Annotation of src/usr.bin/mandoc/read.c, Revision 1.155
1.155 ! schwarze 1: /* $OpenBSD: read.c,v 1.154 2017/06/25 17:42:37 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.129 schwarze 4: * Copyright (c) 2010-2017 Ingo Schwarze <schwarze@openbsd.org>
1.20 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 schwarze 6: *
7: * Permission to use, copy, modify, and distribute this software for any
8: * purpose with or without fee is hereby granted, provided that the above
9: * copyright notice and this permission notice appear in all copies.
10: *
1.108 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.108 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18: */
1.58 schwarze 19: #include <sys/types.h>
20: #include <sys/mman.h>
1.1 schwarze 21: #include <sys/stat.h>
22:
23: #include <assert.h>
24: #include <ctype.h>
1.118 schwarze 25: #include <err.h>
1.18 schwarze 26: #include <errno.h>
1.1 schwarze 27: #include <fcntl.h>
28: #include <stdarg.h>
29: #include <stdio.h>
30: #include <stdlib.h>
31: #include <string.h>
32: #include <unistd.h>
1.115 schwarze 33: #include <zlib.h>
1.1 schwarze 34:
1.108 schwarze 35: #include "mandoc_aux.h"
1.1 schwarze 36: #include "mandoc.h"
1.108 schwarze 37: #include "roff.h"
1.1 schwarze 38: #include "mdoc.h"
39: #include "man.h"
1.108 schwarze 40: #include "libmandoc.h"
1.114 schwarze 41: #include "roff_int.h"
1.1 schwarze 42:
43: #define REPARSE_LIMIT 1000
44:
45: struct mparse {
1.133 schwarze 46: struct roff *roff; /* roff parser (!NULL) */
1.109 schwarze 47: struct roff_man *man; /* man parser */
1.23 schwarze 48: char *sodest; /* filename pointed to by .so */
1.59 schwarze 49: const char *file; /* filename of current input file */
50: struct buf *primary; /* buffer currently being parsed */
51: struct buf *secondary; /* preprocessed copy of input */
1.151 schwarze 52: const char *os_s; /* default operating system */
1.59 schwarze 53: mandocmsg mmsg; /* warning/error message handler */
54: enum mandoclevel file_status; /* status of current parse */
1.151 schwarze 55: enum mandocerr mmin; /* ignore messages below this */
1.59 schwarze 56: int options; /* parser options */
1.115 schwarze 57: int gzip; /* current input file is gzipped */
1.70 schwarze 58: int filenc; /* encoding of the current file */
1.1 schwarze 59: int reparse_count; /* finite interp. stack */
1.59 schwarze 60: int line; /* line number in the file */
1.1 schwarze 61: };
62:
1.60 schwarze 63: static void choose_parser(struct mparse *);
1.1 schwarze 64: static void resize_buf(struct buf *, size_t);
1.135 schwarze 65: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.18 schwarze 66: static int read_whole_file(struct mparse *, const char *, int,
67: struct buf *, int *);
1.1 schwarze 68: static void mparse_end(struct mparse *);
1.15 schwarze 69: static void mparse_parse_buffer(struct mparse *, struct buf,
70: const char *);
1.1 schwarze 71:
72: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
73: MANDOCERR_OK,
1.137 schwarze 74: MANDOCERR_STYLE,
1.1 schwarze 75: MANDOCERR_WARNING,
76: MANDOCERR_ERROR,
1.87 schwarze 77: MANDOCERR_UNSUPP,
1.1 schwarze 78: MANDOCERR_MAX,
79: MANDOCERR_MAX
80: };
81:
82: static const char * const mandocerrs[MANDOCERR_MAX] = {
83: "ok",
84:
1.151 schwarze 85: "base system convention",
1.138 schwarze 86:
1.148 schwarze 87: "Mdocdate found",
88: "Mdocdate missing",
1.153 schwarze 89: "unknown architecture",
1.152 schwarze 90: "operating system explicitly specified",
1.151 schwarze 91: "RCS id missing",
92:
93: "generic style suggestion",
94:
1.148 schwarze 95: "legacy man(7) date format",
1.150 schwarze 96: "duplicate RCS id",
1.154 schwarze 97: "typo in section name",
1.138 schwarze 98: "useless macro",
1.139 schwarze 99: "consider using OS macro",
1.144 schwarze 100: "errnos out of order",
101: "duplicate errno",
1.140 schwarze 102: "description line ends with a full stop",
1.146 schwarze 103: "no blank before trailing delimiter",
