Annotation of src/usr.bin/mandoc/read.c, Revision 1.107
1.107 ! schwarze 1: /* $OpenBSD: read.c,v 1.106 2015/03/11 13:04:10 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.82 schwarze 4: * Copyright (c) 2010-2015 Ingo Schwarze <schwarze@openbsd.org>
1.20 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 schwarze 6: *
7: * Permission to use, copy, modify, and distribute this software for any
8: * purpose with or without fee is hereby granted, provided that the above
9: * copyright notice and this permission notice appear in all copies.
10: *
11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18: */
1.58 schwarze 19: #include <sys/types.h>
20: #include <sys/mman.h>
1.1 schwarze 21: #include <sys/stat.h>
1.58 schwarze 22: #include <sys/wait.h>
1.1 schwarze 23:
24: #include <assert.h>
25: #include <ctype.h>
1.18 schwarze 26: #include <errno.h>
1.1 schwarze 27: #include <fcntl.h>
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:
35: #include "mandoc.h"
1.24 schwarze 36: #include "mandoc_aux.h"
1.1 schwarze 37: #include "libmandoc.h"
38: #include "mdoc.h"
39: #include "man.h"
40:
41: #define REPARSE_LIMIT 1000
42:
43: struct mparse {
44: struct man *pman; /* persistent man parser */
45: struct mdoc *pmdoc; /* persistent mdoc parser */
46: struct man *man; /* man parser */
47: struct mdoc *mdoc; /* mdoc parser */
48: struct roff *roff; /* roff parser (!NULL) */
1.69 schwarze 49: const struct mchars *mchars; /* character table */
1.23 schwarze 50: char *sodest; /* filename pointed to by .so */
1.59 schwarze 51: const char *file; /* filename of current input file */
52: struct buf *primary; /* buffer currently being parsed */
53: struct buf *secondary; /* preprocessed copy of input */
54: const char *defos; /* default operating system */
55: mandocmsg mmsg; /* warning/error message handler */
56: enum mandoclevel file_status; /* status of current parse */
57: enum mandoclevel wlevel; /* ignore messages below this */
58: int options; /* parser options */
1.70 schwarze 59: int filenc; /* encoding of the current file */
1.1 schwarze 60: int reparse_count; /* finite interp. stack */
1.59 schwarze 61: int line; /* line number in the file */
1.73 schwarze 62: pid_t child; /* the gunzip(1) process */
1.1 schwarze 63: };
64:
1.60 schwarze 65: static void choose_parser(struct mparse *);
1.1 schwarze 66: static void resize_buf(struct buf *, size_t);
1.71 schwarze 67: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.18 schwarze 68: static int read_whole_file(struct mparse *, const char *, int,
69: struct buf *, int *);
1.1 schwarze 70: static void mparse_end(struct mparse *);
1.15 schwarze 71: static void mparse_parse_buffer(struct mparse *, struct buf,
72: const char *);
1.1 schwarze 73:
74: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
75: MANDOCERR_OK,
76: MANDOCERR_WARNING,
77: MANDOCERR_WARNING,
78: MANDOCERR_ERROR,
1.87 schwarze 79: MANDOCERR_UNSUPP,
1.1 schwarze 80: MANDOCERR_MAX,
81: MANDOCERR_MAX
82: };
83:
84: static const char * const mandocerrs[MANDOCERR_MAX] = {
85: "ok",
86:
87: "generic warning",
88:
89: /* related to the prologue */
1.57 schwarze 90: "missing manual title, using UNTITLED",
91: "missing manual title, using \"\"",
1.32 schwarze 92: "lower case character in document title",
1.57 schwarze 93: "missing manual section, using \"\"",
1.1 schwarze 94: "unknown manual section",
1.32 schwarze 95: "missing date, using today's date",
1.1 schwarze 96: "cannot parse date, using it verbatim",
1.57 schwarze 97: "missing Os macro, using \"\"",
98: "duplicate prologue macro",
99: "late prologue macro",
100: "skipping late title macro",
1.1 schwarze 101: "prologue macros out of order",
102:
103: /* related to document structure */
104: ".so is fragile, better use ln(1)",
1.28 schwarze 105: "no document body",
1.32 schwarze 106: "content before first section header",
