Annotation of src/usr.bin/mandoc/read.c, Revision 1.108
1.108 ! schwarze 1: /* $OpenBSD: read.c,v 1.107 2015/03/17 07:32:21 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.82 schwarze 4: * Copyright (c) 2010-2015 Ingo Schwarze <schwarze@openbsd.org>
1.20 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 schwarze 6: *
7: * Permission to use, copy, modify, and distribute this software for any
8: * purpose with or without fee is hereby granted, provided that the above
9: * copyright notice and this permission notice appear in all copies.
10: *
1.108 ! schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.108 ! schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18: */
1.58 schwarze 19: #include <sys/types.h>
20: #include <sys/mman.h>
1.1 schwarze 21: #include <sys/stat.h>
1.58 schwarze 22: #include <sys/wait.h>
1.1 schwarze 23:
24: #include <assert.h>
25: #include <ctype.h>
1.18 schwarze 26: #include <errno.h>
1.1 schwarze 27: #include <fcntl.h>
1.107 schwarze 28: #include <signal.h>
1.1 schwarze 29: #include <stdarg.h>
30: #include <stdio.h>
31: #include <stdlib.h>
32: #include <string.h>
33: #include <unistd.h>
34:
1.108 ! schwarze 35: #include "mandoc_aux.h"
1.1 schwarze 36: #include "mandoc.h"
1.108 ! schwarze 37: #include "roff.h"
1.1 schwarze 38: #include "mdoc.h"
39: #include "man.h"
1.108 ! schwarze 40: #include "libmandoc.h"
1.1 schwarze 41:
42: #define REPARSE_LIMIT 1000
43:
44: struct mparse {
45: struct man *pman; /* persistent man parser */
46: struct mdoc *pmdoc; /* persistent mdoc parser */
47: struct man *man; /* man parser */
48: struct mdoc *mdoc; /* mdoc parser */
49: struct roff *roff; /* roff parser (!NULL) */
1.69 schwarze 50: const struct mchars *mchars; /* character table */
1.23 schwarze 51: char *sodest; /* filename pointed to by .so */
1.59 schwarze 52: const char *file; /* filename of current input file */
53: struct buf *primary; /* buffer currently being parsed */
54: struct buf *secondary; /* preprocessed copy of input */
55: const char *defos; /* default operating system */
56: mandocmsg mmsg; /* warning/error message handler */
57: enum mandoclevel file_status; /* status of current parse */
58: enum mandoclevel wlevel; /* ignore messages below this */
59: int options; /* parser options */
1.70 schwarze 60: int filenc; /* encoding of the current file */
1.1 schwarze 61: int reparse_count; /* finite interp. stack */
1.59 schwarze 62: int line; /* line number in the file */
1.73 schwarze 63: pid_t child; /* the gunzip(1) process */
1.1 schwarze 64: };
65:
1.60 schwarze 66: static void choose_parser(struct mparse *);
1.1 schwarze 67: static void resize_buf(struct buf *, size_t);
1.71 schwarze 68: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.18 schwarze 69: static int read_whole_file(struct mparse *, const char *, int,
70: struct buf *, int *);
1.1 schwarze 71: static void mparse_end(struct mparse *);
1.15 schwarze 72: static void mparse_parse_buffer(struct mparse *, struct buf,
73: const char *);
1.1 schwarze 74:
75: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
76: MANDOCERR_OK,
77: MANDOCERR_WARNING,
78: MANDOCERR_WARNING,
79: MANDOCERR_ERROR,
1.87 schwarze 80: MANDOCERR_UNSUPP,
1.1 schwarze 81: MANDOCERR_MAX,
82: MANDOCERR_MAX
83: };
84:
85: static const char * const mandocerrs[MANDOCERR_MAX] = {
86: "ok",
87:
88: "generic warning",
89:
90: /* related to the prologue */
1.57 schwarze 91: "missing manual title, using UNTITLED",
92: "missing manual title, using \"\"",
1.32 schwarze 93: "lower case character in document title",
1.57 schwarze 94: "missing manual section, using \"\"",
1.1 schwarze 95: "unknown manual section",
1.32 schwarze 96: "missing date, using today's date",
1.1 schwarze 97: "cannot parse date, using it verbatim",
1.57 schwarze 98: "missing Os macro, using \"\"",
99: "duplicate prologue macro",
100: "late prologue macro",
101: "skipping late title macro",
1.1 schwarze 102: "prologue macros out of order",
103:
104: /* related to document structure */
105: ".so is fragile, better use ln(1)",
1.28 schwarze 106: "no document body",
1.32 schwarze 107: "content before first section header",
108: "first section is not \"NAME\"",
1.103 schwarze 109: "NAME section without name",
110: "NAME section without description",
111: "description not at the end of NAME",
112: "bad NAME section content",
1.96 schwarze 113: "missing description line, using \"\"",
1.1 schwarze 114: "sections out of conventional order",
1.32 schwarze 115: "duplicate section title",
116: "unexpected section",
1.63 schwarze 117: "unusual Xr order",
118: "unusual Xr punctuation",
1.62 schwarze 119: "AUTHORS section without An macro",
1.1 schwarze 120:
121: /* related to macros and nesting */
1.33 schwarze 122: "obsolete macro",
1.78 schwarze 123: "macro neither callable nor escaped",
1.1 schwarze 124: "skipping paragraph macro",
1.10 schwarze 125: "moving paragraph macro out of list",
1.1 schwarze 126: "skipping no-space macro",
127: "blocks badly nested",
128: "nested displays are not portable",
1.35 schwarze 129: "moving content out of list",
130: ".Vt block has child macro",
1.56 schwarze 131: "fill mode already enabled, skipping",
132: "fill mode already disabled, skipping",
1.1 schwarze 133: "line scope broken",
134:
135: /* related to missing macro arguments */
1.36 schwarze 136: "skipping empty request",
137: "conditional request controls empty scope",
1.1 schwarze 138: "skipping empty macro",
1.99 schwarze 139: "empty block",
1.40 schwarze 140: "empty argument, using 0n",
1.38 schwarze 141: "missing display type, using -ragged",
142: "list type is not the first argument",
143: "missing -width in -tag list, using 8n",
1.56 schwarze 144: "missing utility name, using \"\"",
1.98 schwarze 145: "missing function name, using \"\"",
1.38 schwarze 146: "empty head in list item",
147: "empty list item",
1.39 schwarze 148: "missing font type, using \\fR",
149: "unknown font type, using \\fR",
1.79 schwarze 150: "nothing follows prefix",
1.97 schwarze 151: "empty reference block",
1.38 schwarze 152: "missing -std argument, adding it",
1.100 schwarze 153: "missing option string, using \"\"",
154: "missing resource identifier, using \"\"",
1.66 schwarze 155: "missing eqn box, using \"\"",
1.1 schwarze 156:
157: /* related to bad macro arguments */
1.42 schwarze 158: "unterminated quoted argument",
1.1 schwarze 159: "duplicate argument",
1.54 schwarze 160: "skipping duplicate argument",
1.41 schwarze 161: "skipping duplicate display type",
162: "skipping duplicate list type",
1.54 schwarze 163: "skipping -width argument",
1.101 schwarze 164: "wrong number of cells",
1.1 schwarze 165: "unknown AT&T UNIX version",
1.64 schwarze 166: "comma in function argument",
1.65 schwarze 167: "parenthesis in function name",
1.45 schwarze 168: "invalid content in Rs block",
1.41 schwarze 169: "invalid Boolean argument",
170: "unknown font, skipping request",
1.101 schwarze 171: "odd number of characters in request",
1.1 schwarze 172:
173: /* related to plain text */
1.42 schwarze 174: "blank line in fill mode, using .sp",
175: "tab in filled text",
176: "whitespace at end of input line",
1.1 schwarze 177: "bad comment style",
1.42 schwarze 178: "invalid escape sequence",
179: "undefined string, using \"\"",
1.3 schwarze 180:
1.93 schwarze 181: /* related to tables */
182: "tbl line starts with span",
183: "tbl column starts with span",
184: "skipping vertical bar in tbl layout",
185:
1.1 schwarze 186: "generic error",
1.3 schwarze 187:
1.1 schwarze 188: /* related to tables */
1.91 schwarze 189: "non-alphabetic character in tbl options",
190: "skipping unknown tbl option",
191: "missing tbl option argument",
192: "wrong tbl option argument size",
1.93 schwarze 193: "empty tbl layout",
194: "invalid character in tbl layout",
195: "unmatched parenthesis in tbl layout",
1.94 schwarze 196: "tbl without any data cells",
197: "ignoring data in spanned tbl cell",
