Annotation of mandoc/read.c, Revision 1.128
1.128 ! schwarze 1: /* $Id: read.c,v 1.127 2015/02/20 22:40:38 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\"",
1.128 ! schwarze 112: "NAME section without name",
! 113: "NAME section without description",
! 114: "description not at the end of NAME",
! 115: "bad NAME section content",
1.121 schwarze 116: "missing description line, using \"\"",
1.7 kristaps 117: "sections out of conventional order",
1.54 schwarze 118: "duplicate section title",
119: "unexpected section",
1.87 schwarze 120: "unusual Xr order",
121: "unusual Xr punctuation",
1.86 schwarze 122: "AUTHORS section without An macro",
1.7 kristaps 123:
124: /* related to macros and nesting */
1.55 schwarze 125: "obsolete macro",
1.102 schwarze 126: "macro neither callable nor escaped",
1.7 kristaps 127: "skipping paragraph macro",
1.31 schwarze 128: "moving paragraph macro out of list",
1.7 kristaps 129: "skipping no-space macro",
130: "blocks badly nested",
131: "nested displays are not portable",
1.57 schwarze 132: "moving content out of list",
133: ".Vt block has child macro",
1.78 schwarze 134: "fill mode already enabled, skipping",
135: "fill mode already disabled, skipping",
1.7 kristaps 136: "line scope broken",
137:
138: /* related to missing macro arguments */
1.58 schwarze 139: "skipping empty request",
140: "conditional request controls empty scope",
1.7 kristaps 141: "skipping empty macro",
1.124 schwarze 142: "empty block",
1.62 schwarze 143: "empty argument, using 0n",
1.60 schwarze 144: "missing display type, using -ragged",
145: "list type is not the first argument",
146: "missing -width in -tag list, using 8n",
1.78 schwarze 147: "missing utility name, using \"\"",
1.123 schwarze 148: "missing function name, using \"\"",
1.60 schwarze 149: "empty head in list item",
150: "empty list item",
1.61 schwarze 151: "missing font type, using \\fR",
152: "unknown font type, using \\fR",
1.103 schwarze 153: "nothing follows prefix",
1.122 schwarze 154: "empty reference block",
1.60 schwarze 155: "missing -std argument, adding it",
1.125 schwarze 156: "missing option string, using \"\"",
157: "missing resource identifier, using \"\"",
1.90 schwarze 158: "missing eqn box, using \"\"",
1.7 kristaps 159:
160: /* related to bad macro arguments */
1.64 schwarze 161: "unterminated quoted argument",
1.7 kristaps 162: "duplicate argument",
1.76 schwarze 163: "skipping duplicate argument",
1.63 schwarze 164: "skipping duplicate display type",
165: "skipping duplicate list type",
1.76 schwarze 166: "skipping -width argument",
1.126 schwarze 167: "wrong number of cells",
1.7 kristaps 168: "unknown AT&T UNIX version",
1.88 schwarze 169: "comma in function argument",
1.89 schwarze 170: "parenthesis in function name",
1.67 schwarze 171: "invalid content in Rs block",
1.63 schwarze 172: "invalid Boolean argument",
173: "unknown font, skipping request",
1.126 schwarze 174: "odd number of characters in request",
1.7 kristaps 175:
176: /* related to plain text */
1.64 schwarze 177: "blank line in fill mode, using .sp",
178: "tab in filled text",
179: "whitespace at end of input line",
1.7 kristaps 180: "bad comment style",
1.64 schwarze 181: "invalid escape sequence",
182: "undefined string, using \"\"",
1.16 kristaps 183:
1.118 schwarze 184: /* related to tables */
185: "tbl line starts with span",
186: "tbl column starts with span",
187: "skipping vertical bar in tbl layout",
188:
1.7 kristaps 189: "generic error",
190:
191: /* related to tables */
1.116 schwarze 192: "non-alphabetic character in tbl options",
193: "skipping unknown tbl option",
194: "missing tbl option argument",
195: "wrong tbl option argument size",
1.118 schwarze 196: "empty tbl layout",
197: "invalid character in tbl layout",
198: "unmatched parenthesis in tbl layout",
1.119 schwarze 199: "tbl without any data cells",
200: "ignoring data in spanned tbl cell",
201: "ignoring extra tbl data cells",
202: "data block open at end of tbl",
1.7 kristaps 203:
1.68 schwarze 204: /* related to document structure and macros */
1.108 schwarze 205: NULL,
1.7 kristaps 206: "input stack limit exceeded, infinite loop?",
207: "skipping bad character",
1.68 schwarze 208: "skipping unknown macro",
1.112 schwarze 209: "skipping insecure request",
1.73 schwarze 210: "skipping item outside list",
1.68 schwarze 211: "skipping column outside column list",
212: "skipping end of block that is not open",
1.115 schwarze 213: "fewer RS blocks open, skipping",
1.68 schwarze 214: "inserting missing end of block",
215: "appending missing end of block",
216:
217: /* related to request and macro arguments */
1.7 kristaps 218: "escaped character not allowed in a name",
1.99 schwarze 219: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 220: "missing list type, using -item",
1.70 schwarze 221: "missing manual name, using \"\"",
1.71 schwarze 222: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 223: "unknown standard specifier",
1.71 schwarze 224: "skipping request without numeric argument",
1.109 schwarze 225: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 226: ".so request failed",
1.61 schwarze 227: "skipping all arguments",
228: "skipping excess arguments",
1.92 kristaps 229: "divide by zero",
1.112 schwarze 230:
231: "unsupported feature",
232: "input too large",
1.114 schwarze 233: "unsupported control character",
1.112 schwarze 234: "unsupported roff request",
1.119 schwarze 235: "eqn delim option in tbl",
1.118 schwarze 236: "unsupported tbl layout modifier",
1.112 schwarze 237: "ignoring macro in table",
1.7 kristaps 238: };
239:
240: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
241: "SUCCESS",
242: "RESERVED",
243: "WARNING",
244: "ERROR",
1.112 schwarze 245: "UNSUPP",
1.7 kristaps 246: "BADARG",
247: "SYSERR"
248: };
249:
1.47 schwarze 250:
1.1 kristaps 251: static void
252: resize_buf(struct buf *buf, size_t initial)
253: {
254:
255: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
256: buf->buf = mandoc_realloc(buf->buf, buf->sz);
257: }
258:
259: static void
1.84 schwarze 260: choose_parser(struct mparse *curp)
1.1 kristaps 261: {
1.83 schwarze 262: char *cp, *ep;
263: int format;
1.1 kristaps 264:
1.83 schwarze 265: /*
266: * If neither command line arguments -mdoc or -man select
267: * a parser nor the roff parser found a .Dd or .TH macro
268: * yet, look ahead in the main input buffer.
269: */
270:
271: if ((format = roff_getformat(curp->roff)) == 0) {
272: cp = curp->primary->buf;
273: ep = cp + curp->primary->sz;
274: while (cp < ep) {
1.85 schwarze 275: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 276: cp++;
277: if (cp[0] == 'D' && cp[1] == 'd') {
278: format = MPARSE_MDOC;
279: break;
280: }
281: if (cp[0] == 'T' && cp[1] == 'H') {
282: format = MPARSE_MAN;
283: break;
284: }
285: }
286: cp = memchr(cp, '\n', ep - cp);
287: if (cp == NULL)
288: break;
289: cp++;
290: }
1.1 kristaps 291: }
292:
1.83 schwarze 293: if (format == MPARSE_MDOC) {
1.47 schwarze 294: if (NULL == curp->pmdoc)
1.44 schwarze 295: curp->pmdoc = mdoc_alloc(
296: curp->roff, curp, curp->defos,
297: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 298: assert(curp->pmdoc);
299: curp->mdoc = curp->pmdoc;
300: return;
1.47 schwarze 301: }
1.1 kristaps 302:
1.83 schwarze 303: /* Fall back to man(7) as a last resort. */
304:
1.47 schwarze 305: if (NULL == curp->pman)
1.106 schwarze 306: curp->pman = man_alloc(
307: curp->roff, curp, curp->defos,
1.44 schwarze 308: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 309: assert(curp->pman);
310: curp->man = curp->pman;
311: }
312:
313: /*
1.95 schwarze 314: * Main parse routine for a buffer.
315: * It assumes encoding and line numbering are already set up.
316: * It can recurse directly (for invocations of user-defined
317: * macros, inline equations, and input line traps)
318: * and indirectly (for .so file inclusion).
1.1 kristaps 319: */
320: static void
1.95 schwarze 321: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 322: {
323: const struct tbl_span *span;
324: struct buf ln;
1.113 schwarze 325: const char *save_file;
1.110 schwarze 326: char *cp;
1.95 schwarze 327: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 328: enum rofferr rr;
1.100 schwarze 329: int of;
1.1 kristaps 330: int lnn; /* line number in the real file */
1.110 schwarze 331: int fd;
332: pid_t save_child;
1.1 kristaps 333: unsigned char c;
334:
1.95 schwarze 335: memset(&ln, 0, sizeof(ln));
1.1 kristaps 336:
1.47 schwarze 337: lnn = curp->line;
338: pos = 0;
1.1 kristaps 339:
1.95 schwarze 340: while (i < blk.sz) {
1.1 kristaps 341: if (0 == pos && '\0' == blk.buf[i])
342: break;
343:
344: if (start) {
345: curp->line = lnn;
346: curp->reparse_count = 0;
1.93 schwarze 347:
348: if (lnn < 3 &&
349: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 350: curp->filenc & MPARSE_LATIN1)
351: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 352: }
353:
1.95 schwarze 354: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 355:
356: /*
357: * When finding an unescaped newline character,
358: * leave the character loop to process the line.
359: * Skip a preceding carriage return, if any.
360: */
361:
1.95 schwarze 362: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 363: '\n' == blk.buf[i + 1])
364: ++i;
365: if ('\n' == blk.buf[i]) {
366: ++i;
367: ++lnn;
368: break;
369: }
370:
1.35 schwarze 371: /*
1.93 schwarze 372: * Make sure we have space for the worst
373: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 374: */
375:
1.95 schwarze 376: if (pos + 11 > ln.sz)
1.35 schwarze 377: resize_buf(&ln, 256);
378:
1.47 schwarze 379: /*
1.93 schwarze 380: * Encode 8-bit input.
1.1 kristaps 381: */
382:
1.93 schwarze 383: c = blk.buf[i];
384: if (c & 0x80) {
1.95 schwarze 385: if ( ! (curp->filenc && preconv_encode(
386: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 387: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
388: curp->line, pos, "0x%x", c);
1.93 schwarze 389: ln.buf[pos++] = '?';
390: i++;
391: }
392: continue;
393: }
394:
395: /*
396: * Exclude control characters.
397: */
1.1 kristaps 398:
1.93 schwarze 399: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 400: mandoc_vmsg(c == 0x00 || c == 0x04 ||
401: c > 0x0a ? MANDOCERR_CHAR_BAD :
402: MANDOCERR_CHAR_UNSUPP,
403: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 404: i++;
1.127 schwarze 405: if (c != '\r')
406: ln.buf[pos++] = '?';
1.1 kristaps 407: continue;
408: }
409:
410: /* Trailing backslash = a plain char. */
411:
1.95 schwarze 412: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 413: ln.buf[pos++] = blk.buf[i++];
414: continue;
415: }
416:
417: /*
418: * Found escape and at least one other character.
419: * When it's a newline character, skip it.
420: * When there is a carriage return in between,
421: * skip that one as well.
422: */
423:
1.95 schwarze 424: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 425: '\n' == blk.buf[i + 2])
426: ++i;
427: if ('\n' == blk.buf[i + 1]) {
428: i += 2;
429: ++lnn;
430: continue;
431: }
432:
1.13 kristaps 433: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 434: i += 2;
435: /* Comment, skip to end of line */
1.95 schwarze 436: for (; i < blk.sz; ++i) {
1.1 kristaps 437: if ('\n' == blk.buf[i]) {
438: ++i;
439: ++lnn;
440: break;
441: }
442: }
443:
444: /* Backout trailing whitespaces */
445: for (; pos > 0; --pos) {
446: if (ln.buf[pos - 1] != ' ')
447: break;
448: if (pos > 2 && ln.buf[pos - 2] == '\\')
449: break;
450: }
451: break;
452: }
453:
1.35 schwarze 454: /* Catch escaped bogus characters. */
455:
456: c = (unsigned char) blk.buf[i+1];
457:
1.47 schwarze 458: if ( ! (isascii(c) &&
459: (isgraph(c) || isblank(c)))) {
1.114 schwarze 460: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 461: curp->line, pos, "0x%x", c);
1.35 schwarze 462: i += 2;
463: ln.buf[pos++] = '?';
464: continue;
465: }
466:
1.1 kristaps 467: /* Some other escape sequence, copy & cont. */
468:
469: ln.buf[pos++] = blk.buf[i++];
470: ln.buf[pos++] = blk.buf[i++];
471: }
472:
1.95 schwarze 473: if (pos >= ln.sz)
1.1 kristaps 474: resize_buf(&ln, 256);
475:
476: ln.buf[pos] = '\0';
477:
478: /*
479: * A significant amount of complexity is contained by
480: * the roff preprocessor. It's line-oriented but can be
481: * expressed on one line, so we need at times to
482: * readjust our starting point and re-run it. The roff
483: * preprocessor can also readjust the buffers with new
484: * data, so we pass them in wholesale.
485: */
486:
487: of = 0;
488:
1.24 kristaps 489: /*
490: * Maintain a lookaside buffer of all parsed lines. We
491: * only do this if mparse_keep() has been invoked (the
492: * buffer may be accessed with mparse_getkeep()).
493: */
494:
495: if (curp->secondary) {
1.47 schwarze 496: curp->secondary->buf = mandoc_realloc(
497: curp->secondary->buf,
498: curp->secondary->sz + pos + 2);
499: memcpy(curp->secondary->buf +
500: curp->secondary->sz,
501: ln.buf, pos);
1.24 kristaps 502: curp->secondary->sz += pos;
503: curp->secondary->buf
504: [curp->secondary->sz] = '\n';
505: curp->secondary->sz++;
506: curp->secondary->buf
507: [curp->secondary->sz] = '\0';
508: }
1.1 kristaps 509: rerun:
1.96 schwarze 510: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 511:
512: switch (rr) {
1.47 schwarze 513: case ROFF_REPARSE:
1.1 kristaps 514: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 515: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 516: else
1.3 kristaps 517: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 518: curp->line, pos, NULL);
1.1 kristaps 519: pos = 0;
520: continue;
1.47 schwarze 521: case ROFF_APPEND:
1.95 schwarze 522: pos = strlen(ln.buf);
1.1 kristaps 523: continue;
1.47 schwarze 524: case ROFF_RERUN:
1.1 kristaps 525: goto rerun;
1.47 schwarze 526: case ROFF_IGN:
1.1 kristaps 527: pos = 0;
528: continue;
1.47 schwarze 529: case ROFF_SO:
1.95 schwarze 530: if ( ! (curp->options & MPARSE_SO) &&
531: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 532: curp->sodest = mandoc_strdup(ln.buf + of);
533: free(ln.buf);
534: return;
535: }
1.24 kristaps 536: /*
537: * We remove `so' clauses from our lookaside
538: * buffer because we're going to descend into
539: * the file recursively.
540: */
1.47 schwarze 541: if (curp->secondary)
1.25 kristaps 542: curp->secondary->sz -= pos + 1;
1.113 schwarze 543: save_file = curp->file;
1.110 schwarze 544: save_child = curp->child;
545: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 schwarze 546: MANDOCLEVEL_OK) {
1.110 schwarze 547: mparse_readfd(curp, fd, ln.buf + of);
1.113 schwarze 548: curp->file = save_file;
549: } else {
550: curp->file = save_file;
1.52 schwarze 551: mandoc_vmsg(MANDOCERR_SO_FAIL,
552: curp, curp->line, pos,
553: ".so %s", ln.buf + of);
1.110 schwarze 554: ln.sz = mandoc_asprintf(&cp,
555: ".sp\nSee the file %s.\n.sp",
556: ln.buf + of);
557: free(ln.buf);
558: ln.buf = cp;
559: of = 0;
560: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 561: }
1.110 schwarze 562: curp->child = save_child;
1.1 kristaps 563: pos = 0;
564: continue;
565: default:
566: break;
567: }
568:
569: /*
570: * If input parsers have not been allocated, do so now.
1.14 kristaps 571: * We keep these instanced between parsers, but set them
1.1 kristaps 572: * locally per parse routine since we can use different
573: * parsers with each one.
574: */
575:
576: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 577: choose_parser(curp);
1.1 kristaps 578:
1.47 schwarze 579: /*
1.84 schwarze 580: * Lastly, push down into the parsers themselves.
1.1 kristaps 581: * If libroff returns ROFF_TBL, then add it to the
582: * currently open parse. Since we only get here if
583: * there does exist data (see tbl_data.c), we're
584: * guaranteed that something's been allocated.
585: * Do the same for ROFF_EQN.
586: */
587:
1.100 schwarze 588: if (rr == ROFF_TBL) {
589: while ((span = roff_span(curp->roff)) != NULL)
590: if (curp->man == NULL)
591: mdoc_addspan(curp->mdoc, span);
592: else
593: man_addspan(curp->man, span);
594: } else if (rr == ROFF_EQN) {
595: if (curp->man == NULL)
596: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
597: else
598: man_addeqn(curp->man, roff_eqn(curp->roff));
599: } else if ((curp->man == NULL ?
600: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
601: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
602: break;
1.1 kristaps 603:
604: /* Temporary buffers typically are not full. */
605:
606: if (0 == start && '\0' == blk.buf[i])
607: break;
608:
609: /* Start the next input line. */
610:
611: pos = 0;
612: }
613:
614: free(ln.buf);
615: }
616:
617: static int
1.40 schwarze 618: read_whole_file(struct mparse *curp, const char *file, int fd,
619: struct buf *fb, int *with_mmap)
1.1 kristaps 620: {
621: size_t off;
622: ssize_t ssz;
623:
1.81 schwarze 624: #if HAVE_MMAP
1.15 kristaps 625: struct stat st;
1.1 kristaps 626: if (-1 == fstat(fd, &st)) {
1.107 schwarze 627: perror(file);
628: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 629: }
630:
631: /*
632: * If we're a regular file, try just reading in the whole entry
633: * via mmap(). This is faster than reading it into blocks, and
634: * since each file is only a few bytes to begin with, I'm not
635: * concerned that this is going to tank any machines.
636: */
637:
638: if (S_ISREG(st.st_mode)) {
639: if (st.st_size >= (1U << 31)) {
1.111 schwarze 640: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 641: return(0);
642: }
643: *with_mmap = 1;
644: fb->sz = (size_t)st.st_size;
1.37 schwarze 645: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 646: if (fb->buf != MAP_FAILED)
647: return(1);
648: }
1.15 kristaps 649: #endif
1.1 kristaps 650:
651: /*
652: * If this isn't a regular file (like, say, stdin), then we must
653: * go the old way and just read things in bit by bit.
654: */
655:
656: *with_mmap = 0;
657: off = 0;
658: fb->sz = 0;
659: fb->buf = NULL;
660: for (;;) {
661: if (off == fb->sz) {
662: if (fb->sz == (1U << 31)) {
1.111 schwarze 663: mandoc_msg(MANDOCERR_TOOLARGE, curp,
664: 0, 0, NULL);
1.1 kristaps 665: break;
666: }
667: resize_buf(fb, 65536);
668: }
669: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
670: if (ssz == 0) {
671: fb->sz = off;
672: return(1);
673: }
674: if (ssz == -1) {
1.107 schwarze 675: perror(file);
676: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 677: }
678: off += (size_t)ssz;
679: }
680:
681: free(fb->buf);
682: fb->buf = NULL;
683: return(0);
684: }
685:
686: static void
687: mparse_end(struct mparse *curp)
688: {
689:
1.72 schwarze 690: if (curp->mdoc == NULL &&
691: curp->man == NULL &&
692: curp->sodest == NULL) {
693: if (curp->options & MPARSE_MDOC)
694: curp->mdoc = curp->pmdoc;
695: else {
696: if (curp->pman == NULL)
1.106 schwarze 697: curp->pman = man_alloc(
698: curp->roff, curp, curp->defos,
1.72 schwarze 699: curp->options & MPARSE_QUICK ? 1 : 0);
700: curp->man = curp->pman;
701: }
702: }
1.111 schwarze 703: if (curp->mdoc)
704: mdoc_endparse(curp->mdoc);
705: if (curp->man)
706: man_endparse(curp->man);
1.1 kristaps 707: roff_endparse(curp->roff);
708: }
709:
710: static void
1.36 schwarze 711: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 712: {
1.85 schwarze 713: struct buf *svprimary;
1.28 joerg 714: const char *svfile;
1.95 schwarze 715: size_t offset;
1.36 schwarze 716: static int recursion_depth;
717:
718: if (64 < recursion_depth) {
719: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
720: return;
721: }
1.28 joerg 722:
723: /* Line number is per-file. */
724: svfile = curp->file;
725: curp->file = file;
1.85 schwarze 726: svprimary = curp->primary;
1.83 schwarze 727: curp->primary = &blk;
1.28 joerg 728: curp->line = 1;
1.36 schwarze 729: recursion_depth++;
1.28 joerg 730:
1.93 schwarze 731: /* Skip an UTF-8 byte order mark. */
732: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
733: (unsigned char)blk.buf[0] == 0xef &&
734: (unsigned char)blk.buf[1] == 0xbb &&
735: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 736: offset = 3;
1.93 schwarze 737: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 738: } else
739: offset = 0;
1.93 schwarze 740:
1.95 schwarze 741: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 742:
1.111 schwarze 743: if (--recursion_depth == 0)
1.28 joerg 744: mparse_end(curp);
745:
1.85 schwarze 746: curp->primary = svprimary;
1.28 joerg 747: curp->file = svfile;
748: }
749:
750: enum mandoclevel
1.104 schwarze 751: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 752: const char *file)
753: {
754: struct buf blk;
755:
1.104 schwarze 756: blk.buf = buf;
1.28 joerg 757: blk.sz = len;
758:
1.36 schwarze 759: mparse_parse_buffer(curp, blk, file);
1.28 joerg 760: return(curp->file_status);
761: }
762:
1.98 schwarze 763: /*
764: * Read the whole file into memory and call the parsers.
765: * Called recursively when an .so request is encountered.
766: */
1.36 schwarze 767: enum mandoclevel
768: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 769: {
1.28 joerg 770: struct buf blk;
771: int with_mmap;
1.93 schwarze 772: int save_filenc;
1.1 kristaps 773:
1.91 schwarze 774: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 775: save_filenc = curp->filenc;
776: curp->filenc = curp->options &
777: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 778: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 779: curp->filenc = save_filenc;
1.81 schwarze 780: #if HAVE_MMAP
1.91 schwarze 781: if (with_mmap)
782: munmap(blk.buf, blk.sz);
783: else
1.28 joerg 784: #endif
1.91 schwarze 785: free(blk.buf);
786: }
1.1 kristaps 787:
1.98 schwarze 788: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 789: perror(file);
1.91 schwarze 790:
1.98 schwarze 791: mparse_wait(curp);
1.1 kristaps 792: return(curp->file_status);
1.82 schwarze 793: }
794:
795: enum mandoclevel
1.97 schwarze 796: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 797: {
798: int pfd[2];
1.98 schwarze 799: int save_errno;
1.82 schwarze 800: char *cp;
801:
802: curp->file = file;
1.98 schwarze 803:
804: /* Unless zipped, try to just open the file. */
805:
1.82 schwarze 806: if ((cp = strrchr(file, '.')) == NULL ||
807: strcmp(cp + 1, "gz")) {
1.97 schwarze 808: curp->child = 0;
1.98 schwarze 809: if ((*fd = open(file, O_RDONLY)) != -1)
810: return(MANDOCLEVEL_OK);
811:
812: /* Open failed; try to append ".gz". */
813:
814: mandoc_asprintf(&cp, "%s.gz", file);
815: file = cp;
816: } else
817: cp = NULL;
818:
819: /* Before forking, make sure the file can be read. */
820:
821: save_errno = errno;
822: if (access(file, R_OK) == -1) {
823: if (cp != NULL)
824: errno = save_errno;
1.108 schwarze 825: free(cp);
826: *fd = -1;
827: curp->child = 0;
828: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
829: return(MANDOCLEVEL_ERROR);
1.82 schwarze 830: }
831:
1.98 schwarze 832: /* Run gunzip(1). */
833:
1.82 schwarze 834: if (pipe(pfd) == -1) {
1.107 schwarze 835: perror("pipe");
836: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 837: }
838:
1.97 schwarze 839: switch (curp->child = fork()) {
1.82 schwarze 840: case -1:
1.107 schwarze 841: perror("fork");
842: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 843: case 0:
844: close(pfd[0]);
845: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 846: perror("dup");
847: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 848: }
849: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 850: perror("exec");
851: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 852: default:
853: close(pfd[1]);
854: *fd = pfd[0];
855: return(MANDOCLEVEL_OK);
856: }
857: }
858:
859: enum mandoclevel
1.97 schwarze 860: mparse_wait(struct mparse *curp)
1.82 schwarze 861: {
862: int status;
863:
1.97 schwarze 864: if (curp->child == 0)
865: return(MANDOCLEVEL_OK);
866:
867: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 868: perror("wait");
869: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 870: }
871: if (WIFSIGNALED(status)) {
1.108 schwarze 872: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
873: "gunzip died from signal %d", WTERMSIG(status));
874: return(MANDOCLEVEL_ERROR);
1.82 schwarze 875: }
876: if (WEXITSTATUS(status)) {
1.108 schwarze 877: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
878: "gunzip failed with code %d", WEXITSTATUS(status));
879: return(MANDOCLEVEL_ERROR);
1.82 schwarze 880: }
881: return(MANDOCLEVEL_OK);
1.1 kristaps 882: }
883:
884: struct mparse *
1.94 schwarze 885: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
886: const struct mchars *mchars, const char *defos)
1.1 kristaps 887: {
888: struct mparse *curp;
1.10 kristaps 889:
1.1 kristaps 890: curp = mandoc_calloc(1, sizeof(struct mparse));
891:
1.44 schwarze 892: curp->options = options;
1.3 kristaps 893: curp->wlevel = wlevel;
1.1 kristaps 894: curp->mmsg = mmsg;
1.29 schwarze 895: curp->defos = defos;
1.1 kristaps 896:
1.94 schwarze 897: curp->mchars = mchars;
898: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 899: if (curp->options & MPARSE_MDOC)
900: curp->pmdoc = mdoc_alloc(
901: curp->roff, curp, curp->defos,
902: curp->options & MPARSE_QUICK ? 1 : 0);
903: if (curp->options & MPARSE_MAN)
1.106 schwarze 904: curp->pman = man_alloc(
905: curp->roff, curp, curp->defos,
1.72 schwarze 906: curp->options & MPARSE_QUICK ? 1 : 0);
907:
1.1 kristaps 908: return(curp);
909: }
910:
911: void
912: mparse_reset(struct mparse *curp)
913: {
914:
915: roff_reset(curp->roff);
916:
917: if (curp->mdoc)
918: mdoc_reset(curp->mdoc);
919: if (curp->man)
920: man_reset(curp->man);
1.24 kristaps 921: if (curp->secondary)
922: curp->secondary->sz = 0;
1.1 kristaps 923:
924: curp->file_status = MANDOCLEVEL_OK;
925: curp->mdoc = NULL;
926: curp->man = NULL;
1.45 schwarze 927:
928: free(curp->sodest);
929: curp->sodest = NULL;
1.1 kristaps 930: }
931:
932: void
933: mparse_free(struct mparse *curp)
934: {
935:
936: if (curp->pmdoc)
937: mdoc_free(curp->pmdoc);
938: if (curp->pman)
939: man_free(curp->pman);
940: if (curp->roff)
941: roff_free(curp->roff);
1.24 kristaps 942: if (curp->secondary)
943: free(curp->secondary->buf);
1.1 kristaps 944:
1.24 kristaps 945: free(curp->secondary);
1.45 schwarze 946: free(curp->sodest);
1.1 kristaps 947: free(curp);
948: }
949:
950: void
1.45 schwarze 951: mparse_result(struct mparse *curp,
952: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 953: {
954:
1.45 schwarze 955: if (sodest && NULL != (*sodest = curp->sodest)) {
956: *mdoc = NULL;
957: *man = NULL;
958: return;
959: }
1.9 kristaps 960: if (mdoc)
961: *mdoc = curp->mdoc;
962: if (man)
963: *man = curp->man;
1.3 kristaps 964: }
965:
966: void
967: mandoc_vmsg(enum mandocerr t, struct mparse *m,
968: int ln, int pos, const char *fmt, ...)
969: {
970: char buf[256];
971: va_list ap;
972:
973: va_start(ap, fmt);
1.48 schwarze 974: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 975: va_end(ap);
976:
977: mandoc_msg(t, m, ln, pos, buf);
978: }
979:
980: void
1.47 schwarze 981: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 982: int ln, int col, const char *msg)
983: {
984: enum mandoclevel level;
985:
1.112 schwarze 986: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 987: while (er < mandoclimits[level])
988: level--;
989:
1.108 schwarze 990: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 991: return;
992:
1.8 kristaps 993: if (m->mmsg)
994: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 995:
996: if (m->file_status < level)
997: m->file_status = level;
1.7 kristaps 998: }
999:
1000: const char *
1001: mparse_strerror(enum mandocerr er)
1002: {
1003:
1004: return(mandocerrs[er]);
1005: }
1006:
1007: const char *
1008: mparse_strlevel(enum mandoclevel lvl)
1009: {
1010: return(mandoclevels[lvl]);
1.24 kristaps 1011: }
1012:
1013: void
1014: mparse_keep(struct mparse *p)
1015: {
1016:
1017: assert(NULL == p->secondary);
1018: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1019: }
1020:
1021: const char *
1022: mparse_getkeep(const struct mparse *p)
1023: {
1024:
1025: assert(p->secondary);
1026: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1027: }
CVSweb