1.147 schwarze 104: "function name without markup",
1.138 schwarze 105:
1.1 schwarze 106: "generic warning",
107:
108: /* related to the prologue */
1.57 schwarze 109: "missing manual title, using UNTITLED",
110: "missing manual title, using \"\"",
1.32 schwarze 111: "lower case character in document title",
1.57 schwarze 112: "missing manual section, using \"\"",
1.1 schwarze 113: "unknown manual section",
1.32 schwarze 114: "missing date, using today's date",
1.1 schwarze 115: "cannot parse date, using it verbatim",
1.57 schwarze 116: "missing Os macro, using \"\"",
117: "duplicate prologue macro",
118: "late prologue macro",
119: "skipping late title macro",
1.1 schwarze 120: "prologue macros out of order",
121:
122: /* related to document structure */
123: ".so is fragile, better use ln(1)",
1.28 schwarze 124: "no document body",
1.32 schwarze 125: "content before first section header",
126: "first section is not \"NAME\"",
1.129 schwarze 127: "NAME section without Nm before Nd",
1.103 schwarze 128: "NAME section without description",
129: "description not at the end of NAME",
130: "bad NAME section content",
1.129 schwarze 131: "missing comma before name",
1.96 schwarze 132: "missing description line, using \"\"",
1.134 schwarze 133: "description line outside NAME section",
1.1 schwarze 134: "sections out of conventional order",
1.32 schwarze 135: "duplicate section title",
136: "unexpected section",
1.63 schwarze 137: "unusual Xr order",
138: "unusual Xr punctuation",
1.62 schwarze 139: "AUTHORS section without An macro",
1.1 schwarze 140:
141: /* related to macros and nesting */
1.33 schwarze 142: "obsolete macro",
1.78 schwarze 143: "macro neither callable nor escaped",
1.1 schwarze 144: "skipping paragraph macro",
1.10 schwarze 145: "moving paragraph macro out of list",
1.1 schwarze 146: "skipping no-space macro",
147: "blocks badly nested",
148: "nested displays are not portable",
1.35 schwarze 149: "moving content out of list",
1.155 ! schwarze 150: "first macro on line",
1.56 schwarze 151: "fill mode already enabled, skipping",
152: "fill mode already disabled, skipping",
1.1 schwarze 153: "line scope broken",
1.141 schwarze 154: "skipping blank line in line scope",
1.1 schwarze 155:
156: /* related to missing macro arguments */
1.36 schwarze 157: "skipping empty request",
158: "conditional request controls empty scope",
1.1 schwarze 159: "skipping empty macro",
1.99 schwarze 160: "empty block",
1.40 schwarze 161: "empty argument, using 0n",
1.38 schwarze 162: "missing display type, using -ragged",
163: "list type is not the first argument",
1.125 schwarze 164: "missing -width in -tag list, using 6n",
1.56 schwarze 165: "missing utility name, using \"\"",
1.98 schwarze 166: "missing function name, using \"\"",
1.38 schwarze 167: "empty head in list item",
168: "empty list item",
1.155 ! schwarze 169: "missing argument, using next line",
1.39 schwarze 170: "missing font type, using \\fR",
171: "unknown font type, using \\fR",
1.79 schwarze 172: "nothing follows prefix",
1.97 schwarze 173: "empty reference block",
1.128 schwarze 174: "missing section argument",
1.38 schwarze 175: "missing -std argument, adding it",
1.100 schwarze 176: "missing option string, using \"\"",
177: "missing resource identifier, using \"\"",
1.66 schwarze 178: "missing eqn box, using \"\"",
1.1 schwarze 179:
180: /* related to bad macro arguments */
1.42 schwarze 181: "unterminated quoted argument",
1.1 schwarze 182: "duplicate argument",
1.54 schwarze 183: "skipping duplicate argument",
1.41 schwarze 184: "skipping duplicate display type",
185: "skipping duplicate list type",
1.54 schwarze 186: "skipping -width argument",
1.101 schwarze 187: "wrong number of cells",
1.1 schwarze 188: "unknown AT&T UNIX version",
1.64 schwarze 189: "comma in function argument",
1.65 schwarze 190: "parenthesis in function name",
1.145 schwarze 191: "unknown library name",
1.45 schwarze 192: "invalid content in Rs block",
1.41 schwarze 193: "invalid Boolean argument",
194: "unknown font, skipping request",
1.101 schwarze 195: "odd number of characters in request",
1.1 schwarze 196:
197: /* related to plain text */
1.42 schwarze 198: "blank line in fill mode, using .sp",
199: "tab in filled text",
200: "whitespace at end of input line",
1.131 schwarze 201: "new sentence, new line",
1.1 schwarze 202: "bad comment style",
1.42 schwarze 203: "invalid escape sequence",
204: "undefined string, using \"\"",
1.3 schwarze 205:
1.93 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:
1.1 schwarze 211: "generic error",
1.3 schwarze 212:
1.1 schwarze 213: /* related to tables */
1.91 schwarze 214: "non-alphabetic character in tbl options",
215: "skipping unknown tbl option",
216: "missing tbl option argument",
217: "wrong tbl option argument size",
1.93 schwarze 218: "empty tbl layout",
219: "invalid character in tbl layout",
220: "unmatched parenthesis in tbl layout",
1.94 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",
1.1 schwarze 225:
1.46 schwarze 226: /* related to document structure and macros */
1.83 schwarze 227: NULL,
1.1 schwarze 228: "input stack limit exceeded, infinite loop?",
229: "skipping bad character",
1.46 schwarze 230: "skipping unknown macro",
1.87 schwarze 231: "skipping insecure request",
1.51 schwarze 232: "skipping item outside list",
1.46 schwarze 233: "skipping column outside column list",
234: "skipping end of block that is not open",
1.90 schwarze 235: "fewer RS blocks open, skipping",
1.46 schwarze 236: "inserting missing end of block",
237: "appending missing end of block",
238:
239: /* related to request and macro arguments */
1.1 schwarze 240: "escaped character not allowed in a name",
1.75 schwarze 241: "NOT IMPLEMENTED: Bd -file",
1.120 schwarze 242: "skipping display without arguments",
1.49 schwarze 243: "missing list type, using -item",
1.143 schwarze 244: "argument is not numeric, using 1",
1.48 schwarze 245: "missing manual name, using \"\"",
1.49 schwarze 246: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 247: "unknown standard specifier",
1.49 schwarze 248: "skipping request without numeric argument",
1.84 schwarze 249: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 250: ".so request failed",
1.39 schwarze 251: "skipping all arguments",
252: "skipping excess arguments",
1.68 schwarze 253: "divide by zero",
1.87 schwarze 254:
255: "unsupported feature",
256: "input too large",
1.89 schwarze 257: "unsupported control character",
1.87 schwarze 258: "unsupported roff request",
1.94 schwarze 259: "eqn delim option in tbl",
1.93 schwarze 260: "unsupported tbl layout modifier",
1.87 schwarze 261: "ignoring macro in table",
1.1 schwarze 262: };
263:
264: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
265: "SUCCESS",
1.137 schwarze 266: "STYLE",
1.1 schwarze 267: "WARNING",
268: "ERROR",
1.87 schwarze 269: "UNSUPP",
1.1 schwarze 270: "BADARG",
271: "SYSERR"
272: };
273:
1.25 schwarze 274:
1.1 schwarze 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
1.60 schwarze 284: choose_parser(struct mparse *curp)
1.1 schwarze 285: {
1.59 schwarze 286: char *cp, *ep;
287: int format;
1.1 schwarze 288:
1.59 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) {
1.61 schwarze 299: if (*cp == '.' || *cp == '\'') {
1.59 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 schwarze 315: }
316:
1.59 schwarze 317: if (format == MPARSE_MDOC) {
1.112 schwarze 318: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 319: if (curp->man->mdocmac == NULL)
320: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 321: } else {
322: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 323: if (curp->man->manmac == NULL)
324: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 325: }
1.136 schwarze 326: curp->man->first->tok = TOKEN_NONE;
1.1 schwarze 327: }
328:
329: /*
1.71 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 schwarze 335: */
1.135 schwarze 336: static int
1.71 schwarze 337: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 338: {
339: const struct tbl_span *span;
340: struct buf ln;
1.88 schwarze 341: const char *save_file;
1.85 schwarze 342: char *cp;
1.71 schwarze 343: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 344: enum rofferr rr;
1.76 schwarze 345: int of;
1.1 schwarze 346: int lnn; /* line number in the real file */
1.85 schwarze 347: int fd;
1.1 schwarze 348: unsigned char c;
349:
1.71 schwarze 350: memset(&ln, 0, sizeof(ln));
1.1 schwarze 351:
1.25 schwarze 352: lnn = curp->line;
353: pos = 0;
1.1 schwarze 354:
1.71 schwarze 355: while (i < blk.sz) {
1.1 schwarze 356: if (0 == pos && '\0' == blk.buf[i])
357: break;
358:
359: if (start) {
360: curp->line = lnn;
361: curp->reparse_count = 0;
1.70 schwarze 362:
363: if (lnn < 3 &&
364: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 365: curp->filenc & MPARSE_LATIN1)
366: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 367: }
368:
1.71 schwarze 369: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 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:
1.71 schwarze 377: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 378: '\n' == blk.buf[i + 1])
379: ++i;
380: if ('\n' == blk.buf[i]) {
381: ++i;
382: ++lnn;
383: break;
384: }
385:
1.13 schwarze 386: /*
1.70 schwarze 387: * Make sure we have space for the worst
388: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 389: */
390:
1.71 schwarze 391: if (pos + 11 > ln.sz)
1.13 schwarze 392: resize_buf(&ln, 256);
393:
1.25 schwarze 394: /*
1.70 schwarze 395: * Encode 8-bit input.
1.1 schwarze 396: */
397:
1.70 schwarze 398: c = blk.buf[i];
399: if (c & 0x80) {
1.71 schwarze 400: if ( ! (curp->filenc && preconv_encode(
401: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 402: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
403: curp->line, pos, "0x%x", c);
1.70 schwarze 404: ln.buf[pos++] = '?';
405: i++;
406: }
407: continue;
408: }
409:
410: /*
411: * Exclude control characters.
412: */
1.1 schwarze 413:
1.70 schwarze 414: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 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 schwarze 419: i++;
1.102 schwarze 420: if (c != '\r')
421: ln.buf[pos++] = '?';
1.1 schwarze 422: continue;
423: }
424:
425: ln.buf[pos++] = blk.buf[i++];
426: }
427:
1.142 schwarze 428: if (pos + 1 >= ln.sz)
1.1 schwarze 429: resize_buf(&ln, 256);
430:
1.142 schwarze 431: if (i == blk.sz || blk.buf[i] == '\0')
432: ln.buf[pos++] = '\n';
1.1 schwarze 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:
1.4 schwarze 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) {
1.25 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);
1.4 schwarze 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 schwarze 466: rerun:
1.72 schwarze 467: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 468:
469: switch (rr) {
1.25 schwarze 470: case ROFF_REPARSE:
1.135 schwarze 471: if (++curp->reparse_count > REPARSE_LIMIT)
1.1 schwarze 472: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 473: curp->line, pos, NULL);
1.135 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;
1.25 schwarze 481: case ROFF_APPEND:
1.71 schwarze 482: pos = strlen(ln.buf);
1.1 schwarze 483: continue;
1.25 schwarze 484: case ROFF_RERUN:
1.1 schwarze 485: goto rerun;
1.25 schwarze 486: case ROFF_IGN:
1.1 schwarze 487: pos = 0;
488: continue;
1.25 schwarze 489: case ROFF_SO:
1.71 schwarze 490: if ( ! (curp->options & MPARSE_SO) &&
491: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 492: curp->sodest = mandoc_strdup(ln.buf + of);
493: free(ln.buf);
1.135 schwarze 494: return 1;
1.23 schwarze 495: }
1.4 schwarze 496: /*
497: * We remove `so' clauses from our lookaside
498: * buffer because we're going to descend into
499: * the file recursively.