198: "ignoring extra tbl data cells",
199: "data block open at end of tbl",
1.1 schwarze 200:
1.46 schwarze 201: /* related to document structure and macros */
1.83 schwarze 202: NULL,
1.1 schwarze 203: "input stack limit exceeded, infinite loop?",
204: "skipping bad character",
1.46 schwarze 205: "skipping unknown macro",
1.87 schwarze 206: "skipping insecure request",
1.51 schwarze 207: "skipping item outside list",
1.46 schwarze 208: "skipping column outside column list",
209: "skipping end of block that is not open",
1.90 schwarze 210: "fewer RS blocks open, skipping",
1.46 schwarze 211: "inserting missing end of block",
212: "appending missing end of block",
213:
214: /* related to request and macro arguments */
1.1 schwarze 215: "escaped character not allowed in a name",
1.75 schwarze 216: "NOT IMPLEMENTED: Bd -file",
1.49 schwarze 217: "missing list type, using -item",
1.48 schwarze 218: "missing manual name, using \"\"",
1.49 schwarze 219: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 220: "unknown standard specifier",
1.49 schwarze 221: "skipping request without numeric argument",
1.84 schwarze 222: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 223: ".so request failed",
1.39 schwarze 224: "skipping all arguments",
225: "skipping excess arguments",
1.68 schwarze 226: "divide by zero",
1.87 schwarze 227:
228: "unsupported feature",
229: "input too large",
1.89 schwarze 230: "unsupported control character",
1.87 schwarze 231: "unsupported roff request",
1.94 schwarze 232: "eqn delim option in tbl",
1.93 schwarze 233: "unsupported tbl layout modifier",
1.87 schwarze 234: "ignoring macro in table",
1.1 schwarze 235: };
236:
237: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
238: "SUCCESS",
239: "RESERVED",
240: "WARNING",
241: "ERROR",
1.87 schwarze 242: "UNSUPP",
1.1 schwarze 243: "BADARG",
244: "SYSERR"
245: };
246:
1.25 schwarze 247:
1.1 schwarze 248: static void
249: resize_buf(struct buf *buf, size_t initial)
250: {
251:
252: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
253: buf->buf = mandoc_realloc(buf->buf, buf->sz);
254: }
255:
256: static void
1.60 schwarze 257: choose_parser(struct mparse *curp)
1.1 schwarze 258: {
1.59 schwarze 259: char *cp, *ep;
260: int format;
1.1 schwarze 261:
1.59 schwarze 262: /*
263: * If neither command line arguments -mdoc or -man select
264: * a parser nor the roff parser found a .Dd or .TH macro
265: * yet, look ahead in the main input buffer.
266: */
267:
268: if ((format = roff_getformat(curp->roff)) == 0) {
269: cp = curp->primary->buf;
270: ep = cp + curp->primary->sz;
271: while (cp < ep) {
1.61 schwarze 272: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 273: cp++;
274: if (cp[0] == 'D' && cp[1] == 'd') {
275: format = MPARSE_MDOC;
276: break;
277: }
278: if (cp[0] == 'T' && cp[1] == 'H') {
279: format = MPARSE_MAN;
280: break;
281: }
282: }
283: cp = memchr(cp, '\n', ep - cp);
284: if (cp == NULL)
285: break;
286: cp++;
287: }
1.1 schwarze 288: }
289:
1.59 schwarze 290: if (format == MPARSE_MDOC) {
1.25 schwarze 291: if (NULL == curp->pmdoc)
1.22 schwarze 292: curp->pmdoc = mdoc_alloc(
293: curp->roff, curp, curp->defos,
294: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 schwarze 295: assert(curp->pmdoc);
296: curp->mdoc = curp->pmdoc;
297: return;
1.25 schwarze 298: }
1.1 schwarze 299:
1.59 schwarze 300: /* Fall back to man(7) as a last resort. */
301:
1.25 schwarze 302: if (NULL == curp->pman)
1.81 schwarze 303: curp->pman = man_alloc(
304: curp->roff, curp, curp->defos,
1.22 schwarze 305: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 schwarze 306: assert(curp->pman);
307: curp->man = curp->pman;
308: }
309:
310: /*
1.71 schwarze 311: * Main parse routine for a buffer.
312: * It assumes encoding and line numbering are already set up.
313: * It can recurse directly (for invocations of user-defined
314: * macros, inline equations, and input line traps)
315: * and indirectly (for .so file inclusion).
1.1 schwarze 316: */
317: static void
1.71 schwarze 318: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 319: {
320: const struct tbl_span *span;
321: struct buf ln;
1.88 schwarze 322: const char *save_file;
1.85 schwarze 323: char *cp;
1.71 schwarze 324: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 325: enum rofferr rr;
1.76 schwarze 326: int of;
1.1 schwarze 327: int lnn; /* line number in the real file */
1.85 schwarze 328: int fd;
329: pid_t save_child;
1.1 schwarze 330: unsigned char c;
331:
1.71 schwarze 332: memset(&ln, 0, sizeof(ln));
1.1 schwarze 333:
1.25 schwarze 334: lnn = curp->line;
335: pos = 0;
1.1 schwarze 336:
1.71 schwarze 337: while (i < blk.sz) {
1.1 schwarze 338: if (0 == pos && '\0' == blk.buf[i])
339: break;
340:
341: if (start) {
342: curp->line = lnn;
343: curp->reparse_count = 0;
1.70 schwarze 344:
345: if (lnn < 3 &&
346: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 347: curp->filenc & MPARSE_LATIN1)
348: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 349: }
350:
1.71 schwarze 351: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 352:
353: /*
354: * When finding an unescaped newline character,
355: * leave the character loop to process the line.
356: * Skip a preceding carriage return, if any.
357: */
358:
1.71 schwarze 359: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 360: '\n' == blk.buf[i + 1])
361: ++i;
362: if ('\n' == blk.buf[i]) {
363: ++i;
364: ++lnn;
365: break;
366: }
367:
1.13 schwarze 368: /*
1.70 schwarze 369: * Make sure we have space for the worst
370: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 371: */
372:
1.71 schwarze 373: if (pos + 11 > ln.sz)
1.13 schwarze 374: resize_buf(&ln, 256);
375:
1.25 schwarze 376: /*
1.70 schwarze 377: * Encode 8-bit input.
1.1 schwarze 378: */
379:
1.70 schwarze 380: c = blk.buf[i];
381: if (c & 0x80) {
1.71 schwarze 382: if ( ! (curp->filenc && preconv_encode(
383: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 384: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
385: curp->line, pos, "0x%x", c);
1.70 schwarze 386: ln.buf[pos++] = '?';
387: i++;
388: }
389: continue;
390: }
391:
392: /*
393: * Exclude control characters.
394: */
1.1 schwarze 395:
1.70 schwarze 396: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 397: mandoc_vmsg(c == 0x00 || c == 0x04 ||
398: c > 0x0a ? MANDOCERR_CHAR_BAD :
399: MANDOCERR_CHAR_UNSUPP,
400: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 401: i++;
1.102 schwarze 402: if (c != '\r')
403: ln.buf[pos++] = '?';
1.1 schwarze 404: continue;
405: }
406:
407: /* Trailing backslash = a plain char. */
408:
1.71 schwarze 409: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 schwarze 410: ln.buf[pos++] = blk.buf[i++];
411: continue;
412: }
413:
414: /*
415: * Found escape and at least one other character.
416: * When it's a newline character, skip it.
417: * When there is a carriage return in between,
418: * skip that one as well.
419: */
420:
1.71 schwarze 421: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 schwarze 422: '\n' == blk.buf[i + 2])
423: ++i;
424: if ('\n' == blk.buf[i + 1]) {
425: i += 2;
426: ++lnn;
427: continue;
428: }
429:
430: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
431: i += 2;
432: /* Comment, skip to end of line */
1.71 schwarze 433: for (; i < blk.sz; ++i) {
1.1 schwarze 434: if ('\n' == blk.buf[i]) {
435: ++i;
436: ++lnn;
437: break;
438: }
439: }
440:
441: /* Backout trailing whitespaces */
442: for (; pos > 0; --pos) {
443: if (ln.buf[pos - 1] != ' ')
444: break;
445: if (pos > 2 && ln.buf[pos - 2] == '\\')
446: break;
447: }
448: break;
449: }
450:
1.13 schwarze 451: /* Catch escaped bogus characters. */
452:
453: c = (unsigned char) blk.buf[i+1];
454:
1.25 schwarze 455: if ( ! (isascii(c) &&
456: (isgraph(c) || isblank(c)))) {
1.89 schwarze 457: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.56 schwarze 458: curp->line, pos, "0x%x", c);
1.13 schwarze 459: i += 2;
460: ln.buf[pos++] = '?';
461: continue;
462: }
463:
1.1 schwarze 464: /* Some other escape sequence, copy & cont. */
465:
466: ln.buf[pos++] = blk.buf[i++];
467: ln.buf[pos++] = blk.buf[i++];
468: }
469:
1.71 schwarze 470: if (pos >= ln.sz)
1.1 schwarze 471: resize_buf(&ln, 256);
472:
473: ln.buf[pos] = '\0';
474:
475: /*
476: * A significant amount of complexity is contained by
477: * the roff preprocessor. It's line-oriented but can be
478: * expressed on one line, so we need at times to
479: * readjust our starting point and re-run it. The roff
480: * preprocessor can also readjust the buffers with new
481: * data, so we pass them in wholesale.
482: */
483:
484: of = 0;
485:
1.4 schwarze 486: /*
487: * Maintain a lookaside buffer of all parsed lines. We
488: * only do this if mparse_keep() has been invoked (the
489: * buffer may be accessed with mparse_getkeep()).
490: */
491:
492: if (curp->secondary) {
1.25 schwarze 493: curp->secondary->buf = mandoc_realloc(
494: curp->secondary->buf,
495: curp->secondary->sz + pos + 2);
496: memcpy(curp->secondary->buf +
497: curp->secondary->sz,
498: ln.buf, pos);
1.4 schwarze 499: curp->secondary->sz += pos;
500: curp->secondary->buf
501: [curp->secondary->sz] = '\n';
502: curp->secondary->sz++;
503: curp->secondary->buf
504: [curp->secondary->sz] = '\0';
505: }
1.1 schwarze 506: rerun:
1.72 schwarze 507: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 508:
509: switch (rr) {
1.25 schwarze 510: case ROFF_REPARSE:
1.1 schwarze 511: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.71 schwarze 512: mparse_buf_r(curp, ln, of, 0);
1.1 schwarze 513: else
514: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 515: curp->line, pos, NULL);
1.1 schwarze 516: pos = 0;
517: continue;
1.25 schwarze 518: case ROFF_APPEND:
1.71 schwarze 519: pos = strlen(ln.buf);
1.1 schwarze 520: continue;
1.25 schwarze 521: case ROFF_RERUN:
1.1 schwarze 522: goto rerun;
1.25 schwarze 523: case ROFF_IGN:
1.1 schwarze 524: pos = 0;
525: continue;
1.25 schwarze 526: case ROFF_SO:
1.71 schwarze 527: if ( ! (curp->options & MPARSE_SO) &&
528: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 529: curp->sodest = mandoc_strdup(ln.buf + of);
530: free(ln.buf);
531: return;
532: }
1.4 schwarze 533: /*
534: * We remove `so' clauses from our lookaside
535: * buffer because we're going to descend into
536: * the file recursively.
537: */
1.25 schwarze 538: if (curp->secondary)
1.4 schwarze 539: curp->secondary->sz -= pos + 1;
1.88 schwarze 540: save_file = curp->file;
1.85 schwarze 541: save_child = curp->child;
542: if (mparse_open(curp, &fd, ln.buf + of) ==
1.88 schwarze 543: MANDOCLEVEL_OK) {
1.85 schwarze 544: mparse_readfd(curp, fd, ln.buf + of);
1.88 schwarze 545: curp->file = save_file;
546: } else {
547: curp->file = save_file;
1.30 schwarze 548: mandoc_vmsg(MANDOCERR_SO_FAIL,
549: curp, curp->line, pos,
550: ".so %s", ln.buf + of);
1.85 schwarze 551: ln.sz = mandoc_asprintf(&cp,
552: ".sp\nSee the file %s.\n.sp",
553: ln.buf + of);
554: free(ln.buf);
555: ln.buf = cp;
556: of = 0;
557: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 558: }
1.85 schwarze 559: curp->child = save_child;
1.1 schwarze 560: pos = 0;
561: continue;
562: default:
563: break;
564: }
565:
566: /*
567: * If input parsers have not been allocated, do so now.
1.2 schwarze 568: * We keep these instanced between parsers, but set them
1.1 schwarze 569: * locally per parse routine since we can use different
570: * parsers with each one.
571: */
572:
573: if ( ! (curp->man || curp->mdoc))
1.60 schwarze 574: choose_parser(curp);
1.1 schwarze 575:
1.25 schwarze 576: /*
1.60 schwarze 577: * Lastly, push down into the parsers themselves.
1.1 schwarze 578: * If libroff returns ROFF_TBL, then add it to the
579: * currently open parse. Since we only get here if
580: * there does exist data (see tbl_data.c), we're
581: * guaranteed that something's been allocated.
582: * Do the same for ROFF_EQN.
583: */
584:
1.76 schwarze 585: if (rr == ROFF_TBL) {
586: while ((span = roff_span(curp->roff)) != NULL)
587: if (curp->man == NULL)
588: mdoc_addspan(curp->mdoc, span);
589: else
590: man_addspan(curp->man, span);
591: } else if (rr == ROFF_EQN) {
592: if (curp->man == NULL)
593: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
594: else
595: man_addeqn(curp->man, roff_eqn(curp->roff));
596: } else if ((curp->man == NULL ?
597: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
598: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
599: break;
1.1 schwarze 600:
601: /* Temporary buffers typically are not full. */
602:
603: if (0 == start && '\0' == blk.buf[i])
604: break;
605:
606: /* Start the next input line. */
607:
608: pos = 0;
609: }
610:
611: free(ln.buf);
612: }
613:
614: static int
1.18 schwarze 615: read_whole_file(struct mparse *curp, const char *file, int fd,
616: struct buf *fb, int *with_mmap)
1.1 schwarze 617: {
618: struct stat st;
619: size_t off;
620: ssize_t ssz;
621:
622: if (-1 == fstat(fd, &st)) {
1.82 schwarze 623: perror(file);
624: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 625: }
626:
627: /*
628: * If we're a regular file, try just reading in the whole entry
629: * via mmap(). This is faster than reading it into blocks, and
630: * since each file is only a few bytes to begin with, I'm not
631: * concerned that this is going to tank any machines.
632: */
633:
634: if (S_ISREG(st.st_mode)) {
1.106 schwarze 635: if (st.st_size > 0x7fffffff) {
1.86 schwarze 636: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 schwarze 637: return(0);
638: }
639: *with_mmap = 1;
640: fb->sz = (size_t)st.st_size;
1.15 schwarze 641: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 642: if (fb->buf != MAP_FAILED)
643: return(1);
644: }
645:
646: /*
647: * If this isn't a regular file (like, say, stdin), then we must
648: * go the old way and just read things in bit by bit.
649: */
650:
651: *with_mmap = 0;
652: off = 0;
653: fb->sz = 0;
654: fb->buf = NULL;
655: for (;;) {
656: if (off == fb->sz) {
657: if (fb->sz == (1U << 31)) {
1.86 schwarze 658: mandoc_msg(MANDOCERR_TOOLARGE, curp,
659: 0, 0, NULL);
1.1 schwarze 660: break;
661: }
662: resize_buf(fb, 65536);
663: }
664: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
665: if (ssz == 0) {
666: fb->sz = off;
667: return(1);
668: }
669: if (ssz == -1) {
1.82 schwarze 670: perror(file);
671: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 672: }
673: off += (size_t)ssz;
674: }
675:
676: free(fb->buf);
677: fb->buf = NULL;
678: return(0);
679: }
680:
681: static void
682: mparse_end(struct mparse *curp)
683: {
684:
1.50 schwarze 685: if (curp->mdoc == NULL &&
686: curp->man == NULL &&
687: curp->sodest == NULL) {
688: if (curp->options & MPARSE_MDOC)
689: curp->mdoc = curp->pmdoc;
690: else {
691: if (curp->pman == NULL)
1.81 schwarze 692: curp->pman = man_alloc(
693: curp->roff, curp, curp->defos,
1.50 schwarze 694: curp->options & MPARSE_QUICK ? 1 : 0);
695: curp->man = curp->pman;
696: }
697: }
1.86 schwarze 698: if (curp->mdoc)
699: mdoc_endparse(curp->mdoc);
700: if (curp->man)
701: man_endparse(curp->man);
1.1 schwarze 702: roff_endparse(curp->roff);
703: }
704:
1.15 schwarze 705: static void
706: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 707: {
1.61 schwarze 708: struct buf *svprimary;
1.1 schwarze 709: const char *svfile;
1.71 schwarze 710: size_t offset;
1.14 schwarze 711: static int recursion_depth;
712:
713: if (64 < recursion_depth) {
714: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 715: return;
1.14 schwarze 716: }
1.1 schwarze 717:
1.15 schwarze 718: /* Line number is per-file. */
719: svfile = curp->file;
720: curp->file = file;
1.61 schwarze 721: svprimary = curp->primary;
1.59 schwarze 722: curp->primary = &blk;
1.15 schwarze 723: curp->line = 1;
724: recursion_depth++;
725:
1.70 schwarze 726: /* Skip an UTF-8 byte order mark. */
727: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
728: (unsigned char)blk.buf[0] == 0xef &&
729: (unsigned char)blk.buf[1] == 0xbb &&
730: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 731: offset = 3;
1.70 schwarze 732: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 733: } else
734: offset = 0;
1.70 schwarze 735:
1.71 schwarze 736: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 737:
1.86 schwarze 738: if (--recursion_depth == 0)
1.15 schwarze 739: mparse_end(curp);
740:
1.61 schwarze 741: curp->primary = svprimary;
1.15 schwarze 742: curp->file = svfile;
743: }
744:
1.74 schwarze 745: /*
746: * Read the whole file into memory and call the parsers.
747: * Called recursively when an .so request is encountered.
748: */
1.15 schwarze 749: enum mandoclevel
750: mparse_readfd(struct mparse *curp, int fd, const char *file)
751: {
752: struct buf blk;
753: int with_mmap;
1.70 schwarze 754: int save_filenc;
1.1 schwarze 755:
1.67 schwarze 756: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 757: save_filenc = curp->filenc;
758: curp->filenc = curp->options &
759: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 760: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 761: curp->filenc = save_filenc;
1.67 schwarze 762: if (with_mmap)
763: munmap(blk.buf, blk.sz);
764: else
765: free(blk.buf);
766: }
1.1 schwarze 767:
1.74 schwarze 768: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 schwarze 769: perror(file);
1.67 schwarze 770:
1.74 schwarze 771: mparse_wait(curp);
1.1 schwarze 772: return(curp->file_status);
1.58 schwarze 773: }
774:
775: enum mandoclevel
1.73 schwarze 776: mparse_open(struct mparse *curp, int *fd, const char *file)
1.58 schwarze 777: {
778: int pfd[2];
1.74 schwarze 779: int save_errno;
1.58 schwarze 780: char *cp;
781:
782: curp->file = file;
1.74 schwarze 783:
784: /* Unless zipped, try to just open the file. */
785:
1.58 schwarze 786: if ((cp = strrchr(file, '.')) == NULL ||
787: strcmp(cp + 1, "gz")) {
1.73 schwarze 788: curp->child = 0;
1.74 schwarze 789: if ((*fd = open(file, O_RDONLY)) != -1)
790: return(MANDOCLEVEL_OK);
791:
792: /* Open failed; try to append ".gz". */
793:
794: mandoc_asprintf(&cp, "%s.gz", file);
795: file = cp;
796: } else
797: cp = NULL;
798:
799: /* Before forking, make sure the file can be read. */
800:
801: save_errno = errno;
802: if (access(file, R_OK) == -1) {
803: if (cp != NULL)
804: errno = save_errno;
1.83 schwarze 805: free(cp);
806: *fd = -1;
807: curp->child = 0;
808: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
809: return(MANDOCLEVEL_ERROR);
1.58 schwarze 810: }
811:
1.74 schwarze 812: /* Run gunzip(1). */
813:
1.58 schwarze 814: if (pipe(pfd) == -1) {
1.82 schwarze 815: perror("pipe");
816: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 817: }
818:
1.73 schwarze 819: switch (curp->child = fork()) {
1.58 schwarze 820: case -1:
1.82 schwarze 821: perror("fork");
822: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 823: case 0:
824: close(pfd[0]);
825: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.82 schwarze 826: perror("dup");
827: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 828: }
1.107 schwarze 829: signal(SIGPIPE, SIG_DFL);
1.58 schwarze 830: execlp("gunzip", "gunzip", "-c", file, NULL);
1.82 schwarze 831: perror("exec");
832: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 833: default:
834: close(pfd[1]);
835: *fd = pfd[0];
836: return(MANDOCLEVEL_OK);
837: }
838: }
839:
840: enum mandoclevel
1.73 schwarze 841: mparse_wait(struct mparse *curp)
1.58 schwarze 842: {
843: int status;
844:
1.73 schwarze 845: if (curp->child == 0)
846: return(MANDOCLEVEL_OK);
847:
848: if (waitpid(curp->child, &status, 0) == -1) {
1.82 schwarze 849: perror("wait");
850: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 851: }
1.104 schwarze 852: curp->child = 0;
1.58 schwarze 853: if (WIFSIGNALED(status)) {
1.83 schwarze 854: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
855: "gunzip died from signal %d", WTERMSIG(status));
856: return(MANDOCLEVEL_ERROR);
1.58 schwarze 857: }
858: if (WEXITSTATUS(status)) {
1.83 schwarze 859: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
860: "gunzip failed with code %d", WEXITSTATUS(status));
861: return(MANDOCLEVEL_ERROR);
1.58 schwarze 862: }
863: return(MANDOCLEVEL_OK);
1.1 schwarze 864: }
865:
866: struct mparse *
1.69 schwarze 867: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
868: const struct mchars *mchars, const char *defos)
1.1 schwarze 869: {
870: struct mparse *curp;
871:
872: curp = mandoc_calloc(1, sizeof(struct mparse));
873:
1.22 schwarze 874: curp->options = options;
1.1 schwarze 875: curp->wlevel = wlevel;
876: curp->mmsg = mmsg;
1.7 schwarze 877: curp->defos = defos;
1.1 schwarze 878:
1.69 schwarze 879: curp->mchars = mchars;
880: curp->roff = roff_alloc(curp, curp->mchars, options);
1.50 schwarze 881: if (curp->options & MPARSE_MDOC)
882: curp->pmdoc = mdoc_alloc(
883: curp->roff, curp, curp->defos,
884: curp->options & MPARSE_QUICK ? 1 : 0);
885: if (curp->options & MPARSE_MAN)
1.81 schwarze 886: curp->pman = man_alloc(
887: curp->roff, curp, curp->defos,
1.50 schwarze 888: curp->options & MPARSE_QUICK ? 1 : 0);
889:
1.1 schwarze 890: return(curp);
891: }
892:
893: void
894: mparse_reset(struct mparse *curp)
895: {
896:
897: roff_reset(curp->roff);
898:
899: if (curp->mdoc)
900: mdoc_reset(curp->mdoc);
901: if (curp->man)
902: man_reset(curp->man);
1.4 schwarze 903: if (curp->secondary)
904: curp->secondary->sz = 0;
1.1 schwarze 905:
906: curp->file_status = MANDOCLEVEL_OK;
907: curp->mdoc = NULL;
908: curp->man = NULL;
1.23 schwarze 909:
910: free(curp->sodest);
911: curp->sodest = NULL;
1.1 schwarze 912: }
913:
914: void
915: mparse_free(struct mparse *curp)
916: {
917:
918: if (curp->pmdoc)
919: mdoc_free(curp->pmdoc);
920: if (curp->pman)
921: man_free(curp->pman);
922: if (curp->roff)
923: roff_free(curp->roff);
1.4 schwarze 924: if (curp->secondary)
925: free(curp->secondary->buf);
1.1 schwarze 926:
1.4 schwarze 927: free(curp->secondary);
1.23 schwarze 928: free(curp->sodest);
1.1 schwarze 929: free(curp);
930: }
931:
932: void
1.23 schwarze 933: mparse_result(struct mparse *curp,
934: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 schwarze 935: {
936:
1.23 schwarze 937: if (sodest && NULL != (*sodest = curp->sodest)) {
938: *mdoc = NULL;
939: *man = NULL;
940: return;
941: }
1.1 schwarze 942: if (mdoc)
943: *mdoc = curp->mdoc;
944: if (man)
945: *man = curp->man;
946: }
947:
948: void
949: mandoc_vmsg(enum mandocerr t, struct mparse *m,
950: int ln, int pos, const char *fmt, ...)
951: {
952: char buf[256];
953: va_list ap;
954:
955: va_start(ap, fmt);
1.26 schwarze 956: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 957: va_end(ap);
958:
959: mandoc_msg(t, m, ln, pos, buf);
960: }
961:
962: void
1.25 schwarze 963: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 964: int ln, int col, const char *msg)
965: {
966: enum mandoclevel level;
967:
1.87 schwarze 968: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 969: while (er < mandoclimits[level])
970: level--;
971:
1.83 schwarze 972: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 973: return;
974:
975: if (m->mmsg)
976: (*m->mmsg)(er, level, m->file, ln, col, msg);
977:
978: if (m->file_status < level)
979: m->file_status = level;
980: }
981:
982: const char *
983: mparse_strerror(enum mandocerr er)
984: {
985:
986: return(mandocerrs[er]);
987: }
988:
989: const char *
990: mparse_strlevel(enum mandoclevel lvl)
991: {
992: return(mandoclevels[lvl]);
1.4 schwarze 993: }
994:
995: void
996: mparse_keep(struct mparse *p)
997: {
998:
999: assert(NULL == p->secondary);
1000: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1001: }
1002:
1003: const char *
1004: mparse_getkeep(const struct mparse *p)
1005: {
1006:
1007: assert(p->secondary);
1008: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 schwarze 1009: }
![[BACK]](/icons/back.gif){kind=icon}
Return to read.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc