Annotation of src/usr.bin/mandoc/read.c, Revision 1.157
1.157 ! schwarze 1: /* $OpenBSD: read.c,v 1.156 2017/07/01 09:47:23 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",
1.140 schwarze 103: "description line ends with a full stop",
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",
1.57 schwarze 117: "missing Os macro, using \"\"",
118: "duplicate prologue macro",
119: "late prologue macro",
120: "skipping late title macro",
1.1 schwarze 121: "prologue macros out of order",
122:
123: /* related to document structure */
124: ".so is fragile, better use ln(1)",
1.28 schwarze 125: "no document body",
1.32 schwarze 126: "content before first section header",
127: "first section is not \"NAME\"",
1.129 schwarze 128: "NAME section without Nm before Nd",
1.103 schwarze 129: "NAME section without description",
130: "description not at the end of NAME",
131: "bad NAME section content",
1.129 schwarze 132: "missing comma before name",
1.96 schwarze 133: "missing description line, using \"\"",
1.134 schwarze 134: "description line outside NAME section",
1.1 schwarze 135: "sections out of conventional order",
1.32 schwarze 136: "duplicate section title",
137: "unexpected section",
1.157 ! schwarze 138: "cross reference to self",
1.63 schwarze 139: "unusual Xr order",
140: "unusual Xr punctuation",
1.62 schwarze 141: "AUTHORS section without An macro",
1.1 schwarze 142:
143: /* related to macros and nesting */
1.33 schwarze 144: "obsolete macro",
1.78 schwarze 145: "macro neither callable nor escaped",
1.1 schwarze 146: "skipping paragraph macro",
1.10 schwarze 147: "moving paragraph macro out of list",
1.1 schwarze 148: "skipping no-space macro",
149: "blocks badly nested",
150: "nested displays are not portable",
1.35 schwarze 151: "moving content out of list",
1.155 schwarze 152: "first macro on line",
1.56 schwarze 153: "fill mode already enabled, skipping",
154: "fill mode already disabled, skipping",
1.1 schwarze 155: "line scope broken",
1.141 schwarze 156: "skipping blank line in line scope",
1.1 schwarze 157:
158: /* related to missing macro arguments */
1.36 schwarze 159: "skipping empty request",
160: "conditional request controls empty scope",
1.1 schwarze 161: "skipping empty macro",
1.99 schwarze 162: "empty block",
1.40 schwarze 163: "empty argument, using 0n",
1.38 schwarze 164: "missing display type, using -ragged",
165: "list type is not the first argument",
1.125 schwarze 166: "missing -width in -tag list, using 6n",
1.56 schwarze 167: "missing utility name, using \"\"",
1.98 schwarze 168: "missing function name, using \"\"",
1.38 schwarze 169: "empty head in list item",
170: "empty list item",
1.155 schwarze 171: "missing argument, using next line",
1.39 schwarze 172: "missing font type, using \\fR",
173: "unknown font type, using \\fR",
1.79 schwarze 174: "nothing follows prefix",
1.97 schwarze 175: "empty reference block",
1.128 schwarze 176: "missing section argument",
1.38 schwarze 177: "missing -std argument, adding it",
1.100 schwarze 178: "missing option string, using \"\"",
179: "missing resource identifier, using \"\"",
1.66 schwarze 180: "missing eqn box, using \"\"",
1.1 schwarze 181:
182: /* related to bad macro arguments */
1.42 schwarze 183: "unterminated quoted argument",
1.1 schwarze 184: "duplicate argument",
1.54 schwarze 185: "skipping duplicate argument",
1.41 schwarze 186: "skipping duplicate display type",
187: "skipping duplicate list type",
1.54 schwarze 188: "skipping -width argument",
1.101 schwarze 189: "wrong number of cells",
1.1 schwarze 190: "unknown AT&T UNIX version",
1.64 schwarze 191: "comma in function argument",
1.65 schwarze 192: "parenthesis in function name",
1.145 schwarze 193: "unknown library name",
1.45 schwarze 194: "invalid content in Rs block",
