Annotation of mandoc/read.c, Revision 1.178
1.178 ! schwarze 1: /* $Id: read.c,v 1.177 2017/06/17 22:43:14 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:
1.176 schwarze 91: "Mdocdate found",
92: "Mdocdate missing",
93: "legacy man(7) date format",
1.177 schwarze 94: "RCS id missing",
1.178 ! schwarze 95: "duplicate RCS id",
1.166 schwarze 96: "useless macro",
1.167 schwarze 97: "consider using OS macro",
1.172 schwarze 98: "errnos out of order",
99: "duplicate errno",
1.168 schwarze 100: "description line ends with a full stop",
1.174 schwarze 101: "no blank before trailing delimiter",
1.175 schwarze 102: "function name without markup",
1.166 schwarze 103:
1.7 kristaps 104: "generic warning",
105:
106: /* related to the prologue */
1.79 schwarze 107: "missing manual title, using UNTITLED",
108: "missing manual title, using \"\"",
1.54 schwarze 109: "lower case character in document title",
1.79 schwarze 110: "missing manual section, using \"\"",
1.7 kristaps 111: "unknown manual section",
1.54 schwarze 112: "missing date, using today's date",
1.7 kristaps 113: "cannot parse date, using it verbatim",
1.79 schwarze 114: "missing Os macro, using \"\"",
115: "duplicate prologue macro",
116: "late prologue macro",
117: "skipping late title macro",
1.7 kristaps 118: "prologue macros out of order",
119:
120: /* related to document structure */
121: ".so is fragile, better use ln(1)",
1.50 schwarze 122: "no document body",
1.54 schwarze 123: "content before first section header",
124: "first section is not \"NAME\"",
1.156 schwarze 125: "NAME section without Nm before Nd",
1.128 schwarze 126: "NAME section without description",
127: "description not at the end of NAME",
128: "bad NAME section content",
1.156 schwarze 129: "missing comma before name",
1.121 schwarze 130: "missing description line, using \"\"",
1.162 schwarze 131: "description line outside NAME section",
1.7 kristaps 132: "sections out of conventional order",
1.54 schwarze 133: "duplicate section title",
134: "unexpected section",
1.87 schwarze 135: "unusual Xr order",
136: "unusual Xr punctuation",
1.86 schwarze 137: "AUTHORS section without An macro",
1.7 kristaps 138:
139: /* related to macros and nesting */
1.55 schwarze 140: "obsolete macro",
1.102 schwarze 141: "macro neither callable nor escaped",
1.7 kristaps 142: "skipping paragraph macro",
1.31 schwarze 143: "moving paragraph macro out of list",
1.7 kristaps 144: "skipping no-space macro",
145: "blocks badly nested",
146: "nested displays are not portable",
1.57 schwarze 147: "moving content out of list",
1.78 schwarze 148: "fill mode already enabled, skipping",
149: "fill mode already disabled, skipping",
1.7 kristaps 150: "line scope broken",
1.169 schwarze 151: "skipping blank line in line scope",
1.7 kristaps 152:
153: /* related to missing macro arguments */
1.58 schwarze 154: "skipping empty request",
155: "conditional request controls empty scope",
1.7 kristaps 156: "skipping empty macro",
1.124 schwarze 157: "empty block",
1.62 schwarze 158: "empty argument, using 0n",
1.60 schwarze 159: "missing display type, using -ragged",
160: "list type is not the first argument",
1.152 schwarze 161: "missing -width in -tag list, using 6n",
1.78 schwarze 162: "missing utility name, using \"\"",
1.123 schwarze 163: "missing function name, using \"\"",
1.60 schwarze 164: "empty head in list item",
165: "empty list item",
1.61 schwarze 166: "missing font type, using \\fR",
167: "unknown font type, using \\fR",
1.103 schwarze 168: "nothing follows prefix",
1.122 schwarze 169: "empty reference block",
1.155 schwarze 170: "missing section argument",
1.60 schwarze 171: "missing -std argument, adding it",
1.125 schwarze 172: "missing option string, using \"\"",
173: "missing resource identifier, using \"\"",
1.90 schwarze 174: "missing eqn box, using \"\"",
1.7 kristaps 175:
176: /* related to bad macro arguments */
1.64 schwarze 177: "unterminated quoted argument",
1.7 kristaps 178: "duplicate argument",
1.76 schwarze 179: "skipping duplicate argument",
1.63 schwarze 180: "skipping duplicate display type",
181: "skipping duplicate list type",
1.76 schwarze 182: "skipping -width argument",
1.126 schwarze 183: "wrong number of cells",
1.7 kristaps 184: "unknown AT&T UNIX version",
1.88 schwarze 185: "comma in function argument",
1.89 schwarze 186: "parenthesis in function name",
1.173 schwarze 187: "unknown library name",
1.67 schwarze 188: "invalid content in Rs block",
1.63 schwarze 189: "invalid Boolean argument",
190: "unknown font, skipping request",
1.126 schwarze 191: "odd number of characters in request",
1.7 kristaps 192:
193: /* related to plain text */
1.64 schwarze 194: "blank line in fill mode, using .sp",
195: "tab in filled text",
196: "whitespace at end of input line",
1.158 schwarze 197: "new sentence, new line",
1.7 kristaps 198: "bad comment style",
1.64 schwarze 199: "invalid escape sequence",
200: "undefined string, using \"\"",
1.16 kristaps 201:
1.118 schwarze 202: /* related to tables */
203: "tbl line starts with span",
204: "tbl column starts with span",
205: "skipping vertical bar in tbl layout",
206:
1.7 kristaps 207: "generic error",
208:
209: /* related to tables */
1.116 schwarze 210: "non-alphabetic character in tbl options",
211: "skipping unknown tbl option",
212: "missing tbl option argument",
213: "wrong tbl option argument size",
1.118 schwarze 214: "empty tbl layout",
215: "invalid character in tbl layout",
216: "unmatched parenthesis in tbl layout",
1.119 schwarze 217: "tbl without any data cells",
218: "ignoring data in spanned tbl cell",
219: "ignoring extra tbl data cells",
220: "data block open at end of tbl",
1.7 kristaps 221:
1.68 schwarze 222: /* related to document structure and macros */
1.108 schwarze 223: NULL,
1.7 kristaps 224: "input stack limit exceeded, infinite loop?",
225: "skipping bad character",
1.68 schwarze 226: "skipping unknown macro",
1.112 schwarze 227: "skipping insecure request",
1.73 schwarze 228: "skipping item outside list",
1.68 schwarze 229: "skipping column outside column list",
230: "skipping end of block that is not open",
1.115 schwarze 231: "fewer RS blocks open, skipping",
1.68 schwarze 232: "inserting missing end of block",
233: "appending missing end of block",
234:
235: /* related to request and macro arguments */
1.7 kristaps 236: "escaped character not allowed in a name",
1.99 schwarze 237: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 238: "skipping display without arguments",
1.71 schwarze 239: "missing list type, using -item",
1.171 schwarze 240: "argument is not numeric, using 1",
1.70 schwarze 241: "missing manual name, using \"\"",
1.71 schwarze 242: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 243: "unknown standard specifier",
1.71 schwarze 244: "skipping request without numeric argument",
1.109 schwarze 245: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 246: ".so request failed",
1.61 schwarze 247: "skipping all arguments",
248: "skipping excess arguments",
1.92 kristaps 249: "divide by zero",
1.112 schwarze 250:
251: "unsupported feature",
252: "input too large",
1.114 schwarze 253: "unsupported control character",
1.112 schwarze 254: "unsupported roff request",
1.119 schwarze 255: "eqn delim option in tbl",
1.118 schwarze 256: "unsupported tbl layout modifier",
1.112 schwarze 257: "ignoring macro in table",
1.7 kristaps 258: };
259:
260: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
261: "SUCCESS",
1.165 schwarze 262: "STYLE",
1.7 kristaps 263: "WARNING",
264: "ERROR",
1.112 schwarze 265: "UNSUPP",
1.7 kristaps 266: "BADARG",
267: "SYSERR"
268: };
269:
1.47 schwarze 270:
1.1 kristaps 271: static void
272: resize_buf(struct buf *buf, size_t initial)
273: {
274:
275: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
276: buf->buf = mandoc_realloc(buf->buf, buf->sz);
277: }
278:
279: static void
1.84 schwarze 280: choose_parser(struct mparse *curp)
1.1 kristaps 281: {
1.83 schwarze 282: char *cp, *ep;
283: int format;
1.1 kristaps 284:
1.83 schwarze 285: /*
286: * If neither command line arguments -mdoc or -man select
287: * a parser nor the roff parser found a .Dd or .TH macro
288: * yet, look ahead in the main input buffer.
289: */
290:
291: if ((format = roff_getformat(curp->roff)) == 0) {
292: cp = curp->primary->buf;
293: ep = cp + curp->primary->sz;
294: while (cp < ep) {
1.85 schwarze 295: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 296: cp++;
297: if (cp[0] == 'D' && cp[1] == 'd') {
298: format = MPARSE_MDOC;
299: break;
300: }
301: if (cp[0] == 'T' && cp[1] == 'H') {
302: format = MPARSE_MAN;
303: break;
304: }
305: }
306: cp = memchr(cp, '\n', ep - cp);
307: if (cp == NULL)
308: break;
309: cp++;
310: }
1.1 kristaps 311: }
312:
1.83 schwarze 313: if (format == MPARSE_MDOC) {
1.137 schwarze 314: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 315: if (curp->man->mdocmac == NULL)
316: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 317: } else {
318: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 319: if (curp->man->manmac == NULL)
320: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 321: }
1.164 schwarze 322: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 323: }
324:
325: /*
1.95 schwarze 326: * Main parse routine for a buffer.
327: * It assumes encoding and line numbering are already set up.
328: * It can recurse directly (for invocations of user-defined
329: * macros, inline equations, and input line traps)
330: * and indirectly (for .so file inclusion).
1.1 kristaps 331: */
1.163 schwarze 332: static int
1.95 schwarze 333: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 334: {
335: const struct tbl_span *span;
336: struct buf ln;
1.113 schwarze 337: const char *save_file;
1.110 schwarze 338: char *cp;
1.95 schwarze 339: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 340: enum rofferr rr;
1.100 schwarze 341: int of;
1.1 kristaps 342: int lnn; /* line number in the real file */
1.110 schwarze 343: int fd;
1.1 kristaps 344: unsigned char c;
345:
1.95 schwarze 346: memset(&ln, 0, sizeof(ln));
1.1 kristaps 347:
1.47 schwarze 348: lnn = curp->line;
349: pos = 0;
1.1 kristaps 350:
1.95 schwarze 351: while (i < blk.sz) {
1.1 kristaps 352: if (0 == pos && '\0' == blk.buf[i])
353: break;
354:
355: if (start) {
356: curp->line = lnn;
357: curp->reparse_count = 0;
1.93 schwarze 358:
359: if (lnn < 3 &&
360: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 361: curp->filenc & MPARSE_LATIN1)
362: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 363: }
364:
1.95 schwarze 365: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 366:
367: /*
368: * When finding an unescaped newline character,
369: * leave the character loop to process the line.
370: * Skip a preceding carriage return, if any.
371: */
372:
1.95 schwarze 373: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 374: '\n' == blk.buf[i + 1])
375: ++i;
376: if ('\n' == blk.buf[i]) {
377: ++i;
378: ++lnn;
379: break;
380: }
381:
1.35 schwarze 382: /*
1.93 schwarze 383: * Make sure we have space for the worst
384: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 385: */
386:
1.95 schwarze 387: if (pos + 11 > ln.sz)
1.35 schwarze 388: resize_buf(&ln, 256);
389:
1.47 schwarze 390: /*
1.93 schwarze 391: * Encode 8-bit input.
1.1 kristaps 392: */
393:
1.93 schwarze 394: c = blk.buf[i];
395: if (c & 0x80) {
1.95 schwarze 396: if ( ! (curp->filenc && preconv_encode(
397: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 398: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
399: curp->line, pos, "0x%x", c);
1.93 schwarze 400: ln.buf[pos++] = '?';
401: i++;
402: }
403: continue;
404: }
405:
406: /*
407: * Exclude control characters.
408: */
1.1 kristaps 409:
1.93 schwarze 410: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 411: mandoc_vmsg(c == 0x00 || c == 0x04 ||
412: c > 0x0a ? MANDOCERR_CHAR_BAD :
413: MANDOCERR_CHAR_UNSUPP,
414: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 415: i++;
1.127 schwarze 416: if (c != '\r')
417: ln.buf[pos++] = '?';
1.1 kristaps 418: continue;
419: }
420:
421: ln.buf[pos++] = blk.buf[i++];
422: }
423:
1.170 schwarze 424: if (pos + 1 >= ln.sz)
1.1 kristaps 425: resize_buf(&ln, 256);
426:
1.170 schwarze 427: if (i == blk.sz || blk.buf[i] == '\0')
428: ln.buf[pos++] = '\n';
1.1 kristaps 429: ln.buf[pos] = '\0';
430:
431: /*
432: * A significant amount of complexity is contained by
433: * the roff preprocessor. It's line-oriented but can be
434: * expressed on one line, so we need at times to
435: * readjust our starting point and re-run it. The roff
436: * preprocessor can also readjust the buffers with new
437: * data, so we pass them in wholesale.
438: */
439:
440: of = 0;
441:
1.24 kristaps 442: /*
443: * Maintain a lookaside buffer of all parsed lines. We
444: * only do this if mparse_keep() has been invoked (the
445: * buffer may be accessed with mparse_getkeep()).
446: */
447:
448: if (curp->secondary) {
1.47 schwarze 449: curp->secondary->buf = mandoc_realloc(
450: curp->secondary->buf,
451: curp->secondary->sz + pos + 2);
452: memcpy(curp->secondary->buf +
453: curp->secondary->sz,
454: ln.buf, pos);
1.24 kristaps 455: curp->secondary->sz += pos;
456: curp->secondary->buf
457: [curp->secondary->sz] = '\n';
458: curp->secondary->sz++;
459: curp->secondary->buf
460: [curp->secondary->sz] = '\0';
461: }
1.1 kristaps 462: rerun:
1.96 schwarze 463: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 464:
465: switch (rr) {
1.47 schwarze 466: case ROFF_REPARSE:
1.163 schwarze 467: if (++curp->reparse_count > REPARSE_LIMIT)
1.3 kristaps 468: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 469: curp->line, pos, NULL);
1.163 schwarze 470: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
471: start == 1) {
472: pos = 0;
473: continue;
474: }
475: free(ln.buf);
476: return 0;
1.47 schwarze 477: case ROFF_APPEND:
1.95 schwarze 478: pos = strlen(ln.buf);
1.1 kristaps 479: continue;
1.47 schwarze 480: case ROFF_RERUN:
1.1 kristaps 481: goto rerun;
1.47 schwarze 482: case ROFF_IGN:
1.1 kristaps 483: pos = 0;
484: continue;
1.47 schwarze 485: case ROFF_SO:
1.95 schwarze 486: if ( ! (curp->options & MPARSE_SO) &&
487: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 488: curp->sodest = mandoc_strdup(ln.buf + of);
489: free(ln.buf);
1.163 schwarze 490: return 1;
1.45 schwarze 491: }
1.24 kristaps 492: /*
493: * We remove `so' clauses from our lookaside
494: * buffer because we're going to descend into
495: * the file recursively.
496: */
1.47 schwarze 497: if (curp->secondary)
1.25 kristaps 498: curp->secondary->sz -= pos + 1;
1.113 schwarze 499: save_file = curp->file;
1.148 schwarze 500: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 501: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 502: close(fd);
1.113 schwarze 503: curp->file = save_file;
504: } else {
505: curp->file = save_file;
1.52 schwarze 506: mandoc_vmsg(MANDOCERR_SO_FAIL,
507: curp, curp->line, pos,
508: ".so %s", ln.buf + of);
1.110 schwarze 509: ln.sz = mandoc_asprintf(&cp,
510: ".sp\nSee the file %s.\n.sp",
511: ln.buf + of);
512: free(ln.buf);
513: ln.buf = cp;
514: of = 0;
515: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 516: }
1.1 kristaps 517: pos = 0;
518: continue;
519: default:
520: break;
521: }
522:
1.150 schwarze 523: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 524: choose_parser(curp);
1.1 kristaps 525:
1.47 schwarze 526: /*
1.84 schwarze 527: * Lastly, push down into the parsers themselves.
1.1 kristaps 528: * If libroff returns ROFF_TBL, then add it to the
529: * currently open parse. Since we only get here if
530: * there does exist data (see tbl_data.c), we're
531: * guaranteed that something's been allocated.
532: * Do the same for ROFF_EQN.
533: */
534:
1.139 schwarze 535: if (rr == ROFF_TBL)
1.100 schwarze 536: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 537: roff_addtbl(curp->man, span);
538: else if (rr == ROFF_EQN)
539: roff_addeqn(curp->man, roff_eqn(curp->roff));
540: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 541: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 542: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
543: break;
1.1 kristaps 544:
545: /* Temporary buffers typically are not full. */
546:
547: if (0 == start && '\0' == blk.buf[i])
548: break;
549:
550: /* Start the next input line. */
551:
552: pos = 0;
553: }
554:
555: free(ln.buf);
1.163 schwarze 556: return 1;
1.1 kristaps 557: }
558:
559: static int
1.40 schwarze 560: read_whole_file(struct mparse *curp, const char *file, int fd,
561: struct buf *fb, int *with_mmap)
1.1 kristaps 562: {
1.161 schwarze 563: struct stat st;
1.140 schwarze 564: gzFile gz;
1.1 kristaps 565: size_t off;
566: ssize_t ssz;
1.143 schwarze 567:
568: if (fstat(fd, &st) == -1)
569: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 570:
571: /*
572: * If we're a regular file, try just reading in the whole entry
573: * via mmap(). This is faster than reading it into blocks, and
574: * since each file is only a few bytes to begin with, I'm not
575: * concerned that this is going to tank any machines.
576: */
577:
1.140 schwarze 578: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 579: if (st.st_size > 0x7fffffff) {
1.111 schwarze 580: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 581: return 0;
1.1 kristaps 582: }
583: *with_mmap = 1;
584: fb->sz = (size_t)st.st_size;
1.37 schwarze 585: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 586: if (fb->buf != MAP_FAILED)
1.142 schwarze 587: return 1;
1.1 kristaps 588: }
589:
1.140 schwarze 590: if (curp->gzip) {
1.143 schwarze 591: if ((gz = gzdopen(fd, "rb")) == NULL)
592: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 593: } else
594: gz = NULL;
595:
1.1 kristaps 596: /*
597: * If this isn't a regular file (like, say, stdin), then we must
598: * go the old way and just read things in bit by bit.
599: */
600:
601: *with_mmap = 0;
602: off = 0;
603: fb->sz = 0;
604: fb->buf = NULL;
605: for (;;) {
606: if (off == fb->sz) {
607: if (fb->sz == (1U << 31)) {
1.111 schwarze 608: mandoc_msg(MANDOCERR_TOOLARGE, curp,
609: 0, 0, NULL);
1.1 kristaps 610: break;
611: }
612: resize_buf(fb, 65536);
613: }
1.140 schwarze 614: ssz = curp->gzip ?
615: gzread(gz, fb->buf + (int)off, fb->sz - off) :
616: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 617: if (ssz == 0) {
618: fb->sz = off;
1.142 schwarze 619: return 1;
1.1 kristaps 620: }
1.143 schwarze 621: if (ssz == -1)
622: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 623: off += (size_t)ssz;
624: }
625:
626: free(fb->buf);
627: fb->buf = NULL;
1.142 schwarze 628: return 0;
1.1 kristaps 629: }
630:
631: static void
632: mparse_end(struct mparse *curp)
633: {
1.135 schwarze 634: if (curp->man->macroset == MACROSET_NONE)
635: curp->man->macroset = MACROSET_MAN;
636: if (curp->man->macroset == MACROSET_MDOC)
637: mdoc_endparse(curp->man);
638: else
1.111 schwarze 639: man_endparse(curp->man);
1.1 kristaps 640: roff_endparse(curp->roff);
641: }
642:
643: static void
1.36 schwarze 644: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 645: {
1.85 schwarze 646: struct buf *svprimary;
1.28 joerg 647: const char *svfile;
1.95 schwarze 648: size_t offset;
1.36 schwarze 649: static int recursion_depth;
650:
651: if (64 < recursion_depth) {
652: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
653: return;
654: }
1.28 joerg 655:
656: /* Line number is per-file. */
657: svfile = curp->file;
658: curp->file = file;
1.85 schwarze 659: svprimary = curp->primary;
1.83 schwarze 660: curp->primary = &blk;
1.28 joerg 661: curp->line = 1;
1.36 schwarze 662: recursion_depth++;
1.28 joerg 663:
1.93 schwarze 664: /* Skip an UTF-8 byte order mark. */
665: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
666: (unsigned char)blk.buf[0] == 0xef &&
667: (unsigned char)blk.buf[1] == 0xbb &&
668: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 669: offset = 3;
1.93 schwarze 670: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 671: } else
672: offset = 0;
1.93 schwarze 673:
1.95 schwarze 674: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 675:
1.111 schwarze 676: if (--recursion_depth == 0)
1.28 joerg 677: mparse_end(curp);
678:
1.85 schwarze 679: curp->primary = svprimary;
1.28 joerg 680: curp->file = svfile;
681: }
682:
683: enum mandoclevel
1.104 schwarze 684: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 685: const char *file)
686: {
687: struct buf blk;
688:
1.104 schwarze 689: blk.buf = buf;
1.28 joerg 690: blk.sz = len;
691:
1.36 schwarze 692: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 693: return curp->file_status;
1.28 joerg 694: }
695:
1.98 schwarze 696: /*
697: * Read the whole file into memory and call the parsers.
698: * Called recursively when an .so request is encountered.
699: */
1.36 schwarze 700: enum mandoclevel
701: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 702: {
1.28 joerg 703: struct buf blk;
704: int with_mmap;
1.93 schwarze 705: int save_filenc;
1.1 kristaps 706:
1.91 schwarze 707: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 708: save_filenc = curp->filenc;
709: curp->filenc = curp->options &
710: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 711: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 712: curp->filenc = save_filenc;
1.91 schwarze 713: if (with_mmap)
714: munmap(blk.buf, blk.sz);
715: else
716: free(blk.buf);
717: }
1.142 schwarze 718: return curp->file_status;
1.82 schwarze 719: }
720:
1.148 schwarze 721: int
722: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 723: {
724: char *cp;
1.148 schwarze 725: int fd;
1.82 schwarze 726:
727: curp->file = file;
1.140 schwarze 728: cp = strrchr(file, '.');
729: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 730:
1.140 schwarze 731: /* First try to use the filename as it is. */
1.98 schwarze 732:
1.148 schwarze 733: if ((fd = open(file, O_RDONLY)) != -1)
734: return fd;
1.98 schwarze 735:
1.140 schwarze 736: /*
737: * If that doesn't work and the filename doesn't
738: * already end in .gz, try appending .gz.
739: */
1.98 schwarze 740:
1.140 schwarze 741: if ( ! curp->gzip) {
1.98 schwarze 742: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 743: fd = open(cp, O_RDONLY);
1.108 schwarze 744: free(cp);
1.148 schwarze 745: if (fd != -1) {
1.140 schwarze 746: curp->gzip = 1;
1.148 schwarze 747: return fd;
1.82 schwarze 748: }
749: }
750:
1.140 schwarze 751: /* Neither worked, give up. */
1.97 schwarze 752:
1.140 schwarze 753: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 754: return -1;
1.1 kristaps 755: }
756:
757: struct mparse *
1.94 schwarze 758: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.144 schwarze 759: const char *defos)
1.1 kristaps 760: {
761: struct mparse *curp;
1.10 kristaps 762:
1.1 kristaps 763: curp = mandoc_calloc(1, sizeof(struct mparse));
764:
1.44 schwarze 765: curp->options = options;
1.3 kristaps 766: curp->wlevel = wlevel;
1.1 kristaps 767: curp->mmsg = mmsg;
1.29 schwarze 768: curp->defos = defos;
1.1 kristaps 769:
1.144 schwarze 770: curp->roff = roff_alloc(curp, options);
1.137 schwarze 771: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
772: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 773: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 774: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 775: if (curp->man->mdocmac == NULL)
776: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 777: } else if (curp->options & MPARSE_MAN) {
778: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 779: if (curp->man->manmac == NULL)
780: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 781: }
1.138 schwarze 782: curp->man->first->tok = TOKEN_NONE;
1.142 schwarze 783: return curp;
1.1 kristaps 784: }
785:
786: void
787: mparse_reset(struct mparse *curp)
788: {
789: roff_reset(curp->roff);
1.150 schwarze 790: roff_man_reset(curp->man);
1.160 schwarze 791:
792: free(curp->sodest);
793: curp->sodest = NULL;
794:
1.24 kristaps 795: if (curp->secondary)
796: curp->secondary->sz = 0;
1.1 kristaps 797:
798: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 799: curp->gzip = 0;
1.1 kristaps 800: }
801:
802: void
803: mparse_free(struct mparse *curp)
804: {
805:
1.164 schwarze 806: roffhash_free(curp->man->mdocmac);
807: roffhash_free(curp->man->manmac);
1.137 schwarze 808: roff_man_free(curp->man);
1.160 schwarze 809: roff_free(curp->roff);
1.24 kristaps 810: if (curp->secondary)
811: free(curp->secondary->buf);
1.1 kristaps 812:
1.24 kristaps 813: free(curp->secondary);
1.45 schwarze 814: free(curp->sodest);
1.1 kristaps 815: free(curp);
816: }
817:
818: void
1.135 schwarze 819: mparse_result(struct mparse *curp, struct roff_man **man,
820: char **sodest)
1.1 kristaps 821: {
822:
1.45 schwarze 823: if (sodest && NULL != (*sodest = curp->sodest)) {
824: *man = NULL;
825: return;
826: }
1.9 kristaps 827: if (man)
828: *man = curp->man;
1.157 schwarze 829: }
830:
831: void
832: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
833: {
834: if (curp->file_status > *rc)
835: *rc = curp->file_status;
1.3 kristaps 836: }
837:
838: void
839: mandoc_vmsg(enum mandocerr t, struct mparse *m,
840: int ln, int pos, const char *fmt, ...)
841: {
842: char buf[256];
843: va_list ap;
844:
845: va_start(ap, fmt);
1.48 schwarze 846: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 847: va_end(ap);
848:
849: mandoc_msg(t, m, ln, pos, buf);
850: }
851:
852: void
1.47 schwarze 853: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 854: int ln, int col, const char *msg)
855: {
856: enum mandoclevel level;
857:
1.112 schwarze 858: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 859: while (er < mandoclimits[level])
860: level--;
861:
1.108 schwarze 862: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 863: return;
864:
1.8 kristaps 865: if (m->mmsg)
866: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 867:
868: if (m->file_status < level)
869: m->file_status = level;
1.7 kristaps 870: }
871:
872: const char *
873: mparse_strerror(enum mandocerr er)
874: {
875:
1.142 schwarze 876: return mandocerrs[er];
1.7 kristaps 877: }
878:
879: const char *
880: mparse_strlevel(enum mandoclevel lvl)
881: {
1.142 schwarze 882: return mandoclevels[lvl];
1.24 kristaps 883: }
884:
885: void
886: mparse_keep(struct mparse *p)
887: {
888:
889: assert(NULL == p->secondary);
890: p->secondary = mandoc_calloc(1, sizeof(struct buf));
891: }
892:
893: const char *
894: mparse_getkeep(const struct mparse *p)
895: {
896:
897: assert(p->secondary);
1.142 schwarze 898: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 899: }
CVSweb