Annotation of mandoc/read.c, Revision 1.114
1.114 ! schwarze 1: /* $Id: read.c,v 1.113 2015/01/22 19:26:50 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",
113: "sections out of conventional order",
1.54 schwarze 114: "duplicate section title",
115: "unexpected section",
1.87 schwarze 116: "unusual Xr order",
117: "unusual Xr punctuation",
1.86 schwarze 118: "AUTHORS section without An macro",
1.7 kristaps 119:
120: /* related to macros and nesting */
1.55 schwarze 121: "obsolete macro",
1.102 schwarze 122: "macro neither callable nor escaped",
1.7 kristaps 123: "skipping paragraph macro",
1.31 schwarze 124: "moving paragraph macro out of list",
1.7 kristaps 125: "skipping no-space macro",
126: "blocks badly nested",
127: "nested displays are not portable",
1.57 schwarze 128: "moving content out of list",
129: ".Vt block has child macro",
1.78 schwarze 130: "fill mode already enabled, skipping",
131: "fill mode already disabled, skipping",
1.7 kristaps 132: "line scope broken",
133:
134: /* related to missing macro arguments */
1.58 schwarze 135: "skipping empty request",
136: "conditional request controls empty scope",
1.7 kristaps 137: "skipping empty macro",
1.62 schwarze 138: "empty argument, using 0n",
1.7 kristaps 139: "argument count wrong",
1.60 schwarze 140: "missing display type, using -ragged",
141: "list type is not the first argument",
142: "missing -width in -tag list, using 8n",
1.78 schwarze 143: "missing utility name, using \"\"",
1.60 schwarze 144: "empty head in list item",
145: "empty list item",
1.61 schwarze 146: "missing font type, using \\fR",
147: "unknown font type, using \\fR",
1.103 schwarze 148: "nothing follows prefix",
1.60 schwarze 149: "missing -std argument, adding it",
1.90 schwarze 150: "missing eqn box, using \"\"",
1.7 kristaps 151:
152: /* related to bad macro arguments */
1.64 schwarze 153: "unterminated quoted argument",
1.7 kristaps 154: "duplicate argument",
1.76 schwarze 155: "skipping duplicate argument",
1.63 schwarze 156: "skipping duplicate display type",
157: "skipping duplicate list type",
1.76 schwarze 158: "skipping -width argument",
1.7 kristaps 159: "unknown AT&T UNIX version",
1.88 schwarze 160: "comma in function argument",
1.89 schwarze 161: "parenthesis in function name",
1.67 schwarze 162: "invalid content in Rs block",
1.63 schwarze 163: "invalid Boolean argument",
164: "unknown font, skipping request",
1.7 kristaps 165:
166: /* related to plain text */
1.64 schwarze 167: "blank line in fill mode, using .sp",
168: "tab in filled text",
169: "whitespace at end of input line",
1.7 kristaps 170: "bad comment style",
1.64 schwarze 171: "invalid escape sequence",
172: "undefined string, using \"\"",
1.16 kristaps 173:
1.7 kristaps 174: "generic error",
1.17 kristaps 175:
176: /* related to equations */
1.20 kristaps 177: "unexpected equation scope closure",
178: "equation scope open on exit",
1.21 kristaps 179: "overlapping equation scopes",
180: "unexpected end of equation",
1.7 kristaps 181:
182: /* related to tables */
183: "no table layout cells specified",
184: "no table data cells specified",
185: "ignore data in cell",
186: "data block still open",
187: "ignoring extra data cells",
188:
1.68 schwarze 189: /* related to document structure and macros */
1.108 schwarze 190: NULL,
1.7 kristaps 191: "input stack limit exceeded, infinite loop?",
192: "skipping bad character",
1.68 schwarze 193: "skipping unknown macro",
1.112 schwarze 194: "skipping insecure request",
1.73 schwarze 195: "skipping item outside list",
1.68 schwarze 196: "skipping column outside column list",
197: "skipping end of block that is not open",
198: "inserting missing end of block",
199: "appending missing end of block",
200:
201: /* related to request and macro arguments */
1.7 kristaps 202: "escaped character not allowed in a name",
203: "argument count wrong",
1.99 schwarze 204: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 205: "missing list type, using -item",
1.70 schwarze 206: "missing manual name, using \"\"",
1.71 schwarze 207: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 208: "unknown standard specifier",