1.41 schwarze 195: "invalid Boolean argument",
196: "unknown font, skipping request",
1.101 schwarze 197: "odd number of characters in request",
1.1 schwarze 198:
199: /* related to plain text */
1.42 schwarze 200: "blank line in fill mode, using .sp",
201: "tab in filled text",
202: "whitespace at end of input line",
1.131 schwarze 203: "new sentence, new line",
1.1 schwarze 204: "bad comment style",
1.42 schwarze 205: "invalid escape sequence",
206: "undefined string, using \"\"",
1.3 schwarze 207:
1.93 schwarze 208: /* related to tables */
209: "tbl line starts with span",
210: "tbl column starts with span",
211: "skipping vertical bar in tbl layout",
212:
1.1 schwarze 213: "generic error",
1.3 schwarze 214:
1.1 schwarze 215: /* related to tables */
1.91 schwarze 216: "non-alphabetic character in tbl options",
217: "skipping unknown tbl option",
218: "missing tbl option argument",
219: "wrong tbl option argument size",
1.93 schwarze 220: "empty tbl layout",
221: "invalid character in tbl layout",
222: "unmatched parenthesis in tbl layout",
1.94 schwarze 223: "tbl without any data cells",
224: "ignoring data in spanned tbl cell",
225: "ignoring extra tbl data cells",
226: "data block open at end of tbl",
1.1 schwarze 227:
1.46 schwarze 228: /* related to document structure and macros */
1.83 schwarze 229: NULL,
1.1 schwarze 230: "input stack limit exceeded, infinite loop?",
231: "skipping bad character",
1.46 schwarze 232: "skipping unknown macro",
1.87 schwarze 233: "skipping insecure request",
1.51 schwarze 234: "skipping item outside list",
1.46 schwarze 235: "skipping column outside column list",
236: "skipping end of block that is not open",
1.90 schwarze 237: "fewer RS blocks open, skipping",
1.46 schwarze 238: "inserting missing end of block",
239: "appending missing end of block",
240:
241: /* related to request and macro arguments */
1.1 schwarze 242: "escaped character not allowed in a name",
1.75 schwarze 243: "NOT IMPLEMENTED: Bd -file",
1.120 schwarze 244: "skipping display without arguments",
1.49 schwarze 245: "missing list type, using -item",
1.143 schwarze 246: "argument is not numeric, using 1",
1.48 schwarze 247: "missing manual name, using \"\"",
1.49 schwarze 248: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 249: "unknown standard specifier",
1.49 schwarze 250: "skipping request without numeric argument",
1.84 schwarze 251: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 252: ".so request failed",
1.39 schwarze 253: "skipping all arguments",
254: "skipping excess arguments",
1.68 schwarze 255: "divide by zero",
1.87 schwarze 256:
257: "unsupported feature",
258: "input too large",
1.89 schwarze 259: "unsupported control character",
1.87 schwarze 260: "unsupported roff request",
1.94 schwarze 261: "eqn delim option in tbl",
1.93 schwarze 262: "unsupported tbl layout modifier",
1.87 schwarze 263: "ignoring macro in table",
1.1 schwarze 264: };
265:
266: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
267: "SUCCESS",
1.137 schwarze 268: "STYLE",
1.1 schwarze 269: "WARNING",
270: "ERROR",
1.87 schwarze 271: "UNSUPP",
1.1 schwarze 272: "BADARG",
273: "SYSERR"
274: };
275:
1.25 schwarze 276:
1.1 schwarze 277: static void
278: resize_buf(struct buf *buf, size_t initial)
279: {
280:
281: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
282: buf->buf = mandoc_realloc(buf->buf, buf->sz);
283: }
284:
285: static void
1.60 schwarze 286: choose_parser(struct mparse *curp)
1.1 schwarze 287: {
1.59 schwarze 288: char *cp, *ep;
289: int format;
1.1 schwarze 290:
1.59 schwarze 291: /*
292: * If neither command line arguments -mdoc or -man select
293: * a parser nor the roff parser found a .Dd or .TH macro
294: * yet, look ahead in the main input buffer.
