Annotation of mandoc/read.c, Revision 1.190
1.190 ! schwarze 1: /* $Id: read.c,v 1.189 2017/07/06 22:59:48 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,
1.188 schwarze 78: MANDOCERR_OK,
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",
1.189 schwarze 101: "lower case character in document title",
1.178 schwarze 102: "duplicate RCS id",
1.182 schwarze 103: "typo in section name",
1.189 schwarze 104: "unterminated quoted argument",
1.166 schwarze 105: "useless macro",
1.167 schwarze 106: "consider using OS macro",
1.172 schwarze 107: "errnos out of order",
108: "duplicate errno",
1.187 schwarze 109: "trailing delimiter",
1.174 schwarze 110: "no blank before trailing delimiter",
1.189 schwarze 111: "fill mode already enabled, skipping",
112: "fill mode already disabled, skipping",
1.175 schwarze 113: "function name without markup",
1.189 schwarze 114: "whitespace at end of input line",
115: "bad comment style",
1.166 schwarze 116:
1.7 kristaps 117: "generic warning",
118:
119: /* related to the prologue */
1.79 schwarze 120: "missing manual title, using UNTITLED",
121: "missing manual title, using \"\"",
122: "missing manual section, using \"\"",
1.7 kristaps 123: "unknown manual section",
1.54 schwarze 124: "missing date, using today's date",
1.7 kristaps 125: "cannot parse date, using it verbatim",
1.186 schwarze 126: "date in the future, using it anyway",
1.79 schwarze 127: "missing Os macro, using \"\"",
128: "late prologue macro",
1.7 kristaps 129: "prologue macros out of order",
130:
131: /* related to document structure */
132: ".so is fragile, better use ln(1)",
1.50 schwarze 133: "no document body",
1.54 schwarze 134: "content before first section header",
135: "first section is not \"NAME\"",
1.156 schwarze 136: "NAME section without Nm before Nd",
1.128 schwarze 137: "NAME section without description",
138: "description not at the end of NAME",
139: "bad NAME section content",
1.156 schwarze 140: "missing comma before name",
1.121 schwarze 141: "missing description line, using \"\"",
1.162 schwarze 142: "description line outside NAME section",
1.7 kristaps 143: "sections out of conventional order",
1.54 schwarze 144: "duplicate section title",
145: "unexpected section",
1.185 schwarze 146: "cross reference to self",
1.87 schwarze 147: "unusual Xr order",
148: "unusual Xr punctuation",
1.86 schwarze 149: "AUTHORS section without An macro",
1.7 kristaps 150:
151: /* related to macros and nesting */
1.55 schwarze 152: "obsolete macro",
1.102 schwarze 153: "macro neither callable nor escaped",
1.7 kristaps 154: "skipping paragraph macro",
1.31 schwarze 155: "moving paragraph macro out of list",
1.7 kristaps 156: "skipping no-space macro",
157: "blocks badly nested",
158: "nested displays are not portable",
1.57 schwarze 159: "moving content out of list",
1.183 schwarze 160: "first macro on line",
1.7 kristaps 161: "line scope broken",
1.169 schwarze 162: "skipping blank line in line scope",
1.7 kristaps 163:
164: /* related to missing macro arguments */
1.58 schwarze 165: "skipping empty request",
166: "conditional request controls empty scope",
1.7 kristaps 167: "skipping empty macro",
1.124 schwarze 168: "empty block",
1.62 schwarze 169: "empty argument, using 0n",
1.60 schwarze 170: "missing display type, using -ragged",
171: "list type is not the first argument",
1.152 schwarze 172: "missing -width in -tag list, using 6n",
1.78 schwarze 173: "missing utility name, using \"\"",
1.123 schwarze 174: "missing function name, using \"\"",
1.60 schwarze 175: "empty head in list item",
176: "empty list item",
1.183 schwarze 177: "missing argument, using next line",
1.61 schwarze 178: "missing font type, using \\fR",
179: "unknown font type, using \\fR",
1.103 schwarze 180: "nothing follows prefix",
