Annotation of mandoc/read.c, Revision 1.116
1.116 ! schwarze 1: /* $Id: read.c,v 1.115 2015/01/24 01:58:33 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 */
1.116 ! schwarze 183: "non-alphabetic character in tbl options",
! 184: "skipping unknown tbl option",
! 185: "missing tbl option argument",
! 186: "wrong tbl option argument size",
1.7 kristaps 187: "no table layout cells specified",
188: "no table data cells specified",
189: "ignore data in cell",
190: "data block still open",
191: "ignoring extra data cells",
192:
1.68 schwarze 193: /* related to document structure and macros */
1.108 schwarze 194: NULL,
1.7 kristaps 195: "input stack limit exceeded, infinite loop?",
196: "skipping bad character",
1.68 schwarze 197: "skipping unknown macro",
1.112 schwarze 198: "skipping insecure request",
1.73 schwarze 199: "skipping item outside list",
1.68 schwarze 200: "skipping column outside column list",
201: "skipping end of block that is not open",
1.115 schwarze 202: "fewer RS blocks open, skipping",
1.68 schwarze 203: "inserting missing end of block",
204: "appending missing end of block",
205:
206: /* related to request and macro arguments */
1.7 kristaps 207: "escaped character not allowed in a name",
208: "argument count wrong",
1.99 schwarze 209: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 210: "missing list type, using -item",
1.70 schwarze 211: "missing manual name, using \"\"",
1.71 schwarze 212: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 213: "unknown standard specifier",
1.71 schwarze 214: "skipping request without numeric argument",
1.109 schwarze 215: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 216: ".so request failed",
1.61 schwarze 217: "skipping all arguments",
218: "skipping excess arguments",
1.92 kristaps 219: "divide by zero",
1.112 schwarze 220:
221: "unsupported feature",
222: "input too large",
1.114 schwarze 223: "unsupported control character",
1.112 schwarze 224: "unsupported roff request",
225: "unsupported table layout",
226: "ignoring macro in table",
1.7 kristaps 227: };
228:
229: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
230: "SUCCESS",
231: "RESERVED",
232: "WARNING",
233: "ERROR",
1.112 schwarze 234: "UNSUPP",
1.7 kristaps 235: "BADARG",
236: "SYSERR"
237: };
238:
1.47 schwarze 239:
1.1 kristaps 240: static void
241: resize_buf(struct buf *buf, size_t initial)
242: {
243:
244: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
245: buf->buf = mandoc_realloc(buf->buf, buf->sz);
246: }
247:
248: static void
1.84 schwarze 249: choose_parser(struct mparse *curp)
1.1 kristaps 250: {
1.83 schwarze 251: char *cp, *ep;
252: int format;
1.1 kristaps 253:
1.83 schwarze 254: /*
255: * If neither command line arguments -mdoc or -man select
256: * a parser nor the roff parser found a .Dd or .TH macro
257: * yet, look ahead in the main input buffer.
258: */
259:
260: if ((format = roff_getformat(curp->roff)) == 0) {
261: cp = curp->primary->buf;
262: ep = cp + curp->primary->sz;
263: while (cp < ep) {
1.85 schwarze 264: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 265: cp++;
266: if (cp[0] == 'D' && cp[1] == 'd') {
267: format = MPARSE_MDOC;
268: break;
269: }
270: if (cp[0] == 'T' && cp[1] == 'H') {
271: format = MPARSE_MAN;
272: break;
273: }
274: }
275: cp = memchr(cp, '\n', ep - cp);
276: if (cp == NULL)
277: break;
278: cp++;
279: }
1.1 kristaps 280: }
281:
1.83 schwarze 282: if (format == MPARSE_MDOC) {
1.47 schwarze 283: if (NULL == curp->pmdoc)
1.44 schwarze 284: curp->pmdoc = mdoc_alloc(
285: curp->roff, curp, curp->defos,
286: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 287: assert(curp->pmdoc);
288: curp->mdoc = curp->pmdoc;
289: return;
1.47 schwarze 290: }
1.1 kristaps 291:
1.83 schwarze 292: /* Fall back to man(7) as a last resort. */
293:
1.47 schwarze 294: if (NULL == curp->pman)
1.106 schwarze 295: curp->pman = man_alloc(
296: curp->roff, curp, curp->defos,
1.44 schwarze 297: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 298: assert(curp->pman);
299: curp->man = curp->pman;
300: }
301:
302: /*
1.95 schwarze 303: * Main parse routine for a buffer.
304: * It assumes encoding and line numbering are already set up.
305: * It can recurse directly (for invocations of user-defined
306: * macros, inline equations, and input line traps)
307: * and indirectly (for .so file inclusion).
1.1 kristaps 308: */
309: static void
1.95 schwarze 310: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 311: {
312: const struct tbl_span *span;
313: struct buf ln;
1.113 schwarze 314: const char *save_file;
1.110 schwarze 315: char *cp;
1.95 schwarze 316: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 317: enum rofferr rr;
1.100 schwarze 318: int of;
1.1 kristaps 319: int lnn; /* line number in the real file */
1.110 schwarze 320: int fd;
321: pid_t save_child;
1.1 kristaps 322: unsigned char c;
323:
1.95 schwarze 324: memset(&ln, 0, sizeof(ln));
1.1 kristaps 325:
1.47 schwarze 326: lnn = curp->line;
327: pos = 0;
1.1 kristaps 328:
1.95 schwarze 329: while (i < blk.sz) {
1.1 kristaps 330: if (0 == pos && '\0' == blk.buf[i])
331: break;
332:
333: if (start) {
334: curp->line = lnn;
335: curp->reparse_count = 0;
1.93 schwarze 336:
337: if (lnn < 3 &&
338: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 339: curp->filenc & MPARSE_LATIN1)
340: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 341: }
342:
1.95 schwarze 343: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 344:
345: /*
346: * When finding an unescaped newline character,
347: * leave the character loop to process the line.
348: * Skip a preceding carriage return, if any.
349: */
350:
1.95 schwarze 351: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 352: '\n' == blk.buf[i + 1])
353: ++i;
354: if ('\n' == blk.buf[i]) {
355: ++i;
356: ++lnn;
357: break;
358: }
359:
1.35 schwarze 360: /*
1.93 schwarze 361: * Make sure we have space for the worst
362: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 363: */
364:
1.95 schwarze 365: if (pos + 11 > ln.sz)
1.35 schwarze 366: resize_buf(&ln, 256);
367:
1.47 schwarze 368: /*
1.93 schwarze 369: * Encode 8-bit input.
1.1 kristaps 370: */
371:
1.93 schwarze 372: c = blk.buf[i];
373: if (c & 0x80) {
1.95 schwarze 374: if ( ! (curp->filenc && preconv_encode(
375: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 376: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
377: curp->line, pos, "0x%x", c);
1.93 schwarze 378: ln.buf[pos++] = '?';
379: i++;
380: }
381: continue;
382: }
383:
384: /*
385: * Exclude control characters.
386: */
1.1 kristaps 387:
1.93 schwarze 388: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 389: mandoc_vmsg(c == 0x00 || c == 0x04 ||
390: c > 0x0a ? MANDOCERR_CHAR_BAD :
391: MANDOCERR_CHAR_UNSUPP,
392: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 393: i++;
1.27 joerg 394: ln.buf[pos++] = '?';
1.1 kristaps 395: continue;
396: }
397:
398: /* Trailing backslash = a plain char. */
399:
1.95 schwarze 400: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 401: ln.buf[pos++] = blk.buf[i++];
402: continue;
403: }
404:
405: /*
406: * Found escape and at least one other character.
407: * When it's a newline character, skip it.
408: * When there is a carriage return in between,
409: * skip that one as well.
410: */
411:
1.95 schwarze 412: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 413: '\n' == blk.buf[i + 2])
414: ++i;
415: if ('\n' == blk.buf[i + 1]) {
416: i += 2;
417: ++lnn;
418: continue;
419: }
420:
1.13 kristaps 421: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 422: i += 2;
423: /* Comment, skip to end of line */
1.95 schwarze 424: for (; i < blk.sz; ++i) {
1.1 kristaps 425: if ('\n' == blk.buf[i]) {
426: ++i;
427: ++lnn;
428: break;
429: }
430: }
431:
432: /* Backout trailing whitespaces */
433: for (; pos > 0; --pos) {
434: if (ln.buf[pos - 1] != ' ')
435: break;
436: if (pos > 2 && ln.buf[pos - 2] == '\\')
437: break;
438: }
439: break;
440: }
441:
1.35 schwarze 442: /* Catch escaped bogus characters. */
443:
444: c = (unsigned char) blk.buf[i+1];
445:
1.47 schwarze 446: if ( ! (isascii(c) &&
447: (isgraph(c) || isblank(c)))) {
1.114 schwarze 448: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 449: curp->line, pos, "0x%x", c);
1.35 schwarze 450: i += 2;
451: ln.buf[pos++] = '?';
452: continue;
453: }
454:
1.1 kristaps 455: /* Some other escape sequence, copy & cont. */
456:
457: ln.buf[pos++] = blk.buf[i++];
458: ln.buf[pos++] = blk.buf[i++];
459: }
460:
1.95 schwarze 461: if (pos >= ln.sz)
1.1 kristaps 462: resize_buf(&ln, 256);
463:
464: ln.buf[pos] = '\0';
465:
466: /*
467: * A significant amount of complexity is contained by
468: * the roff preprocessor. It's line-oriented but can be
469: * expressed on one line, so we need at times to
470: * readjust our starting point and re-run it. The roff
471: * preprocessor can also readjust the buffers with new
472: * data, so we pass them in wholesale.
473: */
474:
475: of = 0;
476:
1.24 kristaps 477: /*
478: * Maintain a lookaside buffer of all parsed lines. We
479: * only do this if mparse_keep() has been invoked (the
480: * buffer may be accessed with mparse_getkeep()).
481: */
482:
483: if (curp->secondary) {
1.47 schwarze 484: curp->secondary->buf = mandoc_realloc(
485: curp->secondary->buf,
486: curp->secondary->sz + pos + 2);
487: memcpy(curp->secondary->buf +
488: curp->secondary->sz,
489: ln.buf, pos);
1.24 kristaps 490: curp->secondary->sz += pos;
491: curp->secondary->buf
492: [curp->secondary->sz] = '\n';
493: curp->secondary->sz++;
494: curp->secondary->buf
495: [curp->secondary->sz] = '\0';
496: }
1.1 kristaps 497: rerun:
1.96 schwarze 498: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 499:
500: switch (rr) {
1.47 schwarze 501: case ROFF_REPARSE:
1.1 kristaps 502: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 503: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 504: else
1.3 kristaps 505: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 506: curp->line, pos, NULL);
1.1 kristaps 507: pos = 0;
508: continue;
1.47 schwarze 509: case ROFF_APPEND:
1.95 schwarze 510: pos = strlen(ln.buf);
1.1 kristaps 511: continue;
1.47 schwarze 512: case ROFF_RERUN:
1.1 kristaps 513: goto rerun;
1.47 schwarze 514: case ROFF_IGN:
1.1 kristaps 515: pos = 0;
516: continue;
1.47 schwarze 517: case ROFF_SO:
1.95 schwarze 518: if ( ! (curp->options & MPARSE_SO) &&
519: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 520: curp->sodest = mandoc_strdup(ln.buf + of);
521: free(ln.buf);
522: return;
523: }
1.24 kristaps 524: /*
525: * We remove `so' clauses from our lookaside
526: * buffer because we're going to descend into
527: * the file recursively.
528: */
1.47 schwarze 529: if (curp->secondary)
1.25 kristaps 530: curp->secondary->sz -= pos + 1;
1.113 schwarze 531: save_file = curp->file;
1.110 schwarze 532: save_child = curp->child;
533: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 schwarze 534: MANDOCLEVEL_OK) {
1.110 schwarze 535: mparse_readfd(curp, fd, ln.buf + of);
1.113 schwarze 536: curp->file = save_file;
537: } else {
538: curp->file = save_file;
1.52 schwarze 539: mandoc_vmsg(MANDOCERR_SO_FAIL,
540: curp, curp->line, pos,
541: ".so %s", ln.buf + of);
1.110 schwarze 542: ln.sz = mandoc_asprintf(&cp,
543: ".sp\nSee the file %s.\n.sp",
544: ln.buf + of);
545: free(ln.buf);
546: ln.buf = cp;
547: of = 0;
548: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 549: }
1.110 schwarze 550: curp->child = save_child;
1.1 kristaps 551: pos = 0;
552: continue;
553: default:
554: break;
555: }
556:
557: /*
558: * If input parsers have not been allocated, do so now.
1.14 kristaps 559: * We keep these instanced between parsers, but set them
1.1 kristaps 560: * locally per parse routine since we can use different
561: * parsers with each one.
562: */
563:
564: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 565: choose_parser(curp);
1.1 kristaps 566:
1.47 schwarze 567: /*
1.84 schwarze 568: * Lastly, push down into the parsers themselves.
1.1 kristaps 569: * If libroff returns ROFF_TBL, then add it to the
570: * currently open parse. Since we only get here if
571: * there does exist data (see tbl_data.c), we're
572: * guaranteed that something's been allocated.
573: * Do the same for ROFF_EQN.
574: */
575:
1.100 schwarze 576: if (rr == ROFF_TBL) {
577: while ((span = roff_span(curp->roff)) != NULL)
578: if (curp->man == NULL)
579: mdoc_addspan(curp->mdoc, span);
580: else
581: man_addspan(curp->man, span);
582: } else if (rr == ROFF_EQN) {
583: if (curp->man == NULL)
584: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
585: else
586: man_addeqn(curp->man, roff_eqn(curp->roff));
587: } else if ((curp->man == NULL ?
588: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
589: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
590: break;
1.1 kristaps 591:
592: /* Temporary buffers typically are not full. */
593:
594: if (0 == start && '\0' == blk.buf[i])
595: break;
596:
597: /* Start the next input line. */
598:
599: pos = 0;
600: }
601:
602: free(ln.buf);
603: }
604:
605: static int
1.40 schwarze 606: read_whole_file(struct mparse *curp, const char *file, int fd,
607: struct buf *fb, int *with_mmap)
1.1 kristaps 608: {
609: size_t off;
610: ssize_t ssz;
611:
1.81 schwarze 612: #if HAVE_MMAP
1.15 kristaps 613: struct stat st;
1.1 kristaps 614: if (-1 == fstat(fd, &st)) {
1.107 schwarze 615: perror(file);
616: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 617: }
618:
619: /*
620: * If we're a regular file, try just reading in the whole entry
621: * via mmap(). This is faster than reading it into blocks, and
622: * since each file is only a few bytes to begin with, I'm not
623: * concerned that this is going to tank any machines.
624: */
625:
626: if (S_ISREG(st.st_mode)) {
627: if (st.st_size >= (1U << 31)) {
1.111 schwarze 628: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 629: return(0);
630: }
631: *with_mmap = 1;
632: fb->sz = (size_t)st.st_size;
1.37 schwarze 633: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 634: if (fb->buf != MAP_FAILED)
635: return(1);
636: }
1.15 kristaps 637: #endif
1.1 kristaps 638:
639: /*
640: * If this isn't a regular file (like, say, stdin), then we must
641: * go the old way and just read things in bit by bit.
642: */
643:
644: *with_mmap = 0;
645: off = 0;
646: fb->sz = 0;
647: fb->buf = NULL;
648: for (;;) {
649: if (off == fb->sz) {
650: if (fb->sz == (1U << 31)) {
1.111 schwarze 651: mandoc_msg(MANDOCERR_TOOLARGE, curp,
652: 0, 0, NULL);
1.1 kristaps 653: break;
654: }
655: resize_buf(fb, 65536);
656: }
657: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
658: if (ssz == 0) {
659: fb->sz = off;
660: return(1);
661: }
662: if (ssz == -1) {
1.107 schwarze 663: perror(file);
664: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 665: }
666: off += (size_t)ssz;
667: }
668:
669: free(fb->buf);
670: fb->buf = NULL;
671: return(0);
672: }
673:
674: static void
675: mparse_end(struct mparse *curp)
676: {
677:
1.72 schwarze 678: if (curp->mdoc == NULL &&
679: curp->man == NULL &&
680: curp->sodest == NULL) {
681: if (curp->options & MPARSE_MDOC)
682: curp->mdoc = curp->pmdoc;
683: else {
684: if (curp->pman == NULL)
1.106 schwarze 685: curp->pman = man_alloc(
686: curp->roff, curp, curp->defos,
1.72 schwarze 687: curp->options & MPARSE_QUICK ? 1 : 0);
688: curp->man = curp->pman;
689: }
690: }
1.111 schwarze 691: if (curp->mdoc)
692: mdoc_endparse(curp->mdoc);
693: if (curp->man)
694: man_endparse(curp->man);
1.1 kristaps 695: roff_endparse(curp->roff);
696: }
697:
698: static void
1.36 schwarze 699: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 700: {
1.85 schwarze 701: struct buf *svprimary;
1.28 joerg 702: const char *svfile;
1.95 schwarze 703: size_t offset;
1.36 schwarze 704: static int recursion_depth;
705:
706: if (64 < recursion_depth) {
707: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
708: return;
709: }
1.28 joerg 710:
711: /* Line number is per-file. */
712: svfile = curp->file;
713: curp->file = file;
1.85 schwarze 714: svprimary = curp->primary;
1.83 schwarze 715: curp->primary = &blk;
1.28 joerg 716: curp->line = 1;
1.36 schwarze 717: recursion_depth++;
1.28 joerg 718:
1.93 schwarze 719: /* Skip an UTF-8 byte order mark. */
720: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
721: (unsigned char)blk.buf[0] == 0xef &&
722: (unsigned char)blk.buf[1] == 0xbb &&
723: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 724: offset = 3;
1.93 schwarze 725: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 726: } else
727: offset = 0;
1.93 schwarze 728:
1.95 schwarze 729: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 730:
1.111 schwarze 731: if (--recursion_depth == 0)
1.28 joerg 732: mparse_end(curp);
733:
1.85 schwarze 734: curp->primary = svprimary;
1.28 joerg 735: curp->file = svfile;
736: }
737:
738: enum mandoclevel
1.104 schwarze 739: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 740: const char *file)
741: {
742: struct buf blk;
743:
1.104 schwarze 744: blk.buf = buf;
1.28 joerg 745: blk.sz = len;
746:
1.36 schwarze 747: mparse_parse_buffer(curp, blk, file);
1.28 joerg 748: return(curp->file_status);
749: }
750:
1.98 schwarze 751: /*
752: * Read the whole file into memory and call the parsers.
753: * Called recursively when an .so request is encountered.
754: */
1.36 schwarze 755: enum mandoclevel
756: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 757: {
1.28 joerg 758: struct buf blk;
759: int with_mmap;
1.93 schwarze 760: int save_filenc;
1.1 kristaps 761:
1.91 schwarze 762: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 763: save_filenc = curp->filenc;
764: curp->filenc = curp->options &
765: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 766: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 767: curp->filenc = save_filenc;
1.81 schwarze 768: #if HAVE_MMAP
1.91 schwarze 769: if (with_mmap)
770: munmap(blk.buf, blk.sz);
771: else
1.28 joerg 772: #endif
1.91 schwarze 773: free(blk.buf);
774: }
1.1 kristaps 775:
1.98 schwarze 776: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 777: perror(file);
1.91 schwarze 778:
1.98 schwarze 779: mparse_wait(curp);
1.1 kristaps 780: return(curp->file_status);
1.82 schwarze 781: }
782:
783: enum mandoclevel
1.97 schwarze 784: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 785: {
786: int pfd[2];
1.98 schwarze 787: int save_errno;
1.82 schwarze 788: char *cp;
789:
790: curp->file = file;
1.98 schwarze 791:
792: /* Unless zipped, try to just open the file. */
793:
1.82 schwarze 794: if ((cp = strrchr(file, '.')) == NULL ||
795: strcmp(cp + 1, "gz")) {
1.97 schwarze 796: curp->child = 0;
1.98 schwarze 797: if ((*fd = open(file, O_RDONLY)) != -1)
798: return(MANDOCLEVEL_OK);
799:
800: /* Open failed; try to append ".gz". */
801:
802: mandoc_asprintf(&cp, "%s.gz", file);
803: file = cp;
804: } else
805: cp = NULL;
806:
807: /* Before forking, make sure the file can be read. */
808:
809: save_errno = errno;
810: if (access(file, R_OK) == -1) {
811: if (cp != NULL)
812: errno = save_errno;
1.108 schwarze 813: free(cp);
814: *fd = -1;
815: curp->child = 0;
816: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
817: return(MANDOCLEVEL_ERROR);
1.82 schwarze 818: }
819:
1.98 schwarze 820: /* Run gunzip(1). */
821:
1.82 schwarze 822: if (pipe(pfd) == -1) {
1.107 schwarze 823: perror("pipe");
824: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 825: }
826:
1.97 schwarze 827: switch (curp->child = fork()) {
1.82 schwarze 828: case -1:
1.107 schwarze 829: perror("fork");
830: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 831: case 0:
832: close(pfd[0]);
833: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 834: perror("dup");
835: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 836: }
837: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 838: perror("exec");
839: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 840: default:
841: close(pfd[1]);
842: *fd = pfd[0];
843: return(MANDOCLEVEL_OK);
844: }
845: }
846:
847: enum mandoclevel
1.97 schwarze 848: mparse_wait(struct mparse *curp)
1.82 schwarze 849: {
850: int status;
851:
1.97 schwarze 852: if (curp->child == 0)
853: return(MANDOCLEVEL_OK);
854:
855: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 856: perror("wait");
857: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 858: }
859: if (WIFSIGNALED(status)) {
1.108 schwarze 860: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
861: "gunzip died from signal %d", WTERMSIG(status));
862: return(MANDOCLEVEL_ERROR);
1.82 schwarze 863: }
864: if (WEXITSTATUS(status)) {
1.108 schwarze 865: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
866: "gunzip failed with code %d", WEXITSTATUS(status));
867: return(MANDOCLEVEL_ERROR);
1.82 schwarze 868: }
869: return(MANDOCLEVEL_OK);
1.1 kristaps 870: }
871:
872: struct mparse *
1.94 schwarze 873: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
874: const struct mchars *mchars, const char *defos)
1.1 kristaps 875: {
876: struct mparse *curp;
1.10 kristaps 877:
1.1 kristaps 878: curp = mandoc_calloc(1, sizeof(struct mparse));
879:
1.44 schwarze 880: curp->options = options;
1.3 kristaps 881: curp->wlevel = wlevel;
1.1 kristaps 882: curp->mmsg = mmsg;
1.29 schwarze 883: curp->defos = defos;
1.1 kristaps 884:
1.94 schwarze 885: curp->mchars = mchars;
886: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 887: if (curp->options & MPARSE_MDOC)
888: curp->pmdoc = mdoc_alloc(
889: curp->roff, curp, curp->defos,
890: curp->options & MPARSE_QUICK ? 1 : 0);
891: if (curp->options & MPARSE_MAN)
1.106 schwarze 892: curp->pman = man_alloc(
893: curp->roff, curp, curp->defos,
1.72 schwarze 894: curp->options & MPARSE_QUICK ? 1 : 0);
895:
1.1 kristaps 896: return(curp);
897: }
898:
899: void
900: mparse_reset(struct mparse *curp)
901: {
902:
903: roff_reset(curp->roff);
904:
905: if (curp->mdoc)
906: mdoc_reset(curp->mdoc);
907: if (curp->man)
908: man_reset(curp->man);
1.24 kristaps 909: if (curp->secondary)
910: curp->secondary->sz = 0;
1.1 kristaps 911:
912: curp->file_status = MANDOCLEVEL_OK;
913: curp->mdoc = NULL;
914: curp->man = NULL;
1.45 schwarze 915:
916: free(curp->sodest);
917: curp->sodest = NULL;
1.1 kristaps 918: }
919:
920: void
921: mparse_free(struct mparse *curp)
922: {
923:
924: if (curp->pmdoc)
925: mdoc_free(curp->pmdoc);
926: if (curp->pman)
927: man_free(curp->pman);
928: if (curp->roff)
929: roff_free(curp->roff);
1.24 kristaps 930: if (curp->secondary)
931: free(curp->secondary->buf);
1.1 kristaps 932:
1.24 kristaps 933: free(curp->secondary);
1.45 schwarze 934: free(curp->sodest);
1.1 kristaps 935: free(curp);
936: }
937:
938: void
1.45 schwarze 939: mparse_result(struct mparse *curp,
940: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 941: {
942:
1.45 schwarze 943: if (sodest && NULL != (*sodest = curp->sodest)) {
944: *mdoc = NULL;
945: *man = NULL;
946: return;
947: }
1.9 kristaps 948: if (mdoc)
949: *mdoc = curp->mdoc;
950: if (man)
951: *man = curp->man;
1.3 kristaps 952: }
953:
954: void
955: mandoc_vmsg(enum mandocerr t, struct mparse *m,
956: int ln, int pos, const char *fmt, ...)
957: {
958: char buf[256];
959: va_list ap;
960:
961: va_start(ap, fmt);
1.48 schwarze 962: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 963: va_end(ap);
964:
965: mandoc_msg(t, m, ln, pos, buf);
966: }
967:
968: void
1.47 schwarze 969: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 970: int ln, int col, const char *msg)
971: {
972: enum mandoclevel level;
973:
1.112 schwarze 974: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 975: while (er < mandoclimits[level])
976: level--;
977:
1.108 schwarze 978: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 979: return;
980:
1.8 kristaps 981: if (m->mmsg)
982: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 983:
984: if (m->file_status < level)
985: m->file_status = level;
1.7 kristaps 986: }
987:
988: const char *
989: mparse_strerror(enum mandocerr er)
990: {
991:
992: return(mandocerrs[er]);
993: }
994:
995: const char *
996: mparse_strlevel(enum mandoclevel lvl)
997: {
998: return(mandoclevels[lvl]);
1.24 kristaps 999: }
1000:
1001: void
1002: mparse_keep(struct mparse *p)
1003: {
1004:
1005: assert(NULL == p->secondary);
1006: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1007: }
1008:
1009: const char *
1010: mparse_getkeep(const struct mparse *p)
1011: {
1012:
1013: assert(p->secondary);
1014: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1015: }
CVSweb