Annotation of src/usr.bin/mandoc/read.c, Revision 1.102
1.102 ! schwarze 1: /* $OpenBSD: read.c,v 1.101 2015/02/06 16:05:51 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.82 schwarze 4: * Copyright (c) 2010-2015 Ingo Schwarze <schwarze@openbsd.org>
1.20 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 schwarze 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: */
1.58 schwarze 19: #include <sys/types.h>
20: #include <sys/mman.h>
1.1 schwarze 21: #include <sys/stat.h>
1.58 schwarze 22: #include <sys/wait.h>
1.1 schwarze 23:
24: #include <assert.h>
25: #include <ctype.h>
1.18 schwarze 26: #include <errno.h>
1.1 schwarze 27: #include <fcntl.h>
28: #include <stdarg.h>
29: #include <stdio.h>
30: #include <stdlib.h>
31: #include <string.h>
32: #include <unistd.h>
33:
34: #include "mandoc.h"
1.24 schwarze 35: #include "mandoc_aux.h"
1.1 schwarze 36: #include "libmandoc.h"
37: #include "mdoc.h"
38: #include "man.h"
39:
40: #define REPARSE_LIMIT 1000
41:
42: struct mparse {
43: struct man *pman; /* persistent man parser */
44: struct mdoc *pmdoc; /* persistent mdoc parser */
45: struct man *man; /* man parser */
46: struct mdoc *mdoc; /* mdoc parser */
47: struct roff *roff; /* roff parser (!NULL) */
1.69 schwarze 48: const struct mchars *mchars; /* character table */
1.23 schwarze 49: char *sodest; /* filename pointed to by .so */
1.59 schwarze 50: const char *file; /* filename of current input file */
51: struct buf *primary; /* buffer currently being parsed */
52: struct buf *secondary; /* preprocessed copy of input */
53: const char *defos; /* default operating system */
54: mandocmsg mmsg; /* warning/error message handler */
55: enum mandoclevel file_status; /* status of current parse */
56: enum mandoclevel wlevel; /* ignore messages below this */
57: int options; /* parser options */
1.70 schwarze 58: int filenc; /* encoding of the current file */
1.1 schwarze 59: int reparse_count; /* finite interp. stack */
1.59 schwarze 60: int line; /* line number in the file */
1.73 schwarze 61: pid_t child; /* the gunzip(1) process */
1.1 schwarze 62: };
63:
1.60 schwarze 64: static void choose_parser(struct mparse *);
1.1 schwarze 65: static void resize_buf(struct buf *, size_t);
1.71 schwarze 66: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.18 schwarze 67: static int read_whole_file(struct mparse *, const char *, int,
68: struct buf *, int *);
1.1 schwarze 69: static void mparse_end(struct mparse *);
1.15 schwarze 70: static void mparse_parse_buffer(struct mparse *, struct buf,
71: const char *);
1.1 schwarze 72:
73: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
74: MANDOCERR_OK,
75: MANDOCERR_WARNING,
76: MANDOCERR_WARNING,
77: MANDOCERR_ERROR,
1.87 schwarze 78: MANDOCERR_UNSUPP,
1.1 schwarze 79: MANDOCERR_MAX,
80: MANDOCERR_MAX
81: };
82:
83: static const char * const mandocerrs[MANDOCERR_MAX] = {
84: "ok",
85:
86: "generic warning",
87:
88: /* related to the prologue */
1.57 schwarze 89: "missing manual title, using UNTITLED",
90: "missing manual title, using \"\"",
1.32 schwarze 91: "lower case character in document title",
1.57 schwarze 92: "missing manual section, using \"\"",
1.1 schwarze 93: "unknown manual section",
1.32 schwarze 94: "missing date, using today's date",
1.1 schwarze 95: "cannot parse date, using it verbatim",
1.57 schwarze 96: "missing Os macro, using \"\"",
97: "duplicate prologue macro",
98: "late prologue macro",
99: "skipping late title macro",
1.1 schwarze 100: "prologue macros out of order",
101:
102: /* related to document structure */
103: ".so is fragile, better use ln(1)",
1.28 schwarze 104: "no document body",