107: "first section is not \"NAME\"",
1.103 schwarze 108: "NAME section without name",
109: "NAME section without description",
110: "description not at the end of NAME",
111: "bad NAME section content",
1.96 schwarze 112: "missing description line, using \"\"",
1.1 schwarze 113: "sections out of conventional order",
1.32 schwarze 114: "duplicate section title",
115: "unexpected section",
1.63 schwarze 116: "unusual Xr order",
117: "unusual Xr punctuation",
1.62 schwarze 118: "AUTHORS section without An macro",
1.1 schwarze 119:
120: /* related to macros and nesting */
1.33 schwarze 121: "obsolete macro",
1.78 schwarze 122: "macro neither callable nor escaped",
1.1 schwarze 123: "skipping paragraph macro",
1.10 schwarze 124: "moving paragraph macro out of list",
1.1 schwarze 125: "skipping no-space macro",
126: "blocks badly nested",
127: "nested displays are not portable",
1.35 schwarze 128: "moving content out of list",
129: ".Vt block has child macro",
1.56 schwarze 130: "fill mode already enabled, skipping",
131: "fill mode already disabled, skipping",
1.1 schwarze 132: "line scope broken",
133:
134: /* related to missing macro arguments */
1.36 schwarze 135: "skipping empty request",
136: "conditional request controls empty scope",
1.1 schwarze 137: "skipping empty macro",
1.99 schwarze 138: "empty block",
1.40 schwarze 139: "empty argument, using 0n",
1.38 schwarze 140: "missing display type, using -ragged",
141: "list type is not the first argument",
142: "missing -width in -tag list, using 8n",
1.56 schwarze 143: "missing utility name, using \"\"",
1.98 schwarze 144: "missing function name, using \"\"",
1.38 schwarze 145: "empty head in list item",
146: "empty list item",
1.39 schwarze 147: "missing font type, using \\fR",
148: "unknown font type, using \\fR",
1.79 schwarze 149: "nothing follows prefix",
1.97 schwarze 150: "empty reference block",
1.38 schwarze 151: "missing -std argument, adding it",
1.100 schwarze 152: "missing option string, using \"\"",
153: "missing resource identifier, using \"\"",
1.66 schwarze 154: "missing eqn box, using \"\"",
1.1 schwarze 155:
156: /* related to bad macro arguments */
1.42 schwarze 157: "unterminated quoted argument",
1.1 schwarze 158: "duplicate argument",
1.54 schwarze 159: "skipping duplicate argument",
1.41 schwarze 160: "skipping duplicate display type",
161: "skipping duplicate list type",
1.54 schwarze 162: "skipping -width argument",
1.101 schwarze 163: "wrong number of cells",
1.1 schwarze 164: "unknown AT&T UNIX version",
1.64 schwarze 165: "comma in function argument",
1.65 schwarze 166: "parenthesis in function name",
1.45 schwarze 167: "invalid content in Rs block",
1.41 schwarze 168: "invalid Boolean argument",
169: "unknown font, skipping request",
1.101 schwarze 170: "odd number of characters in request",
1.1 schwarze 171:
172: /* related to plain text */
1.42 schwarze 173: "blank line in fill mode, using .sp",
174: "tab in filled text",
175: "whitespace at end of input line",
1.1 schwarze 176: "bad comment style",
1.42 schwarze 177: "invalid escape sequence",
178: "undefined string, using \"\"",
1.3 schwarze 179:
1.93 schwarze 180: /* related to tables */
181: "tbl line starts with span",
182: "tbl column starts with span",
183: "skipping vertical bar in tbl layout",
184:
1.1 schwarze 185: "generic error",
1.3 schwarze 186:
1.1 schwarze 187: /* related to tables */
1.91 schwarze 188: "non-alphabetic character in tbl options",
189: "skipping unknown tbl option",
190: "missing tbl option argument",
191: "wrong tbl option argument size",
1.93 schwarze 192: "empty tbl layout",
193: "invalid character in tbl layout",
194: "unmatched parenthesis in tbl layout",
1.94 schwarze 195: "tbl without any data cells",
196: "ignoring data in spanned tbl cell",
197: "ignoring extra tbl data cells",
198: "data block open at end of tbl",
1.1 schwarze 199:
