Annotation of mandoc/read.c, Revision 1.150
1.150 ! schwarze 1: /* $Id: read.c,v 1.149 2016/07/10 13:34:30 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.148 schwarze 4: * Copyright (c) 2010-2016 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>
1.81 schwarze 22: #if HAVE_MMAP
1.82 schwarze 23: #include <sys/mman.h>
1.80 schwarze 24: #include <sys/stat.h>
1.15 kristaps 25: #endif
1.1 kristaps 26:
27: #include <assert.h>
28: #include <ctype.h>
1.146 schwarze 29: #if HAVE_ERR
1.143 schwarze 30: #include <err.h>
1.146 schwarze 31: #endif
1.40 schwarze 32: #include <errno.h>
1.1 kristaps 33: #include <fcntl.h>
1.3 kristaps 34: #include <stdarg.h>
1.28 joerg 35: #include <stdint.h>
1.1 kristaps 36: #include <stdio.h>
37: #include <stdlib.h>
38: #include <string.h>
39: #include <unistd.h>
1.140 schwarze 40: #include <zlib.h>
1.1 kristaps 41:
1.133 schwarze 42: #include "mandoc_aux.h"
1.1 kristaps 43: #include "mandoc.h"
1.133 schwarze 44: #include "roff.h"
1.1 kristaps 45: #include "mdoc.h"
46: #include "man.h"
1.133 schwarze 47: #include "libmandoc.h"
1.139 schwarze 48: #include "roff_int.h"
1.1 kristaps 49:
50: #define REPARSE_LIMIT 1000
51:
52: struct mparse {
1.134 schwarze 53: struct roff_man *man; /* man parser */
1.1 kristaps 54: struct roff *roff; /* roff parser (!NULL) */
1.45 schwarze 55: char *sodest; /* filename pointed to by .so */
1.83 schwarze 56: const char *file; /* filename of current input file */
57: struct buf *primary; /* buffer currently being parsed */
58: struct buf *secondary; /* preprocessed copy of input */
59: const char *defos; /* default operating system */
60: mandocmsg mmsg; /* warning/error message handler */
61: enum mandoclevel file_status; /* status of current parse */
62: enum mandoclevel wlevel; /* ignore messages below this */
63: int options; /* parser options */
1.140 schwarze 64: int gzip; /* current input file is gzipped */
1.93 schwarze 65: int filenc; /* encoding of the current file */
1.1 kristaps 66: int reparse_count; /* finite interp. stack */
1.83 schwarze 67: int line; /* line number in the file */
1.1 kristaps 68: };
69:
1.84 schwarze 70: static void choose_parser(struct mparse *);
1.1 kristaps 71: static void resize_buf(struct buf *, size_t);
1.95 schwarze 72: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 73: static int read_whole_file(struct mparse *, const char *, int,
74: struct buf *, int *);
1.1 kristaps 75: static void mparse_end(struct mparse *);
1.37 schwarze 76: static void mparse_parse_buffer(struct mparse *, struct buf,
77: const char *);
1.1 kristaps 78:
1.3 kristaps 79: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
80: MANDOCERR_OK,
81: MANDOCERR_WARNING,
82: MANDOCERR_WARNING,
83: MANDOCERR_ERROR,
1.112 schwarze 84: MANDOCERR_UNSUPP,
1.3 kristaps 85: MANDOCERR_MAX,
86: MANDOCERR_MAX
87: };
88:
1.7 kristaps 89: static const char * const mandocerrs[MANDOCERR_MAX] = {
90: "ok",
91:
92: "generic warning",
93:
94: /* related to the prologue */
1.79 schwarze 95: "missing manual title, using UNTITLED",
96: "missing manual title, using \"\"",
1.54 schwarze 97: "lower case character in document title",
1.79 schwarze 98: "missing manual section, using \"\"",
1.7 kristaps 99: "unknown manual section",
1.54 schwarze 100: "missing date, using today's date",
1.7 kristaps 101: "cannot parse date, using it verbatim",
1.79 schwarze 102: "missing Os macro, using \"\"",
103: "duplicate prologue macro",
104: "late prologue macro",
105: "skipping late title macro",
1.7 kristaps 106: "prologue macros out of order",
107:
108: /* related to document structure */
109: ".so is fragile, better use ln(1)",
1.50 schwarze 110: "no document body",
1.54 schwarze 111: "content before first section header",
112: "first section is not \"NAME\"",
1.128 schwarze 113: "NAME section without name",
114: "NAME section without description",
115: "description not at the end of NAME",
116: "bad NAME section content",
1.121 schwarze 117: "missing description line, using \"\"",
1.7 kristaps 118: "sections out of conventional order",
1.54 schwarze 119: "duplicate section title",
120: "unexpected section",
1.87 schwarze 121: "unusual Xr order",
122: "unusual Xr punctuation",
1.86 schwarze 123: "AUTHORS section without An macro",
1.7 kristaps 124:
125: /* related to macros and nesting */
1.55 schwarze 126: "obsolete macro",
1.102 schwarze 127: "macro neither callable nor escaped",
1.7 kristaps 128: "skipping paragraph macro",
1.31 schwarze 129: "moving paragraph macro out of list",
1.7 kristaps 130: "skipping no-space macro",
131: "blocks badly nested",
132: "nested displays are not portable",
1.57 schwarze 133: "moving content out of list",
1.78 schwarze 134: "fill mode already enabled, skipping",
135: "fill mode already disabled, skipping",
1.7 kristaps 136: "line scope broken",
137:
138: /* related to missing macro arguments */
1.58 schwarze 139: "skipping empty request",
140: "conditional request controls empty scope",
1.7 kristaps 141: "skipping empty macro",
1.124 schwarze 142: "empty block",
1.62 schwarze 143: "empty argument, using 0n",
1.60 schwarze 144: "missing display type, using -ragged",
145: "list type is not the first argument",
146: "missing -width in -tag list, using 8n",
1.78 schwarze 147: "missing utility name, using \"\"",
1.123 schwarze 148: "missing function name, using \"\"",
1.60 schwarze 149: "empty head in list item",
150: "empty list item",
1.61 schwarze 151: "missing font type, using \\fR",
152: "unknown font type, using \\fR",
1.103 schwarze 153: "nothing follows prefix",
1.122 schwarze 154: "empty reference block",
1.60 schwarze 155: "missing -std argument, adding it",
1.125 schwarze 156: "missing option string, using \"\"",
157: "missing resource identifier, using \"\"",
1.90 schwarze 158: "missing eqn box, using \"\"",
1.7 kristaps 159:
160: /* related to bad macro arguments */
1.64 schwarze 161: "unterminated quoted argument",
1.7 kristaps 162: "duplicate argument",
1.76 schwarze 163: "skipping duplicate argument",
1.63 schwarze 164: "skipping duplicate display type",
165: "skipping duplicate list type",
1.76 schwarze 166: "skipping -width argument",
1.126 schwarze 167: "wrong number of cells",
1.7 kristaps 168: "unknown AT&T UNIX version",
1.88 schwarze 169: "comma in function argument",
1.89 schwarze 170: "parenthesis in function name",
1.67 schwarze 171: "invalid content in Rs block",
1.63 schwarze 172: "invalid Boolean argument",
173: "unknown font, skipping request",
1.126 schwarze 174: "odd number of characters in request",
1.7 kristaps 175:
176: /* related to plain text */
1.64 schwarze 177: "blank line in fill mode, using .sp",
178: "tab in filled text",
179: "whitespace at end of input line",
1.7 kristaps 180: "bad comment style",
1.64 schwarze 181: "invalid escape sequence",
182: "undefined string, using \"\"",
1.16 kristaps 183:
1.118 schwarze 184: /* related to tables */
185: "tbl line starts with span",
186: "tbl column starts with span",
187: "skipping vertical bar in tbl layout",
188:
1.7 kristaps 189: "generic error",
190:
191: /* related to tables */
1.116 schwarze 192: "non-alphabetic character in tbl options",
193: "skipping unknown tbl option",
194: "missing tbl option argument",
195: "wrong tbl option argument size",
1.118 schwarze 196: "empty tbl layout",
197: "invalid character in tbl layout",
198: "unmatched parenthesis in tbl layout",
1.119 schwarze 199: "tbl without any data cells",
200: "ignoring data in spanned tbl cell",
201: "ignoring extra tbl data cells",
202: "data block open at end of tbl",
1.7 kristaps 203:
1.68 schwarze 204: /* related to document structure and macros */
1.108 schwarze 205: NULL,
1.7 kristaps 206: "input stack limit exceeded, infinite loop?",
207: "skipping bad character",
1.68 schwarze 208: "skipping unknown macro",
1.112 schwarze 209: "skipping insecure request",
1.73 schwarze 210: "skipping item outside list",
1.68 schwarze 211: "skipping column outside column list",
212: "skipping end of block that is not open",
1.115 schwarze 213: "fewer RS blocks open, skipping",
1.68 schwarze 214: "inserting missing end of block",
215: "appending missing end of block",
216:
217: /* related to request and macro arguments */
1.7 kristaps 218: "escaped character not allowed in a name",
1.99 schwarze 219: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 220: "skipping display without arguments",
1.71 schwarze 221: "missing list type, using -item",
1.70 schwarze 222: "missing manual name, using \"\"",
1.71 schwarze 223: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 224: "unknown standard specifier",
1.71 schwarze 225: "skipping request without numeric argument",
1.109 schwarze 226: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 227: ".so request failed",
1.61 schwarze 228: "skipping all arguments",
229: "skipping excess arguments",
1.92 kristaps 230: "divide by zero",
1.112 schwarze 231:
232: "unsupported feature",
233: "input too large",
1.114 schwarze 234: "unsupported control character",
1.112 schwarze 235: "unsupported roff request",
1.119 schwarze 236: "eqn delim option in tbl",
1.118 schwarze 237: "unsupported tbl layout modifier",
1.112 schwarze 238: "ignoring macro in table",
1.7 kristaps 239: };
240:
241: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
242: "SUCCESS",
243: "RESERVED",
244: "WARNING",
245: "ERROR",
1.112 schwarze 246: "UNSUPP",
1.7 kristaps 247: "BADARG",
248: "SYSERR"
249: };
250:
1.47 schwarze 251:
1.1 kristaps 252: static void
253: resize_buf(struct buf *buf, size_t initial)
254: {
255:
256: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
257: buf->buf = mandoc_realloc(buf->buf, buf->sz);
258: }
259:
260: static void
1.84 schwarze 261: choose_parser(struct mparse *curp)
1.1 kristaps 262: {
1.83 schwarze 263: char *cp, *ep;
264: int format;
1.1 kristaps 265:
1.83 schwarze 266: /*
267: * If neither command line arguments -mdoc or -man select
268: * a parser nor the roff parser found a .Dd or .TH macro
269: * yet, look ahead in the main input buffer.
270: */
271:
272: if ((format = roff_getformat(curp->roff)) == 0) {
273: cp = curp->primary->buf;
274: ep = cp + curp->primary->sz;
275: while (cp < ep) {
1.85 schwarze 276: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 277: cp++;
278: if (cp[0] == 'D' && cp[1] == 'd') {
279: format = MPARSE_MDOC;
280: break;
281: }
282: if (cp[0] == 'T' && cp[1] == 'H') {
283: format = MPARSE_MAN;
284: break;
285: }
286: }
287: cp = memchr(cp, '\n', ep - cp);
288: if (cp == NULL)
289: break;
290: cp++;
291: }
1.1 kristaps 292: }
293:
1.83 schwarze 294: if (format == MPARSE_MDOC) {
1.136 schwarze 295: mdoc_hash_init();
1.137 schwarze 296: curp->man->macroset = MACROSET_MDOC;
1.138 schwarze 297: curp->man->first->tok = TOKEN_NONE;
1.137 schwarze 298: } else {
299: man_hash_init();
300: curp->man->macroset = MACROSET_MAN;
1.138 schwarze 301: curp->man->first->tok = TOKEN_NONE;
1.47 schwarze 302: }
1.1 kristaps 303: }
304:
305: /*
1.95 schwarze 306: * Main parse routine for a buffer.
