Annotation of mandoc/read.c, Revision 1.154
1.154 ! schwarze 1: /* $Id: read.c,v 1.153 2016/11/10 12:47:50 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.148 schwarze 4: * Copyright (c) 2010-2016 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>
1.28 joerg 33: #include <stdint.h>
1.1 kristaps 34: #include <stdio.h>
35: #include <stdlib.h>
36: #include <string.h>
37: #include <unistd.h>
1.140 schwarze 38: #include <zlib.h>
1.1 kristaps 39:
1.133 schwarze 40: #include "mandoc_aux.h"
1.1 kristaps 41: #include "mandoc.h"
1.133 schwarze 42: #include "roff.h"
1.1 kristaps 43: #include "mdoc.h"
44: #include "man.h"
1.133 schwarze 45: #include "libmandoc.h"
1.139 schwarze 46: #include "roff_int.h"
1.1 kristaps 47:
48: #define REPARSE_LIMIT 1000
49:
50: struct mparse {
1.134 schwarze 51: struct roff_man *man; /* man parser */
1.1 kristaps 52: struct roff *roff; /* roff parser (!NULL) */
1.45 schwarze 53: char *sodest; /* filename pointed to by .so */
1.83 schwarze 54: const char *file; /* filename of current input file */
55: struct buf *primary; /* buffer currently being parsed */
56: struct buf *secondary; /* preprocessed copy of input */
57: const char *defos; /* default operating system */
58: mandocmsg mmsg; /* warning/error message handler */
59: enum mandoclevel file_status; /* status of current parse */
60: enum mandoclevel wlevel; /* ignore messages below this */
61: int options; /* parser options */
1.140 schwarze 62: int gzip; /* current input file is gzipped */
1.93 schwarze 63: int filenc; /* encoding of the current file */
1.1 kristaps 64: int reparse_count; /* finite interp. stack */
1.83 schwarze 65: int line; /* line number in the file */
1.1 kristaps 66: };
67:
1.84 schwarze 68: static void choose_parser(struct mparse *);
1.1 kristaps 69: static void resize_buf(struct buf *, size_t);
1.95 schwarze 70: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 71: static int read_whole_file(struct mparse *, const char *, int,
72: struct buf *, int *);
1.1 kristaps 73: static void mparse_end(struct mparse *);
1.37 schwarze 74: static void mparse_parse_buffer(struct mparse *, struct buf,
75: const char *);
1.1 kristaps 76:
1.3 kristaps 77: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
78: MANDOCERR_OK,
79: MANDOCERR_WARNING,
80: MANDOCERR_WARNING,
81: MANDOCERR_ERROR,
1.112 schwarze 82: MANDOCERR_UNSUPP,
1.3 kristaps 83: MANDOCERR_MAX,
84: MANDOCERR_MAX
85: };
86:
1.7 kristaps 87: static const char * const mandocerrs[MANDOCERR_MAX] = {
88: "ok",
89:
90: "generic warning",
91:
92: /* related to the prologue */
1.79 schwarze 93: "missing manual title, using UNTITLED",
94: "missing manual title, using \"\"",
1.54 schwarze 95: "lower case character in document title",
1.79 schwarze 96: "missing manual section, using \"\"",
1.7 kristaps 97: "unknown manual section",
1.54 schwarze 98: "missing date, using today's date",
1.7 kristaps 99: "cannot parse date, using it verbatim",
1.79 schwarze 100: "missing Os macro, using \"\"",
101: "duplicate prologue macro",
102: "late prologue macro",
103: "skipping late title macro",
1.7 kristaps 104: "prologue macros out of order",
105:
106: /* related to document structure */
107: ".so is fragile, better use ln(1)",
1.50 schwarze 108: "no document body",
1.54 schwarze 109: "content before first section header",
110: "first section is not \"NAME\"",
1.128 schwarze 111: "NAME section without name",
112: "NAME section without description",
113: "description not at the end of NAME",
114: "bad NAME section content",
1.121 schwarze 115: "missing description line, using \"\"",
1.7 kristaps 116: "sections out of conventional order",
1.54 schwarze 117: "duplicate section title",
118: "unexpected section",
1.87 schwarze 119: "unusual Xr order",
120: "unusual Xr punctuation",
1.86 schwarze 121: "AUTHORS section without An macro",
1.7 kristaps 122:
123: /* related to macros and nesting */
1.55 schwarze 124: "obsolete macro",
1.102 schwarze 125: "macro neither callable nor escaped",
1.7 kristaps 126: "skipping paragraph macro",
1.31 schwarze 127: "moving paragraph macro out of list",
1.7 kristaps 128: "skipping no-space macro",
129: "blocks badly nested",
130: "nested displays are not portable",
1.57 schwarze 131: "moving content out of list",
1.78 schwarze 132: "fill mode already enabled, skipping",
133: "fill mode already disabled, skipping",
1.7 kristaps 134: "line scope broken",
135:
136: /* related to missing macro arguments */
1.58 schwarze 137: "skipping empty request",
138: "conditional request controls empty scope",
1.7 kristaps 139: "skipping empty macro",
1.124 schwarze 140: "empty block",
1.62 schwarze 141: "empty argument, using 0n",
1.60 schwarze 142: "missing display type, using -ragged",
143: "list type is not the first argument",
1.152 schwarze 144: "missing -width in -tag list, using 6n",
1.78 schwarze 145: "missing utility name, using \"\"",
1.123 schwarze 146: "missing function name, using \"\"",
1.60 schwarze 147: "empty head in list item",
148: "empty list item",
1.61 schwarze 149: "missing font type, using \\fR",
150: "unknown font type, using \\fR",
1.103 schwarze 151: "nothing follows prefix",
1.122 schwarze 152: "empty reference block",
1.60 schwarze 153: "missing -std argument, adding it",
1.125 schwarze 154: "missing option string, using \"\"",
155: "missing resource identifier, using \"\"",
1.90 schwarze 156: "missing eqn box, using \"\"",
1.7 kristaps 157:
158: /* related to bad macro arguments */
1.64 schwarze 159: "unterminated quoted argument",
1.7 kristaps 160: "duplicate argument",
1.76 schwarze 161: "skipping duplicate argument",
1.63 schwarze 162: "skipping duplicate display type",
163: "skipping duplicate list type",
1.76 schwarze 164: "skipping -width argument",
1.126 schwarze 165: "wrong number of cells",
1.7 kristaps 166: "unknown AT&T UNIX version",
1.88 schwarze 167: "comma in function argument",
1.89 schwarze 168: "parenthesis in function name",
1.67 schwarze 169: "invalid content in Rs block",
1.63 schwarze 170: "invalid Boolean argument",
171: "unknown font, skipping request",
1.126 schwarze 172: "odd number of characters in request",
1.7 kristaps 173:
174: /* related to plain text */
1.64 schwarze 175: "blank line in fill mode, using .sp",
176: "tab in filled text",
177: "whitespace at end of input line",
1.7 kristaps 178: "bad comment style",
1.64 schwarze 179: "invalid escape sequence",
180: "undefined string, using \"\"",
1.16 kristaps 181:
1.118 schwarze 182: /* related to tables */
183: "tbl line starts with span",
184: "tbl column starts with span",
185: "skipping vertical bar in tbl layout",
186:
1.7 kristaps 187: "generic error",
188:
189: /* related to tables */
1.116 schwarze 190: "non-alphabetic character in tbl options",
191: "skipping unknown tbl option",
192: "missing tbl option argument",
193: "wrong tbl option argument size",
1.118 schwarze 194: "empty tbl layout",
195: "invalid character in tbl layout",
196: "unmatched parenthesis in tbl layout",
1.119 schwarze 197: "tbl without any data cells",
198: "ignoring data in spanned tbl cell",
199: "ignoring extra tbl data cells",
200: "data block open at end of tbl",
1.7 kristaps 201:
1.68 schwarze 202: /* related to document structure and macros */
1.108 schwarze 203: NULL,
1.7 kristaps 204: "input stack limit exceeded, infinite loop?",
205: "skipping bad character",
1.68 schwarze 206: "skipping unknown macro",
1.112 schwarze 207: "skipping insecure request",
1.73 schwarze 208: "skipping item outside list",
1.68 schwarze 209: "skipping column outside column list",
210: "skipping end of block that is not open",
1.115 schwarze 211: "fewer RS blocks open, skipping",
1.68 schwarze 212: "inserting missing end of block",
213: "appending missing end of block",
214:
215: /* related to request and macro arguments */
1.7 kristaps 216: "escaped character not allowed in a name",
1.99 schwarze 217: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 218: "skipping display without arguments",
1.71 schwarze 219: "missing list type, using -item",
1.70 schwarze 220: "missing manual name, using \"\"",
1.71 schwarze 221: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 222: "unknown standard specifier",
1.71 schwarze 223: "skipping request without numeric argument",
1.109 schwarze 224: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 225: ".so request failed",
1.61 schwarze 226: "skipping all arguments",
227: "skipping excess arguments",
1.92 kristaps 228: "divide by zero",
1.112 schwarze 229:
230: "unsupported feature",
231: "input too large",
1.114 schwarze 232: "unsupported control character",
1.112 schwarze 233: "unsupported roff request",
1.119 schwarze 234: "eqn delim option in tbl",
1.118 schwarze 235: "unsupported tbl layout modifier",
1.112 schwarze 236: "ignoring macro in table",
1.7 kristaps 237: };
238:
239: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
240: "SUCCESS",
241: "RESERVED",
242: "WARNING",
243: "ERROR",
1.112 schwarze 244: "UNSUPP",
1.7 kristaps 245: "BADARG",
246: "SYSERR"
247: };
248:
1.47 schwarze 249:
1.1 kristaps 250: static void
251: resize_buf(struct buf *buf, size_t initial)
252: {
253:
254: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
255: buf->buf = mandoc_realloc(buf->buf, buf->sz);
256: }
257:
258: static void
1.84 schwarze 259: choose_parser(struct mparse *curp)
1.1 kristaps 260: {
1.83 schwarze 261: char *cp, *ep;
262: int format;
1.1 kristaps 263:
1.83 schwarze 264: /*
265: * If neither command line arguments -mdoc or -man select
266: * a parser nor the roff parser found a .Dd or .TH macro
267: * yet, look ahead in the main input buffer.
268: */
269:
270: if ((format = roff_getformat(curp->roff)) == 0) {
271: cp = curp->primary->buf;
272: ep = cp + curp->primary->sz;
273: while (cp < ep) {
1.85 schwarze 274: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 275: cp++;
276: if (cp[0] == 'D' && cp[1] == 'd') {
277: format = MPARSE_MDOC;
278: break;
279: }
280: if (cp[0] == 'T' && cp[1] == 'H') {
281: format = MPARSE_MAN;
282: break;
283: }
284: }
285: cp = memchr(cp, '\n', ep - cp);
286: if (cp == NULL)
287: break;
288: cp++;
289: }
1.1 kristaps 290: }
291:
1.83 schwarze 292: if (format == MPARSE_MDOC) {
1.136 schwarze 293: mdoc_hash_init();
1.137 schwarze 294: curp->man->macroset = MACROSET_MDOC;
1.138 schwarze 295: curp->man->first->tok = TOKEN_NONE;
1.137 schwarze 296: } else {
297: man_hash_init();
298: curp->man->macroset = MACROSET_MAN;
1.138 schwarze 299: curp->man->first->tok = TOKEN_NONE;
1.47 schwarze 300: }
1.1 kristaps 301: }
302:
303: /*
1.95 schwarze 304: * Main parse routine for a buffer.
305: * It assumes encoding and line numbering are already set up.
306: * It can recurse directly (for invocations of user-defined
307: * macros, inline equations, and input line traps)
308: * and indirectly (for .so file inclusion).
1.1 kristaps 309: */
310: static void
1.95 schwarze 311: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 312: {
313: const struct tbl_span *span;
314: struct buf ln;
1.113 schwarze 315: const char *save_file;
1.110 schwarze 316: char *cp;
1.95 schwarze 317: size_t pos; /* byte number in the ln buffer */
1.154 ! schwarze 318: size_t j; /* auxiliary byte number in the blk buffer */
1.1 kristaps 319: enum rofferr rr;
1.100 schwarze 320: int of;
1.1 kristaps 321: int lnn; /* line number in the real file */
1.110 schwarze 322: int fd;
1.1 kristaps 323: unsigned char c;
324:
1.95 schwarze 325: memset(&ln, 0, sizeof(ln));
1.1 kristaps 326:
1.47 schwarze 327: lnn = curp->line;
328: pos = 0;
1.1 kristaps 329:
1.95 schwarze 330: while (i < blk.sz) {
1.1 kristaps 331: if (0 == pos && '\0' == blk.buf[i])
332: break;
333:
334: if (start) {
335: curp->line = lnn;
336: curp->reparse_count = 0;
1.93 schwarze 337:
338: if (lnn < 3 &&
339: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 340: curp->filenc & MPARSE_LATIN1)
341: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 342: }
343:
1.95 schwarze 344: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 345:
346: /*
347: * When finding an unescaped newline character,
348: * leave the character loop to process the line.
349: * Skip a preceding carriage return, if any.
350: */
351:
1.95 schwarze 352: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 353: '\n' == blk.buf[i + 1])
354: ++i;
355: if ('\n' == blk.buf[i]) {
356: ++i;
357: ++lnn;
358: break;
359: }
360:
1.35 schwarze 361: /*
1.93 schwarze 362: * Make sure we have space for the worst
363: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 364: */
365:
1.95 schwarze 366: if (pos + 11 > ln.sz)
1.35 schwarze 367: resize_buf(&ln, 256);
368:
1.47 schwarze 369: /*
1.93 schwarze 370: * Encode 8-bit input.
1.1 kristaps 371: */
372:
1.93 schwarze 373: c = blk.buf[i];
374: if (c & 0x80) {
1.95 schwarze 375: if ( ! (curp->filenc && preconv_encode(
376: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 377: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
378: curp->line, pos, "0x%x", c);
1.93 schwarze 379: ln.buf[pos++] = '?';
380: i++;
381: }
382: continue;
383: }
384:
385: /*
386: * Exclude control characters.
387: */
1.1 kristaps 388:
1.93 schwarze 389: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 390: mandoc_vmsg(c == 0x00 || c == 0x04 ||
391: c > 0x0a ? MANDOCERR_CHAR_BAD :
392: MANDOCERR_CHAR_UNSUPP,
393: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 394: i++;
1.127 schwarze 395: if (c != '\r')
396: ln.buf[pos++] = '?';
1.1 kristaps 397: continue;
398: }
399:
400: /* Trailing backslash = a plain char. */
401:
1.95 schwarze 402: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 403: ln.buf[pos++] = blk.buf[i++];
404: continue;
405: }
406:
407: /*
408: * Found escape and at least one other character.
409: * When it's a newline character, skip it.
410: * When there is a carriage return in between,
411: * skip that one as well.
412: */
413:
1.95 schwarze 414: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 415: '\n' == blk.buf[i + 2])
416: ++i;
417: if ('\n' == blk.buf[i + 1]) {
418: i += 2;
419: ++lnn;
420: continue;
421: }
422:
1.13 kristaps 423: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.154 ! schwarze 424: j = i;
1.1 kristaps 425: i += 2;
426: /* Comment, skip to end of line */
1.95 schwarze 427: for (; i < blk.sz; ++i) {
1.153 schwarze 428: if (blk.buf[i] != '\n')
429: continue;
430: if (blk.buf[i - 1] == ' ' ||
431: blk.buf[i - 1] == '\t')
432: mandoc_msg(
433: MANDOCERR_SPACE_EOL,
434: curp, curp->line,
1.154 ! schwarze 435: pos + i-1 - j, NULL);
1.153 schwarze 436: ++i;
437: ++lnn;
438: break;
1.1 kristaps 439: }
440:
441: /* Backout trailing whitespaces */
442: for (; pos > 0; --pos) {
443: if (ln.buf[pos - 1] != ' ')
444: break;
445: if (pos > 2 && ln.buf[pos - 2] == '\\')
446: break;
447: }
448: break;
449: }
450:
1.35 schwarze 451: /* Catch escaped bogus characters. */
452:
453: c = (unsigned char) blk.buf[i+1];
454:
1.47 schwarze 455: if ( ! (isascii(c) &&
456: (isgraph(c) || isblank(c)))) {
1.114 schwarze 457: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 458: curp->line, pos, "0x%x", c);
1.35 schwarze 459: i += 2;
460: ln.buf[pos++] = '?';
461: continue;
462: }
463:
1.1 kristaps 464: /* Some other escape sequence, copy & cont. */
465:
466: ln.buf[pos++] = blk.buf[i++];
467: ln.buf[pos++] = blk.buf[i++];
468: }
469:
1.95 schwarze 470: if (pos >= ln.sz)
1.1 kristaps 471: resize_buf(&ln, 256);
472:
473: ln.buf[pos] = '\0';
474:
475: /*
476: * A significant amount of complexity is contained by
477: * the roff preprocessor. It's line-oriented but can be
478: * expressed on one line, so we need at times to
479: * readjust our starting point and re-run it. The roff
480: * preprocessor can also readjust the buffers with new
481: * data, so we pass them in wholesale.
482: */
483:
484: of = 0;
485:
1.24 kristaps 486: /*
487: * Maintain a lookaside buffer of all parsed lines. We
488: * only do this if mparse_keep() has been invoked (the
489: * buffer may be accessed with mparse_getkeep()).
490: */
491:
492: if (curp->secondary) {
1.47 schwarze 493: curp->secondary->buf = mandoc_realloc(
494: curp->secondary->buf,
495: curp->secondary->sz + pos + 2);
496: memcpy(curp->secondary->buf +
497: curp->secondary->sz,
498: ln.buf, pos);
1.24 kristaps 499: curp->secondary->sz += pos;
500: curp->secondary->buf
501: [curp->secondary->sz] = '\n';
502: curp->secondary->sz++;
503: curp->secondary->buf
504: [curp->secondary->sz] = '\0';
505: }
1.1 kristaps 506: rerun:
1.96 schwarze 507: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 508:
509: switch (rr) {
1.47 schwarze 510: case ROFF_REPARSE:
1.1 kristaps 511: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 512: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 513: else
1.3 kristaps 514: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 515: curp->line, pos, NULL);
1.1 kristaps 516: pos = 0;
517: continue;
1.47 schwarze 518: case ROFF_APPEND:
1.95 schwarze 519: pos = strlen(ln.buf);
1.1 kristaps 520: continue;
1.47 schwarze 521: case ROFF_RERUN:
1.1 kristaps 522: goto rerun;
1.47 schwarze 523: case ROFF_IGN:
1.1 kristaps 524: pos = 0;
525: continue;
1.47 schwarze 526: case ROFF_SO:
1.95 schwarze 527: if ( ! (curp->options & MPARSE_SO) &&
528: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 529: curp->sodest = mandoc_strdup(ln.buf + of);
530: free(ln.buf);
531: return;
532: }
1.24 kristaps 533: /*
534: * We remove `so' clauses from our lookaside
535: * buffer because we're going to descend into
536: * the file recursively.
537: */
1.47 schwarze 538: if (curp->secondary)
1.25 kristaps 539: curp->secondary->sz -= pos + 1;
1.113 schwarze 540: save_file = curp->file;
1.148 schwarze 541: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 542: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 543: close(fd);
1.113 schwarze 544: curp->file = save_file;
545: } else {
546: curp->file = save_file;
1.52 schwarze 547: mandoc_vmsg(MANDOCERR_SO_FAIL,
548: curp, curp->line, pos,
549: ".so %s", ln.buf + of);
1.110 schwarze 550: ln.sz = mandoc_asprintf(&cp,
551: ".sp\nSee the file %s.\n.sp",
552: ln.buf + of);
553: free(ln.buf);
554: ln.buf = cp;
555: of = 0;
556: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 557: }
1.1 kristaps 558: pos = 0;
559: continue;
560: default:
561: break;
562: }
563:
1.150 schwarze 564: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 565: choose_parser(curp);
1.1 kristaps 566:
1.47 schwarze 567: /*
1.84 schwarze 568: * Lastly, push down into the parsers themselves.
1.1 kristaps 569: * If libroff returns ROFF_TBL, then add it to the
570: * currently open parse. Since we only get here if
571: * there does exist data (see tbl_data.c), we're
572: * guaranteed that something's been allocated.
573: * Do the same for ROFF_EQN.
574: */
575:
1.139 schwarze 576: if (rr == ROFF_TBL)
1.100 schwarze 577: while ((span = roff_span(curp->roff)) != NULL)
1.139 schwarze 578: roff_addtbl(curp->man, span);
579: else if (rr == ROFF_EQN)
580: roff_addeqn(curp->man, roff_eqn(curp->roff));
581: else if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 582: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 583: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
584: break;
1.1 kristaps 585:
586: /* Temporary buffers typically are not full. */
587:
588: if (0 == start && '\0' == blk.buf[i])
589: break;
590:
591: /* Start the next input line. */
592:
593: pos = 0;
594: }
595:
596: free(ln.buf);
597: }
598:
599: static int
1.40 schwarze 600: read_whole_file(struct mparse *curp, const char *file, int fd,
601: struct buf *fb, int *with_mmap)
1.1 kristaps 602: {
1.140 schwarze 603: gzFile gz;
1.1 kristaps 604: size_t off;
605: ssize_t ssz;
606:
1.15 kristaps 607: struct stat st;
1.143 schwarze 608:
609: if (fstat(fd, &st) == -1)
610: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 611:
612: /*
613: * If we're a regular file, try just reading in the whole entry
614: * via mmap(). This is faster than reading it into blocks, and
615: * since each file is only a few bytes to begin with, I'm not
616: * concerned that this is going to tank any machines.
617: */
618:
1.140 schwarze 619: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 620: if (st.st_size > 0x7fffffff) {
1.111 schwarze 621: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 622: return 0;
1.1 kristaps 623: }
624: *with_mmap = 1;
625: fb->sz = (size_t)st.st_size;
1.37 schwarze 626: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 627: if (fb->buf != MAP_FAILED)
1.142 schwarze 628: return 1;
1.1 kristaps 629: }
630:
1.140 schwarze 631: if (curp->gzip) {
1.143 schwarze 632: if ((gz = gzdopen(fd, "rb")) == NULL)
633: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.140 schwarze 634: } else
635: gz = NULL;
636:
1.1 kristaps 637: /*
638: * If this isn't a regular file (like, say, stdin), then we must
639: * go the old way and just read things in bit by bit.
640: */
641:
642: *with_mmap = 0;
643: off = 0;
644: fb->sz = 0;
645: fb->buf = NULL;
646: for (;;) {
647: if (off == fb->sz) {
648: if (fb->sz == (1U << 31)) {
1.111 schwarze 649: mandoc_msg(MANDOCERR_TOOLARGE, curp,
650: 0, 0, NULL);
1.1 kristaps 651: break;
652: }
653: resize_buf(fb, 65536);
654: }
1.140 schwarze 655: ssz = curp->gzip ?
656: gzread(gz, fb->buf + (int)off, fb->sz - off) :
657: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 658: if (ssz == 0) {
659: fb->sz = off;
1.142 schwarze 660: return 1;
1.1 kristaps 661: }
1.143 schwarze 662: if (ssz == -1)
663: err((int)MANDOCLEVEL_SYSERR, "%s", file);
1.1 kristaps 664: off += (size_t)ssz;
665: }
666:
667: free(fb->buf);
668: fb->buf = NULL;
1.142 schwarze 669: return 0;
1.1 kristaps 670: }
671:
672: static void
673: mparse_end(struct mparse *curp)
674: {
1.135 schwarze 675: if (curp->man->macroset == MACROSET_NONE)
676: curp->man->macroset = MACROSET_MAN;
677: if (curp->man->macroset == MACROSET_MDOC)
678: mdoc_endparse(curp->man);
679: else
1.111 schwarze 680: man_endparse(curp->man);
1.1 kristaps 681: roff_endparse(curp->roff);
682: }
683:
684: static void
1.36 schwarze 685: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 686: {
1.85 schwarze 687: struct buf *svprimary;
1.28 joerg 688: const char *svfile;
1.95 schwarze 689: size_t offset;
1.36 schwarze 690: static int recursion_depth;
691:
692: if (64 < recursion_depth) {
693: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
694: return;
695: }
1.28 joerg 696:
697: /* Line number is per-file. */
698: svfile = curp->file;
699: curp->file = file;
1.85 schwarze 700: svprimary = curp->primary;
1.83 schwarze 701: curp->primary = &blk;
1.28 joerg 702: curp->line = 1;
1.36 schwarze 703: recursion_depth++;
1.28 joerg 704:
1.93 schwarze 705: /* Skip an UTF-8 byte order mark. */
706: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
707: (unsigned char)blk.buf[0] == 0xef &&
708: (unsigned char)blk.buf[1] == 0xbb &&
709: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 710: offset = 3;
1.93 schwarze 711: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 712: } else
713: offset = 0;
1.93 schwarze 714:
1.95 schwarze 715: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 716:
1.111 schwarze 717: if (--recursion_depth == 0)
1.28 joerg 718: mparse_end(curp);
719:
1.85 schwarze 720: curp->primary = svprimary;
1.28 joerg 721: curp->file = svfile;
722: }
723:
724: enum mandoclevel
1.104 schwarze 725: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 726: const char *file)
727: {
728: struct buf blk;
729:
1.104 schwarze 730: blk.buf = buf;
1.28 joerg 731: blk.sz = len;
732:
1.36 schwarze 733: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 734: return curp->file_status;
1.28 joerg 735: }
736:
1.98 schwarze 737: /*
738: * Read the whole file into memory and call the parsers.
739: * Called recursively when an .so request is encountered.
740: */
1.36 schwarze 741: enum mandoclevel
742: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 743: {
1.28 joerg 744: struct buf blk;
745: int with_mmap;
1.93 schwarze 746: int save_filenc;
1.1 kristaps 747:
1.91 schwarze 748: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 749: save_filenc = curp->filenc;
750: curp->filenc = curp->options &
751: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 752: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 753: curp->filenc = save_filenc;
1.91 schwarze 754: if (with_mmap)
755: munmap(blk.buf, blk.sz);
756: else
757: free(blk.buf);
758: }
1.142 schwarze 759: return curp->file_status;
1.82 schwarze 760: }
761:
1.148 schwarze 762: int
763: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 764: {
765: char *cp;
1.148 schwarze 766: int fd;
1.82 schwarze 767:
768: curp->file = file;
1.140 schwarze 769: cp = strrchr(file, '.');
770: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 771:
1.140 schwarze 772: /* First try to use the filename as it is. */
1.98 schwarze 773:
1.148 schwarze 774: if ((fd = open(file, O_RDONLY)) != -1)
775: return fd;
1.98 schwarze 776:
1.140 schwarze 777: /*
778: * If that doesn't work and the filename doesn't
779: * already end in .gz, try appending .gz.
780: */
1.98 schwarze 781:
1.140 schwarze 782: if ( ! curp->gzip) {
1.98 schwarze 783: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 784: fd = open(cp, O_RDONLY);
1.108 schwarze 785: free(cp);
1.148 schwarze 786: if (fd != -1) {
1.140 schwarze 787: curp->gzip = 1;
1.148 schwarze 788: return fd;
1.82 schwarze 789: }
790: }
791:
1.140 schwarze 792: /* Neither worked, give up. */
1.97 schwarze 793:
1.140 schwarze 794: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 795: return -1;
1.1 kristaps 796: }
797:
798: struct mparse *
1.94 schwarze 799: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
1.144 schwarze 800: const char *defos)
1.1 kristaps 801: {
802: struct mparse *curp;
1.10 kristaps 803:
1.1 kristaps 804: curp = mandoc_calloc(1, sizeof(struct mparse));
805:
1.44 schwarze 806: curp->options = options;
1.3 kristaps 807: curp->wlevel = wlevel;
1.1 kristaps 808: curp->mmsg = mmsg;
1.29 schwarze 809: curp->defos = defos;
1.1 kristaps 810:
1.144 schwarze 811: curp->roff = roff_alloc(curp, options);
1.137 schwarze 812: curp->man = roff_man_alloc( curp->roff, curp, curp->defos,
813: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 814: if (curp->options & MPARSE_MDOC) {
815: mdoc_hash_init();
1.137 schwarze 816: curp->man->macroset = MACROSET_MDOC;
817: } else if (curp->options & MPARSE_MAN) {
1.136 schwarze 818: man_hash_init();
1.137 schwarze 819: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 820: }
1.138 schwarze 821: curp->man->first->tok = TOKEN_NONE;
1.142 schwarze 822: return curp;
1.1 kristaps 823: }
824:
825: void
826: mparse_reset(struct mparse *curp)
827: {
828: roff_reset(curp->roff);
1.150 schwarze 829: roff_man_reset(curp->man);
1.24 kristaps 830: if (curp->secondary)
831: curp->secondary->sz = 0;
1.1 kristaps 832:
833: curp->file_status = MANDOCLEVEL_OK;
1.45 schwarze 834:
835: free(curp->sodest);
836: curp->sodest = NULL;
1.1 kristaps 837: }
838:
839: void
840: mparse_free(struct mparse *curp)
841: {
842:
1.137 schwarze 843: roff_man_free(curp->man);
1.1 kristaps 844: if (curp->roff)
845: roff_free(curp->roff);
1.24 kristaps 846: if (curp->secondary)
847: free(curp->secondary->buf);
1.1 kristaps 848:
1.24 kristaps 849: free(curp->secondary);
1.45 schwarze 850: free(curp->sodest);
1.1 kristaps 851: free(curp);
852: }
853:
854: void
1.135 schwarze 855: mparse_result(struct mparse *curp, struct roff_man **man,
856: char **sodest)
1.1 kristaps 857: {
858:
1.45 schwarze 859: if (sodest && NULL != (*sodest = curp->sodest)) {
860: *man = NULL;
861: return;
862: }
1.9 kristaps 863: if (man)
864: *man = curp->man;
1.3 kristaps 865: }
866:
867: void
868: mandoc_vmsg(enum mandocerr t, struct mparse *m,
869: int ln, int pos, const char *fmt, ...)
870: {
871: char buf[256];
872: va_list ap;
873:
874: va_start(ap, fmt);
1.48 schwarze 875: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 876: va_end(ap);
877:
878: mandoc_msg(t, m, ln, pos, buf);
879: }
880:
881: void
1.47 schwarze 882: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 883: int ln, int col, const char *msg)
884: {
885: enum mandoclevel level;
886:
1.112 schwarze 887: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 888: while (er < mandoclimits[level])
889: level--;
890:
1.108 schwarze 891: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 892: return;
893:
1.8 kristaps 894: if (m->mmsg)
895: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 896:
897: if (m->file_status < level)
898: m->file_status = level;
1.7 kristaps 899: }
900:
901: const char *
902: mparse_strerror(enum mandocerr er)
903: {
904:
1.142 schwarze 905: return mandocerrs[er];
1.7 kristaps 906: }
907:
908: const char *
909: mparse_strlevel(enum mandoclevel lvl)
910: {
1.142 schwarze 911: return mandoclevels[lvl];
1.24 kristaps 912: }
913:
914: void
915: mparse_keep(struct mparse *p)
916: {
917:
918: assert(NULL == p->secondary);
919: p->secondary = mandoc_calloc(1, sizeof(struct buf));
920: }
921:
922: const char *
923: mparse_getkeep(const struct mparse *p)
924: {
925:
926: assert(p->secondary);
1.142 schwarze 927: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 928: }
CVSweb