500: */
1.25 schwarze 501: if (curp->secondary)
1.4 schwarze 502: curp->secondary->sz -= pos + 1;
1.88 schwarze 503: save_file = curp->file;
1.122 schwarze 504: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 505: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 506: close(fd);
1.88 schwarze 507: curp->file = save_file;
508: } else {
509: curp->file = save_file;
1.30 schwarze 510: mandoc_vmsg(MANDOCERR_SO_FAIL,
511: curp, curp->line, pos,
512: ".so %s", ln.buf + of);
1.85 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);
1.30 schwarze 520: }
1.1 schwarze 521: pos = 0;
522: continue;
523: default:
524: break;
525: }
526:
1.124 schwarze 527: if (curp->man->macroset == MACROSET_NONE)
1.60 schwarze 528: choose_parser(curp);
1.1 schwarze 529:
1.25 schwarze 530: /*
1.60 schwarze 531: * Lastly, push down into the parsers themselves.
1.1 schwarze 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:
1.114 schwarze 539: if (rr == ROFF_TBL)
1.76 schwarze 540: while ((span = roff_span(curp->roff)) != NULL)
1.114 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 ?
1.110 schwarze 545: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 546: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
547: break;
1.1 schwarze 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);
1.135 schwarze 560: return 1;
1.1 schwarze 561: }
562:
563: static int
1.18 schwarze 564: read_whole_file(struct mparse *curp, const char *file, int fd,
565: struct buf *fb, int *with_mmap)
1.1 schwarze 566: {
567: struct stat st;
1.115 schwarze 568: gzFile gz;
1.1 schwarze 569: size_t off;
570: ssize_t ssz;
571:
1.118 schwarze 572: if (fstat(fd, &st) == -1)
573: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 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:
1.115 schwarze 582: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 583: if (st.st_size > 0x7fffffff) {
1.86 schwarze 584: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.117 schwarze 585: return 0;
1.1 schwarze 586: }
587: *with_mmap = 1;
588: fb->sz = (size_t)st.st_size;
1.15 schwarze 589: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 590: if (fb->buf != MAP_FAILED)
1.117 schwarze 591: return 1;
1.1 schwarze 592: }
593:
1.115 schwarze 594: if (curp->gzip) {
1.118 schwarze 595: if ((gz = gzdopen(fd, "rb")) == NULL)
596: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.115 schwarze 597: } else
598: gz = NULL;
599:
1.1 schwarze 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)) {
1.86 schwarze 612: mandoc_msg(MANDOCERR_TOOLARGE, curp,
613: 0, 0, NULL);
1.1 schwarze 614: break;
615: }
616: resize_buf(fb, 65536);
617: }
1.115 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 schwarze 621: if (ssz == 0) {
622: fb->sz = off;
1.117 schwarze 623: return 1;
1.1 schwarze 624: }
1.118 schwarze 625: if (ssz == -1)
626: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 627: off += (size_t)ssz;
628: }
629:
630: free(fb->buf);
631: fb->buf = NULL;
1.117 schwarze 632: return 0;
1.1 schwarze 633: }
634:
635: static void
636: mparse_end(struct mparse *curp)
637: {
1.110 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
1.86 schwarze 643: man_endparse(curp->man);
1.1 schwarze 644: roff_endparse(curp->roff);
645: }
646:
1.15 schwarze 647: static void
648: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 649: {
1.61 schwarze 650: struct buf *svprimary;
1.1 schwarze 651: const char *svfile;
1.71 schwarze 652: size_t offset;
1.14 schwarze 653: static int recursion_depth;
654:
655: if (64 < recursion_depth) {
656: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 657: return;
1.14 schwarze 658: }
1.1 schwarze 659:
1.15 schwarze 660: /* Line number is per-file. */
661: svfile = curp->file;
662: curp->file = file;
1.61 schwarze 663: svprimary = curp->primary;
1.59 schwarze 664: curp->primary = &blk;
1.15 schwarze 665: curp->line = 1;
666: recursion_depth++;
667:
1.70 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) {
1.71 schwarze 673: offset = 3;
1.70 schwarze 674: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 675: } else
676: offset = 0;
1.70 schwarze 677:
1.71 schwarze 678: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 679:
1.86 schwarze 680: if (--recursion_depth == 0)
1.15 schwarze 681: mparse_end(curp);
682:
1.61 schwarze 683: curp->primary = svprimary;
1.15 schwarze 684: curp->file = svfile;
685: }
686:
1.74 schwarze 687: /*
688: * Read the whole file into memory and call the parsers.
689: * Called recursively when an .so request is encountered.
690: */
1.15 schwarze 691: enum mandoclevel
692: mparse_readfd(struct mparse *curp, int fd, const char *file)
693: {
694: struct buf blk;
695: int with_mmap;
1.70 schwarze 696: int save_filenc;
1.1 schwarze 697:
1.67 schwarze 698: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 699: save_filenc = curp->filenc;
700: curp->filenc = curp->options &
701: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 702: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 703: curp->filenc = save_filenc;
1.67 schwarze 704: if (with_mmap)
705: munmap(blk.buf, blk.sz);
706: else
707: free(blk.buf);
708: }
1.117 schwarze 709: return curp->file_status;
1.58 schwarze 710: }
711:
1.122 schwarze 712: int
713: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 714: {
715: char *cp;
1.122 schwarze 716: int fd;
1.58 schwarze 717:
718: curp->file = file;
1.115 schwarze 719: cp = strrchr(file, '.');
720: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 721:
1.115 schwarze 722: /* First try to use the filename as it is. */
1.74 schwarze 723:
1.122 schwarze 724: if ((fd = open(file, O_RDONLY)) != -1)
725: return fd;
1.74 schwarze 726:
1.115 schwarze 727: /*
728: * If that doesn't work and the filename doesn't
729: * already end in .gz, try appending .gz.
730: */
1.74 schwarze 731:
1.115 schwarze 732: if ( ! curp->gzip) {
1.74 schwarze 733: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 734: fd = open(cp, O_RDONLY);
1.83 schwarze 735: free(cp);
1.122 schwarze 736: if (fd != -1) {
1.115 schwarze 737: curp->gzip = 1;
1.122 schwarze 738: return fd;
1.58 schwarze 739: }
740: }
741:
1.115 schwarze 742: /* Neither worked, give up. */
1.73 schwarze 743:
1.115 schwarze 744: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.122 schwarze 745: return -1;
1.1 schwarze 746: }
747:
748: struct mparse *
1.151 schwarze 749: mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
750: enum mandoc_os os_e, const char *os_s)
1.1 schwarze 751: {
752: struct mparse *curp;
753:
754: curp = mandoc_calloc(1, sizeof(struct mparse));
755:
1.22 schwarze 756: curp->options = options;
1.151 schwarze 757: curp->mmin = mmin;
1.1 schwarze 758: curp->mmsg = mmsg;
1.151 schwarze 759: curp->os_s = os_s;
1.1 schwarze 760:
1.119 schwarze 761: curp->roff = roff_alloc(curp, options);
1.151 schwarze 762: curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
1.112 schwarze 763: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 764: if (curp->options & MPARSE_MDOC) {
1.112 schwarze 765: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 766: if (curp->man->mdocmac == NULL)
767: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 768: } else if (curp->options & MPARSE_MAN) {
769: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 770: if (curp->man->manmac == NULL)
771: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 772: }
1.113 schwarze 773: curp->man->first->tok = TOKEN_NONE;
1.151 schwarze 774: curp->man->meta.os_e = os_e;
1.117 schwarze 775: return curp;
1.1 schwarze 776: }
777:
778: void
779: mparse_reset(struct mparse *curp)
780: {
781: roff_reset(curp->roff);
1.124 schwarze 782: roff_man_reset(curp->man);
1.133 schwarze 783:
784: free(curp->sodest);
785: curp->sodest = NULL;
786:
1.4 schwarze 787: if (curp->secondary)
788: curp->secondary->sz = 0;
1.1 schwarze 789:
790: curp->file_status = MANDOCLEVEL_OK;
1.132 schwarze 791: curp->gzip = 0;
1.1 schwarze 792: }
793:
794: void
795: mparse_free(struct mparse *curp)
796: {
797:
1.136 schwarze 798: roffhash_free(curp->man->mdocmac);
799: roffhash_free(curp->man->manmac);
1.112 schwarze 800: roff_man_free(curp->man);
1.133 schwarze 801: roff_free(curp->roff);
1.4 schwarze 802: if (curp->secondary)
803: free(curp->secondary->buf);
1.1 schwarze 804:
1.4 schwarze 805: free(curp->secondary);
1.23 schwarze 806: free(curp->sodest);
1.1 schwarze 807: free(curp);
808: }
809:
810: void
1.110 schwarze 811: mparse_result(struct mparse *curp, struct roff_man **man,
812: char **sodest)
1.1 schwarze 813: {
814:
1.23 schwarze 815: if (sodest && NULL != (*sodest = curp->sodest)) {
816: *man = NULL;
817: return;
818: }
1.1 schwarze 819: if (man)
820: *man = curp->man;
1.130 schwarze 821: }
822:
823: void
824: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
825: {
826: if (curp->file_status > *rc)
827: *rc = curp->file_status;
1.1 schwarze 828: }
829:
830: void
831: mandoc_vmsg(enum mandocerr t, struct mparse *m,
832: int ln, int pos, const char *fmt, ...)
833: {
834: char buf[256];
835: va_list ap;
836:
837: va_start(ap, fmt);
1.26 schwarze 838: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 839: va_end(ap);
840:
841: mandoc_msg(t, m, ln, pos, buf);
842: }
843:
844: void
1.25 schwarze 845: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 846: int ln, int col, const char *msg)
847: {
848: enum mandoclevel level;
849:
1.151 schwarze 850: if (er < m->mmin && er != MANDOCERR_FILE)
851: return;
852:
1.87 schwarze 853: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 854: while (er < mandoclimits[level])
855: level--;
856:
857: if (m->mmsg)
858: (*m->mmsg)(er, level, m->file, ln, col, msg);
859:
860: if (m->file_status < level)
861: m->file_status = level;
862: }
863:
864: const char *
865: mparse_strerror(enum mandocerr er)
866: {
867:
1.117 schwarze 868: return mandocerrs[er];
1.1 schwarze 869: }
870:
871: const char *
872: mparse_strlevel(enum mandoclevel lvl)
873: {
1.117 schwarze 874: return mandoclevels[lvl];
1.4 schwarze 875: }
876:
877: void
878: mparse_keep(struct mparse *p)
879: {
880:
881: assert(NULL == p->secondary);
882: p->secondary = mandoc_calloc(1, sizeof(struct buf));
883: }
884:
885: const char *
886: mparse_getkeep(const struct mparse *p)
887: {
888:
889: assert(p->secondary);
1.117 schwarze 890: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 schwarze 891: }