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