Annotation of src/usr.bin/mandoc/read.c, Revision 1.159
1.159 ! schwarze 1: /* $OpenBSD: read.c,v 1.158 2017/07/03 13:40:00 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",
1.156 schwarze 92: "referenced manual not found",
1.151 schwarze 93:
94: "generic style suggestion",
95:
1.148 schwarze 96: "legacy man(7) date format",
1.150 schwarze 97: "duplicate RCS id",
1.154 schwarze 98: "typo in section name",
1.138 schwarze 99: "useless macro",
1.139 schwarze 100: "consider using OS macro",
1.144 schwarze 101: "errnos out of order",
102: "duplicate errno",
1.159 ! schwarze 103: "trailing delimiter",
1.146 schwarze 104: "no blank before trailing delimiter",
1.147 schwarze 105: "function name without markup",
1.138 schwarze 106:
1.1 schwarze 107: "generic warning",
108:
109: /* related to the prologue */
1.57 schwarze 110: "missing manual title, using UNTITLED",
111: "missing manual title, using \"\"",
1.32 schwarze 112: "lower case character in document title",
1.57 schwarze 113: "missing manual section, using \"\"",
1.1 schwarze 114: "unknown manual section",
1.32 schwarze 115: "missing date, using today's date",
1.1 schwarze 116: "cannot parse date, using it verbatim",
1.158 schwarze 117: "date in the future, using it anyway",
1.57 schwarze 118: "missing Os macro, using \"\"",
119: "duplicate prologue macro",
120: "late prologue macro",
121: "skipping late title macro",
1.1 schwarze 122: "prologue macros out of order",
123:
124: /* related to document structure */
125: ".so is fragile, better use ln(1)",
1.28 schwarze 126: "no document body",
1.32 schwarze 127: "content before first section header",
128: "first section is not \"NAME\"",
1.129 schwarze 129: "NAME section without Nm before Nd",
1.103 schwarze 130: "NAME section without description",
131: "description not at the end of NAME",
132: "bad NAME section content",
1.129 schwarze 133: "missing comma before name",
1.96 schwarze 134: "missing description line, using \"\"",
1.134 schwarze 135: "description line outside NAME section",
1.1 schwarze 136: "sections out of conventional order",
1.32 schwarze 137: "duplicate section title",
138: "unexpected section",
1.157 schwarze 139: "cross reference to self",
1.63 schwarze 140: "unusual Xr order",
141: "unusual Xr punctuation",
1.62 schwarze 142: "AUTHORS section without An macro",
1.1 schwarze 143:
144: /* related to macros and nesting */
1.33 schwarze 145: "obsolete macro",
1.78 schwarze 146: "macro neither callable nor escaped",
1.1 schwarze 147: "skipping paragraph macro",
1.10 schwarze 148: "moving paragraph macro out of list",
1.1 schwarze 149: "skipping no-space macro",
150: "blocks badly nested",
151: "nested displays are not portable",
1.35 schwarze 152: "moving content out of list",
1.155 schwarze 153: "first macro on line",
1.56 schwarze 154: "fill mode already enabled, skipping",
155: "fill mode already disabled, skipping",
1.1 schwarze 156: "line scope broken",
1.141 schwarze 157: "skipping blank line in line scope",
1.1 schwarze 158:
159: /* related to missing macro arguments */
1.36 schwarze 160: "skipping empty request",
161: "conditional request controls empty scope",
1.1 schwarze 162: "skipping empty macro",
1.99 schwarze 163: "empty block",
1.40 schwarze 164: "empty argument, using 0n",
1.38 schwarze 165: "missing display type, using -ragged",
166: "list type is not the first argument",
1.125 schwarze 167: "missing -width in -tag list, using 6n",
1.56 schwarze 168: "missing utility name, using \"\"",
1.98 schwarze 169: "missing function name, using \"\"",
1.38 schwarze 170: "empty head in list item",
171: "empty list item",
1.155 schwarze 172: "missing argument, using next line",
1.39 schwarze 173: "missing font type, using \\fR",
174: "unknown font type, using \\fR",
1.79 schwarze 175: "nothing follows prefix",
1.97 schwarze 176: "empty reference block",
1.128 schwarze 177: "missing section argument",
1.38 schwarze 178: "missing -std argument, adding it",
1.100 schwarze 179: "missing option string, using \"\"",
180: "missing resource identifier, using \"\"",
1.66 schwarze 181: "missing eqn box, using \"\"",
1.1 schwarze 182:
183: /* related to bad macro arguments */
1.42 schwarze 184: "unterminated quoted argument",
1.1 schwarze 185: "duplicate argument",
