Annotation of mandoc/read.c, Revision 1.184
1.184 ! schwarze 1: /* $Id: read.c,v 1.183 2017/06/29 15:22:17 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 */
1.179 schwarze 56: const char *os_s; /* default operating system */
1.83 schwarze 57: mandocmsg mmsg; /* warning/error message handler */
58: enum mandoclevel file_status; /* status of current parse */
1.179 schwarze 59: enum mandocerr mmin; /* ignore messages below this */
1.83 schwarze 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:
1.179 schwarze 89: "base system convention",
1.166 schwarze 90:
1.176 schwarze 91: "Mdocdate found",
92: "Mdocdate missing",
1.181 schwarze 93: "unknown architecture",
1.180 schwarze 94: "operating system explicitly specified",
1.179 schwarze 95: "RCS id missing",
1.184 ! schwarze 96: "referenced manual not found",
1.179 schwarze 97:
98: "generic style suggestion",
99:
1.176 schwarze 100: "legacy man(7) date format",
1.178 schwarze 101: "duplicate RCS id",
1.182 schwarze 102: "typo in section name",
1.166 schwarze 103: "useless macro",
1.167 schwarze 104: "consider using OS macro",
1.172 schwarze 105: "errnos out of order",
106: "duplicate errno",
1.168 schwarze 107: "description line ends with a full stop",
1.174 schwarze 108: "no blank before trailing delimiter",
1.175 schwarze 109: "function name without markup",
1.166 schwarze 110:
1.7 kristaps 111: "generic warning",
112:
113: /* related to the prologue */
1.79 schwarze 114: "missing manual title, using UNTITLED",
115: "missing manual title, using \"\"",
1.54 schwarze 116: "lower case character in document title",
1.79 schwarze 117: "missing manual section, using \"\"",
1.7 kristaps 118: "unknown manual section",
1.54 schwarze 119: "missing date, using today's date",
1.7 kristaps 120: "cannot parse date, using it verbatim",
1.79 schwarze 121: "missing Os macro, using \"\"",
122: "duplicate prologue macro",
123: "late prologue macro",
124: "skipping late title macro",
1.7 kristaps 125: "prologue macros out of order",
126:
127: /* related to document structure */
128: ".so is fragile, better use ln(1)",
1.50 schwarze 129: "no document body",
1.54 schwarze 130: "content before first section header",
131: "first section is not \"NAME\"",
1.156 schwarze 132: "NAME section without Nm before Nd",
1.128 schwarze 133: "NAME section without description",
134: "description not at the end of NAME",
135: "bad NAME section content",
1.156 schwarze 136: "missing comma before name",
1.121 schwarze 137: "missing description line, using \"\"",
1.162 schwarze 138: "description line outside NAME section",
1.7 kristaps 139: "sections out of conventional order",
1.54 schwarze 140: "duplicate section title",
141: "unexpected section",
1.87 schwarze 142: "unusual Xr order",
143: "unusual Xr punctuation",
1.86 schwarze 144: "AUTHORS section without An macro",
1.7 kristaps 145:
146: /* related to macros and nesting */
1.55 schwarze 147: "obsolete macro",
1.102 schwarze 148: "macro neither callable nor escaped",
1.7 kristaps 149: "skipping paragraph macro",
1.31 schwarze 150: "moving paragraph macro out of list",
1.7 kristaps 151: "skipping no-space macro",
152: "blocks badly nested",
153: "nested displays are not portable",
1.57 schwarze 154: "moving content out of list",
1.183 schwarze 155: "first macro on line",
1.78 schwarze 156: "fill mode already enabled, skipping",
157: "fill mode already disabled, skipping",
1.7 kristaps 158: "line scope broken",
1.169 schwarze 159: "skipping blank line in line scope",
1.7 kristaps 160:
161: /* related to missing macro arguments */
1.58 schwarze 162: "skipping empty request",
163: "conditional request controls empty scope",
1.7 kristaps 164: "skipping empty macro",
1.124 schwarze 165: "empty block",
1.62 schwarze 166: "empty argument, using 0n",
1.60 schwarze 167: "missing display type, using -ragged",
168: "list type is not the first argument",
1.152 schwarze 169: "missing -width in -tag list, using 6n",
1.78 schwarze 170: "missing utility name, using \"\"",
1.123 schwarze 171: "missing function name, using \"\"",
1.60 schwarze 172: "empty head in list item",
173: "empty list item",
1.183 schwarze 174: "missing argument, using next line",
1.61 schwarze 175: "missing font type, using \\fR",
176: "unknown font type, using \\fR",
1.103 schwarze 177: "nothing follows prefix",
1.122 schwarze 178: "empty reference block",
1.155 schwarze 179: "missing section argument",
1.60 schwarze 180: "missing -std argument, adding it",
1.125 schwarze 181: "missing option string, using \"\"",
182: "missing resource identifier, using \"\"",
1.90 schwarze 183: "missing eqn box, using \"\"",
1.7 kristaps 184:
185: /* related to bad macro arguments */
1.64 schwarze 186: "unterminated quoted argument",
1.7 kristaps 187: "duplicate argument",
1.76 schwarze 188: "skipping duplicate argument",
1.63 schwarze 189: "skipping duplicate display type",
190: "skipping duplicate list type",
1.76 schwarze 191: "skipping -width argument",
1.126 schwarze 192: "wrong number of cells",
1.7 kristaps 193: "unknown AT&T UNIX version",
1.88 schwarze 194: "comma in function argument",
1.89 schwarze 195: "parenthesis in function name",
1.173 schwarze 196: "unknown library name",
1.67 schwarze 197: "invalid content in Rs block",
1.63 schwarze 198: "invalid Boolean argument",
199: "unknown font, skipping request",
1.126 schwarze 200: "odd number of characters in request",
1.7 kristaps 201:
202: /* related to plain text */
1.64 schwarze 203: "blank line in fill mode, using .sp",
204: "tab in filled text",
205: "whitespace at end of input line",
1.158 schwarze 206: "new sentence, new line",
1.7 kristaps 207: "bad comment style",
1.64 schwarze 208: "invalid escape sequence",
209: "undefined string, using \"\"",
1.16 kristaps 210:
1.118 schwarze 211: /* related to tables */
212: "tbl line starts with span",
213: "tbl column starts with span",
214: "skipping vertical bar in tbl layout",
215:
1.7 kristaps 216: "generic error",
217:
218: /* related to tables */
1.116 schwarze 219: "non-alphabetic character in tbl options",
220: "skipping unknown tbl option",
221: "missing tbl option argument",
222: "wrong tbl option argument size",
1.118 schwarze 223: "empty tbl layout",
224: "invalid character in tbl layout",
225: "unmatched parenthesis in tbl layout",
1.119 schwarze 226: "tbl without any data cells",
227: "ignoring data in spanned tbl cell",
228: "ignoring extra tbl data cells",
229: "data block open at end of tbl",
1.7 kristaps 230:
1.68 schwarze 231: /* related to document structure and macros */
1.108 schwarze 232: NULL,
1.7 kristaps 233: "input stack limit exceeded, infinite loop?",
234: "skipping bad character",
1.68 schwarze 235: "skipping unknown macro",
1.112 schwarze 236: "skipping insecure request",
1.73 schwarze 237: "skipping item outside list",
1.68 schwarze 238: "skipping column outside column list",
239: "skipping end of block that is not open",
1.115 schwarze 240: "fewer RS blocks open, skipping",
1.68 schwarze 241: "inserting missing end of block",
242: "appending missing end of block",
243:
244: /* related to request and macro arguments */
1.7 kristaps 245: "escaped character not allowed in a name",
1.99 schwarze 246: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 247: "skipping display without arguments",
1.71 schwarze 248: "missing list type, using -item",
1.171 schwarze 249: "argument is not numeric, using 1",
1.70 schwarze 250: "missing manual name, using \"\"",
1.71 schwarze 251: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 252: "unknown standard specifier",
1.71 schwarze 253: "skipping request without numeric argument",
1.109 schwarze 254: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 255: ".so request failed",
1.61 schwarze 256: "skipping all arguments",
257: "skipping excess arguments",
1.92 kristaps 258: "divide by zero",
1.112 schwarze 259:
260: "unsupported feature",
261: "input too large",
1.114 schwarze 262: "unsupported control character",
1.112 schwarze 263: "unsupported roff request",
1.119 schwarze 264: "eqn delim option in tbl",
1.118 schwarze 265: "unsupported tbl layout modifier",
1.112 schwarze 266: "ignoring macro in table",
1.7 kristaps 267: };
268:
269: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
270: "SUCCESS",
1.165 schwarze 271: "STYLE",
1.7 kristaps 272: "WARNING",
273: "ERROR",
1.112 schwarze 274: "UNSUPP",
1.7 kristaps 275: "BADARG",
276: "SYSERR"
277: };
278:
1.47 schwarze 279:
1.1 kristaps 280: static void
281: resize_buf(struct buf *buf, size_t initial)
282: {
283:
284: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
285: buf->buf = mandoc_realloc(buf->buf, buf->sz);
286: }
287:
288: static void
1.84 schwarze 289: choose_parser(struct mparse *curp)
1.1 kristaps 290: {
1.83 schwarze 291: char *cp, *ep;
292: int format;
1.1 kristaps 293:
1.83 schwarze 294: /*
295: * If neither command line arguments -mdoc or -man select
296: * a parser nor the roff parser found a .Dd or .TH macro
297: * yet, look ahead in the main input buffer.
298: */
299:
300: if ((format = roff_getformat(curp->roff)) == 0) {
301: cp = curp->primary->buf;
302: ep = cp + curp->primary->sz;
303: while (cp < ep) {
1.85 schwarze 304: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 305: cp++;
306: if (cp[0] == 'D' && cp[1] == 'd') {
307: format = MPARSE_MDOC;
308: break;
309: }
310: if (cp[0] == 'T' && cp[1] == 'H') {
311: format = MPARSE_MAN;
312: break;
313: }
314: }
315: cp = memchr(cp, '\n', ep - cp);
316: if (cp == NULL)
317: break;
318: cp++;
319: }
1.1 kristaps 320: }
321:
1.83 schwarze 322: if (format == MPARSE_MDOC) {
1.137 schwarze 323: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 324: if (curp->man->mdocmac == NULL)
325: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 326: } else {
327: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 328: if (curp->man->manmac == NULL)
329: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 330: }
1.164 schwarze 331: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 332: }
333:
334: /*
1.95 schwarze 335: * Main parse routine for a buffer.
336: * It assumes encoding and line numbering are already set up.
337: * It can recurse directly (for invocations of user-defined
338: * macros, inline equations, and input line traps)
339: * and indirectly (for .so file inclusion).
1.1 kristaps 340: */
1.163 schwarze 341: static int
1.95 schwarze 342: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 343: {
344: const struct tbl_span *span;
345: struct buf ln;
1.113 schwarze 346: const char *save_file;
1.110 schwarze 347: char *cp;
1.95 schwarze 348: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 349: enum rofferr rr;
1.100 schwarze 350: int of;
1.1 kristaps 351: int lnn; /* line number in the real file */
1.110 schwarze 352: int fd;
1.1 kristaps 353: unsigned char c;
354:
1.95 schwarze 355: memset(&ln, 0, sizeof(ln));
1.1 kristaps 356:
1.47 schwarze 357: lnn = curp->line;
358: pos = 0;
1.1 kristaps 359:
1.95 schwarze 360: while (i < blk.sz) {
1.1 kristaps 361: if (0 == pos && '\0' == blk.buf[i])
362: break;
363:
364: if (start) {
365: curp->line = lnn;
366: curp->reparse_count = 0;
1.93 schwarze 367:
368: if (lnn < 3 &&
369: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 370: curp->filenc & MPARSE_LATIN1)
371: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 372: }
373:
1.95 schwarze 374: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 375:
376: /*
377: * When finding an unescaped newline character,
378: * leave the character loop to process the line.
379: * Skip a preceding carriage return, if any.
380: */
381:
1.95 schwarze 382: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 383: '\n' == blk.buf[i + 1])
384: ++i;
385: if ('\n' == blk.buf[i]) {
386: ++i;
387: ++lnn;
388: break;
389: }
390:
1.35 schwarze 391: /*
1.93 schwarze 392: * Make sure we have space for the worst
393: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 394: */
395:
1.95 schwarze 396: if (pos + 11 > ln.sz)
1.35 schwarze 397: resize_buf(&ln, 256);
398:
1.47 schwarze 399: /*
1.93 schwarze 400: * Encode 8-bit input.
1.1 kristaps 401: */
402:
1.93 schwarze 403: c = blk.buf[i];
404: if (c & 0x80) {
1.95 schwarze 405: if ( ! (curp->filenc && preconv_encode(
406: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 407: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
408: curp->line, pos, "0x%x", c);
1.93 schwarze 409: ln.buf[pos++] = '?';
410: i++;
411: }
412: continue;
413: }
414:
415: /*
416: * Exclude control characters.
417: */
1.1 kristaps 418:
1.93 schwarze 419: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 420: mandoc_vmsg(c == 0x00 || c == 0x04 ||
421: c > 0x0a ? MANDOCERR_CHAR_BAD :
422: MANDOCERR_CHAR_UNSUPP,
423: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 424: i++;
1.127 schwarze 425: if (c != '\r')
426: ln.buf[pos++] = '?';
1.1 kristaps 427: continue;
428: }
429:
430: ln.buf[pos++] = blk.buf[i++];
431: }
432:
1.170 schwarze 433: if (pos + 1 >= ln.sz)
1.1 kristaps 434: resize_buf(&ln, 256);
435:
1.170 schwarze 436: if (i == blk.sz || blk.buf[i] == '\0')
437: ln.buf[pos++] = '\n';
1.1 kristaps 438: ln.buf[pos] = '\0';
439:
440: /*
441: * A significant amount of complexity is contained by
442: * the roff preprocessor. It's line-oriented but can be
443: * expressed on one line, so we need at times to
444: * readjust our starting point and re-run it. The roff
445: * preprocessor can also readjust the buffers with new
446: * data, so we pass them in wholesale.
447: */
448:
449: of = 0;
450:
1.24 kristaps 451: /*
452: * Maintain a lookaside buffer of all parsed lines. We
453: * only do this if mparse_keep() has been invoked (the
454: * buffer may be accessed with mparse_getkeep()).
455: */
456:
457: if (curp->secondary) {
1.47 schwarze 458: curp->secondary->buf = mandoc_realloc(
459: curp->secondary->buf,
460: curp->secondary->sz + pos + 2);
461: memcpy(curp->secondary->buf +
462: curp->secondary->sz,
463: ln.buf, pos);
1.24 kristaps 464: curp->secondary->sz += pos;
465: curp->secondary->buf
466: [curp->secondary->sz] = '\n';
467: curp->secondary->sz++;
468: curp->secondary->buf
469: [curp->secondary->sz] = '\0';
470: }
1.1 kristaps 471: rerun:
1.96 schwarze 472: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 473:
474: switch (rr) {
1.47 schwarze 475: case ROFF_REPARSE:
1.163 schwarze 476: if (++curp->reparse_count > REPARSE_LIMIT)
1.3 kristaps 477: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 478: curp->line, pos, NULL);
1.163 schwarze 479: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
480: start == 1) {
481: pos = 0;
482: continue;
483: }
484: free(ln.buf);
485: return 0;
1.47 schwarze 486: case ROFF_APPEND:
1.95 schwarze 487: pos = strlen(ln.buf);
1.1 kristaps 488: continue;
1.47 schwarze 489: case ROFF_RERUN:
1.1 kristaps 490: goto rerun;
1.47 schwarze 491: case ROFF_IGN:
1.1 kristaps 492: pos = 0;
493: continue;
1.47 schwarze 494: case ROFF_SO:
1.95 schwarze 495: if ( ! (curp->options & MPARSE_SO) &&
496: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 497: curp->sodest = mandoc_strdup(ln.buf + of);
498: free(ln.buf);
1.163 schwarze 499: return 1;
1.45 schwarze 500: }
1.24 kristaps 501: /*
502: * We remove `so' clauses from our lookaside
503: * buffer because we're going to descend into
504: * the file recursively.
505: */
1.47 schwarze 506: if (curp->secondary)
1.25 kristaps 507: curp->secondary->sz -= pos + 1;
1.113 schwarze 508: save_file = curp->file;
1.148 schwarze 509: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 510: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 511: close(fd);
1.113 schwarze 512: curp->file = save_file;
513: } else {
514: curp->file = save_file;
1.52 schwarze 515: mandoc_vmsg(MANDOCERR_SO_FAIL,
516: curp, curp->line, pos,
517: ".so %s", ln.buf + of);
1.110 schwarze 518: ln.sz = mandoc_asprintf(&cp,
519: ".sp\nSee the file %s.\n.sp",
520: ln.buf + of);
521: free(ln.buf);
522: ln.buf = cp;
523: of = 0;
524: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 525: }
1.1 kristaps 526: pos = 0;
527: continue;
528: default:
529: break;
530: }
531:
1.150 schwarze 532: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 533: choose_parser(curp);
1.1 kristaps 534:
1.47 schwarze 535: /*
1.84 schwarze 536: * Lastly, push down into the parsers themselves.
1.1 kristaps 537: * If libroff returns ROFF_TBL, then add it to the
538: * currently open parse. Since we only get here if
539: * there does exist data (see tbl_data.c), we're
540: * guaranteed that something's been allocated.
541: * Do the same for ROFF_EQN.
542: */
543:
1.139 schwarze 544: if (rr == ROFF_TBL)
1.100 schwarze 545: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 546: roff_addtbl(curp->man, span);
547: else if (rr == ROFF_EQN)
548: roff_addeqn(curp->man, roff_eqn(curp->roff));
549: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 550: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 551: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
552: break;
1.1 kristaps 553:
554: /* Temporary buffers typically are not full. */
555:
556: if (0 == start && '\0' == blk.buf[i])
557: break;
558:
559: /* Start the next input line. */
560:
561: pos = 0;
562: }
563:
564: free(ln.buf);
1.163 schwarze 565: return 1;
1.1 kristaps 566: }
567:
568: static int
1.40 schwarze 569: read_whole_file(struct mparse *curp, const char *file, int fd,
570: struct buf *fb, int *with_mmap)
1.1 kristaps 571: {
1.161 schwarze 572: struct stat st;
1.140 schwarze 573: gzFile gz;
1.1 kristaps 574: size_t off;
575: ssize_t ssz;
1.143 schwarze 576:
577: if (fstat(fd, &st) == -1)
578: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 579:
580: /*
581: * If we're a regular file, try just reading in the whole entry
582: * via mmap(). This is faster than reading it into blocks, and
583: * since each file is only a few bytes to begin with, I'm not
584: * concerned that this is going to tank any machines.
585: */
586:
1.140 schwarze 587: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 588: if (st.st_size > 0x7fffffff) {
1.111 schwarze 589: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 590: return 0;
1.1 kristaps 591: }
592: *with_mmap = 1;
593: fb->sz = (size_t)st.st_size;
1.37 schwarze 594: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 595: if (fb->buf != MAP_FAILED)
1.142 schwarze 596: return 1;
1.1 kristaps 597: }
598:
1.140 schwarze 599: if (curp->gzip) {
1.143 schwarze 600: if ((gz = gzdopen(fd, "rb")) == NULL)
601: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 602: } else
603: gz = NULL;
604:
1.1 kristaps 605: /*
606: * If this isn't a regular file (like, say, stdin), then we must
607: * go the old way and just read things in bit by bit.
608: */
609:
610: *with_mmap = 0;
611: off = 0;
612: fb->sz = 0;
613: fb->buf = NULL;
614: for (;;) {
615: if (off == fb->sz) {
616: if (fb->sz == (1U << 31)) {
1.111 schwarze 617: mandoc_msg(MANDOCERR_TOOLARGE, curp,
618: 0, 0, NULL);
1.1 kristaps 619: break;
620: }
621: resize_buf(fb, 65536);
622: }
1.140 schwarze 623: ssz = curp->gzip ?
624: gzread(gz, fb->buf + (int)off, fb->sz - off) :
625: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 626: if (ssz == 0) {
627: fb->sz = off;
1.142 schwarze 628: return 1;
1.1 kristaps 629: }
1.143 schwarze 630: if (ssz == -1)
631: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 632: off += (size_t)ssz;
633: }
634:
635: free(fb->buf);
636: fb->buf = NULL;
1.142 schwarze 637: return 0;
1.1 kristaps 638: }
639:
640: static void
641: mparse_end(struct mparse *curp)
642: {
1.135 schwarze 643: if (curp->man->macroset == MACROSET_NONE)
644: curp->man->macroset = MACROSET_MAN;
645: if (curp->man->macroset == MACROSET_MDOC)
646: mdoc_endparse(curp->man);
647: else
1.111 schwarze 648: man_endparse(curp->man);
1.1 kristaps 649: roff_endparse(curp->roff);
650: }
651:
652: static void
1.36 schwarze 653: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 654: {
1.85 schwarze 655: struct buf *svprimary;
1.28 joerg 656: const char *svfile;
1.95 schwarze 657: size_t offset;
1.36 schwarze 658: static int recursion_depth;
659:
660: if (64 < recursion_depth) {
661: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
662: return;
663: }
1.28 joerg 664:
665: /* Line number is per-file. */
666: svfile = curp->file;
667: curp->file = file;
1.85 schwarze 668: svprimary = curp->primary;
1.83 schwarze 669: curp->primary = &blk;
1.28 joerg 670: curp->line = 1;
1.36 schwarze 671: recursion_depth++;
1.28 joerg 672:
1.93 schwarze 673: /* Skip an UTF-8 byte order mark. */
674: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
675: (unsigned char)blk.buf[0] == 0xef &&
676: (unsigned char)blk.buf[1] == 0xbb &&
677: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 678: offset = 3;
1.93 schwarze 679: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 680: } else
681: offset = 0;
1.93 schwarze 682:
1.95 schwarze 683: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 684:
1.111 schwarze 685: if (--recursion_depth == 0)
1.28 joerg 686: mparse_end(curp);
687:
1.85 schwarze 688: curp->primary = svprimary;
1.28 joerg 689: curp->file = svfile;
690: }
691:
692: enum mandoclevel
1.104 schwarze 693: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 694: const char *file)
695: {
696: struct buf blk;
697:
1.104 schwarze 698: blk.buf = buf;
1.28 joerg 699: blk.sz = len;
700:
1.36 schwarze 701: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 702: return curp->file_status;
1.28 joerg 703: }
704:
1.98 schwarze 705: /*
706: * Read the whole file into memory and call the parsers.
707: * Called recursively when an .so request is encountered.
708: */
1.36 schwarze 709: enum mandoclevel
710: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 711: {
1.28 joerg 712: struct buf blk;
713: int with_mmap;
1.93 schwarze 714: int save_filenc;
1.1 kristaps 715:
1.91 schwarze 716: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 717: save_filenc = curp->filenc;
718: curp->filenc = curp->options &
719: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 720: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 721: curp->filenc = save_filenc;
1.91 schwarze 722: if (with_mmap)
723: munmap(blk.buf, blk.sz);
724: else
725: free(blk.buf);
726: }
1.142 schwarze 727: return curp->file_status;
1.82 schwarze 728: }
729:
1.148 schwarze 730: int
731: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 732: {
733: char *cp;
1.148 schwarze 734: int fd;
1.82 schwarze 735:
736: curp->file = file;
1.140 schwarze 737: cp = strrchr(file, '.');
738: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 739:
1.140 schwarze 740: /* First try to use the filename as it is. */
1.98 schwarze 741:
1.148 schwarze 742: if ((fd = open(file, O_RDONLY)) != -1)
743: return fd;
1.98 schwarze 744:
1.140 schwarze 745: /*
746: * If that doesn't work and the filename doesn't
747: * already end in .gz, try appending .gz.
748: */
1.98 schwarze 749:
1.140 schwarze 750: if ( ! curp->gzip) {
1.98 schwarze 751: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 752: fd = open(cp, O_RDONLY);
1.108 schwarze 753: free(cp);
1.148 schwarze 754: if (fd != -1) {
1.140 schwarze 755: curp->gzip = 1;
1.148 schwarze 756: return fd;
1.82 schwarze 757: }
758: }
759:
1.140 schwarze 760: /* Neither worked, give up. */
1.97 schwarze 761:
1.140 schwarze 762: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 763: return -1;
1.1 kristaps 764: }
765:
766: struct mparse *
1.179 schwarze 767: mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
768: enum mandoc_os os_e, const char *os_s)
1.1 kristaps 769: {
770: struct mparse *curp;
1.10 kristaps 771:
1.1 kristaps 772: curp = mandoc_calloc(1, sizeof(struct mparse));
773:
1.44 schwarze 774: curp->options = options;
1.179 schwarze 775: curp->mmin = mmin;
1.1 kristaps 776: curp->mmsg = mmsg;
1.179 schwarze 777: curp->os_s = os_s;
1.1 kristaps 778:
1.144 schwarze 779: curp->roff = roff_alloc(curp, options);
1.179 schwarze 780: curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
1.137 schwarze 781: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 782: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 783: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 784: if (curp->man->mdocmac == NULL)
785: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 786: } else if (curp->options & MPARSE_MAN) {
787: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 788: if (curp->man->manmac == NULL)
789: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 790: }
1.138 schwarze 791: curp->man->first->tok = TOKEN_NONE;
1.179 schwarze 792: curp->man->meta.os_e = os_e;
1.142 schwarze 793: return curp;
1.1 kristaps 794: }
795:
796: void
797: mparse_reset(struct mparse *curp)
798: {
799: roff_reset(curp->roff);
1.150 schwarze 800: roff_man_reset(curp->man);
1.160 schwarze 801:
802: free(curp->sodest);
803: curp->sodest = NULL;
804:
1.24 kristaps 805: if (curp->secondary)
806: curp->secondary->sz = 0;
1.1 kristaps 807:
808: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 809: curp->gzip = 0;
1.1 kristaps 810: }
811:
812: void
813: mparse_free(struct mparse *curp)
814: {
815:
1.164 schwarze 816: roffhash_free(curp->man->mdocmac);
817: roffhash_free(curp->man->manmac);
1.137 schwarze 818: roff_man_free(curp->man);
1.160 schwarze 819: roff_free(curp->roff);
1.24 kristaps 820: if (curp->secondary)
821: free(curp->secondary->buf);
1.1 kristaps 822:
1.24 kristaps 823: free(curp->secondary);
1.45 schwarze 824: free(curp->sodest);
1.1 kristaps 825: free(curp);
826: }
827:
828: void
1.135 schwarze 829: mparse_result(struct mparse *curp, struct roff_man **man,
830: char **sodest)
1.1 kristaps 831: {
832:
1.45 schwarze 833: if (sodest && NULL != (*sodest = curp->sodest)) {
834: *man = NULL;
835: return;
836: }
1.9 kristaps 837: if (man)
838: *man = curp->man;
1.157 schwarze 839: }
840:
841: void
842: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
843: {
844: if (curp->file_status > *rc)
845: *rc = curp->file_status;
1.3 kristaps 846: }
847:
848: void
849: mandoc_vmsg(enum mandocerr t, struct mparse *m,
850: int ln, int pos, const char *fmt, ...)
851: {
852: char buf[256];
853: va_list ap;
854:
855: va_start(ap, fmt);
1.48 schwarze 856: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 857: va_end(ap);
858:
859: mandoc_msg(t, m, ln, pos, buf);
860: }
861:
862: void
1.47 schwarze 863: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 864: int ln, int col, const char *msg)
865: {
866: enum mandoclevel level;
867:
1.179 schwarze 868: if (er < m->mmin && er != MANDOCERR_FILE)
869: return;
870:
1.112 schwarze 871: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 872: while (er < mandoclimits[level])
873: level--;
874:
1.8 kristaps 875: if (m->mmsg)
876: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 877:
878: if (m->file_status < level)
879: m->file_status = level;
1.7 kristaps 880: }
881:
882: const char *
883: mparse_strerror(enum mandocerr er)
884: {
885:
1.142 schwarze 886: return mandocerrs[er];
1.7 kristaps 887: }
888:
889: const char *
890: mparse_strlevel(enum mandoclevel lvl)
891: {
1.142 schwarze 892: return mandoclevels[lvl];
1.24 kristaps 893: }
894:
895: void
896: mparse_keep(struct mparse *p)
897: {
898:
899: assert(NULL == p->secondary);
900: p->secondary = mandoc_calloc(1, sizeof(struct buf));
901: }
902:
903: const char *
904: mparse_getkeep(const struct mparse *p)
905: {
906:
907: assert(p->secondary);
1.142 schwarze 908: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 909: }
CVSweb