Annotation of mandoc/read.c, Revision 1.187
1.187 ! schwarze 1: /* $Id: read.c,v 1.186 2017/07/03 13:40:19 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",
1.187 ! schwarze 107: "trailing delimiter",
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",
1.186 schwarze 121: "date in the future, using it anyway",
1.79 schwarze 122: "missing Os macro, using \"\"",
123: "duplicate prologue macro",
124: "late prologue macro",
125: "skipping late title macro",
1.7 kristaps 126: "prologue macros out of order",
127:
128: /* related to document structure */
129: ".so is fragile, better use ln(1)",
1.50 schwarze 130: "no document body",
1.54 schwarze 131: "content before first section header",
132: "first section is not \"NAME\"",
1.156 schwarze 133: "NAME section without Nm before Nd",
1.128 schwarze 134: "NAME section without description",
135: "description not at the end of NAME",
136: "bad NAME section content",
1.156 schwarze 137: "missing comma before name",
1.121 schwarze 138: "missing description line, using \"\"",
1.162 schwarze 139: "description line outside NAME section",
1.7 kristaps 140: "sections out of conventional order",
1.54 schwarze 141: "duplicate section title",
142: "unexpected section",
1.185 schwarze 143: "cross reference to self",
1.87 schwarze 144: "unusual Xr order",
145: "unusual Xr punctuation",
1.86 schwarze 146: "AUTHORS section without An macro",
1.7 kristaps 147:
148: /* related to macros and nesting */
1.55 schwarze 149: "obsolete macro",
1.102 schwarze 150: "macro neither callable nor escaped",
1.7 kristaps 151: "skipping paragraph macro",
1.31 schwarze 152: "moving paragraph macro out of list",
1.7 kristaps 153: "skipping no-space macro",
154: "blocks badly nested",
155: "nested displays are not portable",
1.57 schwarze 156: "moving content out of list",
1.183 schwarze 157: "first macro on line",
1.78 schwarze 158: "fill mode already enabled, skipping",
159: "fill mode already disabled, skipping",
1.7 kristaps 160: "line scope broken",
1.169 schwarze 161: "skipping blank line in line scope",
1.7 kristaps 162:
163: /* related to missing macro arguments */
1.58 schwarze 164: "skipping empty request",
165: "conditional request controls empty scope",
1.7 kristaps 166: "skipping empty macro",
1.124 schwarze 167: "empty block",
1.62 schwarze 168: "empty argument, using 0n",
1.60 schwarze 169: "missing display type, using -ragged",
170: "list type is not the first argument",
1.152 schwarze 171: "missing -width in -tag list, using 6n",
1.78 schwarze 172: "missing utility name, using \"\"",
1.123 schwarze 173: "missing function name, using \"\"",
1.60 schwarze 174: "empty head in list item",
175: "empty list item",
1.183 schwarze 176: "missing argument, using next line",
1.61 schwarze 177: "missing font type, using \\fR",
178: "unknown font type, using \\fR",
1.103 schwarze 179: "nothing follows prefix",
1.122 schwarze 180: "empty reference block",
1.155 schwarze 181: "missing section argument",
1.60 schwarze 182: "missing -std argument, adding it",
1.125 schwarze 183: "missing option string, using \"\"",
184: "missing resource identifier, using \"\"",
1.90 schwarze 185: "missing eqn box, using \"\"",
1.7 kristaps 186:
187: /* related to bad macro arguments */
1.64 schwarze 188: "unterminated quoted argument",
1.7 kristaps 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",
207: "whitespace at end of input line",
1.158 schwarze 208: "new sentence, new line",
1.7 kristaps 209: "bad comment style",
1.64 schwarze 210: "invalid escape sequence",
211: "undefined string, using \"\"",
1.16 kristaps 212:
1.118 schwarze 213: /* related to tables */
214: "tbl line starts with span",
215: "tbl column starts with span",
216: "skipping vertical bar in tbl layout",
217:
1.7 kristaps 218: "generic error",
219:
220: /* related to tables */
1.116 schwarze 221: "non-alphabetic character in tbl options",
222: "skipping unknown tbl option",
223: "missing tbl option argument",
224: "wrong tbl option argument size",
1.118 schwarze 225: "empty tbl layout",
226: "invalid character in tbl layout",
227: "unmatched parenthesis in tbl layout",
1.119 schwarze 228: "tbl without any data cells",
229: "ignoring data in spanned tbl cell",
230: "ignoring extra tbl data cells",
231: "data block open at end of tbl",
1.7 kristaps 232:
1.68 schwarze 233: /* related to document structure and macros */
1.108 schwarze 234: NULL,
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: * Do the same for ROFF_EQN.
544: */
545:
1.139 schwarze 546: if (rr == ROFF_TBL)
1.100 schwarze 547: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 548: roff_addtbl(curp->man, span);
549: else if (rr == ROFF_EQN)
550: roff_addeqn(curp->man, roff_eqn(curp->roff));
551: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 552: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 553: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
554: break;
1.1 kristaps 555:
556: /* Temporary buffers typically are not full. */
557:
558: if (0 == start && '\0' == blk.buf[i])
559: break;
560:
561: /* Start the next input line. */
562:
563: pos = 0;
564: }
565:
566: free(ln.buf);
1.163 schwarze 567: return 1;
1.1 kristaps 568: }
569:
570: static int
1.40 schwarze 571: read_whole_file(struct mparse *curp, const char *file, int fd,
572: struct buf *fb, int *with_mmap)
1.1 kristaps 573: {
1.161 schwarze 574: struct stat st;
1.140 schwarze 575: gzFile gz;
1.1 kristaps 576: size_t off;
577: ssize_t ssz;
1.143 schwarze 578:
579: if (fstat(fd, &st) == -1)
580: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 581:
582: /*
583: * If we're a regular file, try just reading in the whole entry
584: * via mmap(). This is faster than reading it into blocks, and
585: * since each file is only a few bytes to begin with, I'm not
586: * concerned that this is going to tank any machines.
587: */
588:
1.140 schwarze 589: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 590: if (st.st_size > 0x7fffffff) {
1.111 schwarze 591: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 592: return 0;
1.1 kristaps 593: }
594: *with_mmap = 1;
595: fb->sz = (size_t)st.st_size;
1.37 schwarze 596: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 597: if (fb->buf != MAP_FAILED)
1.142 schwarze 598: return 1;
1.1 kristaps 599: }
600:
1.140 schwarze 601: if (curp->gzip) {
1.143 schwarze 602: if ((gz = gzdopen(fd, "rb")) == NULL)
603: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 604: } else
605: gz = NULL;
606:
1.1 kristaps 607: /*
608: * If this isn't a regular file (like, say, stdin), then we must
609: * go the old way and just read things in bit by bit.
610: */
611:
612: *with_mmap = 0;
613: off = 0;
614: fb->sz = 0;
615: fb->buf = NULL;
616: for (;;) {
617: if (off == fb->sz) {
618: if (fb->sz == (1U << 31)) {
1.111 schwarze 619: mandoc_msg(MANDOCERR_TOOLARGE, curp,
620: 0, 0, NULL);
1.1 kristaps 621: break;
622: }
623: resize_buf(fb, 65536);
624: }
1.140 schwarze 625: ssz = curp->gzip ?
626: gzread(gz, fb->buf + (int)off, fb->sz - off) :
627: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 628: if (ssz == 0) {
629: fb->sz = off;
1.142 schwarze 630: return 1;
1.1 kristaps 631: }
1.143 schwarze 632: if (ssz == -1)
633: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 634: off += (size_t)ssz;
635: }
636:
637: free(fb->buf);
638: fb->buf = NULL;
1.142 schwarze 639: return 0;
1.1 kristaps 640: }
641:
642: static void
643: mparse_end(struct mparse *curp)
644: {
1.135 schwarze 645: if (curp->man->macroset == MACROSET_NONE)
646: curp->man->macroset = MACROSET_MAN;
647: if (curp->man->macroset == MACROSET_MDOC)
648: mdoc_endparse(curp->man);
649: else
1.111 schwarze 650: man_endparse(curp->man);
1.1 kristaps 651: roff_endparse(curp->roff);
652: }
653:
654: static void
1.36 schwarze 655: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 656: {
1.85 schwarze 657: struct buf *svprimary;
1.28 joerg 658: const char *svfile;
1.95 schwarze 659: size_t offset;
1.36 schwarze 660: static int recursion_depth;
661:
662: if (64 < recursion_depth) {
663: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
664: return;
665: }
1.28 joerg 666:
667: /* Line number is per-file. */
668: svfile = curp->file;
669: curp->file = file;
1.85 schwarze 670: svprimary = curp->primary;
1.83 schwarze 671: curp->primary = &blk;
1.28 joerg 672: curp->line = 1;
1.36 schwarze 673: recursion_depth++;
1.28 joerg 674:
1.93 schwarze 675: /* Skip an UTF-8 byte order mark. */
676: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
677: (unsigned char)blk.buf[0] == 0xef &&
678: (unsigned char)blk.buf[1] == 0xbb &&
679: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 680: offset = 3;
1.93 schwarze 681: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 682: } else
683: offset = 0;
1.93 schwarze 684:
1.95 schwarze 685: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 686:
1.111 schwarze 687: if (--recursion_depth == 0)
1.28 joerg 688: mparse_end(curp);
689:
1.85 schwarze 690: curp->primary = svprimary;
1.28 joerg 691: curp->file = svfile;
692: }
693:
694: enum mandoclevel
1.104 schwarze 695: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 696: const char *file)
697: {
698: struct buf blk;
699:
1.104 schwarze 700: blk.buf = buf;
1.28 joerg 701: blk.sz = len;
702:
1.36 schwarze 703: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 704: return curp->file_status;
1.28 joerg 705: }
706:
1.98 schwarze 707: /*
708: * Read the whole file into memory and call the parsers.
709: * Called recursively when an .so request is encountered.
710: */
1.36 schwarze 711: enum mandoclevel
712: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 713: {
1.28 joerg 714: struct buf blk;
715: int with_mmap;
1.93 schwarze 716: int save_filenc;
1.1 kristaps 717:
1.91 schwarze 718: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 719: save_filenc = curp->filenc;
720: curp->filenc = curp->options &
721: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 722: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 723: curp->filenc = save_filenc;
1.91 schwarze 724: if (with_mmap)
725: munmap(blk.buf, blk.sz);
726: else
727: free(blk.buf);
728: }
1.142 schwarze 729: return curp->file_status;
1.82 schwarze 730: }
731:
1.148 schwarze 732: int
733: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 734: {
735: char *cp;
1.148 schwarze 736: int fd;
1.82 schwarze 737:
738: curp->file = file;
1.140 schwarze 739: cp = strrchr(file, '.');
740: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 741:
1.140 schwarze 742: /* First try to use the filename as it is. */
1.98 schwarze 743:
1.148 schwarze 744: if ((fd = open(file, O_RDONLY)) != -1)
745: return fd;
1.98 schwarze 746:
1.140 schwarze 747: /*
748: * If that doesn't work and the filename doesn't
749: * already end in .gz, try appending .gz.
750: */
1.98 schwarze 751:
1.140 schwarze 752: if ( ! curp->gzip) {
1.98 schwarze 753: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 754: fd = open(cp, O_RDONLY);
1.108 schwarze 755: free(cp);
1.148 schwarze 756: if (fd != -1) {
1.140 schwarze 757: curp->gzip = 1;
1.148 schwarze 758: return fd;
1.82 schwarze 759: }
760: }
761:
1.140 schwarze 762: /* Neither worked, give up. */
1.97 schwarze 763:
1.140 schwarze 764: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 765: return -1;
1.1 kristaps 766: }
767:
768: struct mparse *
1.179 schwarze 769: mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
770: enum mandoc_os os_e, const char *os_s)
1.1 kristaps 771: {
772: struct mparse *curp;
1.10 kristaps 773:
1.1 kristaps 774: curp = mandoc_calloc(1, sizeof(struct mparse));
775:
1.44 schwarze 776: curp->options = options;
1.179 schwarze 777: curp->mmin = mmin;
1.1 kristaps 778: curp->mmsg = mmsg;
1.179 schwarze 779: curp->os_s = os_s;
1.1 kristaps 780:
1.144 schwarze 781: curp->roff = roff_alloc(curp, options);
1.179 schwarze 782: curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
1.137 schwarze 783: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 784: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 785: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 786: if (curp->man->mdocmac == NULL)
787: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 788: } else if (curp->options & MPARSE_MAN) {
789: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 790: if (curp->man->manmac == NULL)
791: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 792: }
1.138 schwarze 793: curp->man->first->tok = TOKEN_NONE;
1.179 schwarze 794: curp->man->meta.os_e = os_e;
1.142 schwarze 795: return curp;
1.1 kristaps 796: }
797:
798: void
799: mparse_reset(struct mparse *curp)
800: {
801: roff_reset(curp->roff);
1.150 schwarze 802: roff_man_reset(curp->man);
1.160 schwarze 803:
804: free(curp->sodest);
805: curp->sodest = NULL;
806:
1.24 kristaps 807: if (curp->secondary)
808: curp->secondary->sz = 0;
1.1 kristaps 809:
810: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 811: curp->gzip = 0;
1.1 kristaps 812: }
813:
814: void
815: mparse_free(struct mparse *curp)
816: {
817:
1.164 schwarze 818: roffhash_free(curp->man->mdocmac);
819: roffhash_free(curp->man->manmac);
1.137 schwarze 820: roff_man_free(curp->man);
1.160 schwarze 821: roff_free(curp->roff);
1.24 kristaps 822: if (curp->secondary)
823: free(curp->secondary->buf);
1.1 kristaps 824:
1.24 kristaps 825: free(curp->secondary);
1.45 schwarze 826: free(curp->sodest);
1.1 kristaps 827: free(curp);
828: }
829:
830: void
1.135 schwarze 831: mparse_result(struct mparse *curp, struct roff_man **man,
832: char **sodest)
1.1 kristaps 833: {
834:
1.45 schwarze 835: if (sodest && NULL != (*sodest = curp->sodest)) {
836: *man = NULL;
837: return;
838: }
1.9 kristaps 839: if (man)
840: *man = curp->man;
1.157 schwarze 841: }
842:
843: void
844: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
845: {
846: if (curp->file_status > *rc)
847: *rc = curp->file_status;
1.3 kristaps 848: }
849:
850: void
851: mandoc_vmsg(enum mandocerr t, struct mparse *m,
852: int ln, int pos, const char *fmt, ...)
853: {
854: char buf[256];
855: va_list ap;
856:
857: va_start(ap, fmt);
1.48 schwarze 858: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 859: va_end(ap);
860:
861: mandoc_msg(t, m, ln, pos, buf);
862: }
863:
864: void
1.47 schwarze 865: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 866: int ln, int col, const char *msg)
867: {
868: enum mandoclevel level;
869:
1.179 schwarze 870: if (er < m->mmin && er != MANDOCERR_FILE)
871: return;
872:
1.112 schwarze 873: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 874: while (er < mandoclimits[level])
875: level--;
876:
1.8 kristaps 877: if (m->mmsg)
878: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 879:
880: if (m->file_status < level)
881: m->file_status = level;
1.7 kristaps 882: }
883:
884: const char *
885: mparse_strerror(enum mandocerr er)
886: {
887:
1.142 schwarze 888: return mandocerrs[er];
1.7 kristaps 889: }
890:
891: const char *
892: mparse_strlevel(enum mandoclevel lvl)
893: {
1.142 schwarze 894: return mandoclevels[lvl];
1.24 kristaps 895: }
896:
897: void
898: mparse_keep(struct mparse *p)
899: {
900:
901: assert(NULL == p->secondary);
902: p->secondary = mandoc_calloc(1, sizeof(struct buf));
903: }
904:
905: const char *
906: mparse_getkeep(const struct mparse *p)
907: {
908:
909: assert(p->secondary);
1.142 schwarze 910: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 911: }
CVSweb