Annotation of mandoc/read.c, Revision 1.168
1.168 ! schwarze 1: /* $Id: read.c,v 1.167 2017/05/31 15:31:00 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.156 schwarze 4: * Copyright (c) 2010-2017 Ingo Schwarze <schwarze@openbsd.org>
1.42 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 kristaps 6: *
7: * Permission to use, copy, modify, and distribute this software for any
8: * purpose with or without fee is hereby granted, provided that the above
9: * copyright notice and this permission notice appear in all copies.
10: *
1.133 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 kristaps 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.133 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 kristaps 14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18: */
1.11 kristaps 19: #include "config.h"
20:
1.80 schwarze 21: #include <sys/types.h>
1.82 schwarze 22: #include <sys/mman.h>
1.80 schwarze 23: #include <sys/stat.h>
1.1 kristaps 24:
25: #include <assert.h>
26: #include <ctype.h>
1.146 schwarze 27: #if HAVE_ERR
1.143 schwarze 28: #include <err.h>
1.146 schwarze 29: #endif
1.40 schwarze 30: #include <errno.h>
1.1 kristaps 31: #include <fcntl.h>
1.3 kristaps 32: #include <stdarg.h>
1.1 kristaps 33: #include <stdio.h>
34: #include <stdlib.h>
35: #include <string.h>
36: #include <unistd.h>
1.140 schwarze 37: #include <zlib.h>
1.1 kristaps 38:
1.133 schwarze 39: #include "mandoc_aux.h"
1.1 kristaps 40: #include "mandoc.h"
1.133 schwarze 41: #include "roff.h"
1.1 kristaps 42: #include "mdoc.h"
43: #include "man.h"
1.133 schwarze 44: #include "libmandoc.h"
1.139 schwarze 45: #include "roff_int.h"
1.1 kristaps 46:
47: #define REPARSE_LIMIT 1000
48:
49: struct mparse {
1.160 schwarze 50: struct roff *roff; /* roff parser (!NULL) */
1.134 schwarze 51: struct roff_man *man; /* man parser */
1.45 schwarze 52: char *sodest; /* filename pointed to by .so */
1.83 schwarze 53: const char *file; /* filename of current input file */
54: struct buf *primary; /* buffer currently being parsed */
55: struct buf *secondary; /* preprocessed copy of input */
56: const char *defos; /* default operating system */
57: mandocmsg mmsg; /* warning/error message handler */
58: enum mandoclevel file_status; /* status of current parse */
59: enum mandoclevel wlevel; /* ignore messages below this */
60: int options; /* parser options */
1.140 schwarze 61: int gzip; /* current input file is gzipped */
1.93 schwarze 62: int filenc; /* encoding of the current file */
1.1 kristaps 63: int reparse_count; /* finite interp. stack */
1.83 schwarze 64: int line; /* line number in the file */
1.1 kristaps 65: };
66:
1.84 schwarze 67: static void choose_parser(struct mparse *);
1.1 kristaps 68: static void resize_buf(struct buf *, size_t);
1.163 schwarze 69: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 70: static int read_whole_file(struct mparse *, const char *, int,
71: struct buf *, int *);
1.1 kristaps 72: static void mparse_end(struct mparse *);
1.37 schwarze 73: static void mparse_parse_buffer(struct mparse *, struct buf,
74: const char *);
1.1 kristaps 75:
1.3 kristaps 76: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
77: MANDOCERR_OK,
1.165 schwarze 78: MANDOCERR_STYLE,
1.3 kristaps 79: MANDOCERR_WARNING,
80: MANDOCERR_ERROR,
1.112 schwarze 81: MANDOCERR_UNSUPP,
1.3 kristaps 82: MANDOCERR_MAX,
83: MANDOCERR_MAX
84: };
85:
1.7 kristaps 86: static const char * const mandocerrs[MANDOCERR_MAX] = {
87: "ok",
88:
1.165 schwarze 89: "generic style suggestion",
1.166 schwarze 90:
91: "useless macro",
1.167 schwarze 92: "consider using OS macro",
1.168 ! schwarze 93: "description line ends with a full stop",
1.166 schwarze 94:
1.7 kristaps 95: "generic warning",
96:
97: /* related to the prologue */
1.79 schwarze 98: "missing manual title, using UNTITLED",
99: "missing manual title, using \"\"",
1.54 schwarze 100: "lower case character in document title",
1.79 schwarze 101: "missing manual section, using \"\"",
1.7 kristaps 102: "unknown manual section",
1.54 schwarze 103: "missing date, using today's date",
1.7 kristaps 104: "cannot parse date, using it verbatim",
1.79 schwarze 105: "missing Os macro, using \"\"",
106: "duplicate prologue macro",
107: "late prologue macro",
108: "skipping late title macro",
1.7 kristaps 109: "prologue macros out of order",
110:
111: /* related to document structure */
112: ".so is fragile, better use ln(1)",
1.50 schwarze 113: "no document body",
1.54 schwarze 114: "content before first section header",
115: "first section is not \"NAME\"",
1.156 schwarze 116: "NAME section without Nm before Nd",
1.128 schwarze 117: "NAME section without description",
118: "description not at the end of NAME",
119: "bad NAME section content",
1.156 schwarze 120: "missing comma before name",
1.121 schwarze 121: "missing description line, using \"\"",
1.162 schwarze 122: "description line outside NAME section",
1.7 kristaps 123: "sections out of conventional order",
1.54 schwarze 124: "duplicate section title",
125: "unexpected section",
1.87 schwarze 126: "unusual Xr order",
127: "unusual Xr punctuation",
1.86 schwarze 128: "AUTHORS section without An macro",
1.7 kristaps 129:
130: /* related to macros and nesting */
1.55 schwarze 131: "obsolete macro",
1.102 schwarze 132: "macro neither callable nor escaped",
1.7 kristaps 133: "skipping paragraph macro",
1.31 schwarze 134: "moving paragraph macro out of list",
1.7 kristaps 135: "skipping no-space macro",
136: "blocks badly nested",
137: "nested displays are not portable",
1.57 schwarze 138: "moving content out of list",
1.78 schwarze 139: "fill mode already enabled, skipping",
140: "fill mode already disabled, skipping",
1.7 kristaps 141: "line scope broken",
142:
143: /* related to missing macro arguments */
1.58 schwarze 144: "skipping empty request",
145: "conditional request controls empty scope",
1.7 kristaps 146: "skipping empty macro",
1.124 schwarze 147: "empty block",
1.62 schwarze 148: "empty argument, using 0n",
1.60 schwarze 149: "missing display type, using -ragged",
150: "list type is not the first argument",
1.152 schwarze 151: "missing -width in -tag list, using 6n",
1.78 schwarze 152: "missing utility name, using \"\"",
1.123 schwarze 153: "missing function name, using \"\"",
1.60 schwarze 154: "empty head in list item",
155: "empty list item",
1.61 schwarze 156: "missing font type, using \\fR",
157: "unknown font type, using \\fR",
1.103 schwarze 158: "nothing follows prefix",
1.122 schwarze 159: "empty reference block",
1.155 schwarze 160: "missing section argument",
1.60 schwarze 161: "missing -std argument, adding it",
1.125 schwarze 162: "missing option string, using \"\"",
163: "missing resource identifier, using \"\"",
1.90 schwarze 164: "missing eqn box, using \"\"",
1.7 kristaps 165:
166: /* related to bad macro arguments */
1.64 schwarze 167: "unterminated quoted argument",
1.7 kristaps 168: "duplicate argument",
1.76 schwarze 169: "skipping duplicate argument",
1.63 schwarze 170: "skipping duplicate display type",
171: "skipping duplicate list type",
1.76 schwarze 172: "skipping -width argument",
1.126 schwarze 173: "wrong number of cells",
1.7 kristaps 174: "unknown AT&T UNIX version",
1.88 schwarze 175: "comma in function argument",
1.89 schwarze 176: "parenthesis in function name",
1.67 schwarze 177: "invalid content in Rs block",
1.63 schwarze 178: "invalid Boolean argument",
179: "unknown font, skipping request",
1.126 schwarze 180: "odd number of characters in request",
1.7 kristaps 181:
182: /* related to plain text */
1.64 schwarze 183: "blank line in fill mode, using .sp",
184: "tab in filled text",
185: "whitespace at end of input line",
1.158 schwarze 186: "new sentence, new line",
1.7 kristaps 187: "bad comment style",
1.64 schwarze 188: "invalid escape sequence",
189: "undefined string, using \"\"",
1.16 kristaps 190:
1.118 schwarze 191: /* related to tables */
192: "tbl line starts with span",
193: "tbl column starts with span",
194: "skipping vertical bar in tbl layout",
195:
1.7 kristaps 196: "generic error",
197:
198: /* related to tables */
1.116 schwarze 199: "non-alphabetic character in tbl options",
200: "skipping unknown tbl option",
201: "missing tbl option argument",
202: "wrong tbl option argument size",
1.118 schwarze 203: "empty tbl layout",
204: "invalid character in tbl layout",
205: "unmatched parenthesis in tbl layout",
1.119 schwarze 206: "tbl without any data cells",
207: "ignoring data in spanned tbl cell",
208: "ignoring extra tbl data cells",
209: "data block open at end of tbl",
1.7 kristaps 210:
1.68 schwarze 211: /* related to document structure and macros */
1.108 schwarze 212: NULL,
1.7 kristaps 213: "input stack limit exceeded, infinite loop?",
214: "skipping bad character",
1.68 schwarze 215: "skipping unknown macro",
1.112 schwarze 216: "skipping insecure request",
1.73 schwarze 217: "skipping item outside list",
1.68 schwarze 218: "skipping column outside column list",
219: "skipping end of block that is not open",
1.115 schwarze 220: "fewer RS blocks open, skipping",
1.68 schwarze 221: "inserting missing end of block",
222: "appending missing end of block",
223:
224: /* related to request and macro arguments */
1.7 kristaps 225: "escaped character not allowed in a name",
1.99 schwarze 226: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 227: "skipping display without arguments",
1.71 schwarze 228: "missing list type, using -item",
1.70 schwarze 229: "missing manual name, using \"\"",
1.71 schwarze 230: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 231: "unknown standard specifier",
1.71 schwarze 232: "skipping request without numeric argument",
1.109 schwarze 233: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 234: ".so request failed",
1.61 schwarze 235: "skipping all arguments",
236: "skipping excess arguments",
1.92 kristaps 237: "divide by zero",
1.112 schwarze 238:
239: "unsupported feature",
240: "input too large",
1.114 schwarze 241: "unsupported control character",
1.112 schwarze 242: "unsupported roff request",
1.119 schwarze 243: "eqn delim option in tbl",
1.118 schwarze 244: "unsupported tbl layout modifier",
1.112 schwarze 245: "ignoring macro in table",
1.7 kristaps 246: };
247:
248: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
249: "SUCCESS",
1.165 schwarze 250: "STYLE",
1.7 kristaps 251: "WARNING",
252: "ERROR",
1.112 schwarze 253: "UNSUPP",
1.7 kristaps 254: "BADARG",
255: "SYSERR"
256: };
257:
1.47 schwarze 258:
1.1 kristaps 259: static void
260: resize_buf(struct buf *buf, size_t initial)
261: {
262:
263: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
264: buf->buf = mandoc_realloc(buf->buf, buf->sz);
265: }
266:
267: static void
1.84 schwarze 268: choose_parser(struct mparse *curp)
1.1 kristaps 269: {
1.83 schwarze 270: char *cp, *ep;
271: int format;
1.1 kristaps 272:
1.83 schwarze 273: /*
274: * If neither command line arguments -mdoc or -man select
275: * a parser nor the roff parser found a .Dd or .TH macro
276: * yet, look ahead in the main input buffer.
277: */
278:
279: if ((format = roff_getformat(curp->roff)) == 0) {
280: cp = curp->primary->buf;
281: ep = cp + curp->primary->sz;
282: while (cp < ep) {
1.85 schwarze 283: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 284: cp++;
285: if (cp[0] == 'D' && cp[1] == 'd') {
286: format = MPARSE_MDOC;
287: break;
288: }
289: if (cp[0] == 'T' && cp[1] == 'H') {
290: format = MPARSE_MAN;
291: break;
292: }
293: }
294: cp = memchr(cp, '\n', ep - cp);
295: if (cp == NULL)
296: break;
297: cp++;
298: }
1.1 kristaps 299: }
300:
1.83 schwarze 301: if (format == MPARSE_MDOC) {
1.137 schwarze 302: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 303: if (curp->man->mdocmac == NULL)
304: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 305: } else {
306: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 307: if (curp->man->manmac == NULL)
308: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 309: }
1.164 schwarze 310: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 311: }
312:
313: /*
1.95 schwarze 314: * Main parse routine for a buffer.
315: * It assumes encoding and line numbering are already set up.
316: * It can recurse directly (for invocations of user-defined
317: * macros, inline equations, and input line traps)
318: * and indirectly (for .so file inclusion).
1.1 kristaps 319: */
1.163 schwarze 320: static int
1.95 schwarze 321: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 322: {
323: const struct tbl_span *span;
324: struct buf ln;
1.113 schwarze 325: const char *save_file;
1.110 schwarze 326: char *cp;
1.95 schwarze 327: size_t pos; /* byte number in the ln buffer */
1.154 schwarze 328: size_t j; /* auxiliary byte number in the blk buffer */
1.1 kristaps 329: enum rofferr rr;
1.100 schwarze 330: int of;
1.1 kristaps 331: int lnn; /* line number in the real file */
1.110 schwarze 332: int fd;
1.1 kristaps 333: unsigned char c;
334:
1.95 schwarze 335: memset(&ln, 0, sizeof(ln));
1.1 kristaps 336:
1.47 schwarze 337: lnn = curp->line;
338: pos = 0;
1.1 kristaps 339:
1.95 schwarze 340: while (i < blk.sz) {
1.1 kristaps 341: if (0 == pos && '\0' == blk.buf[i])
342: break;
343:
344: if (start) {
345: curp->line = lnn;
346: curp->reparse_count = 0;
1.93 schwarze 347:
348: if (lnn < 3 &&
349: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 350: curp->filenc & MPARSE_LATIN1)
351: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 352: }
353:
1.95 schwarze 354: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 355:
356: /*
357: * When finding an unescaped newline character,
358: * leave the character loop to process the line.
359: * Skip a preceding carriage return, if any.
360: */
361:
1.95 schwarze 362: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 363: '\n' == blk.buf[i + 1])
364: ++i;
365: if ('\n' == blk.buf[i]) {
366: ++i;
367: ++lnn;
368: break;
369: }
370:
1.35 schwarze 371: /*
1.93 schwarze 372: * Make sure we have space for the worst
373: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 374: */
375:
1.95 schwarze 376: if (pos + 11 > ln.sz)
1.35 schwarze 377: resize_buf(&ln, 256);
378:
1.47 schwarze 379: /*
1.93 schwarze 380: * Encode 8-bit input.
1.1 kristaps 381: */
382:
1.93 schwarze 383: c = blk.buf[i];
384: if (c & 0x80) {
1.95 schwarze 385: if ( ! (curp->filenc && preconv_encode(
386: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 387: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
388: curp->line, pos, "0x%x", c);
1.93 schwarze 389: ln.buf[pos++] = '?';
390: i++;
391: }
392: continue;
393: }
394:
395: /*
396: * Exclude control characters.
397: */
1.1 kristaps 398:
1.93 schwarze 399: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 400: mandoc_vmsg(c == 0x00 || c == 0x04 ||
401: c > 0x0a ? MANDOCERR_CHAR_BAD :
402: MANDOCERR_CHAR_UNSUPP,
403: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 404: i++;
1.127 schwarze 405: if (c != '\r')
406: ln.buf[pos++] = '?';
1.1 kristaps 407: continue;
408: }
409:
410: /* Trailing backslash = a plain char. */
411:
1.95 schwarze 412: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 413: ln.buf[pos++] = blk.buf[i++];
414: continue;
415: }
416:
417: /*
418: * Found escape and at least one other character.
419: * When it's a newline character, skip it.
420: * When there is a carriage return in between,
421: * skip that one as well.
422: */
423:
1.95 schwarze 424: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 425: '\n' == blk.buf[i + 2])
426: ++i;
427: if ('\n' == blk.buf[i + 1]) {
428: i += 2;
429: ++lnn;
430: continue;
431: }
432:
1.13 kristaps 433: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.154 schwarze 434: j = i;
1.1 kristaps 435: i += 2;
436: /* Comment, skip to end of line */
1.95 schwarze 437: for (; i < blk.sz; ++i) {
1.153 schwarze 438: if (blk.buf[i] != '\n')
439: continue;
440: if (blk.buf[i - 1] == ' ' ||
441: blk.buf[i - 1] == '\t')
442: mandoc_msg(
443: MANDOCERR_SPACE_EOL,
444: curp, curp->line,
1.154 schwarze 445: pos + i-1 - j, NULL);
1.153 schwarze 446: ++i;
447: ++lnn;
448: break;
1.1 kristaps 449: }
450:
451: /* Backout trailing whitespaces */
452: for (; pos > 0; --pos) {
453: if (ln.buf[pos - 1] != ' ')
454: break;
455: if (pos > 2 && ln.buf[pos - 2] == '\\')
456: break;
457: }
458: break;
459: }
460:
1.35 schwarze 461: /* Catch escaped bogus characters. */
462:
463: c = (unsigned char) blk.buf[i+1];
464:
1.47 schwarze 465: if ( ! (isascii(c) &&
466: (isgraph(c) || isblank(c)))) {
1.114 schwarze 467: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 468: curp->line, pos, "0x%x", c);
1.35 schwarze 469: i += 2;
470: ln.buf[pos++] = '?';
471: continue;
472: }
473:
1.1 kristaps 474: /* Some other escape sequence, copy & cont. */
475:
476: ln.buf[pos++] = blk.buf[i++];
477: ln.buf[pos++] = blk.buf[i++];
478: }
479:
1.95 schwarze 480: if (pos >= ln.sz)
1.1 kristaps 481: resize_buf(&ln, 256);
482:
483: ln.buf[pos] = '\0';
484:
485: /*
486: * A significant amount of complexity is contained by
487: * the roff preprocessor. It's line-oriented but can be
488: * expressed on one line, so we need at times to
489: * readjust our starting point and re-run it. The roff
490: * preprocessor can also readjust the buffers with new
491: * data, so we pass them in wholesale.
492: */
493:
494: of = 0;
495:
1.24 kristaps 496: /*
497: * Maintain a lookaside buffer of all parsed lines. We
498: * only do this if mparse_keep() has been invoked (the
499: * buffer may be accessed with mparse_getkeep()).
500: */
501:
502: if (curp->secondary) {
1.47 schwarze 503: curp->secondary->buf = mandoc_realloc(
504: curp->secondary->buf,
505: curp->secondary->sz + pos + 2);
506: memcpy(curp->secondary->buf +
507: curp->secondary->sz,
508: ln.buf, pos);
1.24 kristaps 509: curp->secondary->sz += pos;
510: curp->secondary->buf
511: [curp->secondary->sz] = '\n';
512: curp->secondary->sz++;
513: curp->secondary->buf
514: [curp->secondary->sz] = '\0';
515: }
1.1 kristaps 516: rerun:
1.96 schwarze 517: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 518:
519: switch (rr) {
1.47 schwarze 520: case ROFF_REPARSE:
1.163 schwarze 521: if (++curp->reparse_count > REPARSE_LIMIT)
1.3 kristaps 522: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 523: curp->line, pos, NULL);
1.163 schwarze 524: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
525: start == 1) {
526: pos = 0;
527: continue;
528: }
529: free(ln.buf);
530: return 0;
1.47 schwarze 531: case ROFF_APPEND:
1.95 schwarze 532: pos = strlen(ln.buf);
1.1 kristaps 533: continue;
1.47 schwarze 534: case ROFF_RERUN:
1.1 kristaps 535: goto rerun;
1.47 schwarze 536: case ROFF_IGN:
1.1 kristaps 537: pos = 0;
538: continue;
1.47 schwarze 539: case ROFF_SO:
1.95 schwarze 540: if ( ! (curp->options & MPARSE_SO) &&
541: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 542: curp->sodest = mandoc_strdup(ln.buf + of);
543: free(ln.buf);
1.163 schwarze 544: return 1;
1.45 schwarze 545: }
1.24 kristaps 546: /*
547: * We remove `so' clauses from our lookaside
548: * buffer because we're going to descend into
549: * the file recursively.
550: */
1.47 schwarze 551: if (curp->secondary)
1.25 kristaps 552: curp->secondary->sz -= pos + 1;
1.113 schwarze 553: save_file = curp->file;
1.148 schwarze 554: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 555: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 556: close(fd);
1.113 schwarze 557: curp->file = save_file;
558: } else {
559: curp->file = save_file;
1.52 schwarze 560: mandoc_vmsg(MANDOCERR_SO_FAIL,
561: curp, curp->line, pos,
562: ".so %s", ln.buf + of);
1.110 schwarze 563: ln.sz = mandoc_asprintf(&cp,
564: ".sp\nSee the file %s.\n.sp",
565: ln.buf + of);
566: free(ln.buf);
567: ln.buf = cp;
568: of = 0;
569: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 570: }
1.1 kristaps 571: pos = 0;
572: continue;
573: default:
574: break;
575: }
576:
1.150 schwarze 577: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 578: choose_parser(curp);
1.1 kristaps 579:
1.47 schwarze 580: /*
1.84 schwarze 581: * Lastly, push down into the parsers themselves.
1.1 kristaps 582: * If libroff returns ROFF_TBL, then add it to the
583: * currently open parse. Since we only get here if
584: * there does exist data (see tbl_data.c), we're
585: * guaranteed that something's been allocated.
586: * Do the same for ROFF_EQN.
587: */
588:
1.139 schwarze 589: if (rr == ROFF_TBL)
1.100 schwarze 590: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 591: roff_addtbl(curp->man, span);
592: else if (rr == ROFF_EQN)
593: roff_addeqn(curp->man, roff_eqn(curp->roff));
594: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 595: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 596: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
597: break;
1.1 kristaps 598:
599: /* Temporary buffers typically are not full. */
600:
601: if (0 == start && '\0' == blk.buf[i])
602: break;
603:
604: /* Start the next input line. */
605:
606: pos = 0;
607: }
608:
609: free(ln.buf);
1.163 schwarze 610: return 1;
1.1 kristaps 611: }
612:
613: static int
1.40 schwarze 614: read_whole_file(struct mparse *curp, const char *file, int fd,
615: struct buf *fb, int *with_mmap)
1.1 kristaps 616: {
1.161 schwarze 617: struct stat st;
1.140 schwarze 618: gzFile gz;
1.1 kristaps 619: size_t off;
620: ssize_t ssz;
1.143 schwarze 621:
622: if (fstat(fd, &st) == -1)
623: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 624:
625: /*
626: * If we're a regular file, try just reading in the whole entry
627: * via mmap(). This is faster than reading it into blocks, and
628: * since each file is only a few bytes to begin with, I'm not
629: * concerned that this is going to tank any machines.
630: */
631:
1.140 schwarze 632: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 633: if (st.st_size > 0x7fffffff) {
1.111 schwarze 634: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 635: return 0;
1.1 kristaps 636: }
637: *with_mmap = 1;
638: fb->sz = (size_t)st.st_size;
1.37 schwarze 639: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 640: if (fb->buf != MAP_FAILED)
1.142 schwarze 641: return 1;
1.1 kristaps 642: }
643:
1.140 schwarze 644: if (curp->gzip) {
1.143 schwarze 645: if ((gz = gzdopen(fd, "rb")) == NULL)
646: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 647: } else
648: gz = NULL;
649:
1.1 kristaps 650: /*
651: * If this isn't a regular file (like, say, stdin), then we must
652: * go the old way and just read things in bit by bit.
653: */
654:
655: *with_mmap = 0;
656: off = 0;
657: fb->sz = 0;
658: fb->buf = NULL;
659: for (;;) {
660: if (off == fb->sz) {
661: if (fb->sz == (1U << 31)) {
1.111 schwarze 662: mandoc_msg(MANDOCERR_TOOLARGE, curp,
663: 0, 0, NULL);
1.1 kristaps 664: break;
665: }
666: resize_buf(fb, 65536);
667: }
1.140 schwarze 668: ssz = curp->gzip ?
669: gzread(gz, fb->buf + (int)off, fb->sz - off) :
670: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 671: if (ssz == 0) {
672: fb->sz = off;
1.142 schwarze 673: return 1;
1.1 kristaps 674: }
1.143 schwarze 675: if (ssz == -1)
676: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 677: off += (size_t)ssz;
678: }
679:
680: free(fb->buf);
681: fb->buf = NULL;
1.142 schwarze 682: return 0;
1.1 kristaps 683: }
684:
685: static void
686: mparse_end(struct mparse *curp)
687: {
1.135 schwarze 688: if (curp->man->macroset == MACROSET_NONE)
689: curp->man->macroset = MACROSET_MAN;
690: if (curp->man->macroset == MACROSET_MDOC)
691: mdoc_endparse(curp->man);
692: else
1.111 schwarze 693: man_endparse(curp->man);
1.1 kristaps 694: roff_endparse(curp->roff);
695: }
696:
697: static void
1.36 schwarze 698: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 699: {
1.85 schwarze 700: struct buf *svprimary;
1.28 joerg 701: const char *svfile;
1.95 schwarze 702: size_t offset;
1.36 schwarze 703: static int recursion_depth;
704:
705: if (64 < recursion_depth) {
706: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
707: return;
708: }
1.28 joerg 709:
710: /* Line number is per-file. */
711: svfile = curp->file;
712: curp->file = file;
1.85 schwarze 713: svprimary = curp->primary;
1.83 schwarze 714: curp->primary = &blk;
1.28 joerg 715: curp->line = 1;
1.36 schwarze 716: recursion_depth++;
1.28 joerg 717:
1.93 schwarze 718: /* Skip an UTF-8 byte order mark. */
719: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
720: (unsigned char)blk.buf[0] == 0xef &&
721: (unsigned char)blk.buf[1] == 0xbb &&
722: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 723: offset = 3;
1.93 schwarze 724: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 725: } else
726: offset = 0;
1.93 schwarze 727:
1.95 schwarze 728: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 729:
1.111 schwarze 730: if (--recursion_depth == 0)
1.28 joerg 731: mparse_end(curp);
732:
1.85 schwarze 733: curp->primary = svprimary;
1.28 joerg 734: curp->file = svfile;
735: }
736:
737: enum mandoclevel
1.104 schwarze 738: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 739: const char *file)
740: {
741: struct buf blk;
742:
1.104 schwarze 743: blk.buf = buf;
1.28 joerg 744: blk.sz = len;
745:
1.36 schwarze 746: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 747: return curp->file_status;
1.28 joerg 748: }
749:
1.98 schwarze 750: /*
751: * Read the whole file into memory and call the parsers.
752: * Called recursively when an .so request is encountered.
753: */
1.36 schwarze 754: enum mandoclevel
755: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 756: {
1.28 joerg 757: struct buf blk;
758: int with_mmap;
1.93 schwarze 759: int save_filenc;
1.1 kristaps 760:
1.91 schwarze 761: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 762: save_filenc = curp->filenc;
763: curp->filenc = curp->options &
764: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 765: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 766: curp->filenc = save_filenc;
1.91 schwarze 767: if (with_mmap)
768: munmap(blk.buf, blk.sz);
769: else
770: free(blk.buf);
771: }
1.142 schwarze 772: return curp->file_status;
1.82 schwarze 773: }
774:
1.148 schwarze 775: int
776: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 777: {
778: char *cp;
1.148 schwarze 779: int fd;
1.82 schwarze 780:
781: curp->file = file;
1.140 schwarze 782: cp = strrchr(file, '.');
783: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 784:
1.140 schwarze 785: /* First try to use the filename as it is. */
1.98 schwarze 786:
1.148 schwarze 787: if ((fd = open(file, O_RDONLY)) != -1)
788: return fd;
1.98 schwarze 789:
1.140 schwarze 790: /*
791: * If that doesn't work and the filename doesn't
792: * already end in .gz, try appending .gz.
793: */
1.98 schwarze 794:
1.140 schwarze 795: if ( ! curp->gzip) {
1.98 schwarze 796: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 797: fd = open(cp, O_RDONLY);
1.108 schwarze 798: free(cp);
1.148 schwarze 799: if (fd != -1) {
1.140 schwarze 800: curp->gzip = 1;
1.148 schwarze 801: return fd;
1.82 schwarze 802: }
803: }
804:
1.140 schwarze 805: /* Neither worked, give up. */
1.97 schwarze 806:
1.140 schwarze 807: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 808: return -1;
1.1 kristaps 809: }
810:
811: struct mparse *
1.94 schwarze 812: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.144 schwarze 813: const char *defos)
1.1 kristaps 814: {
815: struct mparse *curp;
1.10 kristaps 816:
1.1 kristaps 817: curp = mandoc_calloc(1, sizeof(struct mparse));
818:
1.44 schwarze 819: curp->options = options;
1.3 kristaps 820: curp->wlevel = wlevel;
1.1 kristaps 821: curp->mmsg = mmsg;
1.29 schwarze 822: curp->defos = defos;
1.1 kristaps 823:
1.144 schwarze 824: curp->roff = roff_alloc(curp, options);
1.137 schwarze 825: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
826: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 827: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 828: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 829: if (curp->man->mdocmac == NULL)
830: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 831: } else if (curp->options & MPARSE_MAN) {
832: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 833: if (curp->man->manmac == NULL)
834: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 835: }
1.138 schwarze 836: curp->man->first->tok = TOKEN_NONE;
1.142 schwarze 837: return curp;
1.1 kristaps 838: }
839:
840: void
841: mparse_reset(struct mparse *curp)
842: {
843: roff_reset(curp->roff);
1.150 schwarze 844: roff_man_reset(curp->man);
1.160 schwarze 845:
846: free(curp->sodest);
847: curp->sodest = NULL;
848:
1.24 kristaps 849: if (curp->secondary)
850: curp->secondary->sz = 0;
1.1 kristaps 851:
852: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 853: curp->gzip = 0;
1.1 kristaps 854: }
855:
856: void
857: mparse_free(struct mparse *curp)
858: {
859:
1.164 schwarze 860: roffhash_free(curp->man->mdocmac);
861: roffhash_free(curp->man->manmac);
1.137 schwarze 862: roff_man_free(curp->man);
1.160 schwarze 863: roff_free(curp->roff);
1.24 kristaps 864: if (curp->secondary)
865: free(curp->secondary->buf);
1.1 kristaps 866:
1.24 kristaps 867: free(curp->secondary);
1.45 schwarze 868: free(curp->sodest);
1.1 kristaps 869: free(curp);
870: }
871:
872: void
1.135 schwarze 873: mparse_result(struct mparse *curp, struct roff_man **man,
874: char **sodest)
1.1 kristaps 875: {
876:
1.45 schwarze 877: if (sodest && NULL != (*sodest = curp->sodest)) {
878: *man = NULL;
879: return;
880: }
1.9 kristaps 881: if (man)
882: *man = curp->man;
1.157 schwarze 883: }
884:
885: void
886: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
887: {
888: if (curp->file_status > *rc)
889: *rc = curp->file_status;
1.3 kristaps 890: }
891:
892: void
893: mandoc_vmsg(enum mandocerr t, struct mparse *m,
894: int ln, int pos, const char *fmt, ...)
895: {
896: char buf[256];
897: va_list ap;
898:
899: va_start(ap, fmt);
1.48 schwarze 900: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 901: va_end(ap);
902:
903: mandoc_msg(t, m, ln, pos, buf);
904: }
905:
906: void
1.47 schwarze 907: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 908: int ln, int col, const char *msg)
909: {
910: enum mandoclevel level;
911:
1.112 schwarze 912: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 913: while (er < mandoclimits[level])
914: level--;
915:
1.108 schwarze 916: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 917: return;
918:
1.8 kristaps 919: if (m->mmsg)
920: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 921:
922: if (m->file_status < level)
923: m->file_status = level;
1.7 kristaps 924: }
925:
926: const char *
927: mparse_strerror(enum mandocerr er)
928: {
929:
1.142 schwarze 930: return mandocerrs[er];
1.7 kristaps 931: }
932:
933: const char *
934: mparse_strlevel(enum mandoclevel lvl)
935: {
1.142 schwarze 936: return mandoclevels[lvl];
1.24 kristaps 937: }
938:
939: void
940: mparse_keep(struct mparse *p)
941: {
942:
943: assert(NULL == p->secondary);
944: p->secondary = mandoc_calloc(1, sizeof(struct buf));
945: }
946:
947: const char *
948: mparse_getkeep(const struct mparse *p)
949: {
950:
951: assert(p->secondary);
1.142 schwarze 952: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 953: }
CVSweb