1.71 schwarze 209: "skipping request without numeric argument",
1.109 schwarze 210: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 211: ".so request failed",
1.61 schwarze 212: "skipping all arguments",
213: "skipping excess arguments",
1.92 kristaps 214: "divide by zero",
1.112 schwarze 215:
216: "unsupported feature",
217: "input too large",
1.114 ! schwarze 218: "unsupported control character",
1.112 schwarze 219: "unsupported roff request",
220: "unsupported table syntax",
221: "unsupported table option",
222: "unsupported table layout",
223: "ignoring macro in table",
1.7 kristaps 224: };
225:
226: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
227: "SUCCESS",
228: "RESERVED",
229: "WARNING",
230: "ERROR",
1.112 schwarze 231: "UNSUPP",
1.7 kristaps 232: "BADARG",
233: "SYSERR"
234: };
235:
1.47 schwarze 236:
1.1 kristaps 237: static void
238: resize_buf(struct buf *buf, size_t initial)
239: {
240:
241: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
242: buf->buf = mandoc_realloc(buf->buf, buf->sz);
243: }
244:
245: static void
1.84 schwarze 246: choose_parser(struct mparse *curp)
1.1 kristaps 247: {
1.83 schwarze 248: char *cp, *ep;
249: int format;
1.1 kristaps 250:
1.83 schwarze 251: /*
252: * If neither command line arguments -mdoc or -man select
253: * a parser nor the roff parser found a .Dd or .TH macro
254: * yet, look ahead in the main input buffer.
255: */
256:
257: if ((format = roff_getformat(curp->roff)) == 0) {
258: cp = curp->primary->buf;
259: ep = cp + curp->primary->sz;
260: while (cp < ep) {
1.85 schwarze 261: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 262: cp++;
263: if (cp[0] == 'D' && cp[1] == 'd') {
264: format = MPARSE_MDOC;
265: break;
266: }
267: if (cp[0] == 'T' && cp[1] == 'H') {
268: format = MPARSE_MAN;
269: break;
270: }
271: }
272: cp = memchr(cp, '\n', ep - cp);
273: if (cp == NULL)
274: break;
275: cp++;
276: }
1.1 kristaps 277: }
278:
1.83 schwarze 279: if (format == MPARSE_MDOC) {
1.47 schwarze 280: if (NULL == curp->pmdoc)
1.44 schwarze 281: curp->pmdoc = mdoc_alloc(
282: curp->roff, curp, curp->defos,
283: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 284: assert(curp->pmdoc);
285: curp->mdoc = curp->pmdoc;
286: return;
1.47 schwarze 287: }
1.1 kristaps 288:
1.83 schwarze 289: /* Fall back to man(7) as a last resort. */
290:
1.47 schwarze 291: if (NULL == curp->pman)
1.106 schwarze 292: curp->pman = man_alloc(
293: curp->roff, curp, curp->defos,
1.44 schwarze 294: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 295: assert(curp->pman);
296: curp->man = curp->pman;
297: }
298:
299: /*
1.95 schwarze 300: * Main parse routine for a buffer.
301: * It assumes encoding and line numbering are already set up.
302: * It can recurse directly (for invocations of user-defined
303: * macros, inline equations, and input line traps)
304: * and indirectly (for .so file inclusion).
1.1 kristaps 305: */
306: static void
1.95 schwarze 307: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 308: {
309: const struct tbl_span *span;
310: struct buf ln;
1.113 schwarze 311: const char *save_file;
1.110 schwarze 312: char *cp;
1.95 schwarze 313: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 314: enum rofferr rr;
1.100 schwarze 315: int of;
1.1 kristaps 316: int lnn; /* line number in the real file */
1.110 schwarze 317: int fd;
318: pid_t save_child;
1.1 kristaps 319: unsigned char c;
320:
1.95 schwarze 321: memset(&ln, 0, sizeof(ln));
1.1 kristaps 322:
1.47 schwarze 323: lnn = curp->line;
324: pos = 0;
1.1 kristaps 325:
1.95 schwarze 326: while (i < blk.sz) {
1.1 kristaps 327: if (0 == pos && '\0' == blk.buf[i])
328: break;
329:
330: if (start) {
331: curp->line = lnn;
332: curp->reparse_count = 0;
1.93 schwarze 333:
334: if (lnn < 3 &&
335: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 336: curp->filenc & MPARSE_LATIN1)
337: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 338: }
339:
1.95 schwarze 340: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 341:
342: /*
343: * When finding an unescaped newline character,
344: * leave the character loop to process the line.
