Annotation of mandoc/read.c, Revision 1.180
1.180 ! schwarze 1: /* $Id: read.c,v 1.179 2017/06/24 14:38:33 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 */
1.179 schwarze 56: const char *os_s; /* default operating system */
1.83 schwarze 57: mandocmsg mmsg; /* warning/error message handler */
58: enum mandoclevel file_status; /* status of current parse */
1.179 schwarze 59: enum mandocerr mmin; /* ignore messages below this */
1.83 schwarze 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:
1.179 schwarze 89: "base system convention",
1.166 schwarze 90:
1.176 schwarze 91: "Mdocdate found",
92: "Mdocdate missing",
1.180 ! schwarze 93: "operating system explicitly specified",
1.179 schwarze 94: "RCS id missing",
95:
96: "generic style suggestion",
97:
1.176 schwarze 98: "legacy man(7) date format",
1.178 schwarze 99: "duplicate RCS id",
1.166 schwarze 100: "useless macro",
1.167 schwarze 101: "consider using OS macro",
1.172 schwarze 102: "errnos out of order",
103: "duplicate errno",
1.168 schwarze 104: "description line ends with a full stop",
1.174 schwarze 105: "no blank before trailing delimiter",
1.175 schwarze 106: "function name without markup",
1.166 schwarze 107:
1.7 kristaps 108: "generic warning",
109:
110: /* related to the prologue */
1.79 schwarze 111: "missing manual title, using UNTITLED",
112: "missing manual title, using \"\"",
1.54 schwarze 113: "lower case character in document title",
1.79 schwarze 114: "missing manual section, using \"\"",
1.7 kristaps 115: "unknown manual section",
1.54 schwarze 116: "missing date, using today's date",
1.7 kristaps 117: "cannot parse date, using it verbatim",
1.79 schwarze 118: "missing Os macro, using \"\"",
119: "duplicate prologue macro",
120: "late prologue macro",
121: "skipping late title macro",
1.7 kristaps 122: "prologue macros out of order",
123:
124: /* related to document structure */
125: ".so is fragile, better use ln(1)",
1.50 schwarze 126: "no document body",
1.54 schwarze 127: "content before first section header",
128: "first section is not \"NAME\"",
1.156 schwarze 129: "NAME section without Nm before Nd",
1.128 schwarze 130: "NAME section without description",
131: "description not at the end of NAME",
132: "bad NAME section content",
1.156 schwarze 133: "missing comma before name",
1.121 schwarze 134: "missing description line, using \"\"",
1.162 schwarze 135: "description line outside NAME section",
1.7 kristaps 136: "sections out of conventional order",
1.54 schwarze 137: "duplicate section title",
138: "unexpected section",
1.87 schwarze 139: "unusual Xr order",
140: "unusual Xr punctuation",
1.86 schwarze 141: "AUTHORS section without An macro",
1.7 kristaps 142:
143: /* related to macros and nesting */
1.55 schwarze 144: "obsolete macro",
1.102 schwarze 145: "macro neither callable nor escaped",
1.7 kristaps 146: "skipping paragraph macro",
1.31 schwarze 147: "moving paragraph macro out of list",
1.7 kristaps 148: "skipping no-space macro",
149: "blocks badly nested",
150: "nested displays are not portable",
1.57 schwarze 151: "moving content out of list",
1.78 schwarze 152: "fill mode already enabled, skipping",
153: "fill mode already disabled, skipping",
1.7 kristaps 154: "line scope broken",
1.169 schwarze 155: "skipping blank line in line scope",
1.7 kristaps 156:
157: /* related to missing macro arguments */
1.58 schwarze 158: "skipping empty request",
159: "conditional request controls empty scope",
1.7 kristaps 160: "skipping empty macro",
1.124 schwarze 161: "empty block",
1.62 schwarze 162: "empty argument, using 0n",
1.60 schwarze 163: "missing display type, using -ragged",
164: "list type is not the first argument",
1.152 schwarze 165: "missing -width in -tag list, using 6n",
1.78 schwarze 166: "missing utility name, using \"\"",
1.123 schwarze 167: "missing function name, using \"\"",
1.60 schwarze 168: "empty head in list item",
169: "empty list item",
1.61 schwarze 170: "missing font type, using \\fR",
171: "unknown font type, using \\fR",
1.103 schwarze 172: "nothing follows prefix",
1.122 schwarze 173: "empty reference block",
1.155 schwarze 174: "missing section argument",
1.60 schwarze 175: "missing -std argument, adding it",
1.125 schwarze 176: "missing option string, using \"\"",
177: "missing resource identifier, using \"\"",
1.90 schwarze 178: "missing eqn box, using \"\"",