1.54 schwarze 186: "skipping duplicate argument",
1.41 schwarze 187: "skipping duplicate display type",
188: "skipping duplicate list type",
1.54 schwarze 189: "skipping -width argument",
1.101 schwarze 190: "wrong number of cells",
1.1 schwarze 191: "unknown AT&T UNIX version",
1.64 schwarze 192: "comma in function argument",
1.65 schwarze 193: "parenthesis in function name",
1.145 schwarze 194: "unknown library name",
1.45 schwarze 195: "invalid content in Rs block",
1.41 schwarze 196: "invalid Boolean argument",
197: "unknown font, skipping request",
1.101 schwarze 198: "odd number of characters in request",
1.1 schwarze 199:
200: /* related to plain text */
1.42 schwarze 201: "blank line in fill mode, using .sp",
202: "tab in filled text",
203: "whitespace at end of input line",
1.131 schwarze 204: "new sentence, new line",
1.1 schwarze 205: "bad comment style",
1.42 schwarze 206: "invalid escape sequence",
207: "undefined string, using \"\"",
1.3 schwarze 208:
1.93 schwarze 209: /* related to tables */
210: "tbl line starts with span",
211: "tbl column starts with span",
212: "skipping vertical bar in tbl layout",
213:
1.1 schwarze 214: "generic error",
1.3 schwarze 215:
1.1 schwarze 216: /* related to tables */
1.91 schwarze 217: "non-alphabetic character in tbl options",
218: "skipping unknown tbl option",
219: "missing tbl option argument",
220: "wrong tbl option argument size",
1.93 schwarze 221: "empty tbl layout",
222: "invalid character in tbl layout",
223: "unmatched parenthesis in tbl layout",
1.94 schwarze 224: "tbl without any data cells",
225: "ignoring data in spanned tbl cell",
226: "ignoring extra tbl data cells",
227: "data block open at end of tbl",
1.1 schwarze 228:
1.46 schwarze 229: /* related to document structure and macros */
1.83 schwarze 230: NULL,
1.1 schwarze 231: "input stack limit exceeded, infinite loop?",
232: "skipping bad character",
1.46 schwarze 233: "skipping unknown macro",
1.87 schwarze 234: "skipping insecure request",
1.51 schwarze 235: "skipping item outside list",
1.46 schwarze 236: "skipping column outside column list",
237: "skipping end of block that is not open",
1.90 schwarze 238: "fewer RS blocks open, skipping",
1.46 schwarze 239: "inserting missing end of block",
240: "appending missing end of block",
241:
242: /* related to request and macro arguments */
1.1 schwarze 243: "escaped character not allowed in a name",
1.75 schwarze 244: "NOT IMPLEMENTED: Bd -file",
1.120 schwarze 245: "skipping display without arguments",
1.49 schwarze 246: "missing list type, using -item",
1.143 schwarze 247: "argument is not numeric, using 1",
1.48 schwarze 248: "missing manual name, using \"\"",
1.49 schwarze 249: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 250: "unknown standard specifier",
1.49 schwarze 251: "skipping request without numeric argument",
1.84 schwarze 252: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 253: ".so request failed",
1.39 schwarze 254: "skipping all arguments",
255: "skipping excess arguments",
1.68 schwarze 256: "divide by zero",
1.87 schwarze 257:
258: "unsupported feature",
259: "input too large",
1.89 schwarze 260: "unsupported control character",
1.87 schwarze 261: "unsupported roff request",
1.94 schwarze 262: "eqn delim option in tbl",
1.93 schwarze 263: "unsupported tbl layout modifier",
1.87 schwarze 264: "ignoring macro in table",
1.1 schwarze 265: };
266:
267: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
268: "SUCCESS",
1.137 schwarze 269: "STYLE",
1.1 schwarze 270: "WARNING",
271: "ERROR",
1.87 schwarze 272: "UNSUPP",
1.1 schwarze 273: "BADARG",
274: "SYSERR"
275: };
276:
1.25 schwarze 277:
1.1 schwarze 278: static void
279: resize_buf(struct buf *buf, size_t initial)
280: {
281:
282: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
283: buf->buf = mandoc_realloc(buf->buf, buf->sz);
284: }
285:
286: static void
1.60 schwarze 287: choose_parser(struct mparse *curp)
1.1 schwarze 288: {
1.59 schwarze 289: char *cp, *ep;
290: int format;
1.1 schwarze 291:
1.59 schwarze 292: /*
293: * If neither command line arguments -mdoc or -man select
294: * a parser nor the roff parser found a .Dd or .TH macro
295: * yet, look ahead in the main input buffer.
296: */
297:
298: if ((format = roff_getformat(curp->roff)) == 0) {
299: cp = curp->primary->buf;
300: ep = cp + curp->primary->sz;
301: while (cp < ep) {
1.61 schwarze 302: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 303: cp++;
304: if (cp[0] == 'D' && cp[1] == 'd') {
305: format = MPARSE_MDOC;
306: break;
307: }
308: if (cp[0] == 'T' && cp[1] == 'H') {
309: format = MPARSE_MAN;
310: break;
311: }
312: }
313: cp = memchr(cp, '\n', ep - cp);
314: if (cp == NULL)
315: break;
316: cp++;
317: }
1.1 schwarze 318: }
319:
1.59 schwarze 320: if (format == MPARSE_MDOC) {
1.112 schwarze 321: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 322: if (curp->man->mdocmac == NULL)
323: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 324: } else {
325: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 326: if (curp->man->manmac == NULL)
327: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 328: }
1.136 schwarze 329: curp->man->first->tok = TOKEN_NONE;
1.1 schwarze 330: }
331:
332: /*
1.71 schwarze 333: * Main parse routine for a buffer.
334: * It assumes encoding and line numbering are already set up.
335: * It can recurse directly (for invocations of user-defined
336: * macros, inline equations, and input line traps)
337: * and indirectly (for .so file inclusion).
1.1 schwarze 338: */
1.135 schwarze 339: static int
1.71 schwarze 340: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 341: {
342: const struct tbl_span *span;
343: struct buf ln;
1.88 schwarze 344: const char *save_file;
1.85 schwarze 345: char *cp;
1.71 schwarze 346: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 347: enum rofferr rr;
1.76 schwarze 348: int of;
1.1 schwarze 349: int lnn; /* line number in the real file */
1.85 schwarze 350: int fd;
1.1 schwarze 351: unsigned char c;
352:
1.71 schwarze 353: memset(&ln, 0, sizeof(ln));
1.1 schwarze 354:
1.25 schwarze 355: lnn = curp->line;
356: pos = 0;
1.1 schwarze 357:
1.71 schwarze 358: while (i < blk.sz) {
1.1 schwarze 359: if (0 == pos && '\0' == blk.buf[i])
360: break;
361:
362: if (start) {
363: curp->line = lnn;
364: curp->reparse_count = 0;
1.70 schwarze 365:
366: if (lnn < 3 &&
367: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 368: curp->filenc & MPARSE_LATIN1)
369: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 370: }
371:
1.71 schwarze 372: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 373:
374: /*
375: * When finding an unescaped newline character,
376: * leave the character loop to process the line.
377: * Skip a preceding carriage return, if any.
378: */
379:
1.71 schwarze 380: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 381: '\n' == blk.buf[i + 1])
382: ++i;
383: if ('\n' == blk.buf[i]) {
384: ++i;
385: ++lnn;
386: break;
387: }
388:
1.13 schwarze 389: /*
1.70 schwarze 390: * Make sure we have space for the worst
391: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 392: */
393:
1.71 schwarze 394: if (pos + 11 > ln.sz)
1.13 schwarze 395: resize_buf(&ln, 256);
396:
1.25 schwarze 397: /*
1.70 schwarze 398: * Encode 8-bit input.
1.1 schwarze 399: */
400:
1.70 schwarze 401: c = blk.buf[i];
402: if (c & 0x80) {
1.71 schwarze 403: if ( ! (curp->filenc && preconv_encode(
404: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 405: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
406: curp->line, pos, "0x%x", c);
1.70 schwarze 407: ln.buf[pos++] = '?';
408: i++;
409: }
410: continue;
411: }
412:
413: /*
414: * Exclude control characters.
415: */
1.1 schwarze 416:
1.70 schwarze 417: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 418: mandoc_vmsg(c == 0x00 || c == 0x04 ||
419: c > 0x0a ? MANDOCERR_CHAR_BAD :
420: MANDOCERR_CHAR_UNSUPP,
421: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 422: i++;
1.102 schwarze 423: if (c != '\r')
424: ln.buf[pos++] = '?';
1.1 schwarze 425: continue;
426: }
427:
428: ln.buf[pos++] = blk.buf[i++];
429: }
430:
1.142 schwarze 431: if (pos + 1 >= ln.sz)
1.1 schwarze 432: resize_buf(&ln, 256);
433:
1.142 schwarze 434: if (i == blk.sz || blk.buf[i] == '\0')
435: ln.buf[pos++] = '\n';
1.1 schwarze 436: ln.buf[pos] = '\0';
437:
438: /*
439: * A significant amount of complexity is contained by
440: * the roff preprocessor. It's line-oriented but can be
441: * expressed on one line, so we need at times to
442: * readjust our starting point and re-run it. The roff
443: * preprocessor can also readjust the buffers with new
444: * data, so we pass them in wholesale.
445: */
446:
447: of = 0;
448:
1.4 schwarze 449: /*
450: * Maintain a lookaside buffer of all parsed lines. We
451: * only do this if mparse_keep() has been invoked (the
452: * buffer may be accessed with mparse_getkeep()).
453: */
454:
455: if (curp->secondary) {
1.25 schwarze 456: curp->secondary->buf = mandoc_realloc(
457: curp->secondary->buf,
458: curp->secondary->sz + pos + 2);
459: memcpy(curp->secondary->buf +
460: curp->secondary->sz,
461: ln.buf, pos);
1.4 schwarze 462: curp->secondary->sz += pos;
463: curp->secondary->buf
464: [curp->secondary->sz] = '\n';
465: curp->secondary->sz++;
466: curp->secondary->buf
467: [curp->secondary->sz] = '\0';
468: }
1.1 schwarze 469: rerun:
1.72 schwarze 470: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 471:
472: switch (rr) {
1.25 schwarze 473: case ROFF_REPARSE:
1.135 schwarze 474: if (++curp->reparse_count > REPARSE_LIMIT)
1.1 schwarze 475: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 476: curp->line, pos, NULL);
1.135 schwarze 477: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
478: start == 1) {
479: pos = 0;
480: continue;
481: }
482: free(ln.buf);
483: return 0;
1.25 schwarze 484: case ROFF_APPEND:
1.71 schwarze 485: pos = strlen(ln.buf);
1.1 schwarze 486: continue;
1.25 schwarze 487: case ROFF_RERUN:
1.1 schwarze 488: goto rerun;
1.25 schwarze 489: case ROFF_IGN:
1.1 schwarze 490: pos = 0;
491: continue;
1.25 schwarze 492: case ROFF_SO:
1.71 schwarze 493: if ( ! (curp->options & MPARSE_SO) &&
494: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 495: curp->sodest = mandoc_strdup(ln.buf + of);
496: free(ln.buf);
1.135 schwarze 497: return 1;
1.23 schwarze 498: }
1.4 schwarze 499: /*
500: * We remove `so' clauses from our lookaside
501: * buffer because we're going to descend into
502: * the file recursively.
503: */
1.25 schwarze 504: if (curp->secondary)
1.4 schwarze 505: curp->secondary->sz -= pos + 1;
1.88 schwarze 506: save_file = curp->file;
1.122 schwarze 507: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 508: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 509: close(fd);
1.88 schwarze 510: curp->file = save_file;
511: } else {
512: curp->file = save_file;
1.30 schwarze 513: mandoc_vmsg(MANDOCERR_SO_FAIL,
514: curp, curp->line, pos,
515: ".so %s", ln.buf + of);
1.85 schwarze 516: ln.sz = mandoc_asprintf(&cp,
517: ".sp\nSee the file %s.\n.sp",
518: ln.buf + of);
519: free(ln.buf);
520: ln.buf = cp;
521: of = 0;
522: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 523: }
1.1 schwarze 524: pos = 0;
525: continue;
526: default:
527: break;
528: }
529:
1.124 schwarze 530: if (curp->man->macroset == MACROSET_NONE)
1.60 schwarze 531: choose_parser(curp);
1.1 schwarze 532:
1.25 schwarze 533: /*
1.60 schwarze 534: * Lastly, push down into the parsers themselves.
1.1 schwarze 535: * If libroff returns ROFF_TBL, then add it to the
536: * currently open parse. Since we only get here if
537: * there does exist data (see tbl_data.c), we're
538: * guaranteed that something's been allocated.
539: * Do the same for ROFF_EQN.
540: */
541:
1.114 schwarze 542: if (rr == ROFF_TBL)
1.76 schwarze 543: while ((span = roff_span(curp->roff)) != NULL)
1.114 schwarze 544: roff_addtbl(curp->man, span);
545: else if (rr == ROFF_EQN)
546: roff_addeqn(curp->man, roff_eqn(curp->roff));
547: else if ((curp->man->macroset == MACROSET_MDOC ?
1.110 schwarze 548: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 549: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
550: break;
1.1 schwarze 551:
552: /* Temporary buffers typically are not full. */
553:
554: if (0 == start && '\0' == blk.buf[i])
555: break;
556:
557: /* Start the next input line. */
558:
559: pos = 0;
560: }
561:
562: free(ln.buf);
1.135 schwarze 563: return 1;
1.1 schwarze 564: }
565:
566: static int
1.18 schwarze 567: read_whole_file(struct mparse *curp, const char *file, int fd,
568: struct buf *fb, int *with_mmap)
1.1 schwarze 569: {
570: struct stat st;
1.115 schwarze 571: gzFile gz;
1.1 schwarze 572: size_t off;
573: ssize_t ssz;
574:
1.118 schwarze 575: if (fstat(fd, &st) == -1)
576: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 577:
578: /*
579: * If we're a regular file, try just reading in the whole entry
580: * via mmap(). This is faster than reading it into blocks, and
581: * since each file is only a few bytes to begin with, I'm not
582: * concerned that this is going to tank any machines.
583: */
584:
1.115 schwarze 585: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 586: if (st.st_size > 0x7fffffff) {
1.86 schwarze 587: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.117 schwarze 588: return 0;
1.1 schwarze 589: }
590: *with_mmap = 1;
591: fb->sz = (size_t)st.st_size;
1.15 schwarze 592: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 593: if (fb->buf != MAP_FAILED)
1.117 schwarze 594: return 1;
1.1 schwarze 595: }
596:
1.115 schwarze 597: if (curp->gzip) {
1.118 schwarze 598: if ((gz = gzdopen(fd, "rb")) == NULL)
599: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.115 schwarze 600: } else
601: gz = NULL;
602:
1.1 schwarze 603: /*
604: * If this isn't a regular file (like, say, stdin), then we must
605: * go the old way and just read things in bit by bit.
606: */
607:
608: *with_mmap = 0;
609: off = 0;
610: fb->sz = 0;
611: fb->buf = NULL;
612: for (;;) {
613: if (off == fb->sz) {
614: if (fb->sz == (1U << 31)) {
1.86 schwarze 615: mandoc_msg(MANDOCERR_TOOLARGE, curp,
616: 0, 0, NULL);
1.1 schwarze 617: break;
618: }
619: resize_buf(fb, 65536);
620: }
1.115 schwarze 621: ssz = curp->gzip ?
622: gzread(gz, fb->buf + (int)off, fb->sz - off) :
623: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 624: if (ssz == 0) {
625: fb->sz = off;
1.117 schwarze 626: return 1;
1.1 schwarze 627: }
1.118 schwarze 628: if (ssz == -1)
629: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 630: off += (size_t)ssz;
631: }
632:
633: free(fb->buf);
634: fb->buf = NULL;
1.117 schwarze 635: return 0;
1.1 schwarze 636: }
637:
638: static void
639: mparse_end(struct mparse *curp)
640: {
1.110 schwarze 641: if (curp->man->macroset == MACROSET_NONE)
642: curp->man->macroset = MACROSET_MAN;
643: if (curp->man->macroset == MACROSET_MDOC)
644: mdoc_endparse(curp->man);
645: else
1.86 schwarze 646: man_endparse(curp->man);
1.1 schwarze 647: roff_endparse(curp->roff);
648: }
649:
1.15 schwarze 650: static void
651: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 652: {
1.61 schwarze 653: struct buf *svprimary;
1.1 schwarze 654: const char *svfile;
1.71 schwarze 655: size_t offset;
1.14 schwarze 656: static int recursion_depth;
657:
658: if (64 < recursion_depth) {
659: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 660: return;
1.14 schwarze 661: }
1.1 schwarze 662:
1.15 schwarze 663: /* Line number is per-file. */
664: svfile = curp->file;
665: curp->file = file;
1.61 schwarze 666: svprimary = curp->primary;
1.59 schwarze 667: curp->primary = &blk;
1.15 schwarze 668: curp->line = 1;
669: recursion_depth++;
670:
1.70 schwarze 671: /* Skip an UTF-8 byte order mark. */
672: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
673: (unsigned char)blk.buf[0] == 0xef &&
674: (unsigned char)blk.buf[1] == 0xbb &&
675: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 676: offset = 3;
1.70 schwarze 677: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 678: } else
679: offset = 0;
1.70 schwarze 680:
1.71 schwarze 681: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 682:
1.86 schwarze 683: if (--recursion_depth == 0)
1.15 schwarze 684: mparse_end(curp);
685:
1.61 schwarze 686: curp->primary = svprimary;
1.15 schwarze 687: curp->file = svfile;
688: }
689:
1.74 schwarze 690: /*
691: * Read the whole file into memory and call the parsers.
692: * Called recursively when an .so request is encountered.
693: */
1.15 schwarze 694: enum mandoclevel
695: mparse_readfd(struct mparse *curp, int fd, const char *file)
696: {
697: struct buf blk;
698: int with_mmap;
1.70 schwarze 699: int save_filenc;
1.1 schwarze 700:
1.67 schwarze 701: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 702: save_filenc = curp->filenc;
703: curp->filenc = curp->options &
704: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 705: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 706: curp->filenc = save_filenc;
1.67 schwarze 707: if (with_mmap)
708: munmap(blk.buf, blk.sz);
709: else
710: free(blk.buf);
711: }
1.117 schwarze 712: return curp->file_status;
1.58 schwarze 713: }
714:
1.122 schwarze 715: int
716: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 717: {
718: char *cp;
1.122 schwarze 719: int fd;
1.58 schwarze 720:
721: curp->file = file;
1.115 schwarze 722: cp = strrchr(file, '.');
723: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 724:
1.115 schwarze 725: /* First try to use the filename as it is. */
1.74 schwarze 726:
1.122 schwarze 727: if ((fd = open(file, O_RDONLY)) != -1)
728: return fd;
1.74 schwarze 729:
1.115 schwarze 730: /*
731: * If that doesn't work and the filename doesn't
732: * already end in .gz, try appending .gz.