1.32 schwarze 105: "content before first section header",
106: "first section is not \"NAME\"",
1.1 schwarze 107: "bad NAME section contents",
1.96 schwarze 108: "missing description line, using \"\"",
1.1 schwarze 109: "sections out of conventional order",
1.32 schwarze 110: "duplicate section title",
111: "unexpected section",
1.63 schwarze 112: "unusual Xr order",
113: "unusual Xr punctuation",
1.62 schwarze 114: "AUTHORS section without An macro",
1.1 schwarze 115:
116: /* related to macros and nesting */
1.33 schwarze 117: "obsolete macro",
1.78 schwarze 118: "macro neither callable nor escaped",
1.1 schwarze 119: "skipping paragraph macro",
1.10 schwarze 120: "moving paragraph macro out of list",
1.1 schwarze 121: "skipping no-space macro",
122: "blocks badly nested",
123: "nested displays are not portable",
1.35 schwarze 124: "moving content out of list",
125: ".Vt block has child macro",
1.56 schwarze 126: "fill mode already enabled, skipping",
127: "fill mode already disabled, skipping",
1.1 schwarze 128: "line scope broken",
129:
130: /* related to missing macro arguments */
1.36 schwarze 131: "skipping empty request",
132: "conditional request controls empty scope",
1.1 schwarze 133: "skipping empty macro",
1.99 schwarze 134: "empty block",
1.40 schwarze 135: "empty argument, using 0n",
1.38 schwarze 136: "missing display type, using -ragged",
137: "list type is not the first argument",
138: "missing -width in -tag list, using 8n",
1.56 schwarze 139: "missing utility name, using \"\"",
1.98 schwarze 140: "missing function name, using \"\"",
1.38 schwarze 141: "empty head in list item",
142: "empty list item",
1.39 schwarze 143: "missing font type, using \\fR",
144: "unknown font type, using \\fR",
1.79 schwarze 145: "nothing follows prefix",
1.97 schwarze 146: "empty reference block",
1.38 schwarze 147: "missing -std argument, adding it",
1.100 schwarze 148: "missing option string, using \"\"",
149: "missing resource identifier, using \"\"",
1.66 schwarze 150: "missing eqn box, using \"\"",
1.1 schwarze 151:
152: /* related to bad macro arguments */
1.42 schwarze 153: "unterminated quoted argument",
1.1 schwarze 154: "duplicate argument",
1.54 schwarze 155: "skipping duplicate argument",
1.41 schwarze 156: "skipping duplicate display type",
157: "skipping duplicate list type",
1.54 schwarze 158: "skipping -width argument",
1.101 schwarze 159: "wrong number of cells",
1.1 schwarze 160: "unknown AT&T UNIX version",
1.64 schwarze 161: "comma in function argument",
1.65 schwarze 162: "parenthesis in function name",
1.45 schwarze 163: "invalid content in Rs block",
1.41 schwarze 164: "invalid Boolean argument",
165: "unknown font, skipping request",
1.101 schwarze 166: "odd number of characters in request",
1.1 schwarze 167:
168: /* related to plain text */
1.42 schwarze 169: "blank line in fill mode, using .sp",
170: "tab in filled text",
171: "whitespace at end of input line",
1.1 schwarze 172: "bad comment style",
1.42 schwarze 173: "invalid escape sequence",
174: "undefined string, using \"\"",
1.3 schwarze 175:
1.93 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:
1.1 schwarze 181: "generic error",
1.3 schwarze 182:
1.1 schwarze 183: /* related to tables */
1.91 schwarze 184: "non-alphabetic character in tbl options",
185: "skipping unknown tbl option",
186: "missing tbl option argument",
187: "wrong tbl option argument size",
1.93 schwarze 188: "empty tbl layout",
189: "invalid character in tbl layout",
190: "unmatched parenthesis in tbl layout",
1.94 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",
1.1 schwarze 195:
1.46 schwarze 196: /* related to document structure and macros */
1.83 schwarze 197: NULL,
1.1 schwarze 198: "input stack limit exceeded, infinite loop?",
199: "skipping bad character",
1.46 schwarze 200: "skipping unknown macro",
1.87 schwarze 201: "skipping insecure request",
1.51 schwarze 202: "skipping item outside list",
1.46 schwarze 203: "skipping column outside column list",
204: "skipping end of block that is not open",
1.90 schwarze 205: "fewer RS blocks open, skipping",
1.46 schwarze 206: "inserting missing end of block",
207: "appending missing end of block",
208:
209: /* related to request and macro arguments */
1.1 schwarze 210: "escaped character not allowed in a name",
1.75 schwarze 211: "NOT IMPLEMENTED: Bd -file",
1.49 schwarze 212: "missing list type, using -item",
1.48 schwarze 213: "missing manual name, using \"\"",
1.49 schwarze 214: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 215: "unknown standard specifier",
1.49 schwarze 216: "skipping request without numeric argument",
1.84 schwarze 217: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 218: ".so request failed",
1.39 schwarze 219: "skipping all arguments",
220: "skipping excess arguments",
1.68 schwarze 221: "divide by zero",
1.87 schwarze 222:
223: "unsupported feature",
224: "input too large",
1.89 schwarze 225: "unsupported control character",
1.87 schwarze 226: "unsupported roff request",
1.94 schwarze 227: "eqn delim option in tbl",
1.93 schwarze 228: "unsupported tbl layout modifier",
1.87 schwarze 229: "ignoring macro in table",
1.1 schwarze 230: };
231:
232: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
233: "SUCCESS",
234: "RESERVED",
235: "WARNING",
236: "ERROR",
1.87 schwarze 237: "UNSUPP",
1.1 schwarze 238: "BADARG",
239: "SYSERR"
240: };
241:
1.25 schwarze 242:
1.1 schwarze 243: static void
244: resize_buf(struct buf *buf, size_t initial)
245: {
246:
247: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
248: buf->buf = mandoc_realloc(buf->buf, buf->sz);
249: }
250:
251: static void
1.60 schwarze 252: choose_parser(struct mparse *curp)
1.1 schwarze 253: {
1.59 schwarze 254: char *cp, *ep;
255: int format;
1.1 schwarze 256:
1.59 schwarze 257: /*
258: * If neither command line arguments -mdoc or -man select
259: * a parser nor the roff parser found a .Dd or .TH macro
260: * yet, look ahead in the main input buffer.
261: */
262:
263: if ((format = roff_getformat(curp->roff)) == 0) {
264: cp = curp->primary->buf;
265: ep = cp + curp->primary->sz;
266: while (cp < ep) {
1.61 schwarze 267: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 268: cp++;
269: if (cp[0] == 'D' && cp[1] == 'd') {
270: format = MPARSE_MDOC;
271: break;
272: }
273: if (cp[0] == 'T' && cp[1] == 'H') {
274: format = MPARSE_MAN;
275: break;
276: }
277: }
278: cp = memchr(cp, '\n', ep - cp);
279: if (cp == NULL)
280: break;
281: cp++;
282: }
1.1 schwarze 283: }
284:
1.59 schwarze 285: if (format == MPARSE_MDOC) {
1.25 schwarze 286: if (NULL == curp->pmdoc)
1.22 schwarze 287: curp->pmdoc = mdoc_alloc(
288: curp->roff, curp, curp->defos,
289: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 schwarze 290: assert(curp->pmdoc);
291: curp->mdoc = curp->pmdoc;
292: return;
1.25 schwarze 293: }
1.1 schwarze 294:
1.59 schwarze 295: /* Fall back to man(7) as a last resort. */
296:
1.25 schwarze 297: if (NULL == curp->pman)
1.81 schwarze 298: curp->pman = man_alloc(
299: curp->roff, curp, curp->defos,
1.22 schwarze 300: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 schwarze 301: assert(curp->pman);
302: curp->man = curp->pman;
303: }
304:
305: /*
1.71 schwarze 306: * Main parse routine for a buffer.