1.122 schwarze 181: "empty reference block",
1.155 schwarze 182: "missing section argument",
1.60 schwarze 183: "missing -std argument, adding it",
1.125 schwarze 184: "missing option string, using \"\"",
185: "missing resource identifier, using \"\"",
1.90 schwarze 186: "missing eqn box, using \"\"",
1.7 kristaps 187:
188: /* related to bad macro arguments */
189: "duplicate argument",
1.76 schwarze 190: "skipping duplicate argument",
1.63 schwarze 191: "skipping duplicate display type",
192: "skipping duplicate list type",
1.76 schwarze 193: "skipping -width argument",
1.126 schwarze 194: "wrong number of cells",
1.7 kristaps 195: "unknown AT&T UNIX version",
1.88 schwarze 196: "comma in function argument",
1.89 schwarze 197: "parenthesis in function name",
1.173 schwarze 198: "unknown library name",
1.67 schwarze 199: "invalid content in Rs block",
1.63 schwarze 200: "invalid Boolean argument",
201: "unknown font, skipping request",
1.126 schwarze 202: "odd number of characters in request",
1.7 kristaps 203:
204: /* related to plain text */
1.64 schwarze 205: "blank line in fill mode, using .sp",
206: "tab in filled text",
1.158 schwarze 207: "new sentence, new line",
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,
1.189 schwarze 233: "duplicate prologue macro",
234: "skipping late title macro",
1.7 kristaps 235: "input stack limit exceeded, infinite loop?",
236: "skipping bad character",
1.68 schwarze 237: "skipping unknown macro",
1.112 schwarze 238: "skipping insecure request",
1.73 schwarze 239: "skipping item outside list",
1.68 schwarze 240: "skipping column outside column list",
241: "skipping end of block that is not open",
1.115 schwarze 242: "fewer RS blocks open, skipping",
1.68 schwarze 243: "inserting missing end of block",
244: "appending missing end of block",
245:
246: /* related to request and macro arguments */
1.7 kristaps 247: "escaped character not allowed in a name",
1.99 schwarze 248: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 249: "skipping display without arguments",
1.71 schwarze 250: "missing list type, using -item",
1.171 schwarze 251: "argument is not numeric, using 1",
1.70 schwarze 252: "missing manual name, using \"\"",
1.71 schwarze 253: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 254: "unknown standard specifier",
1.71 schwarze 255: "skipping request without numeric argument",
1.109 schwarze 256: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 257: ".so request failed",
1.61 schwarze 258: "skipping all arguments",
259: "skipping excess arguments",
1.92 kristaps 260: "divide by zero",
1.112 schwarze 261:
262: "unsupported feature",
263: "input too large",
1.114 schwarze 264: "unsupported control character",
1.112 schwarze 265: "unsupported roff request",
1.119 schwarze 266: "eqn delim option in tbl",
1.118 schwarze 267: "unsupported tbl layout modifier",
1.112 schwarze 268: "ignoring macro in table",
1.7 kristaps 269: };
270:
271: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
272: "SUCCESS",
1.165 schwarze 273: "STYLE",
1.7 kristaps 274: "WARNING",
275: "ERROR",
1.112 schwarze 276: "UNSUPP",
1.7 kristaps 277: "BADARG",
278: "SYSERR"
279: };
280:
1.47 schwarze 281:
1.1 kristaps 282: static void
283: resize_buf(struct buf *buf, size_t initial)
284: {
285:
286: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
287: buf->buf = mandoc_realloc(buf->buf, buf->sz);
288: }
289:
290: static void
1.84 schwarze 291: choose_parser(struct mparse *curp)
1.1 kristaps 292: {
1.83 schwarze 293: char *cp, *ep;
294: int format;
1.1 kristaps 295:
1.83 schwarze 296: /*
297: * If neither command line arguments -mdoc or -man select
298: * a parser nor the roff parser found a .Dd or .TH macro
299: * yet, look ahead in the main input buffer.
300: */
301:
302: if ((format = roff_getformat(curp->roff)) == 0) {
303: cp = curp->primary->buf;
304: ep = cp + curp->primary->sz;
305: while (cp < ep) {
1.85 schwarze 306: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 307: cp++;
308: if (cp[0] == 'D' && cp[1] == 'd') {
309: format = MPARSE_MDOC;
310: break;
311: }
312: if (cp[0] == 'T' && cp[1] == 'H') {
313: format = MPARSE_MAN;
314: break;
315: }
316: }
317: cp = memchr(cp, '\n', ep - cp);
318: if (cp == NULL)
319: break;
320: cp++;
321: }
1.1 kristaps 322: }
323:
1.83 schwarze 324: if (format == MPARSE_MDOC) {
1.137 schwarze 325: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 326: if (curp->man->mdocmac == NULL)
327: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 328: } else {
329: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 330: if (curp->man->manmac == NULL)
331: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 332: }
1.164 schwarze 333: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 334: }
335:
336: /*
1.95 schwarze 337: * Main parse routine for a buffer.
338: * It assumes encoding and line numbering are already set up.
339: * It can recurse directly (for invocations of user-defined
340: * macros, inline equations, and input line traps)
341: * and indirectly (for .so file inclusion).
1.1 kristaps 342: */
1.163 schwarze 343: static int
1.95 schwarze 344: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 345: {
346: const struct tbl_span *span;
347: struct buf ln;
1.113 schwarze 348: const char *save_file;
1.110 schwarze 349: char *cp;
1.95 schwarze 350: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 351: enum rofferr rr;
1.100 schwarze 352: int of;
1.1 kristaps 353: int lnn; /* line number in the real file */
1.110 schwarze 354: int fd;
1.1 kristaps 355: unsigned char c;
356:
1.95 schwarze 357: memset(&ln, 0, sizeof(ln));
1.1 kristaps 358:
1.47 schwarze 359: lnn = curp->line;
360: pos = 0;
1.1 kristaps 361:
1.95 schwarze 362: while (i < blk.sz) {
1.1 kristaps 363: if (0 == pos && '\0' == blk.buf[i])
364: break;
365:
366: if (start) {
367: curp->line = lnn;
368: curp->reparse_count = 0;
1.93 schwarze 369:
370: if (lnn < 3 &&
371: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 372: curp->filenc & MPARSE_LATIN1)
373: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 374: }
375:
1.95 schwarze 376: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 377:
378: /*
379: * When finding an unescaped newline character,
380: * leave the character loop to process the line.
381: * Skip a preceding carriage return, if any.
382: */
383:
1.95 schwarze 384: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 385: '\n' == blk.buf[i + 1])
386: ++i;
387: if ('\n' == blk.buf[i]) {
388: ++i;
389: ++lnn;
390: break;
391: }
392:
1.35 schwarze 393: /*
1.93 schwarze 394: * Make sure we have space for the worst
395: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 396: */
397:
1.95 schwarze 398: if (pos + 11 > ln.sz)
1.35 schwarze 399: resize_buf(&ln, 256);
400:
1.47 schwarze 401: /*
1.93 schwarze 402: * Encode 8-bit input.
1.1 kristaps 403: */
404:
1.93 schwarze 405: c = blk.buf[i];
406: if (c & 0x80) {
1.95 schwarze 407: if ( ! (curp->filenc && preconv_encode(
408: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 409: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
410: curp->line, pos, "0x%x", c);
1.93 schwarze 411: ln.buf[pos++] = '?';
412: i++;
413: }
414: continue;
415: }
416:
417: /*
418: * Exclude control characters.
419: */
1.1 kristaps 420:
1.93 schwarze 421: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 422: mandoc_vmsg(c == 0x00 || c == 0x04 ||
423: c > 0x0a ? MANDOCERR_CHAR_BAD :
424: MANDOCERR_CHAR_UNSUPP,
425: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 426: i++;
1.127 schwarze 427: if (c != '\r')
428: ln.buf[pos++] = '?';
1.1 kristaps 429: continue;
430: }
431:
432: ln.buf[pos++] = blk.buf[i++];
433: }
434:
1.170 schwarze 435: if (pos + 1 >= ln.sz)
1.1 kristaps 436: resize_buf(&ln, 256);
437:
1.170 schwarze 438: if (i == blk.sz || blk.buf[i] == '\0')
439: ln.buf[pos++] = '\n';
1.1 kristaps 440: ln.buf[pos] = '\0';
441:
442: /*
443: * A significant amount of complexity is contained by
444: * the roff preprocessor. It's line-oriented but can be
445: * expressed on one line, so we need at times to
446: * readjust our starting point and re-run it. The roff
447: * preprocessor can also readjust the buffers with new
448: * data, so we pass them in wholesale.
449: */
450:
451: of = 0;
452:
1.24 kristaps 453: /*
454: * Maintain a lookaside buffer of all parsed lines. We
455: * only do this if mparse_keep() has been invoked (the
456: * buffer may be accessed with mparse_getkeep()).
457: */
458:
459: if (curp->secondary) {
1.47 schwarze 460: curp->secondary->buf = mandoc_realloc(
461: curp->secondary->buf,
462: curp->secondary->sz + pos + 2);
463: memcpy(curp->secondary->buf +
464: curp->secondary->sz,
465: ln.buf, pos);
1.24 kristaps 466: curp->secondary->sz += pos;
467: curp->secondary->buf
468: [curp->secondary->sz] = '\n';
469: curp->secondary->sz++;
470: curp->secondary->buf
471: [curp->secondary->sz] = '\0';
472: }
1.1 kristaps 473: rerun:
1.96 schwarze 474: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 475:
476: switch (rr) {
1.47 schwarze 477: case ROFF_REPARSE:
1.163 schwarze 478: if (++curp->reparse_count > REPARSE_LIMIT)
1.3 kristaps 479: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 480: curp->line, pos, NULL);
1.163 schwarze 481: else if (mparse_buf_r(curp, ln, of, 0) == 1 ||
482: start == 1) {
483: pos = 0;
484: continue;
485: }
486: free(ln.buf);
487: return 0;
1.47 schwarze 488: case ROFF_APPEND:
1.95 schwarze 489: pos = strlen(ln.buf);
1.1 kristaps 490: continue;
1.47 schwarze 491: case ROFF_RERUN:
1.1 kristaps 492: goto rerun;
1.47 schwarze 493: case ROFF_IGN:
1.1 kristaps 494: pos = 0;
495: continue;
1.47 schwarze 496: case ROFF_SO:
1.95 schwarze 497: if ( ! (curp->options & MPARSE_SO) &&
498: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 499: curp->sodest = mandoc_strdup(ln.buf + of);
500: free(ln.buf);
1.163 schwarze 501: return 1;
1.45 schwarze 502: }
1.24 kristaps 503: /*
504: * We remove `so' clauses from our lookaside
505: * buffer because we're going to descend into
506: * the file recursively.
507: */
1.47 schwarze 508: if (curp->secondary)
1.25 kristaps 509: curp->secondary->sz -= pos + 1;
1.113 schwarze 510: save_file = curp->file;
1.148 schwarze 511: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 512: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 513: close(fd);
1.113 schwarze 514: curp->file = save_file;
515: } else {
516: curp->file = save_file;
1.52 schwarze 517: mandoc_vmsg(MANDOCERR_SO_FAIL,
518: curp, curp->line, pos,
519: ".so %s", ln.buf + of);
1.110 schwarze 520: ln.sz = mandoc_asprintf(&cp,
521: ".sp\nSee the file %s.\n.sp",
522: ln.buf + of);
523: free(ln.buf);
524: ln.buf = cp;
525: of = 0;
526: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 527: }
1.1 kristaps 528: pos = 0;
529: continue;
530: default:
531: break;
532: }
533:
1.150 schwarze 534: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 535: choose_parser(curp);
1.1 kristaps 536:
1.47 schwarze 537: /*
1.84 schwarze 538: * Lastly, push down into the parsers themselves.
1.1 kristaps 539: * If libroff returns ROFF_TBL, then add it to the
540: * currently open parse. Since we only get here if
541: * there does exist data (see tbl_data.c), we're
542: * guaranteed that something's been allocated.
543: */
544:
1.139 schwarze 545: if (rr == ROFF_TBL)
1.100 schwarze 546: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 547: roff_addtbl(curp->man, span);
548: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 549: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 550: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
551: break;
1.1 kristaps 552:
553: /* Temporary buffers typically are not full. */
554:
555: if (0 == start && '\0' == blk.buf[i])
556: break;
557:
558: /* Start the next input line. */
559:
560: pos = 0;
561: }
562:
563: free(ln.buf);
1.163 schwarze 564: return 1;
1.1 kristaps 565: }
566:
567: static int
1.40 schwarze 568: read_whole_file(struct mparse *curp, const char *file, int fd,
569: struct buf *fb, int *with_mmap)
1.1 kristaps 570: {
1.161 schwarze 571: struct stat st;
1.140 schwarze 572: gzFile gz;
1.1 kristaps 573: size_t off;
574: ssize_t ssz;
1.143 schwarze 575:
576: if (fstat(fd, &st) == -1)
577: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 578:
579: /*
580: * If we're a regular file, try just reading in the whole entry
581: * via mmap(). This is faster than reading it into blocks, and
582: * since each file is only a few bytes to begin with, I'm not
583: * concerned that this is going to tank any machines.
584: */
585:
1.140 schwarze 586: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 587: if (st.st_size > 0x7fffffff) {
1.111 schwarze 588: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 589: return 0;
1.1 kristaps 590: }
591: *with_mmap = 1;
592: fb->sz = (size_t)st.st_size;
1.37 schwarze 593: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 594: if (fb->buf != MAP_FAILED)
1.142 schwarze 595: return 1;
1.1 kristaps 596: }
597:
1.140 schwarze 598: if (curp->gzip) {
1.143 schwarze 599: if ((gz = gzdopen(fd, "rb")) == NULL)
600: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 601: } else
602: gz = NULL;
603:
1.1 kristaps 604: /*
605: * If this isn't a regular file (like, say, stdin), then we must
606: * go the old way and just read things in bit by bit.
607: */
608:
609: *with_mmap = 0;
610: off = 0;
611: fb->sz = 0;
612: fb->buf = NULL;
613: for (;;) {
614: if (off == fb->sz) {
615: if (fb->sz == (1U << 31)) {
1.111 schwarze 616: mandoc_msg(MANDOCERR_TOOLARGE, curp,
617: 0, 0, NULL);
1.1 kristaps 618: break;
619: }
620: resize_buf(fb, 65536);
621: }
1.140 schwarze 622: ssz = curp->gzip ?
623: gzread(gz, fb->buf + (int)off, fb->sz - off) :
624: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 625: if (ssz == 0) {
626: fb->sz = off;
1.142 schwarze 627: return 1;
1.1 kristaps 628: }
1.143 schwarze 629: if (ssz == -1)
630: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 631: off += (size_t)ssz;
632: }
633:
634: free(fb->buf);
635: fb->buf = NULL;
1.142 schwarze 636: return 0;
1.1 kristaps 637: }
638:
639: static void
640: mparse_end(struct mparse *curp)
641: {
1.135 schwarze 642: if (curp->man->macroset == MACROSET_NONE)
643: curp->man->macroset = MACROSET_MAN;
644: if (curp->man->macroset == MACROSET_MDOC)
645: mdoc_endparse(curp->man);
646: else
1.111 schwarze 647: man_endparse(curp->man);
1.1 kristaps 648: roff_endparse(curp->roff);
649: }
650:
651: static void
1.36 schwarze 652: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 653: {
1.85 schwarze 654: struct buf *svprimary;
1.28 joerg 655: const char *svfile;
1.95 schwarze 656: size_t offset;
1.36 schwarze 657: static int recursion_depth;
658:
659: if (64 < recursion_depth) {
660: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
661: return;
662: }
1.28 joerg 663:
664: /* Line number is per-file. */
665: svfile = curp->file;
666: curp->file = file;
1.85 schwarze 667: svprimary = curp->primary;
1.83 schwarze 668: curp->primary = &blk;
1.28 joerg 669: curp->line = 1;
1.36 schwarze 670: recursion_depth++;
1.28 joerg 671:
1.93 schwarze 672: /* Skip an UTF-8 byte order mark. */
673: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
674: (unsigned char)blk.buf[0] == 0xef &&
675: (unsigned char)blk.buf[1] == 0xbb &&
676: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 677: offset = 3;
1.93 schwarze 678: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 679: } else
680: offset = 0;
1.93 schwarze 681:
1.95 schwarze 682: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 683:
1.111 schwarze 684: if (--recursion_depth == 0)
1.28 joerg 685: mparse_end(curp);
686:
1.85 schwarze 687: curp->primary = svprimary;
1.28 joerg 688: curp->file = svfile;
689: }
690:
691: enum mandoclevel
1.104 schwarze 692: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 693: const char *file)
694: {
695: struct buf blk;
696:
1.104 schwarze 697: blk.buf = buf;
1.28 joerg 698: blk.sz = len;
699:
1.36 schwarze 700: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 701: return curp->file_status;
1.28 joerg 702: }
703:
1.98 schwarze 704: /*
705: * Read the whole file into memory and call the parsers.
706: * Called recursively when an .so request is encountered.
707: */
1.36 schwarze 708: enum mandoclevel
709: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 710: {
1.28 joerg 711: struct buf blk;
712: int with_mmap;
1.93 schwarze 713: int save_filenc;
1.1 kristaps 714:
1.91 schwarze 715: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 716: save_filenc = curp->filenc;
717: curp->filenc = curp->options &
718: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 719: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 720: curp->filenc = save_filenc;
1.91 schwarze 721: if (with_mmap)
722: munmap(blk.buf, blk.sz);
723: else
724: free(blk.buf);
725: }
1.142 schwarze 726: return curp->file_status;
1.82 schwarze 727: }
728:
1.148 schwarze 729: int
730: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 731: {
732: char *cp;
1.148 schwarze 733: int fd;
1.82 schwarze 734:
735: curp->file = file;
1.140 schwarze 736: cp = strrchr(file, '.');
737: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 738:
1.140 schwarze 739: /* First try to use the filename as it is. */
1.98 schwarze 740:
1.148 schwarze 741: if ((fd = open(file, O_RDONLY)) != -1)
742: return fd;
1.98 schwarze 743:
1.140 schwarze 744: /*
745: * If that doesn't work and the filename doesn't
746: * already end in .gz, try appending .gz.
747: */
1.98 schwarze 748:
1.140 schwarze 749: if ( ! curp->gzip) {
1.98 schwarze 750: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 751: fd = open(cp, O_RDONLY);
1.108 schwarze 752: free(cp);
1.148 schwarze 753: if (fd != -1) {
1.140 schwarze 754: curp->gzip = 1;
1.148 schwarze 755: return fd;
1.82 schwarze 756: }
757: }
758:
1.140 schwarze 759: /* Neither worked, give up. */
1.97 schwarze 760:
1.140 schwarze 761: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 762: return -1;
1.1 kristaps 763: }
764:
765: struct mparse *
1.179 schwarze 766: mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
767: enum mandoc_os os_e, const char *os_s)
1.1 kristaps 768: {
769: struct mparse *curp;
1.10 kristaps 770:
1.1 kristaps 771: curp = mandoc_calloc(1, sizeof(struct mparse));
772:
1.44 schwarze 773: curp->options = options;
1.179 schwarze 774: curp->mmin = mmin;
1.1 kristaps 775: curp->mmsg = mmsg;
1.179 schwarze 776: curp->os_s = os_s;
1.1 kristaps 777:
1.144 schwarze 778: curp->roff = roff_alloc(curp, options);
1.179 schwarze 779: curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
1.137 schwarze 780: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 781: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 782: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 783: if (curp->man->mdocmac == NULL)
784: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 785: } else if (curp->options & MPARSE_MAN) {
786: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 787: if (curp->man->manmac == NULL)
788: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 789: }
1.138 schwarze 790: curp->man->first->tok = TOKEN_NONE;
1.179 schwarze 791: curp->man->meta.os_e = os_e;
1.142 schwarze 792: return curp;
1.1 kristaps 793: }
794:
795: void
796: mparse_reset(struct mparse *curp)
797: {
798: roff_reset(curp->roff);
1.150 schwarze 799: roff_man_reset(curp->man);
1.160 schwarze 800:
801: free(curp->sodest);
802: curp->sodest = NULL;
803:
1.24 kristaps 804: if (curp->secondary)
805: curp->secondary->sz = 0;
1.1 kristaps 806:
807: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 808: curp->gzip = 0;
1.1 kristaps 809: }
810:
811: void
812: mparse_free(struct mparse *curp)
813: {
814:
1.164 schwarze 815: roffhash_free(curp->man->mdocmac);
816: roffhash_free(curp->man->manmac);
1.137 schwarze 817: roff_man_free(curp->man);
1.160 schwarze 818: roff_free(curp->roff);
1.24 kristaps 819: if (curp->secondary)
820: free(curp->secondary->buf);
1.1 kristaps 821:
1.24 kristaps 822: free(curp->secondary);
1.45 schwarze 823: free(curp->sodest);
1.1 kristaps 824: free(curp);
825: }
826:
827: void
1.135 schwarze 828: mparse_result(struct mparse *curp, struct roff_man **man,
829: char **sodest)
1.1 kristaps 830: {
831:
1.45 schwarze 832: if (sodest && NULL != (*sodest = curp->sodest)) {
833: *man = NULL;
834: return;
835: }
1.9 kristaps 836: if (man)
837: *man = curp->man;
1.157 schwarze 838: }
839:
840: void
841: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
842: {
843: if (curp->file_status > *rc)
844: *rc = curp->file_status;
1.3 kristaps 845: }
846:
847: void
848: mandoc_vmsg(enum mandocerr t, struct mparse *m,
849: int ln, int pos, const char *fmt, ...)
850: {
851: char buf[256];
852: va_list ap;
853:
854: va_start(ap, fmt);
1.48 schwarze 855: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 856: va_end(ap);
857:
858: mandoc_msg(t, m, ln, pos, buf);
859: }
860:
861: void
1.47 schwarze 862: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 863: int ln, int col, const char *msg)
864: {
865: enum mandoclevel level;
866:
1.179 schwarze 867: if (er < m->mmin && er != MANDOCERR_FILE)
868: return;
869:
1.112 schwarze 870: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 871: while (er < mandoclimits[level])
872: level--;
873:
1.8 kristaps 874: if (m->mmsg)
875: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 876:
877: if (m->file_status < level)
878: m->file_status = level;
1.7 kristaps 879: }
880:
881: const char *
882: mparse_strerror(enum mandocerr er)
883: {
884:
1.142 schwarze 885: return mandocerrs[er];
1.7 kristaps 886: }
887:
888: const char *
889: mparse_strlevel(enum mandoclevel lvl)
890: {
1.142 schwarze 891: return mandoclevels[lvl];
1.24 kristaps 892: }
893:
894: void
895: mparse_keep(struct mparse *p)
896: {
897:
898: assert(NULL == p->secondary);
899: p->secondary = mandoc_calloc(1, sizeof(struct buf));
900: }
901:
902: const char *
903: mparse_getkeep(const struct mparse *p)
904: {
905:
906: assert(p->secondary);
1.142 schwarze 907: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 908: }
CVSweb