295: */
296:
297: if ((format = roff_getformat(curp->roff)) == 0) {
298: cp = curp->primary->buf;
299: ep = cp + curp->primary->sz;
300: while (cp < ep) {
1.61 schwarze 301: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 302: cp++;
303: if (cp[0] == 'D' && cp[1] == 'd') {
304: format = MPARSE_MDOC;
305: break;
306: }
307: if (cp[0] == 'T' && cp[1] == 'H') {
308: format = MPARSE_MAN;
309: break;
310: }
311: }
312: cp = memchr(cp, '\n', ep - cp);
313: if (cp == NULL)
314: break;
315: cp++;
316: }
1.1 schwarze 317: }
318:
1.59 schwarze 319: if (format == MPARSE_MDOC) {
1.112 schwarze 320: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 321: if (curp->man->mdocmac == NULL)
322: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 323: } else {
324: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 325: if (curp->man->manmac == NULL)
326: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 327: }
1.136 schwarze 328: curp->man->first->tok = TOKEN_NONE;
1.1 schwarze 329: }
330:
331: /*
1.71 schwarze 332: * Main parse routine for a buffer.
333: * It assumes encoding and line numbering are already set up.
334: * It can recurse directly (for invocations of user-defined
335: * macros, inline equations, and input line traps)
336: * and indirectly (for .so file inclusion).
1.1 schwarze 337: */
1.135 schwarze 338: static int
1.71 schwarze 339: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 340: {
341: const struct tbl_span *span;
342: struct buf ln;
1.88 schwarze 343: const char *save_file;
1.85 schwarze 344: char *cp;
1.71 schwarze 345: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 346: enum rofferr rr;
1.76 schwarze 347: int of;
1.1 schwarze 348: int lnn; /* line number in the real file */
1.85 schwarze 349: int fd;
1.1 schwarze 350: unsigned char c;
351:
1.71 schwarze 352: memset(&ln, 0, sizeof(ln));
1.1 schwarze 353:
1.25 schwarze 354: lnn = curp->line;
355: pos = 0;
1.1 schwarze 356:
1.71 schwarze 357: while (i < blk.sz) {
1.1 schwarze 358: if (0 == pos && '\0' == blk.buf[i])
359: break;
360:
361: if (start) {
362: curp->line = lnn;
363: curp->reparse_count = 0;
1.70 schwarze 364:
365: if (lnn < 3 &&
366: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 367: curp->filenc & MPARSE_LATIN1)
368: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 369: }
370:
1.71 schwarze 371: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 372:
373: /*
374: * When finding an unescaped newline character,
375: * leave the character loop to process the line.
376: * Skip a preceding carriage return, if any.
377: */
378:
1.71 schwarze 379: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 380: '\n' == blk.buf[i + 1])
381: ++i;
382: if ('\n' == blk.buf[i]) {
383: ++i;
384: ++lnn;
385: break;
386: }
387:
1.13 schwarze 388: /*
1.70 schwarze 389: * Make sure we have space for the worst
390: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 391: */
392:
1.71 schwarze 393: if (pos + 11 > ln.sz)
1.13 schwarze 394: resize_buf(&ln, 256);
395:
1.25 schwarze 396: /*
1.70 schwarze 397: * Encode 8-bit input.
1.1 schwarze 398: */
399:
1.70 schwarze 400: c = blk.buf[i];
401: if (c & 0x80) {
1.71 schwarze 402: if ( ! (curp->filenc && preconv_encode(
403: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 404: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
405: curp->line, pos, "0x%x", c);
1.70 schwarze 406: ln.buf[pos++] = '?';
407: i++;
408: }
409: continue;
410: }
411:
412: /*
413: * Exclude control characters.
414: */
1.1 schwarze 415:
1.70 schwarze 416: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 417: mandoc_vmsg(c == 0x00 || c == 0x04 ||
418: c > 0x0a ? MANDOCERR_CHAR_BAD :
419: MANDOCERR_CHAR_UNSUPP,
420: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 421: i++;
1.102 schwarze 422: if (c != '\r')
423: ln.buf[pos++] = '?';
1.1 schwarze 424: continue;
425: }
426:
427: ln.buf[pos++] = blk.buf[i++];
428: }
429:
1.142 schwarze 430: if (pos + 1 >= ln.sz)
1.1 schwarze 431: resize_buf(&ln, 256);
432:
1.142 schwarze 433: if (i == blk.sz || blk.buf[i] == '\0')
434: ln.buf[pos++] = '\n';
1.1 schwarze 435: ln.buf[pos] = '\0';
436:
437: /*
438: * A significant amount of complexity is contained by
439: * the roff preprocessor. It's line-oriented but can be
440: * expressed on one line, so we need at times to
441: * readjust our starting point and re-run it. The roff
442: * preprocessor can also readjust the buffers with new
443: * data, so we pass them in wholesale.
444: */
445:
446: of = 0;
447:
1.4 schwarze 448: /*
449: * Maintain a lookaside buffer of all parsed lines. We
450: * only do this if mparse_keep() has been invoked (the
451: * buffer may be accessed with mparse_getkeep()).
452: */
453:
454: if (curp->secondary) {
1.25 schwarze 455: curp->secondary->buf = mandoc_realloc(
456: curp->secondary->buf,
457: curp->secondary->sz + pos + 2);
458: memcpy(curp->secondary->buf +
459: curp->secondary->sz,
460: ln.buf, pos);
1.4 schwarze 461: curp->secondary->sz += pos;
462: curp->secondary->buf
463: [curp->secondary->sz] = '\n';
464: curp->secondary->sz++;
465: curp->secondary->buf
466: [curp->secondary->sz] = '\0';
467: }
1.1 schwarze 468: rerun:
1.72 schwarze 469: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 470:
471: switch (rr) {
1.25 schwarze 472: case ROFF_REPARSE:
1.135 schwarze 473: if (++curp->reparse_count > REPARSE_LIMIT)
1.1 schwarze 474: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 475: curp->line, pos, NULL);
1.135 schwarze 476: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
477: start == 1) {
478: pos = 0;
479: continue;
480: }
481: free(ln.buf);
482: return 0;
1.25 schwarze 483: case ROFF_APPEND:
1.71 schwarze 484: pos = strlen(ln.buf);
1.1 schwarze 485: continue;
1.25 schwarze 486: case ROFF_RERUN:
1.1 schwarze 487: goto rerun;
1.25 schwarze 488: case ROFF_IGN:
1.1 schwarze 489: pos = 0;
490: continue;
1.25 schwarze 491: case ROFF_SO:
1.71 schwarze 492: if ( ! (curp->options & MPARSE_SO) &&
493: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 494: curp->sodest = mandoc_strdup(ln.buf + of);
495: free(ln.buf);
1.135 schwarze 496: return 1;
1.23 schwarze 497: }
1.4 schwarze 498: /*
499: * We remove `so' clauses from our lookaside
500: * buffer because we're going to descend into
501: * the file recursively.
502: */
1.25 schwarze 503: if (curp->secondary)
1.4 schwarze 504: curp->secondary->sz -= pos + 1;
1.88 schwarze 505: save_file = curp->file;
1.122 schwarze 506: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 507: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 508: close(fd);
1.88 schwarze 509: curp->file = save_file;
510: } else {
511: curp->file = save_file;
1.30 schwarze 512: mandoc_vmsg(MANDOCERR_SO_FAIL,
513: curp, curp->line, pos,
514: ".so %s", ln.buf + of);
1.85 schwarze 515: ln.sz = mandoc_asprintf(&cp,
516: ".sp\nSee the file %s.\n.sp",
517: ln.buf + of);
518: free(ln.buf);
519: ln.buf = cp;
520: of = 0;
521: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 522: }
1.1 schwarze 523: pos = 0;
524: continue;
525: default:
526: break;
527: }
528:
1.124 schwarze 529: if (curp->man->macroset == MACROSET_NONE)
1.60 schwarze 530: choose_parser(curp);
1.1 schwarze 531:
1.25 schwarze 532: /*
1.60 schwarze 533: * Lastly, push down into the parsers themselves.
