Annotation of mandoc/read.c, Revision 1.177
1.177 ! schwarze 1: /* $Id: read.c,v 1.176 2017/06/11 19:37:01 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.156 schwarze 4: * Copyright (c) 2010-2017 Ingo Schwarze <schwarze@openbsd.org>
1.42 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 kristaps 6: *
7: * Permission to use, copy, modify, and distribute this software for any
8: * purpose with or without fee is hereby granted, provided that the above
9: * copyright notice and this permission notice appear in all copies.
10: *
1.133 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 kristaps 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.133 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 kristaps 14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18: */
1.11 kristaps 19: #include "config.h"
20:
1.80 schwarze 21: #include <sys/types.h>
1.82 schwarze 22: #include <sys/mman.h>
1.80 schwarze 23: #include <sys/stat.h>
1.1 kristaps 24:
25: #include <assert.h>
26: #include <ctype.h>
1.146 schwarze 27: #if HAVE_ERR
1.143 schwarze 28: #include <err.h>
1.146 schwarze 29: #endif
1.40 schwarze 30: #include <errno.h>
1.1 kristaps 31: #include <fcntl.h>
1.3 kristaps 32: #include <stdarg.h>
1.1 kristaps 33: #include <stdio.h>
34: #include <stdlib.h>
35: #include <string.h>
36: #include <unistd.h>
1.140 schwarze 37: #include <zlib.h>
1.1 kristaps 38:
1.133 schwarze 39: #include "mandoc_aux.h"
1.1 kristaps 40: #include "mandoc.h"
1.133 schwarze 41: #include "roff.h"
1.1 kristaps 42: #include "mdoc.h"
43: #include "man.h"
1.133 schwarze 44: #include "libmandoc.h"
1.139 schwarze 45: #include "roff_int.h"
1.1 kristaps 46:
47: #define REPARSE_LIMIT 1000
48:
49: struct mparse {
1.160 schwarze 50: struct roff *roff; /* roff parser (!NULL) */
1.134 schwarze 51: struct roff_man *man; /* man parser */
1.45 schwarze 52: char *sodest; /* filename pointed to by .so */
1.83 schwarze 53: const char *file; /* filename of current input file */
54: struct buf *primary; /* buffer currently being parsed */
55: struct buf *secondary; /* preprocessed copy of input */
56: const char *defos; /* default operating system */
57: mandocmsg mmsg; /* warning/error message handler */
58: enum mandoclevel file_status; /* status of current parse */
59: enum mandoclevel wlevel; /* ignore messages below this */
60: int options; /* parser options */
1.140 schwarze 61: int gzip; /* current input file is gzipped */
1.93 schwarze 62: int filenc; /* encoding of the current file */
1.1 kristaps 63: int reparse_count; /* finite interp. stack */
1.83 schwarze 64: int line; /* line number in the file */
1.1 kristaps 65: };
66:
1.84 schwarze 67: static void choose_parser(struct mparse *);
1.1 kristaps 68: static void resize_buf(struct buf *, size_t);
1.163 schwarze 69: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 70: static int read_whole_file(struct mparse *, const char *, int,
71: struct buf *, int *);
1.1 kristaps 72: static void mparse_end(struct mparse *);
1.37 schwarze 73: static void mparse_parse_buffer(struct mparse *, struct buf,
74: const char *);
1.1 kristaps 75:
1.3 kristaps 76: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
77: MANDOCERR_OK,
1.165 schwarze 78: MANDOCERR_STYLE,
1.3 kristaps 79: MANDOCERR_WARNING,
80: MANDOCERR_ERROR,
1.112 schwarze 81: MANDOCERR_UNSUPP,
1.3 kristaps 82: MANDOCERR_MAX,
83: MANDOCERR_MAX
84: };
85:
1.7 kristaps 86: static const char * const mandocerrs[MANDOCERR_MAX] = {
87: "ok",
88:
1.165 schwarze 89: "generic style suggestion",
1.166 schwarze 90:
1.176 schwarze 91: "Mdocdate found",
92: "Mdocdate missing",
93: "legacy man(7) date format",
1.177 ! schwarze 94: "RCS id missing",
1.166 schwarze 95: "useless macro",
1.167 schwarze 96: "consider using OS macro",
1.172 schwarze 97: "errnos out of order",
98: "duplicate errno",
1.168 schwarze 99: "description line ends with a full stop",
1.174 schwarze 100: "no blank before trailing delimiter",
1.175 schwarze 101: "function name without markup",
1.166 schwarze 102:
1.7 kristaps 103: "generic warning",
104:
105: /* related to the prologue */
1.79 schwarze 106: "missing manual title, using UNTITLED",
107: "missing manual title, using \"\"",
1.54 schwarze 108: "lower case character in document title",
1.79 schwarze 109: "missing manual section, using \"\"",
1.7 kristaps 110: "unknown manual section",
1.54 schwarze 111: "missing date, using today's date",
1.7 kristaps 112: "cannot parse date, using it verbatim",
1.79 schwarze 113: "missing Os macro, using \"\"",
114: "duplicate prologue macro",
115: "late prologue macro",
116: "skipping late title macro",
1.7 kristaps 117: "prologue macros out of order",
118:
119: /* related to document structure */
120: ".so is fragile, better use ln(1)",
1.50 schwarze 121: "no document body",
1.54 schwarze 122: "content before first section header",
123: "first section is not \"NAME\"",
1.156 schwarze 124: "NAME section without Nm before Nd",
1.128 schwarze 125: "NAME section without description",
126: "description not at the end of NAME",
127: "bad NAME section content",
1.156 schwarze 128: "missing comma before name",
1.121 schwarze 129: "missing description line, using \"\"",
1.162 schwarze 130: "description line outside NAME section",
1.7 kristaps 131: "sections out of conventional order",
1.54 schwarze 132: "duplicate section title",
133: "unexpected section",
1.87 schwarze 134: "unusual Xr order",
135: "unusual Xr punctuation",
1.86 schwarze 136: "AUTHORS section without An macro",
1.7 kristaps 137:
138: /* related to macros and nesting */
1.55 schwarze 139: "obsolete macro",
1.102 schwarze 140: "macro neither callable nor escaped",
1.7 kristaps 141: "skipping paragraph macro",
1.31 schwarze 142: "moving paragraph macro out of list",
1.7 kristaps 143: "skipping no-space macro",
144: "blocks badly nested",
145: "nested displays are not portable",
1.57 schwarze 146: "moving content out of list",
1.78 schwarze 147: "fill mode already enabled, skipping",
148: "fill mode already disabled, skipping",
1.7 kristaps 149: "line scope broken",
1.169 schwarze 150: "skipping blank line in line scope",
1.7 kristaps 151:
152: /* related to missing macro arguments */
1.58 schwarze 153: "skipping empty request",
154: "conditional request controls empty scope",
1.7 kristaps 155: "skipping empty macro",
1.124 schwarze 156: "empty block",
1.62 schwarze 157: "empty argument, using 0n",
1.60 schwarze 158: "missing display type, using -ragged",
159: "list type is not the first argument",
1.152 schwarze 160: "missing -width in -tag list, using 6n",
1.78 schwarze 161: "missing utility name, using \"\"",
1.123 schwarze 162: "missing function name, using \"\"",
1.60 schwarze 163: "empty head in list item",
164: "empty list item",
1.61 schwarze 165: "missing font type, using \\fR",
166: "unknown font type, using \\fR",
1.103 schwarze 167: "nothing follows prefix",
1.122 schwarze 168: "empty reference block",
1.155 schwarze 169: "missing section argument",
1.60 schwarze 170: "missing -std argument, adding it",
1.125 schwarze 171: "missing option string, using \"\"",
172: "missing resource identifier, using \"\"",
1.90 schwarze 173: "missing eqn box, using \"\"",
1.7 kristaps 174:
175: /* related to bad macro arguments */
1.64 schwarze 176: "unterminated quoted argument",
1.7 kristaps 177: "duplicate argument",
1.76 schwarze 178: "skipping duplicate argument",
1.63 schwarze 179: "skipping duplicate display type",
180: "skipping duplicate list type",
1.76 schwarze 181: "skipping -width argument",
1.126 schwarze 182: "wrong number of cells",
1.7 kristaps 183: "unknown AT&T UNIX version",
1.88 schwarze 184: "comma in function argument",
1.89 schwarze 185: "parenthesis in function name",
1.173 schwarze 186: "unknown library name",
1.67 schwarze 187: "invalid content in Rs block",
1.63 schwarze 188: "invalid Boolean argument",
189: "unknown font, skipping request",
1.126 schwarze 190: "odd number of characters in request",
1.7 kristaps 191:
192: /* related to plain text */
1.64 schwarze 193: "blank line in fill mode, using .sp",
194: "tab in filled text",
195: "whitespace at end of input line",
1.158 schwarze 196: "new sentence, new line",
1.7 kristaps 197: "bad comment style",
1.64 schwarze 198: "invalid escape sequence",
199: "undefined string, using \"\"",
1.16 kristaps 200:
1.118 schwarze 201: /* related to tables */
202: "tbl line starts with span",
203: "tbl column starts with span",
204: "skipping vertical bar in tbl layout",
205:
1.7 kristaps 206: "generic error",
207:
208: /* related to tables */
1.116 schwarze 209: "non-alphabetic character in tbl options",
210: "skipping unknown tbl option",
211: "missing tbl option argument",
212: "wrong tbl option argument size",
1.118 schwarze 213: "empty tbl layout",
214: "invalid character in tbl layout",
215: "unmatched parenthesis in tbl layout",
1.119 schwarze 216: "tbl without any data cells",
217: "ignoring data in spanned tbl cell",
218: "ignoring extra tbl data cells",
219: "data block open at end of tbl",
1.7 kristaps 220:
1.68 schwarze 221: /* related to document structure and macros */
1.108 schwarze 222: NULL,
1.7 kristaps 223: "input stack limit exceeded, infinite loop?",
224: "skipping bad character",
1.68 schwarze 225: "skipping unknown macro",
1.112 schwarze 226: "skipping insecure request",
1.73 schwarze 227: "skipping item outside list",
1.68 schwarze 228: "skipping column outside column list",
229: "skipping end of block that is not open",
1.115 schwarze 230: "fewer RS blocks open, skipping",
1.68 schwarze 231: "inserting missing end of block",
232: "appending missing end of block",
233:
234: /* related to request and macro arguments */
1.7 kristaps 235: "escaped character not allowed in a name",
1.99 schwarze 236: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 237: "skipping display without arguments",
1.71 schwarze 238: "missing list type, using -item",
1.171 schwarze 239: "argument is not numeric, using 1",
1.70 schwarze 240: "missing manual name, using \"\"",
1.71 schwarze 241: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 242: "unknown standard specifier",
1.71 schwarze 243: "skipping request without numeric argument",
1.109 schwarze 244: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 245: ".so request failed",
1.61 schwarze 246: "skipping all arguments",
247: "skipping excess arguments",
1.92 kristaps 248: "divide by zero",
1.112 schwarze 249:
250: "unsupported feature",
251: "input too large",
1.114 schwarze 252: "unsupported control character",
1.112 schwarze 253: "unsupported roff request",
1.119 schwarze 254: "eqn delim option in tbl",
1.118 schwarze 255: "unsupported tbl layout modifier",
1.112 schwarze 256: "ignoring macro in table",
1.7 kristaps 257: };
258:
259: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
260: "SUCCESS",
1.165 schwarze 261: "STYLE",
1.7 kristaps 262: "WARNING",
263: "ERROR",
1.112 schwarze 264: "UNSUPP",
1.7 kristaps 265: "BADARG",
266: "SYSERR"
267: };
268:
1.47 schwarze 269:
1.1 kristaps 270: static void
271: resize_buf(struct buf *buf, size_t initial)
272: {
273:
274: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
275: buf->buf = mandoc_realloc(buf->buf, buf->sz);
276: }
277:
278: static void
1.84 schwarze 279: choose_parser(struct mparse *curp)
1.1 kristaps 280: {
1.83 schwarze 281: char *cp, *ep;
282: int format;
1.1 kristaps 283:
1.83 schwarze 284: /*
285: * If neither command line arguments -mdoc or -man select
286: * a parser nor the roff parser found a .Dd or .TH macro
287: * yet, look ahead in the main input buffer.
288: */
289:
290: if ((format = roff_getformat(curp->roff)) == 0) {
291: cp = curp->primary->buf;
292: ep = cp + curp->primary->sz;
293: while (cp < ep) {
1.85 schwarze 294: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 295: cp++;
296: if (cp[0] == 'D' && cp[1] == 'd') {
297: format = MPARSE_MDOC;
298: break;
299: }
300: if (cp[0] == 'T' && cp[1] == 'H') {
301: format = MPARSE_MAN;
302: break;
303: }
304: }
305: cp = memchr(cp, '\n', ep - cp);
306: if (cp == NULL)
307: break;
308: cp++;
309: }
1.1 kristaps 310: }
311:
1.83 schwarze 312: if (format == MPARSE_MDOC) {
1.137 schwarze 313: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 314: if (curp->man->mdocmac == NULL)
315: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 316: } else {
317: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 318: if (curp->man->manmac == NULL)
319: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 320: }
1.164 schwarze 321: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 322: }
323:
324: /*
1.95 schwarze 325: * Main parse routine for a buffer.
326: * It assumes encoding and line numbering are already set up.
327: * It can recurse directly (for invocations of user-defined
328: * macros, inline equations, and input line traps)
329: * and indirectly (for .so file inclusion).
1.1 kristaps 330: */
1.163 schwarze 331: static int
1.95 schwarze 332: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 333: {
334: const struct tbl_span *span;
335: struct buf ln;
1.113 schwarze 336: const char *save_file;
1.110 schwarze 337: char *cp;
1.95 schwarze 338: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 339: enum rofferr rr;
1.100 schwarze 340: int of;
1.1 kristaps 341: int lnn; /* line number in the real file */
1.110 schwarze 342: int fd;
1.1 kristaps 343: unsigned char c;
344:
1.95 schwarze 345: memset(&ln, 0, sizeof(ln));
1.1 kristaps 346:
1.47 schwarze 347: lnn = curp->line;
348: pos = 0;
1.1 kristaps 349:
1.95 schwarze 350: while (i < blk.sz) {
1.1 kristaps 351: if (0 == pos && '\0' == blk.buf[i])
352: break;
353:
354: if (start) {
355: curp->line = lnn;
356: curp->reparse_count = 0;
1.93 schwarze 357:
358: if (lnn < 3 &&
359: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 360: curp->filenc & MPARSE_LATIN1)
361: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 362: }
363:
1.95 schwarze 364: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 365:
366: /*
367: * When finding an unescaped newline character,
368: * leave the character loop to process the line.
369: * Skip a preceding carriage return, if any.
370: */
371:
1.95 schwarze 372: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 373: '\n' == blk.buf[i + 1])
374: ++i;
375: if ('\n' == blk.buf[i]) {
376: ++i;
377: ++lnn;
378: break;
379: }
380:
1.35 schwarze 381: /*
1.93 schwarze 382: * Make sure we have space for the worst
383: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 384: */
385:
1.95 schwarze 386: if (pos + 11 > ln.sz)
1.35 schwarze 387: resize_buf(&ln, 256);
388:
1.47 schwarze 389: /*
1.93 schwarze 390: * Encode 8-bit input.
1.1 kristaps 391: */
392:
1.93 schwarze 393: c = blk.buf[i];
394: if (c & 0x80) {
1.95 schwarze 395: if ( ! (curp->filenc && preconv_encode(
396: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 397: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
398: curp->line, pos, "0x%x", c);
1.93 schwarze 399: ln.buf[pos++] = '?';
400: i++;
401: }
402: continue;
403: }
404:
405: /*
406: * Exclude control characters.
407: */
1.1 kristaps 408:
1.93 schwarze 409: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 410: mandoc_vmsg(c == 0x00 || c == 0x04 ||
411: c > 0x0a ? MANDOCERR_CHAR_BAD :
412: MANDOCERR_CHAR_UNSUPP,
413: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 414: i++;
1.127 schwarze 415: if (c != '\r')
416: ln.buf[pos++] = '?';
1.1 kristaps 417: continue;
418: }
419:
420: ln.buf[pos++] = blk.buf[i++];
421: }
422:
1.170 schwarze 423: if (pos + 1 >= ln.sz)
1.1 kristaps 424: resize_buf(&ln, 256);
425:
1.170 schwarze 426: if (i == blk.sz || blk.buf[i] == '\0')
427: ln.buf[pos++] = '\n';
1.1 kristaps 428: ln.buf[pos] = '\0';
429:
430: /*
431: * A significant amount of complexity is contained by
432: * the roff preprocessor. It's line-oriented but can be
433: * expressed on one line, so we need at times to
434: * readjust our starting point and re-run it. The roff
435: * preprocessor can also readjust the buffers with new
436: * data, so we pass them in wholesale.
437: */
438:
439: of = 0;
440:
1.24 kristaps 441: /*
442: * Maintain a lookaside buffer of all parsed lines. We
443: * only do this if mparse_keep() has been invoked (the
444: * buffer may be accessed with mparse_getkeep()).
445: */
446:
447: if (curp->secondary) {
1.47 schwarze 448: curp->secondary->buf = mandoc_realloc(
449: curp->secondary->buf,
450: curp->secondary->sz + pos + 2);
451: memcpy(curp->secondary->buf +
452: curp->secondary->sz,
453: ln.buf, pos);
1.24 kristaps 454: curp->secondary->sz += pos;
455: curp->secondary->buf
456: [curp->secondary->sz] = '\n';
457: curp->secondary->sz++;
458: curp->secondary->buf
459: [curp->secondary->sz] = '\0';
460: }
1.1 kristaps 461: rerun:
1.96 schwarze 462: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 463:
464: switch (rr) {
1.47 schwarze 465: case ROFF_REPARSE:
1.163 schwarze 466: if (++curp->reparse_count > REPARSE_LIMIT)
1.3 kristaps 467: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 468: curp->line, pos, NULL);
1.163 schwarze 469: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
470: start == 1) {
471: pos = 0;
472: continue;
473: }
474: free(ln.buf);
475: return 0;
1.47 schwarze 476: case ROFF_APPEND:
1.95 schwarze 477: pos = strlen(ln.buf);
1.1 kristaps 478: continue;
1.47 schwarze 479: case ROFF_RERUN:
1.1 kristaps 480: goto rerun;
1.47 schwarze 481: case ROFF_IGN:
1.1 kristaps 482: pos = 0;
483: continue;
1.47 schwarze 484: case ROFF_SO:
1.95 schwarze 485: if ( ! (curp->options & MPARSE_SO) &&
486: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 487: curp->sodest = mandoc_strdup(ln.buf + of);
488: free(ln.buf);
1.163 schwarze 489: return 1;
1.45 schwarze 490: }
1.24 kristaps 491: /*
492: * We remove `so' clauses from our lookaside
493: * buffer because we're going to descend into
494: * the file recursively.
495: */
1.47 schwarze 496: if (curp->secondary)
1.25 kristaps 497: curp->secondary->sz -= pos + 1;
1.113 schwarze 498: save_file = curp->file;
1.148 schwarze 499: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 500: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 501: close(fd);
1.113 schwarze 502: curp->file = save_file;
503: } else {
504: curp->file = save_file;
1.52 schwarze 505: mandoc_vmsg(MANDOCERR_SO_FAIL,
506: curp, curp->line, pos,
507: ".so %s", ln.buf + of);
1.110 schwarze 508: ln.sz = mandoc_asprintf(&cp,
509: ".sp\nSee the file %s.\n.sp",
510: ln.buf + of);
511: free(ln.buf);
512: ln.buf = cp;
513: of = 0;
514: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 515: }
1.1 kristaps 516: pos = 0;
517: continue;
518: default:
519: break;
520: }
521:
1.150 schwarze 522: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 523: choose_parser(curp);
1.1 kristaps 524:
1.47 schwarze 525: /*
1.84 schwarze 526: * Lastly, push down into the parsers themselves.
1.1 kristaps 527: * If libroff returns ROFF_TBL, then add it to the
528: * currently open parse. Since we only get here if
529: * there does exist data (see tbl_data.c), we're
530: * guaranteed that something's been allocated.
531: * Do the same for ROFF_EQN.
532: */
533:
1.139 schwarze 534: if (rr == ROFF_TBL)
1.100 schwarze 535: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 536: roff_addtbl(curp->man, span);
537: else if (rr == ROFF_EQN)
538: roff_addeqn(curp->man, roff_eqn(curp->roff));
539: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 540: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 541: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
542: break;
1.1 kristaps 543:
544: /* Temporary buffers typically are not full. */
545:
546: if (0 == start && '\0' == blk.buf[i])
547: break;
548:
549: /* Start the next input line. */
550:
551: pos = 0;
552: }
553:
554: free(ln.buf);
1.163 schwarze 555: return 1;
1.1 kristaps 556: }
557:
558: static int
1.40 schwarze 559: read_whole_file(struct mparse *curp, const char *file, int fd,
560: struct buf *fb, int *with_mmap)
1.1 kristaps 561: {
1.161 schwarze 562: struct stat st;
1.140 schwarze 563: gzFile gz;
1.1 kristaps 564: size_t off;
565: ssize_t ssz;
1.143 schwarze 566:
567: if (fstat(fd, &st) == -1)
568: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 569:
570: /*
571: * If we're a regular file, try just reading in the whole entry
572: * via mmap(). This is faster than reading it into blocks, and
573: * since each file is only a few bytes to begin with, I'm not
574: * concerned that this is going to tank any machines.
575: */
576:
1.140 schwarze 577: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 578: if (st.st_size > 0x7fffffff) {
1.111 schwarze 579: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 580: return 0;
1.1 kristaps 581: }
582: *with_mmap = 1;
583: fb->sz = (size_t)st.st_size;
1.37 schwarze 584: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 585: if (fb->buf != MAP_FAILED)
1.142 schwarze 586: return 1;
1.1 kristaps 587: }
588:
1.140 schwarze 589: if (curp->gzip) {
1.143 schwarze 590: if ((gz = gzdopen(fd, "rb")) == NULL)
591: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 592: } else
593: gz = NULL;
594:
1.1 kristaps 595: /*
596: * If this isn't a regular file (like, say, stdin), then we must
597: * go the old way and just read things in bit by bit.
598: */
599:
600: *with_mmap = 0;
601: off = 0;
602: fb->sz = 0;
603: fb->buf = NULL;
604: for (;;) {
605: if (off == fb->sz) {
606: if (fb->sz == (1U << 31)) {
1.111 schwarze 607: mandoc_msg(MANDOCERR_TOOLARGE, curp,
608: 0, 0, NULL);
1.1 kristaps 609: break;
610: }
611: resize_buf(fb, 65536);
612: }
1.140 schwarze 613: ssz = curp->gzip ?
614: gzread(gz, fb->buf + (int)off, fb->sz - off) :
615: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 616: if (ssz == 0) {
617: fb->sz = off;
1.142 schwarze 618: return 1;
1.1 kristaps 619: }
1.143 schwarze 620: if (ssz == -1)
621: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 622: off += (size_t)ssz;
623: }
624:
625: free(fb->buf);
626: fb->buf = NULL;
1.142 schwarze 627: return 0;
1.1 kristaps 628: }
629:
630: static void
631: mparse_end(struct mparse *curp)
632: {
1.135 schwarze 633: if (curp->man->macroset == MACROSET_NONE)
634: curp->man->macroset = MACROSET_MAN;
635: if (curp->man->macroset == MACROSET_MDOC)
636: mdoc_endparse(curp->man);
637: else
1.111 schwarze 638: man_endparse(curp->man);
1.1 kristaps 639: roff_endparse(curp->roff);
640: }
641:
642: static void
1.36 schwarze 643: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 644: {
1.85 schwarze 645: struct buf *svprimary;
1.28 joerg 646: const char *svfile;
1.95 schwarze 647: size_t offset;
1.36 schwarze 648: static int recursion_depth;
649:
650: if (64 < recursion_depth) {
651: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
652: return;
653: }
1.28 joerg 654:
655: /* Line number is per-file. */
656: svfile = curp->file;
657: curp->file = file;
1.85 schwarze 658: svprimary = curp->primary;
1.83 schwarze 659: curp->primary = &blk;
1.28 joerg 660: curp->line = 1;
1.36 schwarze 661: recursion_depth++;
1.28 joerg 662:
1.93 schwarze 663: /* Skip an UTF-8 byte order mark. */
664: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
665: (unsigned char)blk.buf[0] == 0xef &&
666: (unsigned char)blk.buf[1] == 0xbb &&
667: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 668: offset = 3;
1.93 schwarze 669: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 670: } else
671: offset = 0;
1.93 schwarze 672:
1.95 schwarze 673: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 674:
1.111 schwarze 675: if (--recursion_depth == 0)
1.28 joerg 676: mparse_end(curp);
677:
1.85 schwarze 678: curp->primary = svprimary;
1.28 joerg 679: curp->file = svfile;
680: }
681:
682: enum mandoclevel
1.104 schwarze 683: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 684: const char *file)
685: {
686: struct buf blk;
687:
1.104 schwarze 688: blk.buf = buf;
1.28 joerg 689: blk.sz = len;
690:
1.36 schwarze 691: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 692: return curp->file_status;
1.28 joerg 693: }
694:
1.98 schwarze 695: /*
696: * Read the whole file into memory and call the parsers.
697: * Called recursively when an .so request is encountered.
698: */
1.36 schwarze 699: enum mandoclevel
700: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 701: {
1.28 joerg 702: struct buf blk;
703: int with_mmap;
1.93 schwarze 704: int save_filenc;
1.1 kristaps 705:
1.91 schwarze 706: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 707: save_filenc = curp->filenc;
708: curp->filenc = curp->options &
709: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 710: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 711: curp->filenc = save_filenc;
1.91 schwarze 712: if (with_mmap)
713: munmap(blk.buf, blk.sz);
714: else
715: free(blk.buf);
716: }
1.142 schwarze 717: return curp->file_status;
1.82 schwarze 718: }
719:
1.148 schwarze 720: int
721: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 722: {
723: char *cp;
1.148 schwarze 724: int fd;
1.82 schwarze 725:
726: curp->file = file;
1.140 schwarze 727: cp = strrchr(file, '.');
728: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 729:
1.140 schwarze 730: /* First try to use the filename as it is. */
1.98 schwarze 731:
1.148 schwarze 732: if ((fd = open(file, O_RDONLY)) != -1)
733: return fd;
1.98 schwarze 734:
1.140 schwarze 735: /*
736: * If that doesn't work and the filename doesn't
737: * already end in .gz, try appending .gz.
738: */
1.98 schwarze 739:
1.140 schwarze 740: if ( ! curp->gzip) {
1.98 schwarze 741: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 742: fd = open(cp, O_RDONLY);
1.108 schwarze 743: free(cp);
1.148 schwarze 744: if (fd != -1) {
1.140 schwarze 745: curp->gzip = 1;
1.148 schwarze 746: return fd;
1.82 schwarze 747: }
748: }
749:
1.140 schwarze 750: /* Neither worked, give up. */
1.97 schwarze 751:
1.140 schwarze 752: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 753: return -1;
1.1 kristaps 754: }
755:
756: struct mparse *
1.94 schwarze 757: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.144 schwarze 758: const char *defos)
1.1 kristaps 759: {
760: struct mparse *curp;
1.10 kristaps 761:
1.1 kristaps 762: curp = mandoc_calloc(1, sizeof(struct mparse));
763:
1.44 schwarze 764: curp->options = options;
1.3 kristaps 765: curp->wlevel = wlevel;
1.1 kristaps 766: curp->mmsg = mmsg;
1.29 schwarze 767: curp->defos = defos;
1.1 kristaps 768:
1.144 schwarze 769: curp->roff = roff_alloc(curp, options);
1.137 schwarze 770: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
771: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 772: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 773: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 774: if (curp->man->mdocmac == NULL)
775: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 776: } else if (curp->options & MPARSE_MAN) {
777: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 778: if (curp->man->manmac == NULL)
779: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 780: }
1.138 schwarze 781: curp->man->first->tok = TOKEN_NONE;
1.142 schwarze 782: return curp;
1.1 kristaps 783: }
784:
785: void
786: mparse_reset(struct mparse *curp)
787: {
788: roff_reset(curp->roff);
1.150 schwarze 789: roff_man_reset(curp->man);
1.160 schwarze 790:
791: free(curp->sodest);
792: curp->sodest = NULL;
793:
1.24 kristaps 794: if (curp->secondary)
795: curp->secondary->sz = 0;
1.1 kristaps 796:
797: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 798: curp->gzip = 0;
1.1 kristaps 799: }
800:
801: void
802: mparse_free(struct mparse *curp)
803: {
804:
1.164 schwarze 805: roffhash_free(curp->man->mdocmac);
806: roffhash_free(curp->man->manmac);
1.137 schwarze 807: roff_man_free(curp->man);
1.160 schwarze 808: roff_free(curp->roff);
1.24 kristaps 809: if (curp->secondary)
810: free(curp->secondary->buf);
1.1 kristaps 811:
1.24 kristaps 812: free(curp->secondary);
1.45 schwarze 813: free(curp->sodest);
1.1 kristaps 814: free(curp);
815: }
816:
817: void
1.135 schwarze 818: mparse_result(struct mparse *curp, struct roff_man **man,
819: char **sodest)
1.1 kristaps 820: {
821:
1.45 schwarze 822: if (sodest && NULL != (*sodest = curp->sodest)) {
823: *man = NULL;
824: return;
825: }
1.9 kristaps 826: if (man)
827: *man = curp->man;
1.157 schwarze 828: }
829:
830: void
831: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
832: {
833: if (curp->file_status > *rc)
834: *rc = curp->file_status;
1.3 kristaps 835: }
836:
837: void
838: mandoc_vmsg(enum mandocerr t, struct mparse *m,
839: int ln, int pos, const char *fmt, ...)
840: {
841: char buf[256];
842: va_list ap;
843:
844: va_start(ap, fmt);
1.48 schwarze 845: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 846: va_end(ap);
847:
848: mandoc_msg(t, m, ln, pos, buf);
849: }
850:
851: void
1.47 schwarze 852: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 853: int ln, int col, const char *msg)
854: {
855: enum mandoclevel level;
856:
1.112 schwarze 857: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 858: while (er < mandoclimits[level])
859: level--;
860:
1.108 schwarze 861: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 862: return;
863:
1.8 kristaps 864: if (m->mmsg)
865: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 866:
867: if (m->file_status < level)
868: m->file_status = level;
1.7 kristaps 869: }
870:
871: const char *
872: mparse_strerror(enum mandocerr er)
873: {
874:
1.142 schwarze 875: return mandocerrs[er];
1.7 kristaps 876: }
877:
878: const char *
879: mparse_strlevel(enum mandoclevel lvl)
880: {
1.142 schwarze 881: return mandoclevels[lvl];
1.24 kristaps 882: }
883:
884: void
885: mparse_keep(struct mparse *p)
886: {
887:
888: assert(NULL == p->secondary);
889: p->secondary = mandoc_calloc(1, sizeof(struct buf));
890: }
891:
892: const char *
893: mparse_getkeep(const struct mparse *p)
894: {
895:
896: assert(p->secondary);
1.142 schwarze 897: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 898: }
CVSweb