307: * It assumes encoding and line numbering are already set up.
308: * It can recurse directly (for invocations of user-defined
309: * macros, inline equations, and input line traps)
310: * and indirectly (for .so file inclusion).
1.1 kristaps 311: */
312: static void
1.95 schwarze 313: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 314: {
315: const struct tbl_span *span;
316: struct buf ln;
1.113 schwarze 317: const char *save_file;
1.110 schwarze 318: char *cp;
1.95 schwarze 319: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 320: enum rofferr rr;
1.100 schwarze 321: int of;
1.1 kristaps 322: int lnn; /* line number in the real file */
1.110 schwarze 323: int fd;
1.1 kristaps 324: unsigned char c;
325:
1.95 schwarze 326: memset(&ln, 0, sizeof(ln));
1.1 kristaps 327:
1.47 schwarze 328: lnn = curp->line;
329: pos = 0;
1.1 kristaps 330:
1.95 schwarze 331: while (i < blk.sz) {
1.1 kristaps 332: if (0 == pos && '\0' == blk.buf[i])
333: break;
334:
335: if (start) {
336: curp->line = lnn;
337: curp->reparse_count = 0;
1.93 schwarze 338:
339: if (lnn < 3 &&
340: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 341: curp->filenc & MPARSE_LATIN1)
342: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 343: }
344:
1.95 schwarze 345: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 346:
347: /*
348: * When finding an unescaped newline character,
349: * leave the character loop to process the line.
350: * Skip a preceding carriage return, if any.
351: */
352:
1.95 schwarze 353: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 354: '\n' == blk.buf[i + 1])
355: ++i;
356: if ('\n' == blk.buf[i]) {
357: ++i;
358: ++lnn;
359: break;
360: }
361:
1.35 schwarze 362: /*
1.93 schwarze 363: * Make sure we have space for the worst
364: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 365: */
366:
1.95 schwarze 367: if (pos + 11 > ln.sz)
1.35 schwarze 368: resize_buf(&ln, 256);
369:
1.47 schwarze 370: /*
1.93 schwarze 371: * Encode 8-bit input.
1.1 kristaps 372: */
373:
1.93 schwarze 374: c = blk.buf[i];
375: if (c & 0x80) {
1.95 schwarze 376: if ( ! (curp->filenc && preconv_encode(
377: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 378: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
379: curp->line, pos, "0x%x", c);
1.93 schwarze 380: ln.buf[pos++] = '?';
381: i++;
382: }
383: continue;
384: }
385:
386: /*
387: * Exclude control characters.
388: */
1.1 kristaps 389:
1.93 schwarze 390: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 391: mandoc_vmsg(c == 0x00 || c == 0x04 ||
392: c > 0x0a ? MANDOCERR_CHAR_BAD :
393: MANDOCERR_CHAR_UNSUPP,
394: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 395: i++;
1.127 schwarze 396: if (c != '\r')
397: ln.buf[pos++] = '?';
1.1 kristaps 398: continue;
399: }
400:
401: /* Trailing backslash = a plain char. */
402:
1.95 schwarze 403: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 404: ln.buf[pos++] = blk.buf[i++];
405: continue;
406: }
407:
408: /*
409: * Found escape and at least one other character.
410: * When it's a newline character, skip it.
411: * When there is a carriage return in between,
412: * skip that one as well.
413: */
414:
1.95 schwarze 415: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 416: '\n' == blk.buf[i + 2])
417: ++i;
418: if ('\n' == blk.buf[i + 1]) {
419: i += 2;
420: ++lnn;
421: continue;
422: }
423:
1.13 kristaps 424: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 425: i += 2;
426: /* Comment, skip to end of line */
1.95 schwarze 427: for (; i < blk.sz; ++i) {
1.1 kristaps 428: if ('\n' == blk.buf[i]) {
429: ++i;
430: ++lnn;
431: break;
432: }
433: }
434:
435: /* Backout trailing whitespaces */
436: for (; pos > 0; --pos) {
437: if (ln.buf[pos - 1] != ' ')
438: break;
439: if (pos > 2 && ln.buf[pos - 2] == '\\')
440: break;
441: }
442: break;
443: }
444:
1.35 schwarze 445: /* Catch escaped bogus characters. */
446:
447: c = (unsigned char) blk.buf[i+1];
448:
1.47 schwarze 449: if ( ! (isascii(c) &&
450: (isgraph(c) || isblank(c)))) {
1.114 schwarze 451: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 452: curp->line, pos, "0x%x", c);
1.35 schwarze 453: i += 2;
454: ln.buf[pos++] = '?';
455: continue;
456: }
457:
1.1 kristaps 458: /* Some other escape sequence, copy & cont. */
459:
460: ln.buf[pos++] = blk.buf[i++];
461: ln.buf[pos++] = blk.buf[i++];
462: }
463:
1.95 schwarze 464: if (pos >= ln.sz)
1.1 kristaps 465: resize_buf(&ln, 256);
466:
467: ln.buf[pos] = '\0';
468:
469: /*
470: * A significant amount of complexity is contained by
471: * the roff preprocessor. It's line-oriented but can be
472: * expressed on one line, so we need at times to
473: * readjust our starting point and re-run it. The roff
474: * preprocessor can also readjust the buffers with new
475: * data, so we pass them in wholesale.
476: */
477:
478: of = 0;
479:
1.24 kristaps 480: /*
481: * Maintain a lookaside buffer of all parsed lines. We
482: * only do this if mparse_keep() has been invoked (the
483: * buffer may be accessed with mparse_getkeep()).
484: */
485:
486: if (curp->secondary) {
1.47 schwarze 487: curp->secondary->buf = mandoc_realloc(
488: curp->secondary->buf,
489: curp->secondary->sz + pos + 2);
490: memcpy(curp->secondary->buf +
491: curp->secondary->sz,
492: ln.buf, pos);
1.24 kristaps 493: curp->secondary->sz += pos;
494: curp->secondary->buf
495: [curp->secondary->sz] = '\n';
496: curp->secondary->sz++;
497: curp->secondary->buf
498: [curp->secondary->sz] = '\0';
499: }
1.1 kristaps 500: rerun:
1.96 schwarze 501: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 502:
503: switch (rr) {
1.47 schwarze 504: case ROFF_REPARSE:
1.1 kristaps 505: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 506: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 507: else
1.3 kristaps 508: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 509: curp->line, pos, NULL);
1.1 kristaps 510: pos = 0;
511: continue;
1.47 schwarze 512: case ROFF_APPEND:
1.95 schwarze 513: pos = strlen(ln.buf);
1.1 kristaps 514: continue;
1.47 schwarze 515: case ROFF_RERUN:
1.1 kristaps 516: goto rerun;
1.47 schwarze 517: case ROFF_IGN:
1.1 kristaps 518: pos = 0;
519: continue;
1.47 schwarze 520: case ROFF_SO:
1.95 schwarze 521: if ( ! (curp->options & MPARSE_SO) &&
522: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 523: curp->sodest = mandoc_strdup(ln.buf + of);
524: free(ln.buf);
525: return;
526: }
1.24 kristaps 527: /*
528: * We remove `so' clauses from our lookaside
529: * buffer because we're going to descend into
530: * the file recursively.
531: */
1.47 schwarze 532: if (curp->secondary)
1.25 kristaps 533: curp->secondary->sz -= pos + 1;
1.113 schwarze 534: save_file = curp->file;
1.148 schwarze 535: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 536: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 537: close(fd);
1.113 schwarze 538: curp->file = save_file;
539: } else {
540: curp->file = save_file;
1.52 schwarze 541: mandoc_vmsg(MANDOCERR_SO_FAIL,
542: curp, curp->line, pos,
543: ".so %s", ln.buf + of);
1.110 schwarze 544: ln.sz = mandoc_asprintf(&cp,
545: ".sp\nSee the file %s.\n.sp",
546: ln.buf + of);
547: free(ln.buf);
548: ln.buf = cp;
549: of = 0;
550: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 551: }
1.1 kristaps 552: pos = 0;
553: continue;
554: default:
555: break;
556: }
557:
1.150 ! schwarze 558: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 559: choose_parser(curp);
1.1 kristaps 560:
1.47 schwarze 561: /*
1.84 schwarze 562: * Lastly, push down into the parsers themselves.
1.1 kristaps 563: * If libroff returns ROFF_TBL, then add it to the
564: * currently open parse. Since we only get here if
565: * there does exist data (see tbl_data.c), we're
566: * guaranteed that something's been allocated.
567: * Do the same for ROFF_EQN.
568: */
569:
1.139 schwarze 570: if (rr == ROFF_TBL)
1.100 schwarze 571: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 572: roff_addtbl(curp->man, span);
573: else if (rr == ROFF_EQN)
574: roff_addeqn(curp->man, roff_eqn(curp->roff));
575: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 576: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 577: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
578: break;
1.1 kristaps 579:
580: /* Temporary buffers typically are not full. */
581:
582: if (0 == start && '\0' == blk.buf[i])
583: break;
584:
585: /* Start the next input line. */
586:
587: pos = 0;
588: }
589:
590: free(ln.buf);
591: }
592:
593: static int
1.40 schwarze 594: read_whole_file(struct mparse *curp, const char *file, int fd,
595: struct buf *fb, int *with_mmap)
1.1 kristaps 596: {
1.140 schwarze 597: gzFile gz;
1.1 kristaps 598: size_t off;
599: ssize_t ssz;
600:
1.81 schwarze 601: #if HAVE_MMAP
1.15 kristaps 602: struct stat st;
1.143 schwarze 603:
604: if (fstat(fd, &st) == -1)
605: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 606:
607: /*
608: * If we're a regular file, try just reading in the whole entry
609: * via mmap(). This is faster than reading it into blocks, and
610: * since each file is only a few bytes to begin with, I'm not
611: * concerned that this is going to tank any machines.
612: */
613:
1.140 schwarze 614: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 615: if (st.st_size > 0x7fffffff) {
1.111 schwarze 616: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 617: return 0;
1.1 kristaps 618: }
619: *with_mmap = 1;
620: fb->sz = (size_t)st.st_size;
1.37 schwarze 621: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 622: if (fb->buf != MAP_FAILED)
1.142 schwarze 623: return 1;
1.1 kristaps 624: }
1.15 kristaps 625: #endif
1.1 kristaps 626:
1.140 schwarze 627: if (curp->gzip) {
1.143 schwarze 628: if ((gz = gzdopen(fd, "rb")) == NULL)
629: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 630: } else
631: gz = NULL;
632:
1.1 kristaps 633: /*
634: * If this isn't a regular file (like, say, stdin), then we must
635: * go the old way and just read things in bit by bit.
636: */
637:
638: *with_mmap = 0;
639: off = 0;
640: fb->sz = 0;
641: fb->buf = NULL;
642: for (;;) {
643: if (off == fb->sz) {
644: if (fb->sz == (1U << 31)) {
1.111 schwarze 645: mandoc_msg(MANDOCERR_TOOLARGE, curp,
646: 0, 0, NULL);
1.1 kristaps 647: break;
648: }
649: resize_buf(fb, 65536);
650: }
1.140 schwarze 651: ssz = curp->gzip ?
652: gzread(gz, fb->buf + (int)off, fb->sz - off) :
653: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 654: if (ssz == 0) {
655: fb->sz = off;
1.142 schwarze 656: return 1;
1.1 kristaps 657: }
1.143 schwarze 658: if (ssz == -1)
659: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 660: off += (size_t)ssz;
661: }
662:
663: free(fb->buf);
664: fb->buf = NULL;
1.142 schwarze 665: return 0;
1.1 kristaps 666: }
667:
668: static void
669: mparse_end(struct mparse *curp)
670: {
1.135 schwarze 671: if (curp->man->macroset == MACROSET_NONE)
672: curp->man->macroset = MACROSET_MAN;
673: if (curp->man->macroset == MACROSET_MDOC)
674: mdoc_endparse(curp->man);
675: else
1.111 schwarze 676: man_endparse(curp->man);
1.1 kristaps 677: roff_endparse(curp->roff);
678: }
679:
680: static void
1.36 schwarze 681: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 682: {
1.85 schwarze 683: struct buf *svprimary;
1.28 joerg 684: const char *svfile;
1.95 schwarze 685: size_t offset;
1.36 schwarze 686: static int recursion_depth;
687:
688: if (64 < recursion_depth) {
689: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
690: return;
691: }
1.28 joerg 692:
693: /* Line number is per-file. */
694: svfile = curp->file;
695: curp->file = file;
1.85 schwarze 696: svprimary = curp->primary;
1.83 schwarze 697: curp->primary = &blk;
1.28 joerg 698: curp->line = 1;
1.36 schwarze 699: recursion_depth++;
1.28 joerg 700:
1.93 schwarze 701: /* Skip an UTF-8 byte order mark. */
702: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
703: (unsigned char)blk.buf[0] == 0xef &&
704: (unsigned char)blk.buf[1] == 0xbb &&
705: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 706: offset = 3;
1.93 schwarze 707: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 708: } else
709: offset = 0;
1.93 schwarze 710:
1.95 schwarze 711: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 712:
1.111 schwarze 713: if (--recursion_depth == 0)
1.28 joerg 714: mparse_end(curp);
715:
1.85 schwarze 716: curp->primary = svprimary;
1.28 joerg 717: curp->file = svfile;
718: }
719:
720: enum mandoclevel
1.104 schwarze 721: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 722: const char *file)
723: {
724: struct buf blk;
725:
1.104 schwarze 726: blk.buf = buf;
1.28 joerg 727: blk.sz = len;
728:
1.36 schwarze 729: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 730: return curp->file_status;
1.28 joerg 731: }
732:
1.98 schwarze 733: /*
734: * Read the whole file into memory and call the parsers.
735: * Called recursively when an .so request is encountered.
736: */
1.36 schwarze 737: enum mandoclevel
738: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 739: {
1.28 joerg 740: struct buf blk;
741: int with_mmap;
1.93 schwarze 742: int save_filenc;
1.1 kristaps 743:
1.91 schwarze 744: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 745: save_filenc = curp->filenc;
746: curp->filenc = curp->options &
747: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 748: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 749: curp->filenc = save_filenc;
1.81 schwarze 750: #if HAVE_MMAP
1.91 schwarze 751: if (with_mmap)
752: munmap(blk.buf, blk.sz);
753: else
1.28 joerg 754: #endif
1.91 schwarze 755: free(blk.buf);
756: }
1.142 schwarze 757: return curp->file_status;
1.82 schwarze 758: }
759:
1.148 schwarze 760: int
761: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 762: {
763: char *cp;
1.148 schwarze 764: int fd;
1.82 schwarze 765:
766: curp->file = file;
1.140 schwarze 767: cp = strrchr(file, '.');
768: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 769:
1.140 schwarze 770: /* First try to use the filename as it is. */
1.98 schwarze 771:
1.148 schwarze 772: if ((fd = open(file, O_RDONLY)) != -1)
773: return fd;
1.98 schwarze 774:
1.140 schwarze 775: /*
776: * If that doesn't work and the filename doesn't
777: * already end in .gz, try appending .gz.
778: */
1.98 schwarze 779:
1.140 schwarze 780: if ( ! curp->gzip) {
1.98 schwarze 781: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 782: fd = open(cp, O_RDONLY);
1.108 schwarze 783: free(cp);
1.148 schwarze 784: if (fd != -1) {
1.140 schwarze 785: curp->gzip = 1;
1.148 schwarze 786: return fd;
1.82 schwarze 787: }
788: }
789:
1.140 schwarze 790: /* Neither worked, give up. */
1.97 schwarze 791:
1.140 schwarze 792: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 793: return -1;
1.1 kristaps 794: }
795:
796: struct mparse *
1.94 schwarze 797: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.144 schwarze 798: const char *defos)
1.1 kristaps 799: {
800: struct mparse *curp;
1.10 kristaps 801:
1.1 kristaps 802: curp = mandoc_calloc(1, sizeof(struct mparse));
803:
1.44 schwarze 804: curp->options = options;
1.3 kristaps 805: curp->wlevel = wlevel;
1.1 kristaps 806: curp->mmsg = mmsg;
1.29 schwarze 807: curp->defos = defos;
1.1 kristaps 808:
1.144 schwarze 809: curp->roff = roff_alloc(curp, options);
1.137 schwarze 810: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
811: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 812: if (curp->options & MPARSE_MDOC) {
813: mdoc_hash_init();
1.137 schwarze 814: curp->man->macroset = MACROSET_MDOC;
815: } else if (curp->options & MPARSE_MAN) {
1.136 schwarze 816: man_hash_init();
1.137 schwarze 817: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 818: }
1.138 schwarze 819: curp->man->first->tok = TOKEN_NONE;
1.142 schwarze 820: return curp;
1.1 kristaps 821: }
822:
823: void
824: mparse_reset(struct mparse *curp)
825: {
826: roff_reset(curp->roff);
1.150 ! schwarze 827: roff_man_reset(curp->man);
1.24 kristaps 828: if (curp->secondary)
829: curp->secondary->sz = 0;
1.1 kristaps 830:
831: curp->file_status = MANDOCLEVEL_OK;
1.45 schwarze 832:
833: free(curp->sodest);
834: curp->sodest = NULL;
1.1 kristaps 835: }
836:
837: void
838: mparse_free(struct mparse *curp)
839: {
840:
1.137 schwarze 841: roff_man_free(curp->man);
1.1 kristaps 842: if (curp->roff)
843: roff_free(curp->roff);
1.24 kristaps 844: if (curp->secondary)
845: free(curp->secondary->buf);
1.1 kristaps 846:
1.24 kristaps 847: free(curp->secondary);
1.45 schwarze 848: free(curp->sodest);
1.1 kristaps 849: free(curp);
850: }
851:
852: void
1.135 schwarze 853: mparse_result(struct mparse *curp, struct roff_man **man,
854: char **sodest)
1.1 kristaps 855: {
856:
1.45 schwarze 857: if (sodest && NULL != (*sodest = curp->sodest)) {
858: *man = NULL;
859: return;
860: }
1.9 kristaps 861: if (man)
862: *man = curp->man;
1.3 kristaps 863: }
864:
865: void
866: mandoc_vmsg(enum mandocerr t, struct mparse *m,
867: int ln, int pos, const char *fmt, ...)
868: {
869: char buf[256];
870: va_list ap;
871:
872: va_start(ap, fmt);
1.48 schwarze 873: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 874: va_end(ap);
875:
876: mandoc_msg(t, m, ln, pos, buf);
877: }
878:
879: void
1.47 schwarze 880: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 881: int ln, int col, const char *msg)
882: {
883: enum mandoclevel level;
884:
1.112 schwarze 885: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 886: while (er < mandoclimits[level])
887: level--;
888:
1.108 schwarze 889: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 890: return;
891:
1.8 kristaps 892: if (m->mmsg)
893: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 894:
895: if (m->file_status < level)
896: m->file_status = level;
1.7 kristaps 897: }
898:
899: const char *
900: mparse_strerror(enum mandocerr er)
901: {
902:
1.142 schwarze 903: return mandocerrs[er];
1.7 kristaps 904: }
905:
906: const char *
907: mparse_strlevel(enum mandoclevel lvl)
908: {
1.142 schwarze 909: return mandoclevels[lvl];
1.24 kristaps 910: }
911:
912: void
913: mparse_keep(struct mparse *p)
914: {
915:
916: assert(NULL == p->secondary);
917: p->secondary = mandoc_calloc(1, sizeof(struct buf));
918: }
919:
920: const char *
921: mparse_getkeep(const struct mparse *p)
922: {
923:
924: assert(p->secondary);
1.142 schwarze 925: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 926: }
CVSweb