1.46 schwarze 200: /* related to document structure and macros */
1.83 schwarze 201: NULL,
1.1 schwarze 202: "input stack limit exceeded, infinite loop?",
203: "skipping bad character",
1.46 schwarze 204: "skipping unknown macro",
1.87 schwarze 205: "skipping insecure request",
1.51 schwarze 206: "skipping item outside list",
1.46 schwarze 207: "skipping column outside column list",
208: "skipping end of block that is not open",
1.90 schwarze 209: "fewer RS blocks open, skipping",
1.46 schwarze 210: "inserting missing end of block",
211: "appending missing end of block",
212:
213: /* related to request and macro arguments */
1.1 schwarze 214: "escaped character not allowed in a name",
1.75 schwarze 215: "NOT IMPLEMENTED: Bd -file",
1.49 schwarze 216: "missing list type, using -item",
1.48 schwarze 217: "missing manual name, using \"\"",
1.49 schwarze 218: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 219: "unknown standard specifier",
1.49 schwarze 220: "skipping request without numeric argument",
1.84 schwarze 221: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 222: ".so request failed",
1.39 schwarze 223: "skipping all arguments",
224: "skipping excess arguments",
1.68 schwarze 225: "divide by zero",
1.87 schwarze 226:
227: "unsupported feature",
228: "input too large",
1.89 schwarze 229: "unsupported control character",
1.87 schwarze 230: "unsupported roff request",
1.94 schwarze 231: "eqn delim option in tbl",
1.93 schwarze 232: "unsupported tbl layout modifier",
1.87 schwarze 233: "ignoring macro in table",
1.1 schwarze 234: };
235:
236: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
237: "SUCCESS",
238: "RESERVED",
239: "WARNING",
240: "ERROR",
1.87 schwarze 241: "UNSUPP",
1.1 schwarze 242: "BADARG",
243: "SYSERR"
244: };
245:
1.25 schwarze 246:
1.1 schwarze 247: static void
248: resize_buf(struct buf *buf, size_t initial)
249: {
250:
251: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
252: buf->buf = mandoc_realloc(buf->buf, buf->sz);
253: }
254:
255: static void
1.60 schwarze 256: choose_parser(struct mparse *curp)
1.1 schwarze 257: {
1.59 schwarze 258: char *cp, *ep;
259: int format;
1.1 schwarze 260:
1.59 schwarze 261: /*
262: * If neither command line arguments -mdoc or -man select
263: * a parser nor the roff parser found a .Dd or .TH macro
264: * yet, look ahead in the main input buffer.
265: */
266:
267: if ((format = roff_getformat(curp->roff)) == 0) {
268: cp = curp->primary->buf;
269: ep = cp + curp->primary->sz;
270: while (cp < ep) {
1.61 schwarze 271: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 272: cp++;
273: if (cp[0] == 'D' && cp[1] == 'd') {
274: format = MPARSE_MDOC;
275: break;
276: }
277: if (cp[0] == 'T' && cp[1] == 'H') {
278: format = MPARSE_MAN;
279: break;
280: }
281: }
282: cp = memchr(cp, '\n', ep - cp);
283: if (cp == NULL)
284: break;
285: cp++;
286: }
1.1 schwarze 287: }
288:
1.59 schwarze 289: if (format == MPARSE_MDOC) {
1.25 schwarze 290: if (NULL == curp->pmdoc)
1.22 schwarze 291: curp->pmdoc = mdoc_alloc(
292: curp->roff, curp, curp->defos,
293: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 schwarze 294: assert(curp->pmdoc);
295: curp->mdoc = curp->pmdoc;
296: return;
1.25 schwarze 297: }
1.1 schwarze 298:
1.59 schwarze 299: /* Fall back to man(7) as a last resort. */
300:
1.25 schwarze 301: if (NULL == curp->pman)
1.81 schwarze 302: curp->pman = man_alloc(
303: curp->roff, curp, curp->defos,
1.22 schwarze 304: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 schwarze 305: assert(curp->pman);
306: curp->man = curp->pman;
307: }
308:
309: /*
1.71 schwarze 310: * Main parse routine for a buffer.
311: * It assumes encoding and line numbering are already set up.
312: * It can recurse directly (for invocations of user-defined
313: * macros, inline equations, and input line traps)
314: * and indirectly (for .so file inclusion).
1.1 schwarze 315: */
316: static void
1.71 schwarze 317: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 318: {
319: const struct tbl_span *span;
320: struct buf ln;
1.88 schwarze 321: const char *save_file;
1.85 schwarze 322: char *cp;
1.71 schwarze 323: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 324: enum rofferr rr;
1.76 schwarze 325: int of;
1.1 schwarze 326: int lnn; /* line number in the real file */
1.85 schwarze 327: int fd;
328: pid_t save_child;
1.1 schwarze 329: unsigned char c;
330:
1.71 schwarze 331: memset(&ln, 0, sizeof(ln));
1.1 schwarze 332:
1.25 schwarze 333: lnn = curp->line;
334: pos = 0;
1.1 schwarze 335:
1.71 schwarze 336: while (i < blk.sz) {
1.1 schwarze 337: if (0 == pos && '\0' == blk.buf[i])
338: break;
339:
340: if (start) {
341: curp->line = lnn;
342: curp->reparse_count = 0;
1.70 schwarze 343:
344: if (lnn < 3 &&
345: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 346: curp->filenc & MPARSE_LATIN1)
347: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 348: }
349:
1.71 schwarze 350: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 351:
352: /*
353: * When finding an unescaped newline character,
354: * leave the character loop to process the line.
355: * Skip a preceding carriage return, if any.
356: */
357:
1.71 schwarze 358: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 359: '\n' == blk.buf[i + 1])
360: ++i;
361: if ('\n' == blk.buf[i]) {
362: ++i;
363: ++lnn;
364: break;
365: }
366:
1.13 schwarze 367: /*
1.70 schwarze 368: * Make sure we have space for the worst
369: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 370: */
371:
1.71 schwarze 372: if (pos + 11 > ln.sz)
1.13 schwarze 373: resize_buf(&ln, 256);
374:
1.25 schwarze 375: /*
1.70 schwarze 376: * Encode 8-bit input.
1.1 schwarze 377: */
378:
1.70 schwarze 379: c = blk.buf[i];
380: if (c & 0x80) {
1.71 schwarze 381: if ( ! (curp->filenc && preconv_encode(
382: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 383: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
384: curp->line, pos, "0x%x", c);
1.70 schwarze 385: ln.buf[pos++] = '?';
386: i++;
387: }
388: continue;
389: }
390:
391: /*
392: * Exclude control characters.
393: */
1.1 schwarze 394:
1.70 schwarze 395: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 396: mandoc_vmsg(c == 0x00 || c == 0x04 ||
397: c > 0x0a ? MANDOCERR_CHAR_BAD :
398: MANDOCERR_CHAR_UNSUPP,
399: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 400: i++;
1.102 schwarze 401: if (c != '\r')
402: ln.buf[pos++] = '?';
1.1 schwarze 403: continue;
404: }
405:
406: /* Trailing backslash = a plain char. */
407:
1.71 schwarze 408: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 schwarze 409: ln.buf[pos++] = blk.buf[i++];
410: continue;
411: }
412:
413: /*
414: * Found escape and at least one other character.
415: * When it's a newline character, skip it.
416: * When there is a carriage return in between,
417: * skip that one as well.
418: */
419:
1.71 schwarze 420: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 schwarze 421: '\n' == blk.buf[i + 2])
422: ++i;
423: if ('\n' == blk.buf[i + 1]) {
424: i += 2;
425: ++lnn;
426: continue;
427: }
428:
429: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
430: i += 2;
431: /* Comment, skip to end of line */
1.71 schwarze 432: for (; i < blk.sz; ++i) {
1.1 schwarze 433: if ('\n' == blk.buf[i]) {
434: ++i;
435: ++lnn;
436: break;
437: }
438: }
439:
440: /* Backout trailing whitespaces */
441: for (; pos > 0; --pos) {
442: if (ln.buf[pos - 1] != ' ')
443: break;
444: if (pos > 2 && ln.buf[pos - 2] == '\\')
445: break;
446: }
447: break;
448: }
449:
1.13 schwarze 450: /* Catch escaped bogus characters. */
451:
452: c = (unsigned char) blk.buf[i+1];
453:
1.25 schwarze 454: if ( ! (isascii(c) &&
455: (isgraph(c) || isblank(c)))) {
1.89 schwarze 456: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.56 schwarze 457: curp->line, pos, "0x%x", c);
1.13 schwarze 458: i += 2;
459: ln.buf[pos++] = '?';
460: continue;
461: }
462:
1.1 schwarze 463: /* Some other escape sequence, copy & cont. */
464:
465: ln.buf[pos++] = blk.buf[i++];
466: ln.buf[pos++] = blk.buf[i++];
467: }
468:
1.71 schwarze 469: if (pos >= ln.sz)
1.1 schwarze 470: resize_buf(&ln, 256);
471:
472: ln.buf[pos] = '\0';
473:
474: /*
475: * A significant amount of complexity is contained by
476: * the roff preprocessor. It's line-oriented but can be
477: * expressed on one line, so we need at times to
478: * readjust our starting point and re-run it. The roff
479: * preprocessor can also readjust the buffers with new
480: * data, so we pass them in wholesale.