345: * Skip a preceding carriage return, if any.
346: */
347:
1.95 schwarze 348: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 349: '\n' == blk.buf[i + 1])
350: ++i;
351: if ('\n' == blk.buf[i]) {
352: ++i;
353: ++lnn;
354: break;
355: }
356:
1.35 schwarze 357: /*
1.93 schwarze 358: * Make sure we have space for the worst
359: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 360: */
361:
1.95 schwarze 362: if (pos + 11 > ln.sz)
1.35 schwarze 363: resize_buf(&ln, 256);
364:
1.47 schwarze 365: /*
1.93 schwarze 366: * Encode 8-bit input.
1.1 kristaps 367: */
368:
1.93 schwarze 369: c = blk.buf[i];
370: if (c & 0x80) {
1.95 schwarze 371: if ( ! (curp->filenc && preconv_encode(
372: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 ! schwarze 373: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
! 374: curp->line, pos, "0x%x", c);
1.93 schwarze 375: ln.buf[pos++] = '?';
376: i++;
377: }
378: continue;
379: }
380:
381: /*
382: * Exclude control characters.
383: */
1.1 kristaps 384:
1.93 schwarze 385: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 ! schwarze 386: mandoc_vmsg(c == 0x00 || c == 0x04 ||
! 387: c > 0x0a ? MANDOCERR_CHAR_BAD :
! 388: MANDOCERR_CHAR_UNSUPP,
! 389: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 390: i++;
1.27 joerg 391: ln.buf[pos++] = '?';
1.1 kristaps 392: continue;
393: }
394:
395: /* Trailing backslash = a plain char. */
396:
1.95 schwarze 397: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 398: ln.buf[pos++] = blk.buf[i++];
399: continue;
400: }
401:
402: /*
403: * Found escape and at least one other character.
404: * When it's a newline character, skip it.
405: * When there is a carriage return in between,
406: * skip that one as well.
407: */
408:
1.95 schwarze 409: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 410: '\n' == blk.buf[i + 2])
411: ++i;
412: if ('\n' == blk.buf[i + 1]) {
413: i += 2;
414: ++lnn;
415: continue;
416: }
417:
1.13 kristaps 418: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 419: i += 2;
420: /* Comment, skip to end of line */
1.95 schwarze 421: for (; i < blk.sz; ++i) {
1.1 kristaps 422: if ('\n' == blk.buf[i]) {
423: ++i;
424: ++lnn;
425: break;
426: }
427: }
428:
429: /* Backout trailing whitespaces */
430: for (; pos > 0; --pos) {
431: if (ln.buf[pos - 1] != ' ')
432: break;
433: if (pos > 2 && ln.buf[pos - 2] == '\\')
434: break;
435: }
436: break;
437: }
438:
1.35 schwarze 439: /* Catch escaped bogus characters. */
440:
441: c = (unsigned char) blk.buf[i+1];
442:
1.47 schwarze 443: if ( ! (isascii(c) &&
444: (isgraph(c) || isblank(c)))) {
1.114 ! schwarze 445: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 446: curp->line, pos, "0x%x", c);
1.35 schwarze 447: i += 2;
448: ln.buf[pos++] = '?';
449: continue;
450: }
451:
1.1 kristaps 452: /* Some other escape sequence, copy & cont. */
453:
454: ln.buf[pos++] = blk.buf[i++];
455: ln.buf[pos++] = blk.buf[i++];
456: }
457:
1.95 schwarze 458: if (pos >= ln.sz)
1.1 kristaps 459: resize_buf(&ln, 256);
460:
461: ln.buf[pos] = '\0';
462:
463: /*
464: * A significant amount of complexity is contained by
465: * the roff preprocessor. It's line-oriented but can be
466: * expressed on one line, so we need at times to
467: * readjust our starting point and re-run it. The roff
468: * preprocessor can also readjust the buffers with new
469: * data, so we pass them in wholesale.
470: */
471:
472: of = 0;
473:
1.24 kristaps 474: /*
475: * Maintain a lookaside buffer of all parsed lines. We
476: * only do this if mparse_keep() has been invoked (the
477: * buffer may be accessed with mparse_getkeep()).
478: */
479:
480: if (curp->secondary) {
1.47 schwarze 481: curp->secondary->buf = mandoc_realloc(
482: curp->secondary->buf,
483: curp->secondary->sz + pos + 2);
484: memcpy(curp->secondary->buf +
485: curp->secondary->sz,
486: ln.buf, pos);
1.24 kristaps 487: curp->secondary->sz += pos;
488: curp->secondary->buf
489: [curp->secondary->sz] = '\n';
490: curp->secondary->sz++;
491: curp->secondary->buf
492: [curp->secondary->sz] = '\0';
493: }
1.1 kristaps 494: rerun:
1.96 schwarze 495: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 496:
497: switch (rr) {
1.47 schwarze 498: case ROFF_REPARSE:
1.1 kristaps 499: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 500: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 501: else
1.3 kristaps 502: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 503: curp->line, pos, NULL);
1.1 kristaps 504: pos = 0;
505: continue;
1.47 schwarze 506: case ROFF_APPEND:
1.95 schwarze 507: pos = strlen(ln.buf);
1.1 kristaps 508: continue;
1.47 schwarze 509: case ROFF_RERUN:
1.1 kristaps 510: goto rerun;
1.47 schwarze 511: case ROFF_IGN:
1.1 kristaps 512: pos = 0;
513: continue;
1.47 schwarze 514: case ROFF_SO:
1.95 schwarze 515: if ( ! (curp->options & MPARSE_SO) &&
516: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 517: curp->sodest = mandoc_strdup(ln.buf + of);
518: free(ln.buf);
519: return;
520: }
1.24 kristaps 521: /*
522: * We remove `so' clauses from our lookaside
523: * buffer because we're going to descend into
524: * the file recursively.
525: */
1.47 schwarze 526: if (curp->secondary)
1.25 kristaps 527: curp->secondary->sz -= pos + 1;
1.113 schwarze 528: save_file = curp->file;
1.110 schwarze 529: save_child = curp->child;
530: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 schwarze 531: MANDOCLEVEL_OK) {
1.110 schwarze 532: mparse_readfd(curp, fd, ln.buf + of);
1.113 schwarze 533: curp->file = save_file;
534: } else {
535: curp->file = save_file;
1.52 schwarze 536: mandoc_vmsg(MANDOCERR_SO_FAIL,
537: curp, curp->line, pos,
538: ".so %s", ln.buf + of);
1.110 schwarze 539: ln.sz = mandoc_asprintf(&cp,
540: ".sp\nSee the file %s.\n.sp",
541: ln.buf + of);
542: free(ln.buf);
543: ln.buf = cp;
544: of = 0;
545: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 546: }
1.110 schwarze 547: curp->child = save_child;
1.1 kristaps 548: pos = 0;
549: continue;
550: default:
551: break;
552: }
553:
554: /*
555: * If input parsers have not been allocated, do so now.
1.14 kristaps 556: * We keep these instanced between parsers, but set them
1.1 kristaps 557: * locally per parse routine since we can use different
558: * parsers with each one.
559: */
560:
561: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 562: choose_parser(curp);
1.1 kristaps 563:
1.47 schwarze 564: /*
1.84 schwarze 565: * Lastly, push down into the parsers themselves.
1.1 kristaps 566: * If libroff returns ROFF_TBL, then add it to the
567: * currently open parse. Since we only get here if
568: * there does exist data (see tbl_data.c), we're
569: * guaranteed that something's been allocated.
570: * Do the same for ROFF_EQN.
571: */
572:
1.100 schwarze 573: if (rr == ROFF_TBL) {
574: while ((span = roff_span(curp->roff)) != NULL)
575: if (curp->man == NULL)
576: mdoc_addspan(curp->mdoc, span);
577: else
578: man_addspan(curp->man, span);
579: } else if (rr == ROFF_EQN) {
580: if (curp->man == NULL)
581: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
582: else
583: man_addeqn(curp->man, roff_eqn(curp->roff));
584: } else if ((curp->man == NULL ?