307: * It assumes encoding and line numbering are already set up.
308: * It can recurse directly (for invocations of user-defined
309: * macros, inline equations, and input line traps)
310: * and indirectly (for .so file inclusion).
1.1 schwarze 311: */
312: static void
1.71 schwarze 313: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 314: {
315: const struct tbl_span *span;
316: struct buf ln;
1.88 schwarze 317: const char *save_file;
1.85 schwarze 318: char *cp;
1.71 schwarze 319: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 320: enum rofferr rr;
1.76 schwarze 321: int of;
1.1 schwarze 322: int lnn; /* line number in the real file */
1.85 schwarze 323: int fd;
324: pid_t save_child;
1.1 schwarze 325: unsigned char c;
326:
1.71 schwarze 327: memset(&ln, 0, sizeof(ln));
1.1 schwarze 328:
1.25 schwarze 329: lnn = curp->line;
330: pos = 0;
1.1 schwarze 331:
1.71 schwarze 332: while (i < blk.sz) {
1.1 schwarze 333: if (0 == pos && '\0' == blk.buf[i])
334: break;
335:
336: if (start) {
337: curp->line = lnn;
338: curp->reparse_count = 0;
1.70 schwarze 339:
340: if (lnn < 3 &&
341: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 342: curp->filenc & MPARSE_LATIN1)
343: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 344: }
345:
1.71 schwarze 346: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 347:
348: /*
349: * When finding an unescaped newline character,
350: * leave the character loop to process the line.
351: * Skip a preceding carriage return, if any.
352: */
353:
1.71 schwarze 354: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 355: '\n' == blk.buf[i + 1])
356: ++i;
357: if ('\n' == blk.buf[i]) {
358: ++i;
359: ++lnn;
360: break;
361: }
362:
1.13 schwarze 363: /*
1.70 schwarze 364: * Make sure we have space for the worst
365: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 366: */
367:
1.71 schwarze 368: if (pos + 11 > ln.sz)
1.13 schwarze 369: resize_buf(&ln, 256);
370:
1.25 schwarze 371: /*
1.70 schwarze 372: * Encode 8-bit input.
1.1 schwarze 373: */
374:
1.70 schwarze 375: c = blk.buf[i];
376: if (c & 0x80) {
1.71 schwarze 377: if ( ! (curp->filenc && preconv_encode(
378: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 379: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
380: curp->line, pos, "0x%x", c);
1.70 schwarze 381: ln.buf[pos++] = '?';
382: i++;
383: }
384: continue;
385: }
386:
387: /*
388: * Exclude control characters.
389: */
1.1 schwarze 390:
1.70 schwarze 391: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 392: mandoc_vmsg(c == 0x00 || c == 0x04 ||
393: c > 0x0a ? MANDOCERR_CHAR_BAD :
394: MANDOCERR_CHAR_UNSUPP,
395: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 396: i++;
1.102 ! schwarze 397: if (c != '\r')
! 398: ln.buf[pos++] = '?';
1.1 schwarze 399: continue;
400: }
401:
402: /* Trailing backslash = a plain char. */
403:
1.71 schwarze 404: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 schwarze 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:
1.71 schwarze 416: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 schwarze 417: '\n' == blk.buf[i + 2])
418: ++i;
419: if ('\n' == blk.buf[i + 1]) {
420: i += 2;
421: ++lnn;
422: continue;
423: }
424:
425: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
426: i += 2;
427: /* Comment, skip to end of line */
1.71 schwarze 428: for (; i < blk.sz; ++i) {
1.1 schwarze 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:
1.13 schwarze 446: /* Catch escaped bogus characters. */
447:
448: c = (unsigned char) blk.buf[i+1];
449:
1.25 schwarze 450: if ( ! (isascii(c) &&
451: (isgraph(c) || isblank(c)))) {
