Annotation of mandoc/read.c, Revision 1.122
1.122 ! schwarze 1: /* $Id: read.c,v 1.121 2015/02/04 16:38:56 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.107 schwarze 4: * Copyright (c) 2010-2015 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: *
11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
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>
1.81 schwarze 22: #if HAVE_MMAP
1.82 schwarze 23: #include <sys/mman.h>
1.80 schwarze 24: #include <sys/stat.h>
1.15 kristaps 25: #endif
1.82 schwarze 26: #include <sys/wait.h>
1.1 kristaps 27:
28: #include <assert.h>
29: #include <ctype.h>
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>
38:
39: #include "mandoc.h"
1.46 schwarze 40: #include "mandoc_aux.h"
1.3 kristaps 41: #include "libmandoc.h"
1.1 kristaps 42: #include "mdoc.h"
43: #include "man.h"
44:
45: #define REPARSE_LIMIT 1000
46:
47: struct mparse {
48: struct man *pman; /* persistent man parser */
49: struct mdoc *pmdoc; /* persistent mdoc parser */
50: struct man *man; /* man parser */
51: struct mdoc *mdoc; /* mdoc parser */
52: struct roff *roff; /* roff parser (!NULL) */
1.94 schwarze 53: const struct mchars *mchars; /* character table */
1.45 schwarze 54: char *sodest; /* filename pointed to by .so */
1.83 schwarze 55: const char *file; /* filename of current input file */
56: struct buf *primary; /* buffer currently being parsed */
57: struct buf *secondary; /* preprocessed copy of input */
58: const char *defos; /* default operating system */
59: mandocmsg mmsg; /* warning/error message handler */
60: enum mandoclevel file_status; /* status of current parse */
61: enum mandoclevel wlevel; /* ignore messages below this */
62: int options; /* parser options */
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 */
1.97 schwarze 66: pid_t child; /* the gunzip(1) process */
1.1 kristaps 67: };
68:
1.84 schwarze 69: static void choose_parser(struct mparse *);
1.1 kristaps 70: static void resize_buf(struct buf *, size_t);
1.95 schwarze 71: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 72: static int read_whole_file(struct mparse *, const char *, int,
73: struct buf *, int *);
1.1 kristaps 74: static void mparse_end(struct mparse *);
1.37 schwarze 75: static void mparse_parse_buffer(struct mparse *, struct buf,
76: const char *);
1.1 kristaps 77:
1.3 kristaps 78: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
79: MANDOCERR_OK,
80: MANDOCERR_WARNING,
81: MANDOCERR_WARNING,
82: MANDOCERR_ERROR,
1.112 schwarze 83: MANDOCERR_UNSUPP,
1.3 kristaps 84: MANDOCERR_MAX,
85: MANDOCERR_MAX
86: };
87:
1.7 kristaps 88: static const char * const mandocerrs[MANDOCERR_MAX] = {
89: "ok",
90:
91: "generic warning",
92:
93: /* related to the prologue */
1.79 schwarze 94: "missing manual title, using UNTITLED",
95: "missing manual title, using \"\"",
1.54 schwarze 96: "lower case character in document title",
1.79 schwarze 97: "missing manual section, using \"\"",
1.7 kristaps 98: "unknown manual section",
1.54 schwarze 99: "missing date, using today's date",
1.7 kristaps 100: "cannot parse date, using it verbatim",
1.79 schwarze 101: "missing Os macro, using \"\"",
102: "duplicate prologue macro",
103: "late prologue macro",
104: "skipping late title macro",
1.7 kristaps 105: "prologue macros out of order",
106:
107: /* related to document structure */
108: ".so is fragile, better use ln(1)",
1.50 schwarze 109: "no document body",
1.54 schwarze 110: "content before first section header",
111: "first section is not \"NAME\"",
1.7 kristaps 112: "bad NAME section contents",
1.121 schwarze 113: "missing description line, using \"\"",
1.7 kristaps 114: "sections out of conventional order",
1.54 schwarze 115: "duplicate section title",
116: "unexpected section",
1.87 schwarze 117: "unusual Xr order",
118: "unusual Xr punctuation",
1.86 schwarze 119: "AUTHORS section without An macro",
1.7 kristaps 120:
121: /* related to macros and nesting */
1.55 schwarze 122: "obsolete macro",
1.102 schwarze 123: "macro neither callable nor escaped",
1.7 kristaps 124: "skipping paragraph macro",
1.31 schwarze 125: "moving paragraph macro out of list",
1.7 kristaps 126: "skipping no-space macro",
127: "blocks badly nested",
128: "nested displays are not portable",
1.57 schwarze 129: "moving content out of list",
130: ".Vt block has child macro",
1.78 schwarze 131: "fill mode already enabled, skipping",
132: "fill mode already disabled, skipping",
1.7 kristaps 133: "line scope broken",
134:
135: /* related to missing macro arguments */
1.58 schwarze 136: "skipping empty request",
137: "conditional request controls empty scope",
1.7 kristaps 138: "skipping empty macro",
1.62 schwarze 139: "empty argument, using 0n",
1.7 kristaps 140: "argument count wrong",
1.60 schwarze 141: "missing display type, using -ragged",
142: "list type is not the first argument",
143: "missing -width in -tag list, using 8n",
1.78 schwarze 144: "missing utility name, using \"\"",
1.60 schwarze 145: "empty head in list item",
146: "empty list item",
1.61 schwarze 147: "missing font type, using \\fR",
148: "unknown font type, using \\fR",
1.103 schwarze 149: "nothing follows prefix",
1.122 ! schwarze 150: "empty reference block",
1.60 schwarze 151: "missing -std argument, adding it",
1.90 schwarze 152: "missing eqn box, using \"\"",
1.7 kristaps 153:
154: /* related to bad macro arguments */
1.64 schwarze 155: "unterminated quoted argument",
1.7 kristaps 156: "duplicate argument",
1.76 schwarze 157: "skipping duplicate argument",
1.63 schwarze 158: "skipping duplicate display type",
159: "skipping duplicate list type",
1.76 schwarze 160: "skipping -width argument",
1.7 kristaps 161: "unknown AT&T UNIX version",
1.88 schwarze 162: "comma in function argument",
1.89 schwarze 163: "parenthesis in function name",
1.67 schwarze 164: "invalid content in Rs block",
1.63 schwarze 165: "invalid Boolean argument",
166: "unknown font, skipping request",
1.7 kristaps 167:
168: /* related to plain text */
1.64 schwarze 169: "blank line in fill mode, using .sp",
170: "tab in filled text",
171: "whitespace at end of input line",
1.7 kristaps 172: "bad comment style",
1.64 schwarze 173: "invalid escape sequence",
174: "undefined string, using \"\"",
1.16 kristaps 175:
1.118 schwarze 176: /* related to tables */
177: "tbl line starts with span",
178: "tbl column starts with span",
179: "skipping vertical bar in tbl layout",
180:
1.7 kristaps 181: "generic error",
182:
183: /* related to tables */
1.116 schwarze 184: "non-alphabetic character in tbl options",
185: "skipping unknown tbl option",
186: "missing tbl option argument",
187: "wrong tbl option argument size",
1.118 schwarze 188: "empty tbl layout",
189: "invalid character in tbl layout",
190: "unmatched parenthesis in tbl layout",
1.119 schwarze 191: "tbl without any data cells",
192: "ignoring data in spanned tbl cell",
193: "ignoring extra tbl data cells",
194: "data block open at end of tbl",
1.7 kristaps 195:
1.68 schwarze 196: /* related to document structure and macros */
1.108 schwarze 197: NULL,
1.7 kristaps 198: "input stack limit exceeded, infinite loop?",
199: "skipping bad character",
1.68 schwarze 200: "skipping unknown macro",
1.112 schwarze 201: "skipping insecure request",
1.73 schwarze 202: "skipping item outside list",
1.68 schwarze 203: "skipping column outside column list",
204: "skipping end of block that is not open",
1.115 schwarze 205: "fewer RS blocks open, skipping",
1.68 schwarze 206: "inserting missing end of block",
207: "appending missing end of block",
208:
209: /* related to request and macro arguments */
1.7 kristaps 210: "escaped character not allowed in a name",
211: "argument count wrong",
1.99 schwarze 212: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 213: "missing list type, using -item",
1.70 schwarze 214: "missing manual name, using \"\"",
1.71 schwarze 215: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 216: "unknown standard specifier",
1.71 schwarze 217: "skipping request without numeric argument",
1.109 schwarze 218: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 219: ".so request failed",
1.61 schwarze 220: "skipping all arguments",
221: "skipping excess arguments",
1.92 kristaps 222: "divide by zero",
1.112 schwarze 223:
224: "unsupported feature",
225: "input too large",
1.114 schwarze 226: "unsupported control character",
1.112 schwarze 227: "unsupported roff request",
1.119 schwarze 228: "eqn delim option in tbl",
1.118 schwarze 229: "unsupported tbl layout modifier",
1.112 schwarze 230: "ignoring macro in table",
1.7 kristaps 231: };
232:
233: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
234: "SUCCESS",
235: "RESERVED",
236: "WARNING",
237: "ERROR",
1.112 schwarze 238: "UNSUPP",
1.7 kristaps 239: "BADARG",
240: "SYSERR"
241: };
242:
1.47 schwarze 243:
1.1 kristaps 244: static void
245: resize_buf(struct buf *buf, size_t initial)
246: {
247:
248: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
249: buf->buf = mandoc_realloc(buf->buf, buf->sz);
250: }
251:
252: static void
1.84 schwarze 253: choose_parser(struct mparse *curp)
1.1 kristaps 254: {
1.83 schwarze 255: char *cp, *ep;
256: int format;
1.1 kristaps 257:
1.83 schwarze 258: /*
259: * If neither command line arguments -mdoc or -man select
260: * a parser nor the roff parser found a .Dd or .TH macro
261: * yet, look ahead in the main input buffer.
262: */
263:
264: if ((format = roff_getformat(curp->roff)) == 0) {
265: cp = curp->primary->buf;
266: ep = cp + curp->primary->sz;
267: while (cp < ep) {
1.85 schwarze 268: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 269: cp++;
270: if (cp[0] == 'D' && cp[1] == 'd') {
271: format = MPARSE_MDOC;
272: break;
273: }
274: if (cp[0] == 'T' && cp[1] == 'H') {
275: format = MPARSE_MAN;
276: break;
277: }
278: }
279: cp = memchr(cp, '\n', ep - cp);
280: if (cp == NULL)
281: break;
282: cp++;
283: }
1.1 kristaps 284: }
285:
1.83 schwarze 286: if (format == MPARSE_MDOC) {
1.47 schwarze 287: if (NULL == curp->pmdoc)
1.44 schwarze 288: curp->pmdoc = mdoc_alloc(
289: curp->roff, curp, curp->defos,
290: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 291: assert(curp->pmdoc);
292: curp->mdoc = curp->pmdoc;
293: return;
1.47 schwarze 294: }
1.1 kristaps 295:
1.83 schwarze 296: /* Fall back to man(7) as a last resort. */
297:
1.47 schwarze 298: if (NULL == curp->pman)
1.106 schwarze 299: curp->pman = man_alloc(
300: curp->roff, curp, curp->defos,
1.44 schwarze 301: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 302: assert(curp->pman);
303: curp->man = curp->pman;
304: }
305:
306: /*
1.95 schwarze 307: * Main parse routine for a buffer.
308: * It assumes encoding and line numbering are already set up.
309: * It can recurse directly (for invocations of user-defined
310: * macros, inline equations, and input line traps)
311: * and indirectly (for .so file inclusion).
1.1 kristaps 312: */
313: static void
1.95 schwarze 314: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 315: {
316: const struct tbl_span *span;
317: struct buf ln;
1.113 schwarze 318: const char *save_file;
1.110 schwarze 319: char *cp;
1.95 schwarze 320: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 321: enum rofferr rr;
1.100 schwarze 322: int of;
1.1 kristaps 323: int lnn; /* line number in the real file */
1.110 schwarze 324: int fd;
325: pid_t save_child;
1.1 kristaps 326: unsigned char c;
327:
1.95 schwarze 328: memset(&ln, 0, sizeof(ln));
1.1 kristaps 329:
1.47 schwarze 330: lnn = curp->line;
331: pos = 0;
1.1 kristaps 332:
1.95 schwarze 333: while (i < blk.sz) {
1.1 kristaps 334: if (0 == pos && '\0' == blk.buf[i])
335: break;
336:
337: if (start) {
338: curp->line = lnn;
339: curp->reparse_count = 0;
1.93 schwarze 340:
341: if (lnn < 3 &&
342: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 343: curp->filenc & MPARSE_LATIN1)
344: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 345: }
346:
1.95 schwarze 347: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 348:
349: /*
350: * When finding an unescaped newline character,
351: * leave the character loop to process the line.
352: * Skip a preceding carriage return, if any.
353: */
354:
1.95 schwarze 355: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 356: '\n' == blk.buf[i + 1])
357: ++i;
358: if ('\n' == blk.buf[i]) {
359: ++i;
360: ++lnn;
361: break;
362: }
363:
1.35 schwarze 364: /*
1.93 schwarze 365: * Make sure we have space for the worst
366: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 367: */
368:
1.95 schwarze 369: if (pos + 11 > ln.sz)
1.35 schwarze 370: resize_buf(&ln, 256);
371:
1.47 schwarze 372: /*
1.93 schwarze 373: * Encode 8-bit input.
1.1 kristaps 374: */
375:
1.93 schwarze 376: c = blk.buf[i];
377: if (c & 0x80) {
1.95 schwarze 378: if ( ! (curp->filenc && preconv_encode(
379: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 380: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
381: curp->line, pos, "0x%x", c);
1.93 schwarze 382: ln.buf[pos++] = '?';
383: i++;
384: }
385: continue;
386: }
387:
388: /*
389: * Exclude control characters.
390: */
1.1 kristaps 391:
1.93 schwarze 392: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 393: mandoc_vmsg(c == 0x00 || c == 0x04 ||
394: c > 0x0a ? MANDOCERR_CHAR_BAD :
395: MANDOCERR_CHAR_UNSUPP,
396: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 397: i++;
1.27 joerg 398: ln.buf[pos++] = '?';
1.1 kristaps 399: continue;
400: }
401:
402: /* Trailing backslash = a plain char. */
403:
1.95 schwarze 404: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 405: ln.buf[pos++] = blk.buf[i++];
406: continue;
407: }
408:
409: /*
410: * Found escape and at least one other character.
411: * When it's a newline character, skip it.
412: * When there is a carriage return in between,
413: * skip that one as well.
414: */
415:
1.95 schwarze 416: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 417: '\n' == blk.buf[i + 2])
418: ++i;
419: if ('\n' == blk.buf[i + 1]) {
420: i += 2;
421: ++lnn;
422: continue;
423: }
424:
1.13 kristaps 425: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 426: i += 2;
427: /* Comment, skip to end of line */
1.95 schwarze 428: for (; i < blk.sz; ++i) {
1.1 kristaps 429: if ('\n' == blk.buf[i]) {
430: ++i;
431: ++lnn;
432: break;
433: }
434: }
435:
436: /* Backout trailing whitespaces */
437: for (; pos > 0; --pos) {
438: if (ln.buf[pos - 1] != ' ')
439: break;
440: if (pos > 2 && ln.buf[pos - 2] == '\\')
441: break;
442: }
443: break;
444: }
445:
1.35 schwarze 446: /* Catch escaped bogus characters. */
447:
448: c = (unsigned char) blk.buf[i+1];
449:
1.47 schwarze 450: if ( ! (isascii(c) &&
451: (isgraph(c) || isblank(c)))) {
1.114 schwarze 452: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 453: curp->line, pos, "0x%x", c);
1.35 schwarze 454: i += 2;
455: ln.buf[pos++] = '?';
456: continue;
457: }
458:
1.1 kristaps 459: /* Some other escape sequence, copy & cont. */
460:
461: ln.buf[pos++] = blk.buf[i++];
462: ln.buf[pos++] = blk.buf[i++];
463: }
464:
1.95 schwarze 465: if (pos >= ln.sz)
1.1 kristaps 466: resize_buf(&ln, 256);
467:
468: ln.buf[pos] = '\0';
469:
470: /*
471: * A significant amount of complexity is contained by
472: * the roff preprocessor. It's line-oriented but can be
473: * expressed on one line, so we need at times to
474: * readjust our starting point and re-run it. The roff
475: * preprocessor can also readjust the buffers with new
476: * data, so we pass them in wholesale.
477: */
478:
479: of = 0;
480:
1.24 kristaps 481: /*
482: * Maintain a lookaside buffer of all parsed lines. We
483: * only do this if mparse_keep() has been invoked (the
484: * buffer may be accessed with mparse_getkeep()).
485: */
486:
487: if (curp->secondary) {
1.47 schwarze 488: curp->secondary->buf = mandoc_realloc(
489: curp->secondary->buf,
490: curp->secondary->sz + pos + 2);
491: memcpy(curp->secondary->buf +
492: curp->secondary->sz,
493: ln.buf, pos);
1.24 kristaps 494: curp->secondary->sz += pos;
495: curp->secondary->buf
496: [curp->secondary->sz] = '\n';
497: curp->secondary->sz++;
498: curp->secondary->buf
499: [curp->secondary->sz] = '\0';
500: }
1.1 kristaps 501: rerun:
1.96 schwarze 502: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 503:
504: switch (rr) {
1.47 schwarze 505: case ROFF_REPARSE:
1.1 kristaps 506: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 507: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 508: else
1.3 kristaps 509: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 510: curp->line, pos, NULL);
1.1 kristaps 511: pos = 0;
512: continue;
1.47 schwarze 513: case ROFF_APPEND:
1.95 schwarze 514: pos = strlen(ln.buf);
1.1 kristaps 515: continue;
1.47 schwarze 516: case ROFF_RERUN:
1.1 kristaps 517: goto rerun;
1.47 schwarze 518: case ROFF_IGN:
1.1 kristaps 519: pos = 0;
520: continue;
1.47 schwarze 521: case ROFF_SO:
1.95 schwarze 522: if ( ! (curp->options & MPARSE_SO) &&
523: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 524: curp->sodest = mandoc_strdup(ln.buf + of);
525: free(ln.buf);
526: return;
527: }
1.24 kristaps 528: /*
529: * We remove `so' clauses from our lookaside
530: * buffer because we're going to descend into
531: * the file recursively.
532: */
1.47 schwarze 533: if (curp->secondary)
1.25 kristaps 534: curp->secondary->sz -= pos + 1;
1.113 schwarze 535: save_file = curp->file;
1.110 schwarze 536: save_child = curp->child;
537: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 schwarze 538: MANDOCLEVEL_OK) {
1.110 schwarze 539: mparse_readfd(curp, fd, ln.buf + of);
1.113 schwarze 540: curp->file = save_file;
541: } else {
542: curp->file = save_file;
1.52 schwarze 543: mandoc_vmsg(MANDOCERR_SO_FAIL,
544: curp, curp->line, pos,
545: ".so %s", ln.buf + of);
1.110 schwarze 546: ln.sz = mandoc_asprintf(&cp,
547: ".sp\nSee the file %s.\n.sp",
548: ln.buf + of);
549: free(ln.buf);
550: ln.buf = cp;
551: of = 0;
552: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 553: }
1.110 schwarze 554: curp->child = save_child;
1.1 kristaps 555: pos = 0;
556: continue;
557: default:
558: break;
559: }
560:
561: /*
562: * If input parsers have not been allocated, do so now.
1.14 kristaps 563: * We keep these instanced between parsers, but set them
1.1 kristaps 564: * locally per parse routine since we can use different
565: * parsers with each one.
566: */
567:
568: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 569: choose_parser(curp);
1.1 kristaps 570:
1.47 schwarze 571: /*
1.84 schwarze 572: * Lastly, push down into the parsers themselves.
1.1 kristaps 573: * If libroff returns ROFF_TBL, then add it to the
574: * currently open parse. Since we only get here if
575: * there does exist data (see tbl_data.c), we're
576: * guaranteed that something's been allocated.
577: * Do the same for ROFF_EQN.
578: */
579:
1.100 schwarze 580: if (rr == ROFF_TBL) {
581: while ((span = roff_span(curp->roff)) != NULL)
582: if (curp->man == NULL)
583: mdoc_addspan(curp->mdoc, span);
584: else
585: man_addspan(curp->man, span);
586: } else if (rr == ROFF_EQN) {
587: if (curp->man == NULL)
588: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
589: else
590: man_addeqn(curp->man, roff_eqn(curp->roff));
591: } else if ((curp->man == NULL ?
592: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
593: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
594: break;
1.1 kristaps 595:
596: /* Temporary buffers typically are not full. */
597:
598: if (0 == start && '\0' == blk.buf[i])
599: break;
600:
601: /* Start the next input line. */
602:
603: pos = 0;
604: }
605:
606: free(ln.buf);
607: }
608:
609: static int
1.40 schwarze 610: read_whole_file(struct mparse *curp, const char *file, int fd,
611: struct buf *fb, int *with_mmap)
1.1 kristaps 612: {
613: size_t off;
614: ssize_t ssz;
615:
1.81 schwarze 616: #if HAVE_MMAP
1.15 kristaps 617: struct stat st;
1.1 kristaps 618: if (-1 == fstat(fd, &st)) {
1.107 schwarze 619: perror(file);
620: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 621: }
622:
623: /*
624: * If we're a regular file, try just reading in the whole entry
625: * via mmap(). This is faster than reading it into blocks, and
626: * since each file is only a few bytes to begin with, I'm not
627: * concerned that this is going to tank any machines.
628: */
629:
630: if (S_ISREG(st.st_mode)) {
631: if (st.st_size >= (1U << 31)) {
1.111 schwarze 632: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 633: return(0);
634: }
635: *with_mmap = 1;
636: fb->sz = (size_t)st.st_size;
1.37 schwarze 637: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 638: if (fb->buf != MAP_FAILED)
639: return(1);
640: }
1.15 kristaps 641: #endif
1.1 kristaps 642:
643: /*
644: * If this isn't a regular file (like, say, stdin), then we must
645: * go the old way and just read things in bit by bit.
646: */
647:
648: *with_mmap = 0;
649: off = 0;
650: fb->sz = 0;
651: fb->buf = NULL;
652: for (;;) {
653: if (off == fb->sz) {
654: if (fb->sz == (1U << 31)) {
1.111 schwarze 655: mandoc_msg(MANDOCERR_TOOLARGE, curp,
656: 0, 0, NULL);
1.1 kristaps 657: break;
658: }
659: resize_buf(fb, 65536);
660: }
661: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
662: if (ssz == 0) {
663: fb->sz = off;
664: return(1);
665: }
666: if (ssz == -1) {
1.107 schwarze 667: perror(file);
668: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 669: }
670: off += (size_t)ssz;
671: }
672:
673: free(fb->buf);
674: fb->buf = NULL;
675: return(0);
676: }
677:
678: static void
679: mparse_end(struct mparse *curp)
680: {
681:
1.72 schwarze 682: if (curp->mdoc == NULL &&
683: curp->man == NULL &&
684: curp->sodest == NULL) {
685: if (curp->options & MPARSE_MDOC)
686: curp->mdoc = curp->pmdoc;
687: else {
688: if (curp->pman == NULL)
1.106 schwarze 689: curp->pman = man_alloc(
690: curp->roff, curp, curp->defos,
1.72 schwarze 691: curp->options & MPARSE_QUICK ? 1 : 0);
692: curp->man = curp->pman;
693: }
694: }
1.111 schwarze 695: if (curp->mdoc)
696: mdoc_endparse(curp->mdoc);
697: if (curp->man)
698: man_endparse(curp->man);
1.1 kristaps 699: roff_endparse(curp->roff);
700: }
701:
702: static void
1.36 schwarze 703: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 704: {
1.85 schwarze 705: struct buf *svprimary;
1.28 joerg 706: const char *svfile;
1.95 schwarze 707: size_t offset;
1.36 schwarze 708: static int recursion_depth;
709:
710: if (64 < recursion_depth) {
711: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
712: return;
713: }
1.28 joerg 714:
715: /* Line number is per-file. */
716: svfile = curp->file;
717: curp->file = file;
1.85 schwarze 718: svprimary = curp->primary;
1.83 schwarze 719: curp->primary = &blk;
1.28 joerg 720: curp->line = 1;
1.36 schwarze 721: recursion_depth++;
1.28 joerg 722:
1.93 schwarze 723: /* Skip an UTF-8 byte order mark. */
724: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
725: (unsigned char)blk.buf[0] == 0xef &&
726: (unsigned char)blk.buf[1] == 0xbb &&
727: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 728: offset = 3;
1.93 schwarze 729: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 730: } else
731: offset = 0;
1.93 schwarze 732:
1.95 schwarze 733: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 734:
1.111 schwarze 735: if (--recursion_depth == 0)
1.28 joerg 736: mparse_end(curp);
737:
1.85 schwarze 738: curp->primary = svprimary;
1.28 joerg 739: curp->file = svfile;
740: }
741:
742: enum mandoclevel
1.104 schwarze 743: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 744: const char *file)
745: {
746: struct buf blk;
747:
1.104 schwarze 748: blk.buf = buf;
1.28 joerg 749: blk.sz = len;
750:
1.36 schwarze 751: mparse_parse_buffer(curp, blk, file);
1.28 joerg 752: return(curp->file_status);
753: }
754:
1.98 schwarze 755: /*
756: * Read the whole file into memory and call the parsers.
757: * Called recursively when an .so request is encountered.
758: */
1.36 schwarze 759: enum mandoclevel
760: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 761: {
1.28 joerg 762: struct buf blk;
763: int with_mmap;
1.93 schwarze 764: int save_filenc;
1.1 kristaps 765:
1.91 schwarze 766: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 767: save_filenc = curp->filenc;
768: curp->filenc = curp->options &
769: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 770: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 771: curp->filenc = save_filenc;
1.81 schwarze 772: #if HAVE_MMAP
1.91 schwarze 773: if (with_mmap)
774: munmap(blk.buf, blk.sz);
775: else
1.28 joerg 776: #endif
1.91 schwarze 777: free(blk.buf);
778: }
1.1 kristaps 779:
1.98 schwarze 780: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 781: perror(file);
1.91 schwarze 782:
1.98 schwarze 783: mparse_wait(curp);
1.1 kristaps 784: return(curp->file_status);
1.82 schwarze 785: }
786:
787: enum mandoclevel
1.97 schwarze 788: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 789: {
790: int pfd[2];
1.98 schwarze 791: int save_errno;
1.82 schwarze 792: char *cp;
793:
794: curp->file = file;
1.98 schwarze 795:
796: /* Unless zipped, try to just open the file. */
797:
1.82 schwarze 798: if ((cp = strrchr(file, '.')) == NULL ||
799: strcmp(cp + 1, "gz")) {
1.97 schwarze 800: curp->child = 0;
1.98 schwarze 801: if ((*fd = open(file, O_RDONLY)) != -1)
802: return(MANDOCLEVEL_OK);
803:
804: /* Open failed; try to append ".gz". */
805:
806: mandoc_asprintf(&cp, "%s.gz", file);
807: file = cp;
808: } else
809: cp = NULL;
810:
811: /* Before forking, make sure the file can be read. */
812:
813: save_errno = errno;
814: if (access(file, R_OK) == -1) {
815: if (cp != NULL)
816: errno = save_errno;
1.108 schwarze 817: free(cp);
818: *fd = -1;
819: curp->child = 0;
820: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
821: return(MANDOCLEVEL_ERROR);
1.82 schwarze 822: }
823:
1.98 schwarze 824: /* Run gunzip(1). */
825:
1.82 schwarze 826: if (pipe(pfd) == -1) {
1.107 schwarze 827: perror("pipe");
828: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 829: }
830:
1.97 schwarze 831: switch (curp->child = fork()) {
1.82 schwarze 832: case -1:
1.107 schwarze 833: perror("fork");
834: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 835: case 0:
836: close(pfd[0]);
837: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 838: perror("dup");
839: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 840: }
841: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 842: perror("exec");
843: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 844: default:
845: close(pfd[1]);
846: *fd = pfd[0];
847: return(MANDOCLEVEL_OK);
848: }
849: }
850:
851: enum mandoclevel
1.97 schwarze 852: mparse_wait(struct mparse *curp)
1.82 schwarze 853: {
854: int status;
855:
1.97 schwarze 856: if (curp->child == 0)
857: return(MANDOCLEVEL_OK);
858:
859: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 860: perror("wait");
861: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 862: }
863: if (WIFSIGNALED(status)) {
1.108 schwarze 864: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
865: "gunzip died from signal %d", WTERMSIG(status));
866: return(MANDOCLEVEL_ERROR);
1.82 schwarze 867: }
868: if (WEXITSTATUS(status)) {
1.108 schwarze 869: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
870: "gunzip failed with code %d", WEXITSTATUS(status));
871: return(MANDOCLEVEL_ERROR);
1.82 schwarze 872: }
873: return(MANDOCLEVEL_OK);
1.1 kristaps 874: }
875:
876: struct mparse *
1.94 schwarze 877: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
878: const struct mchars *mchars, const char *defos)
1.1 kristaps 879: {
880: struct mparse *curp;
1.10 kristaps 881:
1.1 kristaps 882: curp = mandoc_calloc(1, sizeof(struct mparse));
883:
1.44 schwarze 884: curp->options = options;
1.3 kristaps 885: curp->wlevel = wlevel;
1.1 kristaps 886: curp->mmsg = mmsg;
1.29 schwarze 887: curp->defos = defos;
1.1 kristaps 888:
1.94 schwarze 889: curp->mchars = mchars;
890: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 891: if (curp->options & MPARSE_MDOC)
892: curp->pmdoc = mdoc_alloc(
893: curp->roff, curp, curp->defos,
894: curp->options & MPARSE_QUICK ? 1 : 0);
895: if (curp->options & MPARSE_MAN)
1.106 schwarze 896: curp->pman = man_alloc(
897: curp->roff, curp, curp->defos,
1.72 schwarze 898: curp->options & MPARSE_QUICK ? 1 : 0);
899:
1.1 kristaps 900: return(curp);
901: }
902:
903: void
904: mparse_reset(struct mparse *curp)
905: {
906:
907: roff_reset(curp->roff);
908:
909: if (curp->mdoc)
910: mdoc_reset(curp->mdoc);
911: if (curp->man)
912: man_reset(curp->man);
1.24 kristaps 913: if (curp->secondary)
914: curp->secondary->sz = 0;
1.1 kristaps 915:
916: curp->file_status = MANDOCLEVEL_OK;
917: curp->mdoc = NULL;
918: curp->man = NULL;
1.45 schwarze 919:
920: free(curp->sodest);
921: curp->sodest = NULL;
1.1 kristaps 922: }
923:
924: void
925: mparse_free(struct mparse *curp)
926: {
927:
928: if (curp->pmdoc)
929: mdoc_free(curp->pmdoc);
930: if (curp->pman)
931: man_free(curp->pman);
932: if (curp->roff)
933: roff_free(curp->roff);
1.24 kristaps 934: if (curp->secondary)
935: free(curp->secondary->buf);
1.1 kristaps 936:
1.24 kristaps 937: free(curp->secondary);
1.45 schwarze 938: free(curp->sodest);
1.1 kristaps 939: free(curp);
940: }
941:
942: void
1.45 schwarze 943: mparse_result(struct mparse *curp,
944: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 945: {
946:
1.45 schwarze 947: if (sodest && NULL != (*sodest = curp->sodest)) {
948: *mdoc = NULL;
949: *man = NULL;
950: return;
951: }
1.9 kristaps 952: if (mdoc)
953: *mdoc = curp->mdoc;
954: if (man)
955: *man = curp->man;
1.3 kristaps 956: }
957:
958: void
959: mandoc_vmsg(enum mandocerr t, struct mparse *m,
960: int ln, int pos, const char *fmt, ...)
961: {
962: char buf[256];
963: va_list ap;
964:
965: va_start(ap, fmt);
1.48 schwarze 966: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 967: va_end(ap);
968:
969: mandoc_msg(t, m, ln, pos, buf);
970: }
971:
972: void
1.47 schwarze 973: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 974: int ln, int col, const char *msg)
975: {
976: enum mandoclevel level;
977:
1.112 schwarze 978: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 979: while (er < mandoclimits[level])
980: level--;
981:
1.108 schwarze 982: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 983: return;
984:
1.8 kristaps 985: if (m->mmsg)
986: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 987:
988: if (m->file_status < level)
989: m->file_status = level;
1.7 kristaps 990: }
991:
992: const char *
993: mparse_strerror(enum mandocerr er)
994: {
995:
996: return(mandocerrs[er]);
997: }
998:
999: const char *
1000: mparse_strlevel(enum mandoclevel lvl)
1001: {
1002: return(mandoclevels[lvl]);
1.24 kristaps 1003: }
1004:
1005: void
1006: mparse_keep(struct mparse *p)
1007: {
1008:
1009: assert(NULL == p->secondary);
1010: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1011: }
1012:
1013: const char *
1014: mparse_getkeep(const struct mparse *p)
1015: {
1016:
1017: assert(p->secondary);
1018: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1019: }
CVSweb