1.7 kristaps 179:
180: /* related to bad macro arguments */
1.64 schwarze 181: "unterminated quoted argument",
1.7 kristaps 182: "duplicate argument",
1.76 schwarze 183: "skipping duplicate argument",
1.63 schwarze 184: "skipping duplicate display type",
185: "skipping duplicate list type",
1.76 schwarze 186: "skipping -width argument",
1.126 schwarze 187: "wrong number of cells",
1.7 kristaps 188: "unknown AT&T UNIX version",
1.88 schwarze 189: "comma in function argument",
1.89 schwarze 190: "parenthesis in function name",
1.173 schwarze 191: "unknown library name",
1.67 schwarze 192: "invalid content in Rs block",
1.63 schwarze 193: "invalid Boolean argument",
194: "unknown font, skipping request",
1.126 schwarze 195: "odd number of characters in request",
1.7 kristaps 196:
197: /* related to plain text */
1.64 schwarze 198: "blank line in fill mode, using .sp",
199: "tab in filled text",
200: "whitespace at end of input line",
1.158 schwarze 201: "new sentence, new line",
1.7 kristaps 202: "bad comment style",
1.64 schwarze 203: "invalid escape sequence",
204: "undefined string, using \"\"",
1.16 kristaps 205:
1.118 schwarze 206: /* related to tables */
207: "tbl line starts with span",
208: "tbl column starts with span",
209: "skipping vertical bar in tbl layout",
210:
1.7 kristaps 211: "generic error",
212:
213: /* related to tables */
1.116 schwarze 214: "non-alphabetic character in tbl options",
215: "skipping unknown tbl option",
216: "missing tbl option argument",
217: "wrong tbl option argument size",
1.118 schwarze 218: "empty tbl layout",
219: "invalid character in tbl layout",
220: "unmatched parenthesis in tbl layout",
1.119 schwarze 221: "tbl without any data cells",
222: "ignoring data in spanned tbl cell",
223: "ignoring extra tbl data cells",
224: "data block open at end of tbl",
1.7 kristaps 225:
1.68 schwarze 226: /* related to document structure and macros */
1.108 schwarze 227: NULL,
1.7 kristaps 228: "input stack limit exceeded, infinite loop?",
229: "skipping bad character",
1.68 schwarze 230: "skipping unknown macro",
1.112 schwarze 231: "skipping insecure request",
1.73 schwarze 232: "skipping item outside list",
1.68 schwarze 233: "skipping column outside column list",
234: "skipping end of block that is not open",
1.115 schwarze 235: "fewer RS blocks open, skipping",
1.68 schwarze 236: "inserting missing end of block",
237: "appending missing end of block",
238:
239: /* related to request and macro arguments */
1.7 kristaps 240: "escaped character not allowed in a name",
1.99 schwarze 241: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 242: "skipping display without arguments",
1.71 schwarze 243: "missing list type, using -item",
1.171 schwarze 244: "argument is not numeric, using 1",
1.70 schwarze 245: "missing manual name, using \"\"",
1.71 schwarze 246: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 247: "unknown standard specifier",
1.71 schwarze 248: "skipping request without numeric argument",
1.109 schwarze 249: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 250: ".so request failed",
1.61 schwarze 251: "skipping all arguments",
252: "skipping excess arguments",
1.92 kristaps 253: "divide by zero",
1.112 schwarze 254:
255: "unsupported feature",
256: "input too large",
1.114 schwarze 257: "unsupported control character",
1.112 schwarze 258: "unsupported roff request",
1.119 schwarze 259: "eqn delim option in tbl",
1.118 schwarze 260: "unsupported tbl layout modifier",
1.112 schwarze 261: "ignoring macro in table",
1.7 kristaps 262: };
263:
264: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
265: "SUCCESS",
1.165 schwarze 266: "STYLE",
1.7 kristaps 267: "WARNING",
268: "ERROR",
1.112 schwarze 269: "UNSUPP",
1.7 kristaps 270: "BADARG",
271: "SYSERR"
272: };
273:
1.47 schwarze 274:
1.1 kristaps 275: static void
276: resize_buf(struct buf *buf, size_t initial)
277: {
278:
279: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
280: buf->buf = mandoc_realloc(buf->buf, buf->sz);
281: }
282:
283: static void
1.84 schwarze 284: choose_parser(struct mparse *curp)
1.1 kristaps 285: {
1.83 schwarze 286: char *cp, *ep;
287: int format;
1.1 kristaps 288:
1.83 schwarze 289: /*
290: * If neither command line arguments -mdoc or -man select
291: * a parser nor the roff parser found a .Dd or .TH macro
292: * yet, look ahead in the main input buffer.
293: */
294:
295: if ((format = roff_getformat(curp->roff)) == 0) {
296: cp = curp->primary->buf;
297: ep = cp + curp->primary->sz;
298: while (cp < ep) {
1.85 schwarze 299: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 300: cp++;
301: if (cp[0] == 'D' && cp[1] == 'd') {
302: format = MPARSE_MDOC;
303: break;
304: }
305: if (cp[0] == 'T' && cp[1] == 'H') {
306: format = MPARSE_MAN;
307: break;
308: }
309: }
310: cp = memchr(cp, '\n', ep - cp);
311: if (cp == NULL)
312: break;
313: cp++;
314: }
1.1 kristaps 315: }
316:
1.83 schwarze 317: if (format == MPARSE_MDOC) {
1.137 schwarze 318: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 319: if (curp->man->mdocmac == NULL)
320: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 321: } else {
322: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 323: if (curp->man->manmac == NULL)
324: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 325: }
1.164 schwarze 326: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 327: }
328:
329: /*
1.95 schwarze 330: * Main parse routine for a buffer.
331: * It assumes encoding and line numbering are already set up.
332: * It can recurse directly (for invocations of user-defined
333: * macros, inline equations, and input line traps)
334: * and indirectly (for .so file inclusion).
1.1 kristaps 335: */
1.163 schwarze 336: static int
1.95 schwarze 337: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 338: {
339: const struct tbl_span *span;
340: struct buf ln;
1.113 schwarze 341: const char *save_file;
1.110 schwarze 342: char *cp;
1.95 schwarze 343: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 344: enum rofferr rr;
1.100 schwarze 345: int of;
1.1 kristaps 346: int lnn; /* line number in the real file */
1.110 schwarze 347: int fd;
1.1 kristaps 348: unsigned char c;
349:
1.95 schwarze 350: memset(&ln, 0, sizeof(ln));
1.1 kristaps 351:
1.47 schwarze 352: lnn = curp->line;
353: pos = 0;
1.1 kristaps 354:
1.95 schwarze 355: while (i < blk.sz) {
1.1 kristaps 356: if (0 == pos && '\0' == blk.buf[i])
357: break;
358:
359: if (start) {
360: curp->line = lnn;
361: curp->reparse_count = 0;
1.93 schwarze 362:
363: if (lnn < 3 &&
364: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 365: curp->filenc & MPARSE_LATIN1)
366: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 367: }
368:
1.95 schwarze 369: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 370:
371: /*
372: * When finding an unescaped newline character,
373: * leave the character loop to process the line.
374: * Skip a preceding carriage return, if any.
375: */
376:
1.95 schwarze 377: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 378: '\n' == blk.buf[i + 1])
379: ++i;
380: if ('\n' == blk.buf[i]) {
381: ++i;
382: ++lnn;
383: break;
384: }
385:
1.35 schwarze 386: /*
1.93 schwarze 387: * Make sure we have space for the worst
388: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 389: */
390:
1.95 schwarze 391: if (pos + 11 > ln.sz)
1.35 schwarze 392: resize_buf(&ln, 256);
393:
1.47 schwarze 394: /*
1.93 schwarze 395: * Encode 8-bit input.
1.1 kristaps 396: */
397:
1.93 schwarze 398: c = blk.buf[i];
399: if (c & 0x80) {
1.95 schwarze 400: if ( ! (curp->filenc && preconv_encode(
401: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 402: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
403: curp->line, pos, "0x%x", c);
1.93 schwarze 404: ln.buf[pos++] = '?';
405: i++;
406: }
407: continue;
408: }
409:
410: /*
411: * Exclude control characters.
412: */
1.1 kristaps 413:
1.93 schwarze 414: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 415: mandoc_vmsg(c == 0x00 || c == 0x04 ||
416: c > 0x0a ? MANDOCERR_CHAR_BAD :
417: MANDOCERR_CHAR_UNSUPP,
418: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 419: i++;
1.127 schwarze 420: if (c != '\r')
421: ln.buf[pos++] = '?';
1.1 kristaps 422: continue;
423: }
424:
425: ln.buf[pos++] = blk.buf[i++];
426: }
427:
1.170 schwarze 428: if (pos + 1 >= ln.sz)
1.1 kristaps 429: resize_buf(&ln, 256);
430:
1.170 schwarze 431: if (i == blk.sz || blk.buf[i] == '\0')
432: ln.buf[pos++] = '\n';
1.1 kristaps 433: ln.buf[pos] = '\0';
434:
435: /*
436: * A significant amount of complexity is contained by
437: * the roff preprocessor. It's line-oriented but can be
438: * expressed on one line, so we need at times to
439: * readjust our starting point and re-run it. The roff
440: * preprocessor can also readjust the buffers with new
441: * data, so we pass them in wholesale.
442: */
443:
444: of = 0;
445:
1.24 kristaps 446: /*
447: * Maintain a lookaside buffer of all parsed lines. We
448: * only do this if mparse_keep() has been invoked (the
449: * buffer may be accessed with mparse_getkeep()).
450: */
451:
452: if (curp->secondary) {
1.47 schwarze 453: curp->secondary->buf = mandoc_realloc(
454: curp->secondary->buf,
455: curp->secondary->sz + pos + 2);
456: memcpy(curp->secondary->buf +
457: curp->secondary->sz,
458: ln.buf, pos);
1.24 kristaps 459: curp->secondary->sz += pos;
460: curp->secondary->buf
461: [curp->secondary->sz] = '\n';
462: curp->secondary->sz++;
463: curp->secondary->buf
464: [curp->secondary->sz] = '\0';
465: }
1.1 kristaps 466: rerun:
1.96 schwarze 467: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 468:
469: switch (rr) {
1.47 schwarze 470: case ROFF_REPARSE:
1.163 schwarze 471: if (++curp->reparse_count > REPARSE_LIMIT)
1.3 kristaps 472: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 473: curp->line, pos, NULL);
1.163 schwarze 474: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
475: start == 1) {
476: pos = 0;
477: continue;
478: }
479: free(ln.buf);
480: return 0;
1.47 schwarze 481: case ROFF_APPEND:
1.95 schwarze 482: pos = strlen(ln.buf);
1.1 kristaps 483: continue;
1.47 schwarze 484: case ROFF_RERUN:
1.1 kristaps 485: goto rerun;
1.47 schwarze 486: case ROFF_IGN:
1.1 kristaps 487: pos = 0;
488: continue;
1.47 schwarze 489: case ROFF_SO:
1.95 schwarze 490: if ( ! (curp->options & MPARSE_SO) &&
491: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 492: curp->sodest = mandoc_strdup(ln.buf + of);
493: free(ln.buf);
1.163 schwarze 494: return 1;
1.45 schwarze 495: }
1.24 kristaps 496: /*
497: * We remove `so' clauses from our lookaside
498: * buffer because we're going to descend into
499: * the file recursively.
500: */
1.47 schwarze 501: if (curp->secondary)
1.25 kristaps 502: curp->secondary->sz -= pos + 1;
1.113 schwarze 503: save_file = curp->file;
1.148 schwarze 504: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 505: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 506: close(fd);
1.113 schwarze 507: curp->file = save_file;
508: } else {
509: curp->file = save_file;
1.52 schwarze 510: mandoc_vmsg(MANDOCERR_SO_FAIL,
511: curp, curp->line, pos,
512: ".so %s", ln.buf + of);
1.110 schwarze 513: ln.sz = mandoc_asprintf(&cp,
514: ".sp\nSee the file %s.\n.sp",
515: ln.buf + of);
516: free(ln.buf);
517: ln.buf = cp;
518: of = 0;
519: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 520: }
1.1 kristaps 521: pos = 0;
522: continue;
523: default:
524: break;
525: }
526:
1.150 schwarze 527: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 528: choose_parser(curp);
1.1 kristaps 529:
1.47 schwarze 530: /*
1.84 schwarze 531: * Lastly, push down into the parsers themselves.
1.1 kristaps 532: * If libroff returns ROFF_TBL, then add it to the
533: * currently open parse. Since we only get here if
534: * there does exist data (see tbl_data.c), we're
535: * guaranteed that something's been allocated.
536: * Do the same for ROFF_EQN.
537: */
538:
1.139 schwarze 539: if (rr == ROFF_TBL)
1.100 schwarze 540: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 541: roff_addtbl(curp->man, span);
542: else if (rr == ROFF_EQN)
543: roff_addeqn(curp->man, roff_eqn(curp->roff));
544: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 545: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 546: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
547: break;
1.1 kristaps 548:
549: /* Temporary buffers typically are not full. */
550:
551: if (0 == start && '\0' == blk.buf[i])
552: break;
553:
554: /* Start the next input line. */
555:
556: pos = 0;
557: }
558:
559: free(ln.buf);
1.163 schwarze 560: return 1;
1.1 kristaps 561: }
562:
563: static int
1.40 schwarze 564: read_whole_file(struct mparse *curp, const char *file, int fd,
565: struct buf *fb, int *with_mmap)
1.1 kristaps 566: {
1.161 schwarze 567: struct stat st;
1.140 schwarze 568: gzFile gz;
1.1 kristaps 569: size_t off;
570: ssize_t ssz;
1.143 schwarze 571:
572: if (fstat(fd, &st) == -1)
573: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 574:
575: /*
576: * If we're a regular file, try just reading in the whole entry
577: * via mmap(). This is faster than reading it into blocks, and
578: * since each file is only a few bytes to begin with, I'm not
579: * concerned that this is going to tank any machines.
580: */
581:
1.140 schwarze 582: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 583: if (st.st_size > 0x7fffffff) {
1.111 schwarze 584: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 585: return 0;
1.1 kristaps 586: }
587: *with_mmap = 1;
588: fb->sz = (size_t)st.st_size;
1.37 schwarze 589: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 590: if (fb->buf != MAP_FAILED)
1.142 schwarze 591: return 1;
1.1 kristaps 592: }
593:
1.140 schwarze 594: if (curp->gzip) {
1.143 schwarze 595: if ((gz = gzdopen(fd, "rb")) == NULL)
596: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 597: } else
598: gz = NULL;
599:
1.1 kristaps 600: /*
601: * If this isn't a regular file (like, say, stdin), then we must
602: * go the old way and just read things in bit by bit.
603: */
604:
605: *with_mmap = 0;
606: off = 0;
607: fb->sz = 0;
608: fb->buf = NULL;
609: for (;;) {
610: if (off == fb->sz) {
611: if (fb->sz == (1U << 31)) {
1.111 schwarze 612: mandoc_msg(MANDOCERR_TOOLARGE, curp,
613: 0, 0, NULL);
1.1 kristaps 614: break;
615: }
616: resize_buf(fb, 65536);
617: }
1.140 schwarze 618: ssz = curp->gzip ?
619: gzread(gz, fb->buf + (int)off, fb->sz - off) :
620: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 621: if (ssz == 0) {
622: fb->sz = off;
1.142 schwarze 623: return 1;
1.1 kristaps 624: }
1.143 schwarze 625: if (ssz == -1)
626: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 627: off += (size_t)ssz;
628: }
629:
630: free(fb->buf);
631: fb->buf = NULL;
1.142 schwarze 632: return 0;
1.1 kristaps 633: }
634:
635: static void
636: mparse_end(struct mparse *curp)
637: {
1.135 schwarze 638: if (curp->man->macroset == MACROSET_NONE)
639: curp->man->macroset = MACROSET_MAN;
640: if (curp->man->macroset == MACROSET_MDOC)
641: mdoc_endparse(curp->man);
642: else
1.111 schwarze 643: man_endparse(curp->man);
1.1 kristaps 644: roff_endparse(curp->roff);
645: }
646:
647: static void
1.36 schwarze 648: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 649: {
1.85 schwarze 650: struct buf *svprimary;
1.28 joerg 651: const char *svfile;
1.95 schwarze 652: size_t offset;
1.36 schwarze 653: static int recursion_depth;
654:
655: if (64 < recursion_depth) {
656: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
657: return;
658: }
1.28 joerg 659:
660: /* Line number is per-file. */
661: svfile = curp->file;
662: curp->file = file;
1.85 schwarze 663: svprimary = curp->primary;
1.83 schwarze 664: curp->primary = &blk;
1.28 joerg 665: curp->line = 1;
1.36 schwarze 666: recursion_depth++;
1.28 joerg 667:
1.93 schwarze 668: /* Skip an UTF-8 byte order mark. */
669: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
670: (unsigned char)blk.buf[0] == 0xef &&
671: (unsigned char)blk.buf[1] == 0xbb &&
672: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 673: offset = 3;
1.93 schwarze 674: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 675: } else
676: offset = 0;
1.93 schwarze 677:
1.95 schwarze 678: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 679:
1.111 schwarze 680: if (--recursion_depth == 0)
1.28 joerg 681: mparse_end(curp);
682:
1.85 schwarze 683: curp->primary = svprimary;
1.28 joerg 684: curp->file = svfile;
685: }
686:
687: enum mandoclevel
1.104 schwarze 688: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 689: const char *file)
690: {
691: struct buf blk;
692:
1.104 schwarze 693: blk.buf = buf;
1.28 joerg 694: blk.sz = len;
695:
1.36 schwarze 696: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 697: return curp->file_status;
1.28 joerg 698: }
699:
1.98 schwarze 700: /*
701: * Read the whole file into memory and call the parsers.
702: * Called recursively when an .so request is encountered.
703: */
1.36 schwarze 704: enum mandoclevel
705: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 706: {
1.28 joerg 707: struct buf blk;
708: int with_mmap;
1.93 schwarze 709: int save_filenc;
1.1 kristaps 710:
1.91 schwarze 711: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 712: save_filenc = curp->filenc;
713: curp->filenc = curp->options &
714: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 715: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 716: curp->filenc = save_filenc;
1.91 schwarze 717: if (with_mmap)
718: munmap(blk.buf, blk.sz);
719: else
720: free(blk.buf);
721: }
1.142 schwarze 722: return curp->file_status;
1.82 schwarze 723: }
724:
1.148 schwarze 725: int
726: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 727: {
728: char *cp;
1.148 schwarze 729: int fd;
1.82 schwarze 730:
731: curp->file = file;
1.140 schwarze 732: cp = strrchr(file, '.');
733: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 734:
1.140 schwarze 735: /* First try to use the filename as it is. */
1.98 schwarze 736:
1.148 schwarze 737: if ((fd = open(file, O_RDONLY)) != -1)
738: return fd;
1.98 schwarze 739:
1.140 schwarze 740: /*
741: * If that doesn't work and the filename doesn't
742: * already end in .gz, try appending .gz.
743: */
1.98 schwarze 744:
1.140 schwarze 745: if ( ! curp->gzip) {
1.98 schwarze 746: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 747: fd = open(cp, O_RDONLY);
1.108 schwarze 748: free(cp);
1.148 schwarze 749: if (fd != -1) {
1.140 schwarze 750: curp->gzip = 1;
1.148 schwarze 751: return fd;
1.82 schwarze 752: }
753: }
754:
1.140 schwarze 755: /* Neither worked, give up. */
1.97 schwarze 756:
1.140 schwarze 757: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 758: return -1;
1.1 kristaps 759: }
760:
761: struct mparse *
1.179 schwarze 762: mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
763: enum mandoc_os os_e, const char *os_s)
1.1 kristaps 764: {
765: struct mparse *curp;
1.10 kristaps 766:
1.1 kristaps 767: curp = mandoc_calloc(1, sizeof(struct mparse));
768:
1.44 schwarze 769: curp->options = options;
1.179 schwarze 770: curp->mmin = mmin;
1.1 kristaps 771: curp->mmsg = mmsg;
1.179 schwarze 772: curp->os_s = os_s;
1.1 kristaps 773:
1.144 schwarze 774: curp->roff = roff_alloc(curp, options);
1.179 schwarze 775: curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
1.137 schwarze 776: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 777: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 778: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 779: if (curp->man->mdocmac == NULL)
780: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 781: } else if (curp->options & MPARSE_MAN) {
782: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 783: if (curp->man->manmac == NULL)
784: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 785: }
1.138 schwarze 786: curp->man->first->tok = TOKEN_NONE;
1.179 schwarze 787: curp->man->meta.os_e = os_e;
1.142 schwarze 788: return curp;
1.1 kristaps 789: }
790:
791: void
792: mparse_reset(struct mparse *curp)
793: {
794: roff_reset(curp->roff);
1.150 schwarze 795: roff_man_reset(curp->man);
1.160 schwarze 796:
797: free(curp->sodest);
798: curp->sodest = NULL;
799:
1.24 kristaps 800: if (curp->secondary)
801: curp->secondary->sz = 0;
1.1 kristaps 802:
803: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 804: curp->gzip = 0;
1.1 kristaps 805: }
806:
807: void
808: mparse_free(struct mparse *curp)
809: {
810:
1.164 schwarze 811: roffhash_free(curp->man->mdocmac);
812: roffhash_free(curp->man->manmac);
1.137 schwarze 813: roff_man_free(curp->man);
1.160 schwarze 814: roff_free(curp->roff);
1.24 kristaps 815: if (curp->secondary)
816: free(curp->secondary->buf);
1.1 kristaps 817:
1.24 kristaps 818: free(curp->secondary);
1.45 schwarze 819: free(curp->sodest);
1.1 kristaps 820: free(curp);
821: }
822:
823: void
1.135 schwarze 824: mparse_result(struct mparse *curp, struct roff_man **man,
825: char **sodest)
1.1 kristaps 826: {
827:
1.45 schwarze 828: if (sodest && NULL != (*sodest = curp->sodest)) {
829: *man = NULL;
830: return;
831: }
1.9 kristaps 832: if (man)
833: *man = curp->man;
1.157 schwarze 834: }
835:
836: void
837: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
838: {
839: if (curp->file_status > *rc)
840: *rc = curp->file_status;
1.3 kristaps 841: }
842:
843: void
844: mandoc_vmsg(enum mandocerr t, struct mparse *m,
845: int ln, int pos, const char *fmt, ...)
846: {
847: char buf[256];
848: va_list ap;
849:
850: va_start(ap, fmt);
1.48 schwarze 851: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 852: va_end(ap);
853:
854: mandoc_msg(t, m, ln, pos, buf);
855: }
856:
857: void
1.47 schwarze 858: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 859: int ln, int col, const char *msg)
860: {
861: enum mandoclevel level;
862:
1.179 schwarze 863: if (er < m->mmin && er != MANDOCERR_FILE)
864: return;
865:
1.112 schwarze 866: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 867: while (er < mandoclimits[level])
868: level--;
869:
1.8 kristaps 870: if (m->mmsg)
871: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 872:
873: if (m->file_status < level)
874: m->file_status = level;
1.7 kristaps 875: }
876:
877: const char *
878: mparse_strerror(enum mandocerr er)
879: {
880:
1.142 schwarze 881: return mandocerrs[er];
1.7 kristaps 882: }
883:
884: const char *
885: mparse_strlevel(enum mandoclevel lvl)
886: {
1.142 schwarze 887: return mandoclevels[lvl];
1.24 kristaps 888: }
889:
890: void
891: mparse_keep(struct mparse *p)
892: {
893:
894: assert(NULL == p->secondary);
895: p->secondary = mandoc_calloc(1, sizeof(struct buf));
896: }
897:
898: const char *
899: mparse_getkeep(const struct mparse *p)
900: {
901:
902: assert(p->secondary);
1.142 schwarze 903: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 904: }
CVSweb