585: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
586: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
587: break;
1.1 kristaps 588:
589: /* Temporary buffers typically are not full. */
590:
591: if (0 == start && '\0' == blk.buf[i])
592: break;
593:
594: /* Start the next input line. */
595:
596: pos = 0;
597: }
598:
599: free(ln.buf);
600: }
601:
602: static int
1.40 schwarze 603: read_whole_file(struct mparse *curp, const char *file, int fd,
604: struct buf *fb, int *with_mmap)
1.1 kristaps 605: {
606: size_t off;
607: ssize_t ssz;
608:
1.81 schwarze 609: #if HAVE_MMAP
1.15 kristaps 610: struct stat st;
1.1 kristaps 611: if (-1 == fstat(fd, &st)) {
1.107 schwarze 612: perror(file);
613: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 614: }
615:
616: /*
617: * If we're a regular file, try just reading in the whole entry
618: * via mmap(). This is faster than reading it into blocks, and
619: * since each file is only a few bytes to begin with, I'm not
620: * concerned that this is going to tank any machines.
621: */
622:
623: if (S_ISREG(st.st_mode)) {
624: if (st.st_size >= (1U << 31)) {
1.111 schwarze 625: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 626: return(0);
627: }
628: *with_mmap = 1;
629: fb->sz = (size_t)st.st_size;
1.37 schwarze 630: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 631: if (fb->buf != MAP_FAILED)
632: return(1);
633: }
1.15 kristaps 634: #endif
1.1 kristaps 635:
636: /*
637: * If this isn't a regular file (like, say, stdin), then we must
638: * go the old way and just read things in bit by bit.
639: */
640:
641: *with_mmap = 0;
642: off = 0;
643: fb->sz = 0;
644: fb->buf = NULL;
645: for (;;) {
646: if (off == fb->sz) {
647: if (fb->sz == (1U << 31)) {
1.111 schwarze 648: mandoc_msg(MANDOCERR_TOOLARGE, curp,
649: 0, 0, NULL);
1.1 kristaps 650: break;
651: }
652: resize_buf(fb, 65536);
653: }
654: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
655: if (ssz == 0) {
656: fb->sz = off;
657: return(1);
658: }
659: if (ssz == -1) {
1.107 schwarze 660: perror(file);
661: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 662: }
663: off += (size_t)ssz;
664: }
665:
666: free(fb->buf);
667: fb->buf = NULL;
668: return(0);
669: }
670:
671: static void
672: mparse_end(struct mparse *curp)
673: {
674:
1.72 schwarze 675: if (curp->mdoc == NULL &&
676: curp->man == NULL &&
677: curp->sodest == NULL) {
678: if (curp->options & MPARSE_MDOC)
679: curp->mdoc = curp->pmdoc;
680: else {
681: if (curp->pman == NULL)
1.106 schwarze 682: curp->pman = man_alloc(
683: curp->roff, curp, curp->defos,
1.72 schwarze 684: curp->options & MPARSE_QUICK ? 1 : 0);
685: curp->man = curp->pman;
686: }
687: }
1.111 schwarze 688: if (curp->mdoc)
689: mdoc_endparse(curp->mdoc);
690: if (curp->man)
691: man_endparse(curp->man);
1.1 kristaps 692: roff_endparse(curp->roff);
693: }
694:
695: static void
1.36 schwarze 696: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 697: {
1.85 schwarze 698: struct buf *svprimary;
1.28 joerg 699: const char *svfile;
1.95 schwarze 700: size_t offset;
1.36 schwarze 701: static int recursion_depth;
702:
703: if (64 < recursion_depth) {
704: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
705: return;
706: }
1.28 joerg 707:
708: /* Line number is per-file. */
709: svfile = curp->file;
710: curp->file = file;
1.85 schwarze 711: svprimary = curp->primary;
1.83 schwarze 712: curp->primary = &blk;
1.28 joerg 713: curp->line = 1;
1.36 schwarze 714: recursion_depth++;
1.28 joerg 715:
1.93 schwarze 716: /* Skip an UTF-8 byte order mark. */
717: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
718: (unsigned char)blk.buf[0] == 0xef &&
719: (unsigned char)blk.buf[1] == 0xbb &&
720: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 721: offset = 3;
1.93 schwarze 722: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 723: } else
724: offset = 0;
1.93 schwarze 725:
1.95 schwarze 726: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 727:
1.111 schwarze 728: if (--recursion_depth == 0)
1.28 joerg 729: mparse_end(curp);
730:
1.85 schwarze 731: curp->primary = svprimary;
1.28 joerg 732: curp->file = svfile;
733: }
734:
735: enum mandoclevel
1.104 schwarze 736: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 737: const char *file)
738: {
739: struct buf blk;
740:
1.104 schwarze 741: blk.buf = buf;
1.28 joerg 742: blk.sz = len;
743:
1.36 schwarze 744: mparse_parse_buffer(curp, blk, file);
1.28 joerg 745: return(curp->file_status);
746: }
747:
1.98 schwarze 748: /*
749: * Read the whole file into memory and call the parsers.
750: * Called recursively when an .so request is encountered.
751: */
1.36 schwarze 752: enum mandoclevel
753: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 754: {
1.28 joerg 755: struct buf blk;
756: int with_mmap;
1.93 schwarze 757: int save_filenc;
1.1 kristaps 758:
1.91 schwarze 759: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 760: save_filenc = curp->filenc;
761: curp->filenc = curp->options &
762: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 763: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 764: curp->filenc = save_filenc;
1.81 schwarze 765: #if HAVE_MMAP
1.91 schwarze 766: if (with_mmap)
767: munmap(blk.buf, blk.sz);
768: else
1.28 joerg 769: #endif
1.91 schwarze 770: free(blk.buf);
771: }
1.1 kristaps 772:
1.98 schwarze 773: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 774: perror(file);
1.91 schwarze 775:
1.98 schwarze 776: mparse_wait(curp);
1.1 kristaps 777: return(curp->file_status);
1.82 schwarze 778: }
779:
780: enum mandoclevel
1.97 schwarze 781: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 782: {
783: int pfd[2];
1.98 schwarze 784: int save_errno;
1.82 schwarze 785: char *cp;
786:
787: curp->file = file;
1.98 schwarze 788:
789: /* Unless zipped, try to just open the file. */
790:
1.82 schwarze 791: if ((cp = strrchr(file, '.')) == NULL ||
792: strcmp(cp + 1, "gz")) {
1.97 schwarze 793: curp->child = 0;
1.98 schwarze 794: if ((*fd = open(file, O_RDONLY)) != -1)
795: return(MANDOCLEVEL_OK);
796:
797: /* Open failed; try to append ".gz". */
798:
799: mandoc_asprintf(&cp, "%s.gz", file);
800: file = cp;
801: } else
802: cp = NULL;
803:
804: /* Before forking, make sure the file can be read. */
805:
806: save_errno = errno;
807: if (access(file, R_OK) == -1) {
808: if (cp != NULL)
809: errno = save_errno;
1.108 schwarze 810: free(cp);
811: *fd = -1;
812: curp->child = 0;
813: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
814: return(MANDOCLEVEL_ERROR);
1.82 schwarze 815: }
816:
1.98 schwarze 817: /* Run gunzip(1). */
818:
1.82 schwarze 819: if (pipe(pfd) == -1) {
1.107 schwarze 820: perror("pipe");
821: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 822: }
823:
1.97 schwarze 824: switch (curp->child = fork()) {
1.82 schwarze 825: case -1:
1.107 schwarze 826: perror("fork");
827: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 828: case 0:
829: close(pfd[0]);
830: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 831: perror("dup");
832: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 833: }
834: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 835: perror("exec");
836: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 837: default:
838: close(pfd[1]);
839: *fd = pfd[0];
840: return(MANDOCLEVEL_OK);
841: }
842: }
843:
844: enum mandoclevel
1.97 schwarze 845: mparse_wait(struct mparse *curp)
1.82 schwarze 846: {
847: int status;
848:
1.97 schwarze 849: if (curp->child == 0)
850: return(MANDOCLEVEL_OK);
851:
852: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 853: perror("wait");
854: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 855: }
856: if (WIFSIGNALED(status)) {
1.108 schwarze 857: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
858: "gunzip died from signal %d", WTERMSIG(status));
859: return(MANDOCLEVEL_ERROR);
1.82 schwarze 860: }
861: if (WEXITSTATUS(status)) {
1.108 schwarze 862: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
863: "gunzip failed with code %d", WEXITSTATUS(status));
864: return(MANDOCLEVEL_ERROR);
1.82 schwarze 865: }
866: return(MANDOCLEVEL_OK);
1.1 kristaps 867: }
868:
869: struct mparse *
1.94 schwarze 870: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
871: const struct mchars *mchars, const char *defos)
1.1 kristaps 872: {
873: struct mparse *curp;
1.10 kristaps 874:
1.1 kristaps 875: curp = mandoc_calloc(1, sizeof(struct mparse));
876:
1.44 schwarze 877: curp->options = options;
1.3 kristaps 878: curp->wlevel = wlevel;
1.1 kristaps 879: curp->mmsg = mmsg;
1.29 schwarze 880: curp->defos = defos;
1.1 kristaps 881:
1.94 schwarze 882: curp->mchars = mchars;
883: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 884: if (curp->options & MPARSE_MDOC)
885: curp->pmdoc = mdoc_alloc(
886: curp->roff, curp, curp->defos,
887: curp->options & MPARSE_QUICK ? 1 : 0);
888: if (curp->options & MPARSE_MAN)
1.106 schwarze 889: curp->pman = man_alloc(
890: curp->roff, curp, curp->defos,
1.72 schwarze 891: curp->options & MPARSE_QUICK ? 1 : 0);
892:
1.1 kristaps 893: return(curp);
894: }
895:
896: void
897: mparse_reset(struct mparse *curp)
898: {
899:
900: roff_reset(curp->roff);
901:
902: if (curp->mdoc)
903: mdoc_reset(curp->mdoc);
904: if (curp->man)
905: man_reset(curp->man);
1.24 kristaps 906: if (curp->secondary)
907: curp->secondary->sz = 0;
1.1 kristaps 908:
909: curp->file_status = MANDOCLEVEL_OK;
910: curp->mdoc = NULL;
911: curp->man = NULL;
1.45 schwarze 912:
913: free(curp->sodest);
914: curp->sodest = NULL;
1.1 kristaps 915: }
916:
917: void
918: mparse_free(struct mparse *curp)
919: {
920:
921: if (curp->pmdoc)
922: mdoc_free(curp->pmdoc);
923: if (curp->pman)
924: man_free(curp->pman);
925: if (curp->roff)
926: roff_free(curp->roff);
1.24 kristaps 927: if (curp->secondary)
928: free(curp->secondary->buf);
1.1 kristaps 929:
1.24 kristaps 930: free(curp->secondary);
1.45 schwarze 931: free(curp->sodest);
1.1 kristaps 932: free(curp);
933: }
934:
935: void
1.45 schwarze 936: mparse_result(struct mparse *curp,
937: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 938: {
939:
1.45 schwarze 940: if (sodest && NULL != (*sodest = curp->sodest)) {
941: *mdoc = NULL;
942: *man = NULL;
943: return;
944: }
1.9 kristaps 945: if (mdoc)
946: *mdoc = curp->mdoc;
947: if (man)
948: *man = curp->man;
1.3 kristaps 949: }
950:
951: void
952: mandoc_vmsg(enum mandocerr t, struct mparse *m,
953: int ln, int pos, const char *fmt, ...)
954: {
955: char buf[256];
956: va_list ap;
957:
958: va_start(ap, fmt);
1.48 schwarze 959: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 960: va_end(ap);
961:
962: mandoc_msg(t, m, ln, pos, buf);
963: }
964:
965: void
1.47 schwarze 966: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 967: int ln, int col, const char *msg)
968: {
969: enum mandoclevel level;
970:
1.112 schwarze 971: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 972: while (er < mandoclimits[level])
973: level--;
974:
1.108 schwarze 975: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 976: return;
977:
1.8 kristaps 978: if (m->mmsg)
979: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 980:
981: if (m->file_status < level)
982: m->file_status = level;
1.7 kristaps 983: }
984:
985: const char *
986: mparse_strerror(enum mandocerr er)
987: {
988:
989: return(mandocerrs[er]);
990: }
991:
992: const char *
993: mparse_strlevel(enum mandoclevel lvl)
994: {
995: return(mandoclevels[lvl]);
1.24 kristaps 996: }
997:
998: void
999: mparse_keep(struct mparse *p)
1000: {
1001:
1002: assert(NULL == p->secondary);
1003: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1004: }
1005:
1006: const char *
1007: mparse_getkeep(const struct mparse *p)
1008: {
1009:
1010: assert(p->secondary);
1011: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1012: }
CVSweb