Annotation of mandoc/read.c, Revision 1.124
1.124 ! schwarze 1: /* $Id: read.c,v 1.123 2015/02/06 03:38:45 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",
1.124 ! schwarze 139: "empty block",
1.62 schwarze 140: "empty argument, using 0n",
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 \"\"",
1.123 schwarze 145: "missing function name, using \"\"",
1.60 schwarze 146: "empty head in list item",
147: "empty list item",
1.61 schwarze 148: "missing font type, using \\fR",
149: "unknown font type, using \\fR",
1.103 schwarze 150: "nothing follows prefix",
1.122 schwarze 151: "empty reference block",
1.60 schwarze 152: "missing -std argument, adding it",
1.90 schwarze 153: "missing eqn box, using \"\"",
1.7 kristaps 154:
155: /* related to bad macro arguments */
1.64 schwarze 156: "unterminated quoted argument",
1.7 kristaps 157: "duplicate argument",
1.76 schwarze 158: "skipping duplicate argument",
1.63 schwarze 159: "skipping duplicate display type",
160: "skipping duplicate list type",
1.76 schwarze 161: "skipping -width argument",
1.7 kristaps 162: "unknown AT&T UNIX version",
1.88 schwarze 163: "comma in function argument",
1.89 schwarze 164: "parenthesis in function name",
1.67 schwarze 165: "invalid content in Rs block",
1.63 schwarze 166: "invalid Boolean argument",
167: "unknown font, skipping request",
1.7 kristaps 168:
169: /* related to plain text */
1.64 schwarze 170: "blank line in fill mode, using .sp",
171: "tab in filled text",
172: "whitespace at end of input line",
1.7 kristaps 173: "bad comment style",
1.64 schwarze 174: "invalid escape sequence",
175: "undefined string, using \"\"",
1.16 kristaps 176:
1.118 schwarze 177: /* related to tables */
178: "tbl line starts with span",
179: "tbl column starts with span",
180: "skipping vertical bar in tbl layout",
181:
1.7 kristaps 182: "generic error",
183:
184: /* related to tables */
1.116 schwarze 185: "non-alphabetic character in tbl options",
186: "skipping unknown tbl option",
187: "missing tbl option argument",
188: "wrong tbl option argument size",
1.118 schwarze 189: "empty tbl layout",
190: "invalid character in tbl layout",
191: "unmatched parenthesis in tbl layout",
1.119 schwarze 192: "tbl without any data cells",
193: "ignoring data in spanned tbl cell",
194: "ignoring extra tbl data cells",
195: "data block open at end of tbl",
1.7 kristaps 196:
1.68 schwarze 197: /* related to document structure and macros */
1.108 schwarze 198: NULL,
1.7 kristaps 199: "input stack limit exceeded, infinite loop?",
200: "skipping bad character",
1.68 schwarze 201: "skipping unknown macro",
1.112 schwarze 202: "skipping insecure request",
1.73 schwarze 203: "skipping item outside list",
1.68 schwarze 204: "skipping column outside column list",
205: "skipping end of block that is not open",
1.115 schwarze 206: "fewer RS blocks open, skipping",
1.68 schwarze 207: "inserting missing end of block",
208: "appending missing end of block",
209:
210: /* related to request and macro arguments */
1.7 kristaps 211: "escaped character not allowed in a name",
212: "argument count wrong",
1.99 schwarze 213: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 214: "missing list type, using -item",
1.70 schwarze 215: "missing manual name, using \"\"",
1.71 schwarze 216: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 217: "unknown standard specifier",
1.71 schwarze 218: "skipping request without numeric argument",
1.109 schwarze 219: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 220: ".so request failed",
1.61 schwarze 221: "skipping all arguments",
222: "skipping excess arguments",
1.92 kristaps 223: "divide by zero",
1.112 schwarze 224:
225: "unsupported feature",
226: "input too large",
1.114 schwarze 227: "unsupported control character",
1.112 schwarze 228: "unsupported roff request",
1.119 schwarze 229: "eqn delim option in tbl",
1.118 schwarze 230: "unsupported tbl layout modifier",
1.112 schwarze 231: "ignoring macro in table",
1.7 kristaps 232: };
233:
234: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
235: "SUCCESS",
236: "RESERVED",
237: "WARNING",
238: "ERROR",
1.112 schwarze 239: "UNSUPP",
1.7 kristaps 240: "BADARG",
241: "SYSERR"
242: };
243:
1.47 schwarze 244:
1.1 kristaps 245: static void
246: resize_buf(struct buf *buf, size_t initial)
247: {
248:
249: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
250: buf->buf = mandoc_realloc(buf->buf, buf->sz);
251: }
252:
253: static void
1.84 schwarze 254: choose_parser(struct mparse *curp)
1.1 kristaps 255: {
1.83 schwarze 256: char *cp, *ep;
257: int format;
1.1 kristaps 258:
1.83 schwarze 259: /*
260: * If neither command line arguments -mdoc or -man select
261: * a parser nor the roff parser found a .Dd or .TH macro
262: * yet, look ahead in the main input buffer.
263: */
264:
265: if ((format = roff_getformat(curp->roff)) == 0) {
266: cp = curp->primary->buf;
267: ep = cp + curp->primary->sz;
268: while (cp < ep) {
1.85 schwarze 269: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 270: cp++;
271: if (cp[0] == 'D' && cp[1] == 'd') {
272: format = MPARSE_MDOC;
273: break;
274: }
275: if (cp[0] == 'T' && cp[1] == 'H') {
276: format = MPARSE_MAN;
277: break;
278: }
279: }
280: cp = memchr(cp, '\n', ep - cp);
281: if (cp == NULL)
282: break;
283: cp++;
284: }
1.1 kristaps 285: }
286:
1.83 schwarze 287: if (format == MPARSE_MDOC) {
1.47 schwarze 288: if (NULL == curp->pmdoc)
1.44 schwarze 289: curp->pmdoc = mdoc_alloc(
290: curp->roff, curp, curp->defos,
291: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 292: assert(curp->pmdoc);
293: curp->mdoc = curp->pmdoc;
294: return;
1.47 schwarze 295: }
1.1 kristaps 296:
1.83 schwarze 297: /* Fall back to man(7) as a last resort. */
298:
1.47 schwarze 299: if (NULL == curp->pman)
1.106 schwarze 300: curp->pman = man_alloc(
301: curp->roff, curp, curp->defos,
1.44 schwarze 302: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 303: assert(curp->pman);
304: curp->man = curp->pman;
305: }
306:
307: /*
1.95 schwarze 308: * Main parse routine for a buffer.
309: * It assumes encoding and line numbering are already set up.
310: * It can recurse directly (for invocations of user-defined
311: * macros, inline equations, and input line traps)
312: * and indirectly (for .so file inclusion).
1.1 kristaps 313: */
314: static void
1.95 schwarze 315: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 316: {
317: const struct tbl_span *span;
318: struct buf ln;
1.113 schwarze 319: const char *save_file;
1.110 schwarze 320: char *cp;
1.95 schwarze 321: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 322: enum rofferr rr;
1.100 schwarze 323: int of;
1.1 kristaps 324: int lnn; /* line number in the real file */
1.110 schwarze 325: int fd;
326: pid_t save_child;
1.1 kristaps 327: unsigned char c;
328:
1.95 schwarze 329: memset(&ln, 0, sizeof(ln));
1.1 kristaps 330:
1.47 schwarze 331: lnn = curp->line;
332: pos = 0;
1.1 kristaps 333:
1.95 schwarze 334: while (i < blk.sz) {
1.1 kristaps 335: if (0 == pos && '\0' == blk.buf[i])
336: break;
337:
338: if (start) {
339: curp->line = lnn;
340: curp->reparse_count = 0;
1.93 schwarze 341:
342: if (lnn < 3 &&
343: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 344: curp->filenc & MPARSE_LATIN1)
345: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 346: }
347:
1.95 schwarze 348: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 349:
350: /*
351: * When finding an unescaped newline character,
352: * leave the character loop to process the line.
353: * Skip a preceding carriage return, if any.
354: */
355:
1.95 schwarze 356: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 357: '\n' == blk.buf[i + 1])
358: ++i;
359: if ('\n' == blk.buf[i]) {
360: ++i;
361: ++lnn;
362: break;
363: }
364:
1.35 schwarze 365: /*
1.93 schwarze 366: * Make sure we have space for the worst
367: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 368: */
369:
1.95 schwarze 370: if (pos + 11 > ln.sz)
1.35 schwarze 371: resize_buf(&ln, 256);
372:
1.47 schwarze 373: /*
1.93 schwarze 374: * Encode 8-bit input.
1.1 kristaps 375: */
376:
1.93 schwarze 377: c = blk.buf[i];
378: if (c & 0x80) {
1.95 schwarze 379: if ( ! (curp->filenc && preconv_encode(
380: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 381: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
382: curp->line, pos, "0x%x", c);
1.93 schwarze 383: ln.buf[pos++] = '?';
384: i++;
385: }
386: continue;
387: }
388:
389: /*
390: * Exclude control characters.
391: */
1.1 kristaps 392:
1.93 schwarze 393: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 394: mandoc_vmsg(c == 0x00 || c == 0x04 ||
395: c > 0x0a ? MANDOCERR_CHAR_BAD :
396: MANDOCERR_CHAR_UNSUPP,
397: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 398: i++;
1.27 joerg 399: ln.buf[pos++] = '?';
1.1 kristaps 400: continue;
401: }
402:
403: /* Trailing backslash = a plain char. */
404:
1.95 schwarze 405: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 406: ln.buf[pos++] = blk.buf[i++];
407: continue;
408: }
409:
410: /*
411: * Found escape and at least one other character.
412: * When it's a newline character, skip it.
413: * When there is a carriage return in between,
414: * skip that one as well.
415: */
416:
1.95 schwarze 417: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 418: '\n' == blk.buf[i + 2])
419: ++i;
420: if ('\n' == blk.buf[i + 1]) {
421: i += 2;
422: ++lnn;
423: continue;
424: }
425:
1.13 kristaps 426: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 427: i += 2;
428: /* Comment, skip to end of line */
1.95 schwarze 429: for (; i < blk.sz; ++i) {
1.1 kristaps 430: if ('\n' == blk.buf[i]) {
431: ++i;
432: ++lnn;
433: break;
434: }
435: }
436:
437: /* Backout trailing whitespaces */
438: for (; pos > 0; --pos) {
439: if (ln.buf[pos - 1] != ' ')
440: break;
441: if (pos > 2 && ln.buf[pos - 2] == '\\')
442: break;
443: }
444: break;
445: }
446:
1.35 schwarze 447: /* Catch escaped bogus characters. */
448:
449: c = (unsigned char) blk.buf[i+1];
450:
1.47 schwarze 451: if ( ! (isascii(c) &&
452: (isgraph(c) || isblank(c)))) {
1.114 schwarze 453: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 454: curp->line, pos, "0x%x", c);
1.35 schwarze 455: i += 2;
456: ln.buf[pos++] = '?';
457: continue;
458: }
459:
1.1 kristaps 460: /* Some other escape sequence, copy & cont. */
461:
462: ln.buf[pos++] = blk.buf[i++];
463: ln.buf[pos++] = blk.buf[i++];
464: }
465:
1.95 schwarze 466: if (pos >= ln.sz)
1.1 kristaps 467: resize_buf(&ln, 256);
468:
469: ln.buf[pos] = '\0';
470:
471: /*
472: * A significant amount of complexity is contained by
473: * the roff preprocessor. It's line-oriented but can be
474: * expressed on one line, so we need at times to
475: * readjust our starting point and re-run it. The roff
476: * preprocessor can also readjust the buffers with new
477: * data, so we pass them in wholesale.
478: */
479:
480: of = 0;
481:
1.24 kristaps 482: /*
483: * Maintain a lookaside buffer of all parsed lines. We
484: * only do this if mparse_keep() has been invoked (the
485: * buffer may be accessed with mparse_getkeep()).
486: */
487:
488: if (curp->secondary) {
1.47 schwarze 489: curp->secondary->buf = mandoc_realloc(
490: curp->secondary->buf,
491: curp->secondary->sz + pos + 2);
492: memcpy(curp->secondary->buf +
493: curp->secondary->sz,
494: ln.buf, pos);
1.24 kristaps 495: curp->secondary->sz += pos;
496: curp->secondary->buf
497: [curp->secondary->sz] = '\n';
498: curp->secondary->sz++;
499: curp->secondary->buf
500: [curp->secondary->sz] = '\0';
501: }
1.1 kristaps 502: rerun:
1.96 schwarze 503: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 504:
505: switch (rr) {
1.47 schwarze 506: case ROFF_REPARSE:
1.1 kristaps 507: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 508: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 509: else
1.3 kristaps 510: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 511: curp->line, pos, NULL);
1.1 kristaps 512: pos = 0;
513: continue;
1.47 schwarze 514: case ROFF_APPEND:
1.95 schwarze 515: pos = strlen(ln.buf);
1.1 kristaps 516: continue;
1.47 schwarze 517: case ROFF_RERUN:
1.1 kristaps 518: goto rerun;
1.47 schwarze 519: case ROFF_IGN:
1.1 kristaps 520: pos = 0;
521: continue;
1.47 schwarze 522: case ROFF_SO:
1.95 schwarze 523: if ( ! (curp->options & MPARSE_SO) &&
524: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 525: curp->sodest = mandoc_strdup(ln.buf + of);
526: free(ln.buf);
527: return;
528: }
1.24 kristaps 529: /*
530: * We remove `so' clauses from our lookaside
531: * buffer because we're going to descend into
532: * the file recursively.
533: */
1.47 schwarze 534: if (curp->secondary)
1.25 kristaps 535: curp->secondary->sz -= pos + 1;
1.113 schwarze 536: save_file = curp->file;
1.110 schwarze 537: save_child = curp->child;
538: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 schwarze 539: MANDOCLEVEL_OK) {
1.110 schwarze 540: mparse_readfd(curp, fd, ln.buf + of);
1.113 schwarze 541: curp->file = save_file;
542: } else {
543: curp->file = save_file;
1.52 schwarze 544: mandoc_vmsg(MANDOCERR_SO_FAIL,
545: curp, curp->line, pos,
546: ".so %s", ln.buf + of);
1.110 schwarze 547: ln.sz = mandoc_asprintf(&cp,
548: ".sp\nSee the file %s.\n.sp",
549: ln.buf + of);
550: free(ln.buf);
551: ln.buf = cp;
552: of = 0;
553: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 554: }
1.110 schwarze 555: curp->child = save_child;
1.1 kristaps 556: pos = 0;
557: continue;
558: default:
559: break;
560: }
561:
562: /*
563: * If input parsers have not been allocated, do so now.
1.14 kristaps 564: * We keep these instanced between parsers, but set them
1.1 kristaps 565: * locally per parse routine since we can use different
566: * parsers with each one.
567: */
568:
569: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 570: choose_parser(curp);
1.1 kristaps 571:
1.47 schwarze 572: /*
1.84 schwarze 573: * Lastly, push down into the parsers themselves.
1.1 kristaps 574: * If libroff returns ROFF_TBL, then add it to the
575: * currently open parse. Since we only get here if
576: * there does exist data (see tbl_data.c), we're
577: * guaranteed that something's been allocated.
578: * Do the same for ROFF_EQN.
579: */
580:
1.100 schwarze 581: if (rr == ROFF_TBL) {
582: while ((span = roff_span(curp->roff)) != NULL)
583: if (curp->man == NULL)
584: mdoc_addspan(curp->mdoc, span);
585: else
586: man_addspan(curp->man, span);
587: } else if (rr == ROFF_EQN) {
588: if (curp->man == NULL)
589: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
590: else
591: man_addeqn(curp->man, roff_eqn(curp->roff));
592: } else if ((curp->man == NULL ?
593: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
594: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
595: break;
1.1 kristaps 596:
597: /* Temporary buffers typically are not full. */
598:
599: if (0 == start && '\0' == blk.buf[i])
600: break;
601:
602: /* Start the next input line. */
603:
604: pos = 0;
605: }
606:
607: free(ln.buf);
608: }
609:
610: static int
1.40 schwarze 611: read_whole_file(struct mparse *curp, const char *file, int fd,
612: struct buf *fb, int *with_mmap)
1.1 kristaps 613: {
614: size_t off;
615: ssize_t ssz;
616:
1.81 schwarze 617: #if HAVE_MMAP
1.15 kristaps 618: struct stat st;
1.1 kristaps 619: if (-1 == fstat(fd, &st)) {
1.107 schwarze 620: perror(file);
621: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 622: }
623:
624: /*
625: * If we're a regular file, try just reading in the whole entry
626: * via mmap(). This is faster than reading it into blocks, and
627: * since each file is only a few bytes to begin with, I'm not
628: * concerned that this is going to tank any machines.
629: */
630:
631: if (S_ISREG(st.st_mode)) {
632: if (st.st_size >= (1U << 31)) {
1.111 schwarze 633: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 634: return(0);
635: }
636: *with_mmap = 1;
637: fb->sz = (size_t)st.st_size;
1.37 schwarze 638: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 639: if (fb->buf != MAP_FAILED)
640: return(1);
641: }
1.15 kristaps 642: #endif
1.1 kristaps 643:
644: /*
645: * If this isn't a regular file (like, say, stdin), then we must
646: * go the old way and just read things in bit by bit.
647: */
648:
649: *with_mmap = 0;
650: off = 0;
651: fb->sz = 0;
652: fb->buf = NULL;
653: for (;;) {
654: if (off == fb->sz) {
655: if (fb->sz == (1U << 31)) {
1.111 schwarze 656: mandoc_msg(MANDOCERR_TOOLARGE, curp,
657: 0, 0, NULL);
1.1 kristaps 658: break;
659: }
660: resize_buf(fb, 65536);
661: }
662: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
663: if (ssz == 0) {
664: fb->sz = off;
665: return(1);
666: }
667: if (ssz == -1) {
1.107 schwarze 668: perror(file);
669: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 670: }
671: off += (size_t)ssz;
672: }
673:
674: free(fb->buf);
675: fb->buf = NULL;
676: return(0);
677: }
678:
679: static void
680: mparse_end(struct mparse *curp)
681: {
682:
1.72 schwarze 683: if (curp->mdoc == NULL &&
684: curp->man == NULL &&
685: curp->sodest == NULL) {
686: if (curp->options & MPARSE_MDOC)
687: curp->mdoc = curp->pmdoc;
688: else {
689: if (curp->pman == NULL)
1.106 schwarze 690: curp->pman = man_alloc(
691: curp->roff, curp, curp->defos,
1.72 schwarze 692: curp->options & MPARSE_QUICK ? 1 : 0);
693: curp->man = curp->pman;
694: }
695: }
1.111 schwarze 696: if (curp->mdoc)
697: mdoc_endparse(curp->mdoc);
698: if (curp->man)
699: man_endparse(curp->man);
1.1 kristaps 700: roff_endparse(curp->roff);
701: }
702:
703: static void
1.36 schwarze 704: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 705: {
1.85 schwarze 706: struct buf *svprimary;
1.28 joerg 707: const char *svfile;
1.95 schwarze 708: size_t offset;
1.36 schwarze 709: static int recursion_depth;
710:
711: if (64 < recursion_depth) {
712: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
713: return;
714: }
1.28 joerg 715:
716: /* Line number is per-file. */
717: svfile = curp->file;
718: curp->file = file;
1.85 schwarze 719: svprimary = curp->primary;
1.83 schwarze 720: curp->primary = &blk;
1.28 joerg 721: curp->line = 1;
1.36 schwarze 722: recursion_depth++;
1.28 joerg 723:
1.93 schwarze 724: /* Skip an UTF-8 byte order mark. */
725: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
726: (unsigned char)blk.buf[0] == 0xef &&
727: (unsigned char)blk.buf[1] == 0xbb &&
728: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 729: offset = 3;
1.93 schwarze 730: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 731: } else
732: offset = 0;
1.93 schwarze 733:
1.95 schwarze 734: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 735:
1.111 schwarze 736: if (--recursion_depth == 0)
1.28 joerg 737: mparse_end(curp);
738:
1.85 schwarze 739: curp->primary = svprimary;
1.28 joerg 740: curp->file = svfile;
741: }
742:
743: enum mandoclevel
1.104 schwarze 744: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 745: const char *file)
746: {
747: struct buf blk;
748:
1.104 schwarze 749: blk.buf = buf;
1.28 joerg 750: blk.sz = len;
751:
1.36 schwarze 752: mparse_parse_buffer(curp, blk, file);
1.28 joerg 753: return(curp->file_status);
754: }
755:
1.98 schwarze 756: /*
757: * Read the whole file into memory and call the parsers.
758: * Called recursively when an .so request is encountered.
759: */
1.36 schwarze 760: enum mandoclevel
761: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 762: {
1.28 joerg 763: struct buf blk;
764: int with_mmap;
1.93 schwarze 765: int save_filenc;
1.1 kristaps 766:
1.91 schwarze 767: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 768: save_filenc = curp->filenc;
769: curp->filenc = curp->options &
770: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 771: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 772: curp->filenc = save_filenc;
1.81 schwarze 773: #if HAVE_MMAP
1.91 schwarze 774: if (with_mmap)
775: munmap(blk.buf, blk.sz);
776: else
1.28 joerg 777: #endif
1.91 schwarze 778: free(blk.buf);
779: }
1.1 kristaps 780:
1.98 schwarze 781: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 782: perror(file);
1.91 schwarze 783:
1.98 schwarze 784: mparse_wait(curp);
1.1 kristaps 785: return(curp->file_status);
1.82 schwarze 786: }
787:
788: enum mandoclevel
1.97 schwarze 789: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 790: {
791: int pfd[2];
1.98 schwarze 792: int save_errno;
1.82 schwarze 793: char *cp;
794:
795: curp->file = file;
1.98 schwarze 796:
797: /* Unless zipped, try to just open the file. */
798:
1.82 schwarze 799: if ((cp = strrchr(file, '.')) == NULL ||
800: strcmp(cp + 1, "gz")) {
1.97 schwarze 801: curp->child = 0;
1.98 schwarze 802: if ((*fd = open(file, O_RDONLY)) != -1)
803: return(MANDOCLEVEL_OK);
804:
805: /* Open failed; try to append ".gz". */
806:
807: mandoc_asprintf(&cp, "%s.gz", file);
808: file = cp;
809: } else
810: cp = NULL;
811:
812: /* Before forking, make sure the file can be read. */
813:
814: save_errno = errno;
815: if (access(file, R_OK) == -1) {
816: if (cp != NULL)
817: errno = save_errno;
1.108 schwarze 818: free(cp);
819: *fd = -1;
820: curp->child = 0;
821: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
822: return(MANDOCLEVEL_ERROR);
1.82 schwarze 823: }
824:
1.98 schwarze 825: /* Run gunzip(1). */
826:
1.82 schwarze 827: if (pipe(pfd) == -1) {
1.107 schwarze 828: perror("pipe");
829: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 830: }
831:
1.97 schwarze 832: switch (curp->child = fork()) {
1.82 schwarze 833: case -1:
1.107 schwarze 834: perror("fork");
835: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 836: case 0:
837: close(pfd[0]);
838: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 839: perror("dup");
840: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 841: }
842: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 843: perror("exec");
844: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 845: default:
846: close(pfd[1]);
847: *fd = pfd[0];
848: return(MANDOCLEVEL_OK);
849: }
850: }
851:
852: enum mandoclevel
1.97 schwarze 853: mparse_wait(struct mparse *curp)
1.82 schwarze 854: {
855: int status;
856:
1.97 schwarze 857: if (curp->child == 0)
858: return(MANDOCLEVEL_OK);
859:
860: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 861: perror("wait");
862: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 863: }
864: if (WIFSIGNALED(status)) {
1.108 schwarze 865: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
866: "gunzip died from signal %d", WTERMSIG(status));
867: return(MANDOCLEVEL_ERROR);
1.82 schwarze 868: }
869: if (WEXITSTATUS(status)) {
1.108 schwarze 870: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
871: "gunzip failed with code %d", WEXITSTATUS(status));
872: return(MANDOCLEVEL_ERROR);
1.82 schwarze 873: }
874: return(MANDOCLEVEL_OK);
1.1 kristaps 875: }
876:
877: struct mparse *
1.94 schwarze 878: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
879: const struct mchars *mchars, const char *defos)
1.1 kristaps 880: {
881: struct mparse *curp;
1.10 kristaps 882:
1.1 kristaps 883: curp = mandoc_calloc(1, sizeof(struct mparse));
884:
1.44 schwarze 885: curp->options = options;
1.3 kristaps 886: curp->wlevel = wlevel;
1.1 kristaps 887: curp->mmsg = mmsg;
1.29 schwarze 888: curp->defos = defos;
1.1 kristaps 889:
1.94 schwarze 890: curp->mchars = mchars;
891: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 892: if (curp->options & MPARSE_MDOC)
893: curp->pmdoc = mdoc_alloc(
894: curp->roff, curp, curp->defos,
895: curp->options & MPARSE_QUICK ? 1 : 0);
896: if (curp->options & MPARSE_MAN)
1.106 schwarze 897: curp->pman = man_alloc(
898: curp->roff, curp, curp->defos,
1.72 schwarze 899: curp->options & MPARSE_QUICK ? 1 : 0);
900:
1.1 kristaps 901: return(curp);
902: }
903:
904: void
905: mparse_reset(struct mparse *curp)
906: {
907:
908: roff_reset(curp->roff);
909:
910: if (curp->mdoc)
911: mdoc_reset(curp->mdoc);
912: if (curp->man)
913: man_reset(curp->man);
1.24 kristaps 914: if (curp->secondary)
915: curp->secondary->sz = 0;
1.1 kristaps 916:
917: curp->file_status = MANDOCLEVEL_OK;
918: curp->mdoc = NULL;
919: curp->man = NULL;
1.45 schwarze 920:
921: free(curp->sodest);
922: curp->sodest = NULL;
1.1 kristaps 923: }
924:
925: void
926: mparse_free(struct mparse *curp)
927: {
928:
929: if (curp->pmdoc)
930: mdoc_free(curp->pmdoc);
931: if (curp->pman)
932: man_free(curp->pman);
933: if (curp->roff)
934: roff_free(curp->roff);
1.24 kristaps 935: if (curp->secondary)
936: free(curp->secondary->buf);
1.1 kristaps 937:
1.24 kristaps 938: free(curp->secondary);
1.45 schwarze 939: free(curp->sodest);
1.1 kristaps 940: free(curp);
941: }
942:
943: void
1.45 schwarze 944: mparse_result(struct mparse *curp,
945: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 946: {
947:
1.45 schwarze 948: if (sodest && NULL != (*sodest = curp->sodest)) {
949: *mdoc = NULL;
950: *man = NULL;
951: return;
952: }
1.9 kristaps 953: if (mdoc)
954: *mdoc = curp->mdoc;
955: if (man)
956: *man = curp->man;
1.3 kristaps 957: }
958:
959: void
960: mandoc_vmsg(enum mandocerr t, struct mparse *m,
961: int ln, int pos, const char *fmt, ...)
962: {
963: char buf[256];
964: va_list ap;
965:
966: va_start(ap, fmt);
1.48 schwarze 967: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 968: va_end(ap);
969:
970: mandoc_msg(t, m, ln, pos, buf);
971: }
972:
973: void
1.47 schwarze 974: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 975: int ln, int col, const char *msg)
976: {
977: enum mandoclevel level;
978:
1.112 schwarze 979: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 980: while (er < mandoclimits[level])
981: level--;
982:
1.108 schwarze 983: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 984: return;
985:
1.8 kristaps 986: if (m->mmsg)
987: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 988:
989: if (m->file_status < level)
990: m->file_status = level;
1.7 kristaps 991: }
992:
993: const char *
994: mparse_strerror(enum mandocerr er)
995: {
996:
997: return(mandocerrs[er]);
998: }
999:
1000: const char *
1001: mparse_strlevel(enum mandoclevel lvl)
1002: {
1003: return(mandoclevels[lvl]);
1.24 kristaps 1004: }
1005:
1006: void
1007: mparse_keep(struct mparse *p)
1008: {
1009:
1010: assert(NULL == p->secondary);
1011: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1012: }
1013:
1014: const char *
1015: mparse_getkeep(const struct mparse *p)
1016: {
1017:
1018: assert(p->secondary);
1019: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1020: }
CVSweb