481: */
482:
483: of = 0;
484:
1.4 schwarze 485: /*
486: * Maintain a lookaside buffer of all parsed lines. We
487: * only do this if mparse_keep() has been invoked (the
488: * buffer may be accessed with mparse_getkeep()).
489: */
490:
491: if (curp->secondary) {
1.25 schwarze 492: curp->secondary->buf = mandoc_realloc(
493: curp->secondary->buf,
494: curp->secondary->sz + pos + 2);
495: memcpy(curp->secondary->buf +
496: curp->secondary->sz,
497: ln.buf, pos);
1.4 schwarze 498: curp->secondary->sz += pos;
499: curp->secondary->buf
500: [curp->secondary->sz] = '\n';
501: curp->secondary->sz++;
502: curp->secondary->buf
503: [curp->secondary->sz] = '\0';
504: }
1.1 schwarze 505: rerun:
1.72 schwarze 506: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 507:
508: switch (rr) {
1.25 schwarze 509: case ROFF_REPARSE:
1.1 schwarze 510: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.71 schwarze 511: mparse_buf_r(curp, ln, of, 0);
1.1 schwarze 512: else
513: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 514: curp->line, pos, NULL);
1.1 schwarze 515: pos = 0;
516: continue;
1.25 schwarze 517: case ROFF_APPEND:
1.71 schwarze 518: pos = strlen(ln.buf);
1.1 schwarze 519: continue;
1.25 schwarze 520: case ROFF_RERUN:
1.1 schwarze 521: goto rerun;
1.25 schwarze 522: case ROFF_IGN:
1.1 schwarze 523: pos = 0;
524: continue;
1.25 schwarze 525: case ROFF_SO:
1.71 schwarze 526: if ( ! (curp->options & MPARSE_SO) &&
527: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 528: curp->sodest = mandoc_strdup(ln.buf + of);
529: free(ln.buf);
530: return;
531: }
1.4 schwarze 532: /*
533: * We remove `so' clauses from our lookaside
534: * buffer because we're going to descend into
535: * the file recursively.
536: */
1.25 schwarze 537: if (curp->secondary)
1.4 schwarze 538: curp->secondary->sz -= pos + 1;
1.88 schwarze 539: save_file = curp->file;
1.85 schwarze 540: save_child = curp->child;
541: if (mparse_open(curp, &fd, ln.buf + of) ==
1.88 schwarze 542: MANDOCLEVEL_OK) {
1.85 schwarze 543: mparse_readfd(curp, fd, ln.buf + of);
1.88 schwarze 544: curp->file = save_file;
545: } else {
546: curp->file = save_file;
1.30 schwarze 547: mandoc_vmsg(MANDOCERR_SO_FAIL,
548: curp, curp->line, pos,
549: ".so %s", ln.buf + of);
1.85 schwarze 550: ln.sz = mandoc_asprintf(&cp,
551: ".sp\nSee the file %s.\n.sp",
552: ln.buf + of);
553: free(ln.buf);
554: ln.buf = cp;
555: of = 0;
556: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 557: }
1.85 schwarze 558: curp->child = save_child;
1.1 schwarze 559: pos = 0;
560: continue;
561: default:
562: break;
563: }
564:
565: /*
566: * If input parsers have not been allocated, do so now.
1.2 schwarze 567: * We keep these instanced between parsers, but set them
1.1 schwarze 568: * locally per parse routine since we can use different
569: * parsers with each one.
570: */
571:
572: if ( ! (curp->man || curp->mdoc))
1.60 schwarze 573: choose_parser(curp);
1.1 schwarze 574:
1.25 schwarze 575: /*
1.60 schwarze 576: * Lastly, push down into the parsers themselves.
1.1 schwarze 577: * If libroff returns ROFF_TBL, then add it to the
578: * currently open parse. Since we only get here if
579: * there does exist data (see tbl_data.c), we're
580: * guaranteed that something's been allocated.
581: * Do the same for ROFF_EQN.
582: */
583:
1.76 schwarze 584: if (rr == ROFF_TBL) {
585: while ((span = roff_span(curp->roff)) != NULL)
586: if (curp->man == NULL)
587: mdoc_addspan(curp->mdoc, span);
588: else
589: man_addspan(curp->man, span);
590: } else if (rr == ROFF_EQN) {
591: if (curp->man == NULL)
592: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
593: else
594: man_addeqn(curp->man, roff_eqn(curp->roff));
595: } else if ((curp->man == NULL ?
596: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
597: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
598: break;
1.1 schwarze 599:
600: /* Temporary buffers typically are not full. */
601:
602: if (0 == start && '\0' == blk.buf[i])
603: break;
604:
605: /* Start the next input line. */
606:
607: pos = 0;
608: }
609:
610: free(ln.buf);
611: }
612:
613: static int
1.18 schwarze 614: read_whole_file(struct mparse *curp, const char *file, int fd,
615: struct buf *fb, int *with_mmap)
1.1 schwarze 616: {
617: struct stat st;
618: size_t off;
619: ssize_t ssz;
620:
621: if (-1 == fstat(fd, &st)) {
1.82 schwarze 622: perror(file);
623: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 624: }
625:
626: /*
627: * If we're a regular file, try just reading in the whole entry
628: * via mmap(). This is faster than reading it into blocks, and
629: * since each file is only a few bytes to begin with, I'm not
630: * concerned that this is going to tank any machines.
631: */
632:
633: if (S_ISREG(st.st_mode)) {
1.106 schwarze 634: if (st.st_size > 0x7fffffff) {
1.86 schwarze 635: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 schwarze 636: return(0);
637: }
638: *with_mmap = 1;
639: fb->sz = (size_t)st.st_size;
1.15 schwarze 640: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 641: if (fb->buf != MAP_FAILED)
642: return(1);
643: }
644:
645: /*
646: * If this isn't a regular file (like, say, stdin), then we must
647: * go the old way and just read things in bit by bit.
648: */
649:
650: *with_mmap = 0;
651: off = 0;
652: fb->sz = 0;
653: fb->buf = NULL;
654: for (;;) {
655: if (off == fb->sz) {
656: if (fb->sz == (1U << 31)) {
1.86 schwarze 657: mandoc_msg(MANDOCERR_TOOLARGE, curp,
658: 0, 0, NULL);
1.1 schwarze 659: break;
660: }
661: resize_buf(fb, 65536);
662: }
663: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
664: if (ssz == 0) {
665: fb->sz = off;
666: return(1);
667: }
668: if (ssz == -1) {
1.82 schwarze 669: perror(file);
670: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 671: }
672: off += (size_t)ssz;
673: }
674:
675: free(fb->buf);
676: fb->buf = NULL;
677: return(0);
678: }
679:
680: static void
681: mparse_end(struct mparse *curp)
682: {
683:
1.50 schwarze 684: if (curp->mdoc == NULL &&
685: curp->man == NULL &&
686: curp->sodest == NULL) {
687: if (curp->options & MPARSE_MDOC)
688: curp->mdoc = curp->pmdoc;
689: else {
690: if (curp->pman == NULL)
1.81 schwarze 691: curp->pman = man_alloc(
692: curp->roff, curp, curp->defos,
1.50 schwarze 693: curp->options & MPARSE_QUICK ? 1 : 0);
694: curp->man = curp->pman;
695: }
696: }
1.86 schwarze 697: if (curp->mdoc)
698: mdoc_endparse(curp->mdoc);
699: if (curp->man)
700: man_endparse(curp->man);
1.1 schwarze 701: roff_endparse(curp->roff);
702: }
703:
1.15 schwarze 704: static void
705: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 706: {
1.61 schwarze 707: struct buf *svprimary;
1.1 schwarze 708: const char *svfile;
1.71 schwarze 709: size_t offset;
1.14 schwarze 710: static int recursion_depth;
711:
712: if (64 < recursion_depth) {
713: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 714: return;
1.14 schwarze 715: }
1.1 schwarze 716:
1.15 schwarze 717: /* Line number is per-file. */
718: svfile = curp->file;
719: curp->file = file;
1.61 schwarze 720: svprimary = curp->primary;
1.59 schwarze 721: curp->primary = &blk;
1.15 schwarze 722: curp->line = 1;
723: recursion_depth++;
724:
1.70 schwarze 725: /* Skip an UTF-8 byte order mark. */
726: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
727: (unsigned char)blk.buf[0] == 0xef &&
728: (unsigned char)blk.buf[1] == 0xbb &&
729: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 730: offset = 3;
1.70 schwarze 731: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 732: } else
733: offset = 0;
1.70 schwarze 734:
1.71 schwarze 735: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 736:
1.86 schwarze 737: if (--recursion_depth == 0)
1.15 schwarze 738: mparse_end(curp);
739:
1.61 schwarze 740: curp->primary = svprimary;
1.15 schwarze 741: curp->file = svfile;
742: }
743:
1.74 schwarze 744: /*
745: * Read the whole file into memory and call the parsers.
746: * Called recursively when an .so request is encountered.
747: */
1.15 schwarze 748: enum mandoclevel
749: mparse_readfd(struct mparse *curp, int fd, const char *file)
750: {
751: struct buf blk;
752: int with_mmap;
1.70 schwarze 753: int save_filenc;
1.1 schwarze 754:
1.67 schwarze 755: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 756: save_filenc = curp->filenc;
757: curp->filenc = curp->options &
758: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 759: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 760: curp->filenc = save_filenc;
1.67 schwarze 761: if (with_mmap)
762: munmap(blk.buf, blk.sz);
763: else
764: free(blk.buf);
765: }
1.1 schwarze 766:
1.74 schwarze 767: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 schwarze 768: perror(file);
1.67 schwarze 769:
1.74 schwarze 770: mparse_wait(curp);
1.1 schwarze 771: return(curp->file_status);
1.58 schwarze 772: }
773:
774: enum mandoclevel
1.73 schwarze 775: mparse_open(struct mparse *curp, int *fd, const char *file)
1.58 schwarze 776: {
777: int pfd[2];
1.74 schwarze 778: int save_errno;
1.58 schwarze 779: char *cp;
780:
781: curp->file = file;
1.74 schwarze 782:
783: /* Unless zipped, try to just open the file. */
784:
1.58 schwarze 785: if ((cp = strrchr(file, '.')) == NULL ||
786: strcmp(cp + 1, "gz")) {
1.73 schwarze 787: curp->child = 0;
1.74 schwarze 788: if ((*fd = open(file, O_RDONLY)) != -1)
789: return(MANDOCLEVEL_OK);
790:
791: /* Open failed; try to append ".gz". */
792:
793: mandoc_asprintf(&cp, "%s.gz", file);
794: file = cp;
795: } else
796: cp = NULL;
797:
798: /* Before forking, make sure the file can be read. */
799:
800: save_errno = errno;
801: if (access(file, R_OK) == -1) {
802: if (cp != NULL)
803: errno = save_errno;
1.83 schwarze 804: free(cp);
805: *fd = -1;
806: curp->child = 0;
807: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
808: return(MANDOCLEVEL_ERROR);
1.58 schwarze 809: }
810:
1.74 schwarze 811: /* Run gunzip(1). */
812:
1.58 schwarze 813: if (pipe(pfd) == -1) {
1.82 schwarze 814: perror("pipe");
815: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 816: }
817:
1.73 schwarze 818: switch (curp->child = fork()) {
1.58 schwarze 819: case -1:
1.82 schwarze 820: perror("fork");
821: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 822: case 0:
823: close(pfd[0]);
824: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.82 schwarze 825: perror("dup");
826: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 827: }
1.107 ! schwarze 828: signal(SIGPIPE, SIG_DFL);
1.58 schwarze 829: execlp("gunzip", "gunzip", "-c", file, NULL);
1.82 schwarze 830: perror("exec");
831: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 832: default:
833: close(pfd[1]);
834: *fd = pfd[0];
835: return(MANDOCLEVEL_OK);
836: }
837: }
838:
839: enum mandoclevel
1.73 schwarze 840: mparse_wait(struct mparse *curp)
1.58 schwarze 841: {
842: int status;
843:
1.73 schwarze 844: if (curp->child == 0)
845: return(MANDOCLEVEL_OK);
846:
847: if (waitpid(curp->child, &status, 0) == -1) {
1.82 schwarze 848: perror("wait");
849: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 850: }
1.104 schwarze 851: curp->child = 0;
1.58 schwarze 852: if (WIFSIGNALED(status)) {
1.83 schwarze 853: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
854: "gunzip died from signal %d", WTERMSIG(status));
855: return(MANDOCLEVEL_ERROR);
1.58 schwarze 856: }
857: if (WEXITSTATUS(status)) {
1.83 schwarze 858: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
859: "gunzip failed with code %d", WEXITSTATUS(status));
860: return(MANDOCLEVEL_ERROR);
1.58 schwarze 861: }
862: return(MANDOCLEVEL_OK);
1.1 schwarze 863: }
864:
865: struct mparse *
1.69 schwarze 866: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
867: const struct mchars *mchars, const char *defos)
1.1 schwarze 868: {
869: struct mparse *curp;
870:
871: curp = mandoc_calloc(1, sizeof(struct mparse));
872:
1.22 schwarze 873: curp->options = options;
1.1 schwarze 874: curp->wlevel = wlevel;
875: curp->mmsg = mmsg;
1.7 schwarze 876: curp->defos = defos;
1.1 schwarze 877:
1.69 schwarze 878: curp->mchars = mchars;
879: curp->roff = roff_alloc(curp, curp->mchars, options);
1.50 schwarze 880: if (curp->options & MPARSE_MDOC)
881: curp->pmdoc = mdoc_alloc(
882: curp->roff, curp, curp->defos,
883: curp->options & MPARSE_QUICK ? 1 : 0);
884: if (curp->options & MPARSE_MAN)
1.81 schwarze 885: curp->pman = man_alloc(
886: curp->roff, curp, curp->defos,
1.50 schwarze 887: curp->options & MPARSE_QUICK ? 1 : 0);
888:
1.1 schwarze 889: return(curp);
890: }
891:
892: void
893: mparse_reset(struct mparse *curp)
894: {
895:
896: roff_reset(curp->roff);
897:
898: if (curp->mdoc)
899: mdoc_reset(curp->mdoc);
900: if (curp->man)
901: man_reset(curp->man);
1.4 schwarze 902: if (curp->secondary)
903: curp->secondary->sz = 0;
1.1 schwarze 904:
905: curp->file_status = MANDOCLEVEL_OK;
906: curp->mdoc = NULL;
907: curp->man = NULL;
1.23 schwarze 908:
909: free(curp->sodest);
910: curp->sodest = NULL;
1.1 schwarze 911: }
912:
913: void
914: mparse_free(struct mparse *curp)
915: {
916:
917: if (curp->pmdoc)
918: mdoc_free(curp->pmdoc);
919: if (curp->pman)
920: man_free(curp->pman);
921: if (curp->roff)
922: roff_free(curp->roff);
1.4 schwarze 923: if (curp->secondary)
924: free(curp->secondary->buf);
1.1 schwarze 925:
1.4 schwarze 926: free(curp->secondary);
1.23 schwarze 927: free(curp->sodest);
1.1 schwarze 928: free(curp);
929: }
930:
931: void
1.23 schwarze 932: mparse_result(struct mparse *curp,
933: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 schwarze 934: {
935:
1.23 schwarze 936: if (sodest && NULL != (*sodest = curp->sodest)) {
937: *mdoc = NULL;
938: *man = NULL;
939: return;
940: }
1.1 schwarze 941: if (mdoc)
942: *mdoc = curp->mdoc;
943: if (man)
944: *man = curp->man;
945: }
946:
947: void
948: mandoc_vmsg(enum mandocerr t, struct mparse *m,
949: int ln, int pos, const char *fmt, ...)
950: {
951: char buf[256];
952: va_list ap;
953:
954: va_start(ap, fmt);
1.26 schwarze 955: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 956: va_end(ap);
957:
958: mandoc_msg(t, m, ln, pos, buf);
959: }
960:
961: void
1.25 schwarze 962: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 963: int ln, int col, const char *msg)
964: {
965: enum mandoclevel level;
966:
1.87 schwarze 967: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 968: while (er < mandoclimits[level])
969: level--;
970:
1.83 schwarze 971: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 972: return;
973:
974: if (m->mmsg)
975: (*m->mmsg)(er, level, m->file, ln, col, msg);
976:
977: if (m->file_status < level)
978: m->file_status = level;
979: }
980:
981: const char *
982: mparse_strerror(enum mandocerr er)
983: {
984:
985: return(mandocerrs[er]);
986: }
987:
988: const char *
989: mparse_strlevel(enum mandoclevel lvl)
990: {
991: return(mandoclevels[lvl]);
1.4 schwarze 992: }
993:
994: void
995: mparse_keep(struct mparse *p)
996: {
997:
998: assert(NULL == p->secondary);
999: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1000: }
1001:
1002: const char *
1003: mparse_getkeep(const struct mparse *p)
1004: {
1005:
1006: assert(p->secondary);
1007: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 schwarze 1008: }
![[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