1.89 schwarze 452: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.56 schwarze 453: curp->line, pos, "0x%x", c);
1.13 schwarze 454: i += 2;
455: ln.buf[pos++] = '?';
456: continue;
457: }
458:
1.1 schwarze 459: /* Some other escape sequence, copy & cont. */
460:
461: ln.buf[pos++] = blk.buf[i++];
462: ln.buf[pos++] = blk.buf[i++];
463: }
464:
1.71 schwarze 465: if (pos >= ln.sz)
1.1 schwarze 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:
1.4 schwarze 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) {
1.25 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);
1.4 schwarze 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 schwarze 501: rerun:
1.72 schwarze 502: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 503:
504: switch (rr) {
1.25 schwarze 505: case ROFF_REPARSE:
1.1 schwarze 506: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.71 schwarze 507: mparse_buf_r(curp, ln, of, 0);
1.1 schwarze 508: else
509: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 510: curp->line, pos, NULL);
1.1 schwarze 511: pos = 0;
512: continue;
1.25 schwarze 513: case ROFF_APPEND:
1.71 schwarze 514: pos = strlen(ln.buf);
1.1 schwarze 515: continue;
1.25 schwarze 516: case ROFF_RERUN:
1.1 schwarze 517: goto rerun;
1.25 schwarze 518: case ROFF_IGN:
1.1 schwarze 519: pos = 0;
520: continue;
1.25 schwarze 521: case ROFF_SO:
1.71 schwarze 522: if ( ! (curp->options & MPARSE_SO) &&
523: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 524: curp->sodest = mandoc_strdup(ln.buf + of);
525: free(ln.buf);
526: return;
527: }
1.4 schwarze 528: /*
529: * We remove `so' clauses from our lookaside
530: * buffer because we're going to descend into
531: * the file recursively.
532: */
1.25 schwarze 533: if (curp->secondary)
1.4 schwarze 534: curp->secondary->sz -= pos + 1;
1.88 schwarze 535: save_file = curp->file;
1.85 schwarze 536: save_child = curp->child;
537: if (mparse_open(curp, &fd, ln.buf + of) ==
1.88 schwarze 538: MANDOCLEVEL_OK) {
1.85 schwarze 539: mparse_readfd(curp, fd, ln.buf + of);
1.88 schwarze 540: curp->file = save_file;
541: } else {
542: curp->file = save_file;
1.30 schwarze 543: mandoc_vmsg(MANDOCERR_SO_FAIL,
544: curp, curp->line, pos,
545: ".so %s", ln.buf + of);
1.85 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);
1.30 schwarze 553: }
1.85 schwarze 554: curp->child = save_child;
1.1 schwarze 555: pos = 0;
556: continue;
557: default:
558: break;
559: }
560:
561: /*
562: * If input parsers have not been allocated, do so now.
1.2 schwarze 563: * We keep these instanced between parsers, but set them
1.1 schwarze 564: * locally per parse routine since we can use different
565: * parsers with each one.
566: */
567:
568: if ( ! (curp->man || curp->mdoc))
1.60 schwarze 569: choose_parser(curp);
1.1 schwarze 570:
1.25 schwarze 571: /*
1.60 schwarze 572: * Lastly, push down into the parsers themselves.
1.1 schwarze 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:
1.76 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 schwarze 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
1.18 schwarze 610: read_whole_file(struct mparse *curp, const char *file, int fd,
611: struct buf *fb, int *with_mmap)
1.1 schwarze 612: {
613: struct stat st;
614: size_t off;
615: ssize_t ssz;
616:
617: if (-1 == fstat(fd, &st)) {
1.82 schwarze 618: perror(file);
619: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 620: }
621:
622: /*
623: * If we're a regular file, try just reading in the whole entry
624: * via mmap(). This is faster than reading it into blocks, and
625: * since each file is only a few bytes to begin with, I'm not
626: * concerned that this is going to tank any machines.
627: */
628:
629: if (S_ISREG(st.st_mode)) {
630: if (st.st_size >= (1U << 31)) {
1.86 schwarze 631: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 schwarze 632: return(0);
633: }
634: *with_mmap = 1;
635: fb->sz = (size_t)st.st_size;
1.15 schwarze 636: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 637: if (fb->buf != MAP_FAILED)
638: return(1);
639: }
640:
641: /*
642: * If this isn't a regular file (like, say, stdin), then we must
643: * go the old way and just read things in bit by bit.
644: */
645:
646: *with_mmap = 0;
647: off = 0;
648: fb->sz = 0;
649: fb->buf = NULL;
650: for (;;) {
651: if (off == fb->sz) {
652: if (fb->sz == (1U << 31)) {
1.86 schwarze 653: mandoc_msg(MANDOCERR_TOOLARGE, curp,
654: 0, 0, NULL);
1.1 schwarze 655: break;
656: }
657: resize_buf(fb, 65536);
658: }
659: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
660: if (ssz == 0) {
661: fb->sz = off;
662: return(1);
663: }
664: if (ssz == -1) {
1.82 schwarze 665: perror(file);
666: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 667: }
668: off += (size_t)ssz;
669: }
670:
671: free(fb->buf);
672: fb->buf = NULL;
673: return(0);
674: }
675:
676: static void
677: mparse_end(struct mparse *curp)
678: {
679:
1.50 schwarze 680: if (curp->mdoc == NULL &&
681: curp->man == NULL &&
682: curp->sodest == NULL) {
683: if (curp->options & MPARSE_MDOC)
684: curp->mdoc = curp->pmdoc;
685: else {
686: if (curp->pman == NULL)
1.81 schwarze 687: curp->pman = man_alloc(
688: curp->roff, curp, curp->defos,
1.50 schwarze 689: curp->options & MPARSE_QUICK ? 1 : 0);
690: curp->man = curp->pman;
691: }
692: }
1.86 schwarze 693: if (curp->mdoc)
694: mdoc_endparse(curp->mdoc);
695: if (curp->man)
696: man_endparse(curp->man);
1.1 schwarze 697: roff_endparse(curp->roff);
698: }
699:
1.15 schwarze 700: static void
701: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 702: {
1.61 schwarze 703: struct buf *svprimary;
1.1 schwarze 704: const char *svfile;
1.71 schwarze 705: size_t offset;
1.14 schwarze 706: static int recursion_depth;
707:
708: if (64 < recursion_depth) {
709: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 710: return;
1.14 schwarze 711: }
1.1 schwarze 712:
1.15 schwarze 713: /* Line number is per-file. */
714: svfile = curp->file;
715: curp->file = file;
1.61 schwarze 716: svprimary = curp->primary;
1.59 schwarze 717: curp->primary = &blk;
1.15 schwarze 718: curp->line = 1;
719: recursion_depth++;
720:
1.70 schwarze 721: /* Skip an UTF-8 byte order mark. */
722: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
723: (unsigned char)blk.buf[0] == 0xef &&
724: (unsigned char)blk.buf[1] == 0xbb &&
725: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 726: offset = 3;
1.70 schwarze 727: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 728: } else
729: offset = 0;
1.70 schwarze 730:
1.71 schwarze 731: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 732:
1.86 schwarze 733: if (--recursion_depth == 0)
1.15 schwarze 734: mparse_end(curp);
735:
1.61 schwarze 736: curp->primary = svprimary;
1.15 schwarze 737: curp->file = svfile;
738: }
739:
1.74 schwarze 740: /*
741: * Read the whole file into memory and call the parsers.
742: * Called recursively when an .so request is encountered.
743: */
1.15 schwarze 744: enum mandoclevel
745: mparse_readfd(struct mparse *curp, int fd, const char *file)
746: {
747: struct buf blk;
748: int with_mmap;
1.70 schwarze 749: int save_filenc;
1.1 schwarze 750:
1.67 schwarze 751: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 752: save_filenc = curp->filenc;
753: curp->filenc = curp->options &
754: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 755: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 756: curp->filenc = save_filenc;
1.67 schwarze 757: if (with_mmap)
758: munmap(blk.buf, blk.sz);
759: else
760: free(blk.buf);
761: }
1.1 schwarze 762:
1.74 schwarze 763: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 schwarze 764: perror(file);
1.67 schwarze 765:
1.74 schwarze 766: mparse_wait(curp);
1.1 schwarze 767: return(curp->file_status);
1.58 schwarze 768: }
769:
770: enum mandoclevel
1.73 schwarze 771: mparse_open(struct mparse *curp, int *fd, const char *file)
1.58 schwarze 772: {
773: int pfd[2];
1.74 schwarze 774: int save_errno;
1.58 schwarze 775: char *cp;
776:
777: curp->file = file;
1.74 schwarze 778:
779: /* Unless zipped, try to just open the file. */
780:
1.58 schwarze 781: if ((cp = strrchr(file, '.')) == NULL ||
782: strcmp(cp + 1, "gz")) {
1.73 schwarze 783: curp->child = 0;
1.74 schwarze 784: if ((*fd = open(file, O_RDONLY)) != -1)
785: return(MANDOCLEVEL_OK);
786:
787: /* Open failed; try to append ".gz". */
788:
789: mandoc_asprintf(&cp, "%s.gz", file);
790: file = cp;
791: } else
792: cp = NULL;
793:
794: /* Before forking, make sure the file can be read. */
795:
796: save_errno = errno;
797: if (access(file, R_OK) == -1) {
798: if (cp != NULL)
799: errno = save_errno;
1.83 schwarze 800: free(cp);
801: *fd = -1;
802: curp->child = 0;
803: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
804: return(MANDOCLEVEL_ERROR);
1.58 schwarze 805: }
806:
1.74 schwarze 807: /* Run gunzip(1). */
808:
1.58 schwarze 809: if (pipe(pfd) == -1) {
1.82 schwarze 810: perror("pipe");
811: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 812: }
813:
1.73 schwarze 814: switch (curp->child = fork()) {
1.58 schwarze 815: case -1:
1.82 schwarze 816: perror("fork");
817: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 818: case 0:
819: close(pfd[0]);
820: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.82 schwarze 821: perror("dup");
822: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 823: }
824: execlp("gunzip", "gunzip", "-c", file, NULL);
1.82 schwarze 825: perror("exec");
826: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 827: default:
828: close(pfd[1]);
829: *fd = pfd[0];
830: return(MANDOCLEVEL_OK);
831: }
832: }
833:
834: enum mandoclevel
1.73 schwarze 835: mparse_wait(struct mparse *curp)
1.58 schwarze 836: {
837: int status;
838:
1.73 schwarze 839: if (curp->child == 0)
840: return(MANDOCLEVEL_OK);
841:
842: if (waitpid(curp->child, &status, 0) == -1) {
1.82 schwarze 843: perror("wait");
844: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 845: }
846: if (WIFSIGNALED(status)) {
1.83 schwarze 847: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
848: "gunzip died from signal %d", WTERMSIG(status));
849: return(MANDOCLEVEL_ERROR);
1.58 schwarze 850: }
851: if (WEXITSTATUS(status)) {
1.83 schwarze 852: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
853: "gunzip failed with code %d", WEXITSTATUS(status));
854: return(MANDOCLEVEL_ERROR);
1.58 schwarze 855: }
856: return(MANDOCLEVEL_OK);
1.1 schwarze 857: }
858:
859: struct mparse *
1.69 schwarze 860: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
861: const struct mchars *mchars, const char *defos)
1.1 schwarze 862: {
863: struct mparse *curp;
864:
865: curp = mandoc_calloc(1, sizeof(struct mparse));
866:
1.22 schwarze 867: curp->options = options;
1.1 schwarze 868: curp->wlevel = wlevel;
869: curp->mmsg = mmsg;
1.7 schwarze 870: curp->defos = defos;
1.1 schwarze 871:
1.69 schwarze 872: curp->mchars = mchars;
873: curp->roff = roff_alloc(curp, curp->mchars, options);
1.50 schwarze 874: if (curp->options & MPARSE_MDOC)
875: curp->pmdoc = mdoc_alloc(
876: curp->roff, curp, curp->defos,
877: curp->options & MPARSE_QUICK ? 1 : 0);
878: if (curp->options & MPARSE_MAN)
1.81 schwarze 879: curp->pman = man_alloc(
880: curp->roff, curp, curp->defos,
1.50 schwarze 881: curp->options & MPARSE_QUICK ? 1 : 0);
882:
1.1 schwarze 883: return(curp);
884: }
885:
886: void
887: mparse_reset(struct mparse *curp)
888: {
889:
890: roff_reset(curp->roff);
891:
892: if (curp->mdoc)
893: mdoc_reset(curp->mdoc);
894: if (curp->man)
895: man_reset(curp->man);
1.4 schwarze 896: if (curp->secondary)
897: curp->secondary->sz = 0;
1.1 schwarze 898:
899: curp->file_status = MANDOCLEVEL_OK;
900: curp->mdoc = NULL;
901: curp->man = NULL;
1.23 schwarze 902:
903: free(curp->sodest);
904: curp->sodest = NULL;
1.1 schwarze 905: }
906:
907: void
908: mparse_free(struct mparse *curp)
909: {
910:
911: if (curp->pmdoc)
912: mdoc_free(curp->pmdoc);
913: if (curp->pman)
914: man_free(curp->pman);
915: if (curp->roff)
916: roff_free(curp->roff);
1.4 schwarze 917: if (curp->secondary)
918: free(curp->secondary->buf);
1.1 schwarze 919:
1.4 schwarze 920: free(curp->secondary);
1.23 schwarze 921: free(curp->sodest);
1.1 schwarze 922: free(curp);
923: }
924:
925: void
1.23 schwarze 926: mparse_result(struct mparse *curp,
927: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 schwarze 928: {
929:
1.23 schwarze 930: if (sodest && NULL != (*sodest = curp->sodest)) {
931: *mdoc = NULL;
932: *man = NULL;
933: return;
934: }
1.1 schwarze 935: if (mdoc)
936: *mdoc = curp->mdoc;
937: if (man)
938: *man = curp->man;
939: }
940:
941: void
942: mandoc_vmsg(enum mandocerr t, struct mparse *m,
943: int ln, int pos, const char *fmt, ...)
944: {
945: char buf[256];
946: va_list ap;
947:
948: va_start(ap, fmt);
1.26 schwarze 949: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 950: va_end(ap);
951:
952: mandoc_msg(t, m, ln, pos, buf);
953: }
954:
955: void
1.25 schwarze 956: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 957: int ln, int col, const char *msg)
958: {
959: enum mandoclevel level;
960:
1.87 schwarze 961: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 962: while (er < mandoclimits[level])
963: level--;
964:
1.83 schwarze 965: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 966: return;
967:
968: if (m->mmsg)
969: (*m->mmsg)(er, level, m->file, ln, col, msg);
970:
971: if (m->file_status < level)
972: m->file_status = level;
973: }
974:
975: const char *
976: mparse_strerror(enum mandocerr er)
977: {
978:
979: return(mandocerrs[er]);
980: }
981:
982: const char *
983: mparse_strlevel(enum mandoclevel lvl)
984: {
985: return(mandoclevels[lvl]);
1.4 schwarze 986: }
987:
988: void
989: mparse_keep(struct mparse *p)
990: {
991:
992: assert(NULL == p->secondary);
993: p->secondary = mandoc_calloc(1, sizeof(struct buf));
994: }
995:
996: const char *
997: mparse_getkeep(const struct mparse *p)
998: {
999:
1000: assert(p->secondary);
1001: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 schwarze 1002: }
![[BACK]](/icons/back.gif){kind=icon}
Return to read.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc