze 307: }
308:
309: /*
1.71 schwarze 310: * Main parse routine for a buffer.
311: * It assumes encoding and line numbering are already set up.
312: * It can recurse directly (for invocations of user-defined
313: * macros, inline equations, and input line traps)
314: * and indirectly (for .so file inclusion).
1.1 schwarze 315: */
1.135 schwarze 316: static int
1.71 schwarze 317: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 318: {
319: const struct tbl_span *span;
320: struct buf ln;
1.88 schwarze 321: const char *save_file;
1.85 schwarze 322: char *cp;
1.71 schwarze 323: size_t pos; /* byte number in the ln buffer */
1.127 schwarze 324: size_t j; /* auxiliary byte number in the blk buffer */
1.1 schwarze 325: enum rofferr rr;
1.76 schwarze 326: int of;
1.1 schwarze 327: int lnn; /* line number in the real file */
1.85 schwarze 328: int fd;
1.1 schwarze 329: unsigned char c;
330:
1.71 schwarze 331: memset(&ln, 0, sizeof(ln));
1.1 schwarze 332:
1.25 schwarze 333: lnn = curp->line;
334: pos = 0;
1.1 schwarze 335:
1.71 schwarze 336: while (i < blk.sz) {
1.1 schwarze 337: if (0 == pos && '\0' == blk.buf[i])
338: break;
339:
340: if (start) {
341: curp->line = lnn;
342: curp->reparse_count = 0;
1.70 schwarze 343:
344: if (lnn < 3 &&
345: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 346: curp->filenc & MPARSE_LATIN1)
347: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 348: }
349:
1.71 schwarze 350: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 351:
352: /*
353: * When finding an unescaped newline character,
354: * leave the character loop to process the line.
355: * Skip a preceding carriage return, if any.
356: */
357:
1.71 schwarze 358: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 359: '\n' == blk.buf[i + 1])
360: ++i;
361: if ('\n' == blk.buf[i]) {
362: ++i;
363: ++lnn;
364: break;
365: }
366:
1.13 schwarze 367: /*
1.70 schwarze 368: * Make sure we have space for the worst
369: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 370: */
371:
1.71 schwarze 372: if (pos + 11 > ln.sz)
1.13 schwarze 373: resize_buf(&ln, 256);
374:
1.25 schwarze 375: /*
1.70 schwarze 376: * Encode 8-bit input.
1.1 schwarze 377: */
378:
1.70 schwarze 379: c = blk.buf[i];
380: if (c & 0x80) {
1.71 schwarze 381: if ( ! (curp->filenc && preconv_encode(
382: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 383: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
384: curp->line, pos, "0x%x", c);
1.70 schwarze 385: ln.buf[pos++] = '?';
386: i++;
387: }
388: continue;
389: }
390:
391: /*
392: * Exclude control characters.
393: */
1.1 schwarze 394:
1.70 schwarze 395: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 396: mandoc_vmsg(c == 0x00 || c == 0x04 ||
397: c > 0x0a ? MANDOCERR_CHAR_BAD :
398: MANDOCERR_CHAR_UNSUPP,
399: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 400: i++;
1.102 schwarze 401: if (c != '\r')
402: ln.buf[pos++] = '?';
1.1 schwarze 403: continue;
404: }
405:
406: /* Trailing backslash = a plain char. */
407:
1.71 schwarze 408: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 schwarze 409: ln.buf[pos++] = blk.buf[i++];
410: continue;
411: }
412:
413: /*
414: * Found escape and at least one other character.
415: * When it's a newline character, skip it.
416: * When there is a carriage return in between,
417: * skip that one as well.
418: */
419:
1.71 schwarze 420: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 schwarze 421: '\n' == blk.buf[i + 2])
422: ++i;
423: if ('\n' == blk.buf[i + 1]) {
424: i += 2;
425: ++lnn;
426: continue;
427: }
428:
429: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.127 schwarze 430: j = i;
1.1 schwarze 431: i += 2;
432: /* Comment, skip to end of line */
1.71 schwarze 433: for (; i < blk.sz; ++i) {
1.126 schwarze 434: if (blk.buf[i] != '\n')
435: continue;
436: if (blk.buf[i - 1] == ' ' ||
437: blk.buf[i - 1] == '\t')
438: mandoc_msg(
439: MANDOCERR_SPACE_EOL,
440: curp, curp->line,
1.127 schwarze 441: pos + i-1 - j, NULL);
1.126 schwarze 442: ++i;
443: ++lnn;
444: break;
1.1 schwarze 445: }
446:
447: /* Backout trailing whitespaces */
448: for (; pos > 0; --pos) {
449: if (ln.buf[pos - 1] != ' ')
450: break;
451: if (pos > 2 && ln.buf[pos - 2] == '\\')
452: break;
453: }
454: break;
455: }
456:
1.13 schwarze 457: /* Catch escaped bogus characters. */
458:
459: c = (unsigned char) blk.buf[i+1];
460:
1.25 schwarze 461: if ( ! (isascii(c) &&
462: (isgraph(c) || isblank(c)))) {
1.89 schwarze 463: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.56 schwarze 464: curp->line, pos, "0x%x", c);
1.13 schwarze 465: i += 2;
466: ln.buf[pos++] = '?';
467: continue;
468: }
469:
1.1 schwarze 470: /* Some other escape sequence, copy & cont. */
471:
472: ln.buf[pos++] = blk.buf[i++];
473: ln.buf[pos++] = blk.buf[i++];
474: }
475:
1.71 schwarze 476: if (pos >= ln.sz)
1.1 schwarze 477: resize_buf(&ln, 256);
478:
479: ln.buf[pos] = '\0';
480:
481: /*
482: * A significant amount of complexity is contained by
483: * the roff preprocessor. It's line-oriented but can be
484: * expressed on one line, so we need at times to
485: * readjust our starting point and re-run it. The roff
486: * preprocessor can also readjust the buffers with new
487: * data, so we pass them in wholesale.
488: */
489:
490: of = 0;
491:
1.4 schwarze 492: /*
493: * Maintain a lookaside buffer of all parsed lines. We
494: * only do this if mparse_keep() has been invoked (the
495: * buffer may be accessed with mparse_getkeep()).
496: */
497:
498: if (curp->secondary) {
1.25 schwarze 499: curp->secondary->buf = mandoc_realloc(
500: curp->secondary->buf,
501: curp->secondary->sz + pos + 2);
502: memcpy(curp->secondary->buf +
503: curp->secondary->sz,
504: ln.buf, pos);
1.4 schwarze 505: curp->secondary->sz += pos;
506: curp->secondary->buf
507: [curp->secondary->sz] = '\n';
508: curp->secondary->sz++;
509: curp->secondary->buf
510: [curp->secondary->sz] = '\0';
511: }
1.1 schwarze 512: rerun:
1.72 schwarze 513: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 514:
515: switch (rr) {
1.25 schwarze 516: case ROFF_REPARSE:
1.135 schwarze 517: if (++curp->reparse_count > REPARSE_LIMIT)
1.1 schwarze 518: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 519: curp->line, pos, NULL);
1.135 schwarze 520: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
521: start == 1) {
522: pos = 0;
523: continue;
524: }
525: free(ln.buf);
526: return 0;
1.25 schwarze 527: case ROFF_APPEND:
1.71 schwarze 528: pos = strlen(ln.buf);
1.1 schwarze 529: continue;
1.25 schwarze 530: case ROFF_RERUN:
1.1 schwarze 531: goto rerun;
1.25 schwarze 532: case ROFF_IGN:
1.1 schwarze 533: pos = 0;
534: continue;
1.25 schwarze 535: case ROFF_SO:
1.71 schwarze 536: if ( ! (curp->options & MPARSE_SO) &&
537: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 538: curp->sodest = mandoc_strdup(ln.buf + of);
539: free(ln.buf);
1.135 schwarze 540: return 1;
1.23 schwarze 541: }
1.4 schwarze 542: /*
543: * We remove `so' clauses from our lookaside
544: * buffer because we're going to descend into
545: * the file recursively.
546: */
1.25 schwarze 547: if (curp->secondary)
1.4 schwarze 548: curp->secondary->sz -= pos + 1;
1.88 schwarze 549: save_file = curp->file;
1.122 schwarze 550: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 551: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 552: close(fd);
1.88 schwarze 553: curp->file = save_file;
554: } else {
555: curp->file = save_file;
1.30 schwarze 556: mandoc_vmsg(MANDOCERR_SO_FAIL,
557: curp, curp->line, pos,
558: ".so %s", ln.buf + of);
1.85 schwarze 559: ln.sz = mandoc_asprintf(&cp,
560: ".sp\nSee the file %s.\n.sp",
561: ln.buf + of);
562: free(ln.buf);
563: ln.buf = cp;
564: of = 0;
565: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 566: }
1.1 schwarze 567: pos = 0;
568: continue;
569: default:
570: break;
571: }
572:
1.124 schwarze 573: if (curp->man->macroset == MACROSET_NONE)
1.60 schwarze 574: choose_parser(curp);
1.1 schwarze 575:
1.25 schwarze 576: /*
1.60 schwarze 577: * Lastly, push down into the parsers themselves.
1.1 schwarze 578: * If libroff returns ROFF_TBL, then add it to the
579: * currently open parse. Since we only get here if
580: * there does exist data (see tbl_data.c), we're
581: * guaranteed that something's been allocated.
582: * Do the same for ROFF_EQN.
583: */
584:
1.114 schwarze 585: if (rr == ROFF_TBL)
1.76 schwarze 586: while ((span = roff_span(curp->roff)) != NULL)
1.114 schwarze 587: roff_addtbl(curp->man, span);
588: else if (rr == ROFF_EQN)
589: roff_addeqn(curp->man, roff_eqn(curp->roff));
590: else if ((curp->man->macroset == MACROSET_MDOC ?
1.110 schwarze 591: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 592: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
593: break;
1.1 schwarze 594:
595: /* Temporary buffers typically are not full. */
596:
597: if (0 == start && '\0' == blk.buf[i])
598: break;
599:
600: /* Start the next input line. */
601:
602: pos = 0;
603: }
604:
605: free(ln.buf);
1.135 schwarze 606: return 1;
1.1 schwarze 607: }
608:
609: static int
1.18 schwarze 610: read_whole_file(struct mparse *curp, const char *file, int fd,
611: struct buf *fb, int *with_mmap)
1.1 schwarze 612: {
613: struct stat st;
1.115 schwarze 614: gzFile gz;
1.1 schwarze 615: size_t off;
616: ssize_t ssz;
617:
1.118 schwarze 618: if (fstat(fd, &st) == -1)
619: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 620:
621: /*
622: * If we're a regular file, try just reading in the whole entry
623: * via mmap(). This is faster than reading it into blocks, and
624: * since each file is only a few bytes to begin with, I'm not
625: * concerned that this is going to tank any machines.
626: */
627:
1.115 schwarze 628: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 629: if (st.st_size > 0x7fffffff) {
1.86 schwarze 630: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.117 schwarze 631: return 0;
1.1 schwarze 632: }
633: *with_mmap = 1;
634: fb->sz = (size_t)st.st_size;
1.15 schwarze 635: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 636: if (fb->buf != MAP_FAILED)
1.117 schwarze 637: return 1;
1.1 schwarze 638: }
639:
1.115 schwarze 640: if (curp->gzip) {
1.118 schwarze 641: if ((gz = gzdopen(fd, "rb")) == NULL)
642: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.115 schwarze 643: } else
644: gz = NULL;
645:
1.1 schwarze 646: /*
647: * If this isn't a regular file (like, say, stdin), then we must
648: * go the old way and just read things in bit by bit.
649: */
650:
651: *with_mmap = 0;
652: off = 0;
653: fb->sz = 0;
654: fb->buf = NULL;
655: for (;;) {
656: if (off == fb->sz) {
657: if (fb->sz == (1U << 31)) {
1.86 schwarze 658: mandoc_msg(MANDOCERR_TOOLARGE, curp,
659: 0, 0, NULL);
1.1 schwarze 660: break;
661: }
662: resize_buf(fb, 65536);
663: }
1.115 schwarze 664: ssz = curp->gzip ?
665: gzread(gz, fb->buf + (int)off, fb->sz - off) :
666: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 667: if (ssz == 0) {
668: fb->sz = off;
1.117 schwarze 669: return 1;
1.1 schwarze 670: }
1.118 schwarze 671: if (ssz == -1)
672: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 673: off += (size_t)ssz;
674: }
675:
676: free(fb->buf);
677: fb->buf = NULL;
1.117 schwarze 678: return 0;
1.1 schwarze 679: }
680:
681: static void
682: mparse_end(struct mparse *curp)
683: {
1.110 schwarze 684: if (curp->man->macroset == MACROSET_NONE)
685: curp->man->macroset = MACROSET_MAN;
686: if (curp->man->macroset == MACROSET_MDOC)
687: mdoc_endparse(curp->man);
688: else
1.86 schwarze 689: man_endparse(curp->man);
1.1 schwarze 690: roff_endparse(curp->roff);
691: }
692:
1.15 schwarze 693: static void
694: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 695: {
1.61 schwarze 696: struct buf *svprimary;
1.1 schwarze 697: const char *svfile;
1.71 schwarze 698: size_t offset;
1.14 schwarze 699: static int recursion_depth;
700:
701: if (64 < recursion_depth) {
702: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 703: return;
1.14 schwarze 704: }
1.1 schwarze 705:
1.15 schwarze 706: /* Line number is per-file. */
707: svfile = curp->file;
708: curp->file = file;
1.61 schwarze 709: svprimary = curp->primary;
1.59 schwarze 710: curp->primary = &blk;
1.15 schwarze 711: curp->line = 1;
712: recursion_depth++;
713:
1.70 schwarze 714: /* Skip an UTF-8 byte order mark. */
715: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
716: (unsigned char)blk.buf[0] == 0xef &&
717: (unsigned char)blk.buf[1] == 0xbb &&
718: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 719: offset = 3;
1.70 schwarze 720: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 721: } else
722: offset = 0;
1.70 schwarze 723:
1.71 schwarze 724: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 725:
1.86 schwarze 726: if (--recursion_depth == 0)
1.15 schwarze 727: mparse_end(curp);
728:
1.61 schwarze 729: curp->primary = svprimary;
1.15 schwarze 730: curp->file = svfile;
731: }
732:
1.74 schwarze 733: /*
734: * Read the whole file into memory and call the parsers.
735: * Called recursively when an .so request is encountered.
736: */
1.15 schwarze 737: enum mandoclevel
738: mparse_readfd(struct mparse *curp, int fd, const char *file)
739: {
740: struct buf blk;
741: int with_mmap;
1.70 schwarze 742: int save_filenc;
1.1 schwarze 743:
1.67 schwarze 744: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 745: save_filenc = curp->filenc;
746: curp->filenc = curp->options &
747: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 748: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 749: curp->filenc = save_filenc;
1.67 schwarze 750: if (with_mmap)
751: munmap(blk.buf, blk.sz);
752: else
753: free(blk.buf);
754: }
1.117 schwarze 755: return curp->file_status;
1.58 schwarze 756: }
757:
1.122 schwarze 758: int
759: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 760: {
761: char *cp;
1.122 schwarze 762: int fd;
1.58 schwarze 763:
764: curp->file = file;
1.115 schwarze 765: cp = strrchr(file, '.');
766: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 767:
1.115 schwarze 768: /* First try to use the filename as it is. */
1.74 schwarze 769:
1.122 schwarze 770: if ((fd = open(file, O_RDONLY)) != -1)
771: return fd;
1.74 schwarze 772:
1.115 schwarze 773: /*
774: * If that doesn't work and the filename doesn't
775: * already end in .gz, try appending .gz.
776: */
1.74 schwarze 777:
1.115 schwarze 778: if ( ! curp->gzip) {
1.74 schwarze 779: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 780: fd = open(cp, O_RDONLY);
1.83 schwarze 781: free(cp);
1.122 schwarze 782: if (fd != -1) {
1.115 schwarze 783: curp->gzip = 1;
1.122 schwarze 784: return fd;
1.58 schwarze 785: }
786: }
787:
1.115 schwarze 788: /* Neither worked, give up. */
1.73 schwarze 789:
1.115 schwarze 790: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.122 schwarze 791: return -1;
1.1 schwarze 792: }
793:
794: struct mparse *
1.69 schwarze 795: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.119 schwarze 796: const char *defos)
1.1 schwarze 797: {
798: struct mparse *curp;
799:
800: curp = mandoc_calloc(1, sizeof(struct mparse));
801:
1.22 schwarze 802: curp->options = options;
1.1 schwarze 803: curp->wlevel = wlevel;
804: curp->mmsg = mmsg;
1.7 schwarze 805: curp->defos = defos;
1.1 schwarze 806:
1.119 schwarze 807: curp->roff = roff_alloc(curp, options);
1.112 schwarze 808: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
809: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 810: if (curp->options & MPARSE_MDOC) {
1.112 schwarze 811: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 812: if (curp->man->mdocmac == NULL)
813: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 814: } else if (curp->options & MPARSE_MAN) {
815: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 816: if (curp->man->manmac == NULL)
817: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 818: }
1.113 schwarze 819: curp->man->first->tok = TOKEN_NONE;
1.117 schwarze 820: return curp;
1.1 schwarze 821: }
822:
823: void
824: mparse_reset(struct mparse *curp)
825: {
826: roff_reset(curp->roff);
1.124 schwarze 827: roff_man_reset(curp->man);
1.133 schwarze 828:
829: free(curp->sodest);
830: curp->sodest = NULL;
831:
1.4 schwarze 832: if (curp->secondary)
833: curp->secondary->sz = 0;
1.1 schwarze 834:
835: curp->file_status = MANDOCLEVEL_OK;
1.132 schwarze 836: curp->gzip = 0;
1.1 schwarze 837: }
838:
839: void
840: mparse_free(struct mparse *curp)
841: {
842:
1.136 schwarze 843: roffhash_free(curp->man->mdocmac);
844: roffhash_free(curp->man->manmac);
1.112 schwarze 845: roff_man_free(curp->man);
1.133 schwarze 846: roff_free(curp->roff);
1.4 schwarze 847: if (curp->secondary)
848: free(curp->secondary->buf);
1.1 schwarze 849:
1.4 schwarze 850: free(curp->secondary);
1.23 schwarze 851: free(curp->sodest);
1.1 schwarze 852: free(curp);
853: }
854:
855: void
1.110 schwarze 856: mparse_result(struct mparse *curp, struct roff_man **man,
857: char **sodest)
1.1 schwarze 858: {
859:
1.23 schwarze 860: if (sodest && NULL != (*sodest = curp->sodest)) {
861: *man = NULL;
862: return;
863: }
1.1 schwarze 864: if (man)
865: *man = curp->man;
1.130 schwarze 866: }
867:
868: void
869: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
870: {
871: if (curp->file_status > *rc)
872: *rc = curp->file_status;
1.1 schwarze 873: }
874:
875: void
876: mandoc_vmsg(enum mandocerr t, struct mparse *m,
877: int ln, int pos, const char *fmt, ...)
878: {
879: char buf[256];
880: va_list ap;
881:
882: va_start(ap, fmt);
1.26 schwarze 883: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 884: va_end(ap);
885:
886: mandoc_msg(t, m, ln, pos, buf);
887: }
888:
889: void
1.25 schwarze 890: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 891: int ln, int col, const char *msg)
892: {
893: enum mandoclevel level;
894:
1.87 schwarze 895: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 896: while (er < mandoclimits[level])
897: level--;
898:
1.83 schwarze 899: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 900: return;
901:
902: if (m->mmsg)
903: (*m->mmsg)(er, level, m->file, ln, col, msg);
904:
905: if (m->file_status < level)
906: m->file_status = level;
907: }
908:
909: const char *
910: mparse_strerror(enum mandocerr er)
911: {
912:
1.117 schwarze 913: return mandocerrs[er];
1.1 schwarze 914: }
915:
916: const char *
917: mparse_strlevel(enum mandoclevel lvl)
918: {
1.117 schwarze 919: return mandoclevels[lvl];
1.4 schwarze 920: }
921:
922: void
923: mparse_keep(struct mparse *p)
924: {
925:
926: assert(NULL == p->secondary);
927: p->secondary = mandoc_calloc(1, sizeof(struct buf));
928: }
929:
930: const char *
931: mparse_getkeep(const struct mparse *p)
932: {
933:
934: assert(p->secondary);
1.117 schwarze 935: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 schwarze 936: }