1.1 schwarze 534: * If libroff returns ROFF_TBL, then add it to the
535: * currently open parse. Since we only get here if
536: * there does exist data (see tbl_data.c), we're
537: * guaranteed that something's been allocated.
538: * Do the same for ROFF_EQN.
539: */
540:
1.114 schwarze 541: if (rr == ROFF_TBL)
1.76 schwarze 542: while ((span = roff_span(curp->roff)) != NULL)
1.114 schwarze 543: roff_addtbl(curp->man, span);
544: else if (rr == ROFF_EQN)
545: roff_addeqn(curp->man, roff_eqn(curp->roff));
546: else if ((curp->man->macroset == MACROSET_MDOC ?
1.110 schwarze 547: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.76 schwarze 548: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
549: break;
1.1 schwarze 550:
551: /* Temporary buffers typically are not full. */
552:
553: if (0 == start && '\0' == blk.buf[i])
554: break;
555:
556: /* Start the next input line. */
557:
558: pos = 0;
559: }
560:
561: free(ln.buf);
1.135 schwarze 562: return 1;
1.1 schwarze 563: }
564:
565: static int
1.18 schwarze 566: read_whole_file(struct mparse *curp, const char *file, int fd,
567: struct buf *fb, int *with_mmap)
1.1 schwarze 568: {
569: struct stat st;
1.115 schwarze 570: gzFile gz;
1.1 schwarze 571: size_t off;
572: ssize_t ssz;
573:
1.118 schwarze 574: if (fstat(fd, &st) == -1)
575: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 576:
577: /*
578: * If we're a regular file, try just reading in the whole entry
579: * via mmap(). This is faster than reading it into blocks, and
580: * since each file is only a few bytes to begin with, I'm not
581: * concerned that this is going to tank any machines.
582: */
583:
1.115 schwarze 584: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 585: if (st.st_size > 0x7fffffff) {
1.86 schwarze 586: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.117 schwarze 587: return 0;
1.1 schwarze 588: }
589: *with_mmap = 1;
590: fb->sz = (size_t)st.st_size;
1.15 schwarze 591: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 592: if (fb->buf != MAP_FAILED)
1.117 schwarze 593: return 1;
1.1 schwarze 594: }
595:
1.115 schwarze 596: if (curp->gzip) {
1.118 schwarze 597: if ((gz = gzdopen(fd, "rb")) == NULL)
598: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.115 schwarze 599: } else
600: gz = NULL;
601:
1.1 schwarze 602: /*
603: * If this isn't a regular file (like, say, stdin), then we must
604: * go the old way and just read things in bit by bit.
605: */
606:
607: *with_mmap = 0;
608: off = 0;
609: fb->sz = 0;
610: fb->buf = NULL;
611: for (;;) {
612: if (off == fb->sz) {
613: if (fb->sz == (1U << 31)) {
1.86 schwarze 614: mandoc_msg(MANDOCERR_TOOLARGE, curp,
615: 0, 0, NULL);
1.1 schwarze 616: break;
617: }
618: resize_buf(fb, 65536);
619: }
1.115 schwarze 620: ssz = curp->gzip ?
621: gzread(gz, fb->buf + (int)off, fb->sz - off) :
622: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 623: if (ssz == 0) {
624: fb->sz = off;
1.117 schwarze 625: return 1;
1.1 schwarze 626: }
1.118 schwarze 627: if (ssz == -1)
628: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 schwarze 629: off += (size_t)ssz;
630: }
631:
632: free(fb->buf);
633: fb->buf = NULL;
1.117 schwarze 634: return 0;
1.1 schwarze 635: }
636:
637: static void
638: mparse_end(struct mparse *curp)
639: {
1.110 schwarze 640: if (curp->man->macroset == MACROSET_NONE)
641: curp->man->macroset = MACROSET_MAN;
642: if (curp->man->macroset == MACROSET_MDOC)
643: mdoc_endparse(curp->man);
644: else
1.86 schwarze 645: man_endparse(curp->man);
1.1 schwarze 646: roff_endparse(curp->roff);
647: }
648:
1.15 schwarze 649: static void
650: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 651: {
1.61 schwarze 652: struct buf *svprimary;
1.1 schwarze 653: const char *svfile;
1.71 schwarze 654: size_t offset;
1.14 schwarze 655: static int recursion_depth;
656:
657: if (64 < recursion_depth) {
658: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 659: return;
1.14 schwarze 660: }
1.1 schwarze 661:
1.15 schwarze 662: /* Line number is per-file. */
663: svfile = curp->file;
664: curp->file = file;
1.61 schwarze 665: svprimary = curp->primary;
1.59 schwarze 666: curp->primary = &blk;
1.15 schwarze 667: curp->line = 1;
668: recursion_depth++;
669:
1.70 schwarze 670: /* Skip an UTF-8 byte order mark. */
671: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
672: (unsigned char)blk.buf[0] == 0xef &&
673: (unsigned char)blk.buf[1] == 0xbb &&
674: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 675: offset = 3;
1.70 schwarze 676: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 677: } else
678: offset = 0;
1.70 schwarze 679:
1.71 schwarze 680: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 681:
1.86 schwarze 682: if (--recursion_depth == 0)
1.15 schwarze 683: mparse_end(curp);
684:
1.61 schwarze 685: curp->primary = svprimary;
1.15 schwarze 686: curp->file = svfile;
687: }
688:
1.74 schwarze 689: /*
690: * Read the whole file into memory and call the parsers.
691: * Called recursively when an .so request is encountered.
692: */
1.15 schwarze 693: enum mandoclevel
694: mparse_readfd(struct mparse *curp, int fd, const char *file)
695: {
696: struct buf blk;
697: int with_mmap;
1.70 schwarze 698: int save_filenc;
1.1 schwarze 699:
1.67 schwarze 700: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 701: save_filenc = curp->filenc;
702: curp->filenc = curp->options &
703: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 704: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 705: curp->filenc = save_filenc;
1.67 schwarze 706: if (with_mmap)
707: munmap(blk.buf, blk.sz);
708: else
709: free(blk.buf);
710: }
1.117 schwarze 711: return curp->file_status;
1.58 schwarze 712: }
713:
1.122 schwarze 714: int
715: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 716: {
717: char *cp;
1.122 schwarze 718: int fd;
1.58 schwarze 719:
720: curp->file = file;
1.115 schwarze 721: cp = strrchr(file, '.');
722: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 723:
1.115 schwarze 724: /* First try to use the filename as it is. */
1.74 schwarze 725:
1.122 schwarze 726: if ((fd = open(file, O_RDONLY)) != -1)
727: return fd;
1.74 schwarze 728:
1.115 schwarze 729: /*
730: * If that doesn't work and the filename doesn't
731: * already end in .gz, try appending .gz.
732: */
1.74 schwarze 733:
1.115 schwarze 734: if ( ! curp->gzip) {
1.74 schwarze 735: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 736: fd = open(cp, O_RDONLY);
1.83 schwarze 737: free(cp);
1.122 schwarze 738: if (fd != -1) {
1.115 schwarze 739: curp->gzip = 1;
1.122 schwarze 740: return fd;
1.58 schwarze 741: }
742: }
743:
1.115 schwarze 744: /* Neither worked, give up. */
1.73 schwarze 745:
1.115 schwarze 746: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.122 schwarze 747: return -1;
1.1 schwarze 748: }
749:
750: struct mparse *
1.151 schwarze 751: mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
752: enum mandoc_os os_e, const char *os_s)
1.1 schwarze 753: {
754: struct mparse *curp;
755:
756: curp = mandoc_calloc(1, sizeof(struct mparse));
757:
1.22 schwarze 758: curp->options = options;
1.151 schwarze 759: curp->mmin = mmin;
1.1 schwarze 760: curp->mmsg = mmsg;
1.151 schwarze 761: curp->os_s = os_s;
1.1 schwarze 762:
1.119 schwarze 763: curp->roff = roff_alloc(curp, options);
1.151 schwarze 764: curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
1.112 schwarze 765: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 766: if (curp->options & MPARSE_MDOC) {
1.112 schwarze 767: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 768: if (curp->man->mdocmac == NULL)
769: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 770: } else if (curp->options & MPARSE_MAN) {
771: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 772: if (curp->man->manmac == NULL)
773: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 774: }
1.113 schwarze 775: curp->man->first->tok = TOKEN_NONE;
1.151 schwarze 776: curp->man->meta.os_e = os_e;
1.117 schwarze 777: return curp;
1.1 schwarze 778: }
779:
780: void
781: mparse_reset(struct mparse *curp)
782: {
783: roff_reset(curp->roff);
1.124 schwarze 784: roff_man_reset(curp->man);
1.133 schwarze 785:
786: free(curp->sodest);
787: curp->sodest = NULL;
788:
1.4 schwarze 789: if (curp->secondary)
790: curp->secondary->sz = 0;
1.1 schwarze 791:
792: curp->file_status = MANDOCLEVEL_OK;
1.132 schwarze 793: curp->gzip = 0;
1.1 schwarze 794: }
795:
796: void
797: mparse_free(struct mparse *curp)
798: {
799:
1.136 schwarze 800: roffhash_free(curp->man->mdocmac);
801: roffhash_free(curp->man->manmac);
1.112 schwarze 802: roff_man_free(curp->man);
1.133 schwarze 803: roff_free(curp->roff);
1.4 schwarze 804: if (curp->secondary)
805: free(curp->secondary->buf);
1.1 schwarze 806:
1.4 schwarze 807: free(curp->secondary);
1.23 schwarze 808: free(curp->sodest);
1.1 schwarze 809: free(curp);
810: }
811:
812: void
1.110 schwarze 813: mparse_result(struct mparse *curp, struct roff_man **man,
814: char **sodest)
1.1 schwarze 815: {
816:
1.23 schwarze 817: if (sodest && NULL != (*sodest = curp->sodest)) {
818: *man = NULL;
819: return;
820: }
1.1 schwarze 821: if (man)
822: *man = curp->man;
1.130 schwarze 823: }
824:
825: void
826: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
827: {
828: if (curp->file_status > *rc)
829: *rc = curp->file_status;
1.1 schwarze 830: }
831:
832: void
833: mandoc_vmsg(enum mandocerr t, struct mparse *m,
834: int ln, int pos, const char *fmt, ...)
835: {
836: char buf[256];
837: va_list ap;
838:
839: va_start(ap, fmt);
1.26 schwarze 840: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 841: va_end(ap);
842:
843: mandoc_msg(t, m, ln, pos, buf);
844: }
845:
846: void
1.25 schwarze 847: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 848: int ln, int col, const char *msg)
849: {
850: enum mandoclevel level;
851:
1.151 schwarze 852: if (er < m->mmin && er != MANDOCERR_FILE)
853: return;
854:
1.87 schwarze 855: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 856: while (er < mandoclimits[level])
857: level--;
858:
859: if (m->mmsg)
860: (*m->mmsg)(er, level, m->file, ln, col, msg);
861:
862: if (m->file_status < level)
863: m->file_status = level;
864: }
865:
866: const char *
867: mparse_strerror(enum mandocerr er)
868: {
869:
1.117 schwarze 870: return mandocerrs[er];
1.1 schwarze 871: }
872:
873: const char *
874: mparse_strlevel(enum mandoclevel lvl)
875: {
1.117 schwarze 876: return mandoclevels[lvl];
1.4 schwarze 877: }
878:
879: void
880: mparse_keep(struct mparse *p)
881: {
882:
883: assert(NULL == p->secondary);
884: p->secondary = mandoc_calloc(1, sizeof(struct buf));
885: }
886:
887: const char *
888: mparse_getkeep(const struct mparse *p)
889: {
890:
891: assert(p->secondary);
1.117 schwarze 892: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 schwarze 893: }