733: */
1.74 schwarze 734:
1.115 schwarze 735: if ( ! curp->gzip) {
1.74 schwarze 736: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 737: fd = open(cp, O_RDONLY);
1.83 schwarze 738: free(cp);
1.122 schwarze 739: if (fd != -1) {
1.115 schwarze 740: curp->gzip = 1;
1.122 schwarze 741: return fd;
1.58 schwarze 742: }
743: }
744:
1.115 schwarze 745: /* Neither worked, give up. */
1.73 schwarze 746:
1.115 schwarze 747: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.122 schwarze 748: return -1;
1.1 schwarze 749: }
750:
751: struct mparse *
1.151 schwarze 752: mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
753: enum mandoc_os os_e, const char *os_s)
1.1 schwarze 754: {
755: struct mparse *curp;
756:
757: curp = mandoc_calloc(1, sizeof(struct mparse));
758:
1.22 schwarze 759: curp->options = options;
1.151 schwarze 760: curp->mmin = mmin;
1.1 schwarze 761: curp->mmsg = mmsg;
1.151 schwarze 762: curp->os_s = os_s;
1.1 schwarze 763:
1.119 schwarze 764: curp->roff = roff_alloc(curp, options);
1.151 schwarze 765: curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
1.112 schwarze 766: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 767: if (curp->options & MPARSE_MDOC) {
1.112 schwarze 768: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 769: if (curp->man->mdocmac == NULL)
770: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 771: } else if (curp->options & MPARSE_MAN) {
772: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 773: if (curp->man->manmac == NULL)
774: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 775: }
1.113 schwarze 776: curp->man->first->tok = TOKEN_NONE;
1.151 schwarze 777: curp->man->meta.os_e = os_e;
1.117 schwarze 778: return curp;
1.1 schwarze 779: }
780:
781: void
782: mparse_reset(struct mparse *curp)
783: {
784: roff_reset(curp->roff);
1.124 schwarze 785: roff_man_reset(curp->man);
1.133 schwarze 786:
787: free(curp->sodest);
788: curp->sodest = NULL;
789:
1.4 schwarze 790: if (curp->secondary)
791: curp->secondary->sz = 0;
1.1 schwarze 792:
793: curp->file_status = MANDOCLEVEL_OK;
1.132 schwarze 794: curp->gzip = 0;
1.1 schwarze 795: }
796:
797: void
798: mparse_free(struct mparse *curp)
799: {
800:
1.136 schwarze 801: roffhash_free(curp->man->mdocmac);
802: roffhash_free(curp->man->manmac);
1.112 schwarze 803: roff_man_free(curp->man);
1.133 schwarze 804: roff_free(curp->roff);
1.4 schwarze 805: if (curp->secondary)
806: free(curp->secondary->buf);
1.1 schwarze 807:
1.4 schwarze 808: free(curp->secondary);
1.23 schwarze 809: free(curp->sodest);
1.1 schwarze 810: free(curp);
811: }
812:
813: void
1.110 schwarze 814: mparse_result(struct mparse *curp, struct roff_man **man,
815: char **sodest)
1.1 schwarze 816: {
817:
1.23 schwarze 818: if (sodest && NULL != (*sodest = curp->sodest)) {
819: *man = NULL;
820: return;
821: }
1.1 schwarze 822: if (man)
823: *man = curp->man;
1.130 schwarze 824: }
825:
826: void
827: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
828: {
829: if (curp->file_status > *rc)
830: *rc = curp->file_status;
1.1 schwarze 831: }
832:
833: void
834: mandoc_vmsg(enum mandocerr t, struct mparse *m,
835: int ln, int pos, const char *fmt, ...)
836: {
837: char buf[256];
838: va_list ap;
839:
840: va_start(ap, fmt);
1.26 schwarze 841: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 842: va_end(ap);
843:
844: mandoc_msg(t, m, ln, pos, buf);
845: }
846:
847: void
1.25 schwarze 848: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 849: int ln, int col, const char *msg)
850: {
851: enum mandoclevel level;
852:
1.151 schwarze 853: if (er < m->mmin && er != MANDOCERR_FILE)
854: return;
855:
1.87 schwarze 856: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 857: while (er < mandoclimits[level])
858: level--;
859:
860: if (m->mmsg)
861: (*m->mmsg)(er, level, m->file, ln, col, msg);
862:
863: if (m->file_status < level)
864: m->file_status = level;
865: }
866:
867: const char *
868: mparse_strerror(enum mandocerr er)
869: {
870:
1.117 schwarze 871: return mandocerrs[er];
1.1 schwarze 872: }
873:
874: const char *
875: mparse_strlevel(enum mandoclevel lvl)
876: {
1.117 schwarze 877: return mandoclevels[lvl];
1.4 schwarze 878: }
879:
880: void
881: mparse_keep(struct mparse *p)
882: {
883:
884: assert(NULL == p->secondary);
885: p->secondary = mandoc_calloc(1, sizeof(struct buf));
886: }
887:
888: const char *
889: mparse_getkeep(const struct mparse *p)
890: {
891:
892: assert(p->secondary);
1.117 schwarze 893: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 schwarze 894: }