Annotation of src/usr.bin/mandoc/read.c, Revision 1.86
1.86 ! schwarze 1: /* $OpenBSD: read.c,v 1.85 2015/01/15 02:29:07 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>
28: #include <stdarg.h>
29: #include <stdio.h>
30: #include <stdlib.h>
31: #include <string.h>
32: #include <unistd.h>
33:
34: #include "mandoc.h"
1.24 schwarze 35: #include "mandoc_aux.h"
1.1 schwarze 36: #include "libmandoc.h"
37: #include "mdoc.h"
38: #include "man.h"
39:
40: #define REPARSE_LIMIT 1000
41:
42: struct mparse {
43: struct man *pman; /* persistent man parser */
44: struct mdoc *pmdoc; /* persistent mdoc parser */
45: struct man *man; /* man parser */
46: struct mdoc *mdoc; /* mdoc parser */
47: struct roff *roff; /* roff parser (!NULL) */
1.69 schwarze 48: const struct mchars *mchars; /* character table */
1.23 schwarze 49: char *sodest; /* filename pointed to by .so */
1.59 schwarze 50: const char *file; /* filename of current input file */
51: struct buf *primary; /* buffer currently being parsed */
52: struct buf *secondary; /* preprocessed copy of input */
53: const char *defos; /* default operating system */
54: mandocmsg mmsg; /* warning/error message handler */
55: enum mandoclevel file_status; /* status of current parse */
56: enum mandoclevel wlevel; /* ignore messages below this */
57: int options; /* parser options */
1.70 schwarze 58: int filenc; /* encoding of the current file */
1.1 schwarze 59: int reparse_count; /* finite interp. stack */
1.59 schwarze 60: int line; /* line number in the file */
1.73 schwarze 61: pid_t child; /* the gunzip(1) process */
1.1 schwarze 62: };
63:
1.60 schwarze 64: static void choose_parser(struct mparse *);
1.1 schwarze 65: static void resize_buf(struct buf *, size_t);
1.71 schwarze 66: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.18 schwarze 67: static int read_whole_file(struct mparse *, const char *, int,
68: struct buf *, int *);
1.1 schwarze 69: static void mparse_end(struct mparse *);
1.15 schwarze 70: static void mparse_parse_buffer(struct mparse *, struct buf,
71: const char *);
1.1 schwarze 72:
73: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
74: MANDOCERR_OK,
75: MANDOCERR_WARNING,
76: MANDOCERR_WARNING,
77: MANDOCERR_ERROR,
1.86 ! schwarze 78: MANDOCERR_MAX,
1.1 schwarze 79: MANDOCERR_MAX,
80: MANDOCERR_MAX
81: };
82:
83: static const char * const mandocerrs[MANDOCERR_MAX] = {
84: "ok",
85:
86: "generic warning",
87:
88: /* related to the prologue */
1.57 schwarze 89: "missing manual title, using UNTITLED",
90: "missing manual title, using \"\"",
1.32 schwarze 91: "lower case character in document title",
1.57 schwarze 92: "missing manual section, using \"\"",
1.1 schwarze 93: "unknown manual section",
1.32 schwarze 94: "missing date, using today's date",
1.1 schwarze 95: "cannot parse date, using it verbatim",
1.57 schwarze 96: "missing Os macro, using \"\"",
97: "duplicate prologue macro",
98: "late prologue macro",
99: "skipping late title macro",
1.1 schwarze 100: "prologue macros out of order",
101:
102: /* related to document structure */
103: ".so is fragile, better use ln(1)",
1.28 schwarze 104: "no document body",
1.32 schwarze 105: "content before first section header",
106: "first section is not \"NAME\"",
1.1 schwarze 107: "bad NAME section contents",
108: "sections out of conventional order",
1.32 schwarze 109: "duplicate section title",
110: "unexpected section",
1.63 schwarze 111: "unusual Xr order",
112: "unusual Xr punctuation",
1.62 schwarze 113: "AUTHORS section without An macro",
1.1 schwarze 114:
115: /* related to macros and nesting */
1.33 schwarze 116: "obsolete macro",
1.78 schwarze 117: "macro neither callable nor escaped",
1.1 schwarze 118: "skipping paragraph macro",
1.10 schwarze 119: "moving paragraph macro out of list",
1.1 schwarze 120: "skipping no-space macro",
121: "blocks badly nested",
122: "nested displays are not portable",
1.35 schwarze 123: "moving content out of list",
124: ".Vt block has child macro",
1.56 schwarze 125: "fill mode already enabled, skipping",
126: "fill mode already disabled, skipping",
1.1 schwarze 127: "line scope broken",
128:
129: /* related to missing macro arguments */
1.36 schwarze 130: "skipping empty request",
131: "conditional request controls empty scope",
1.1 schwarze 132: "skipping empty macro",
1.40 schwarze 133: "empty argument, using 0n",
1.1 schwarze 134: "argument count wrong",
1.38 schwarze 135: "missing display type, using -ragged",
136: "list type is not the first argument",
137: "missing -width in -tag list, using 8n",
1.56 schwarze 138: "missing utility name, using \"\"",
1.38 schwarze 139: "empty head in list item",
140: "empty list item",
1.39 schwarze 141: "missing font type, using \\fR",
142: "unknown font type, using \\fR",
1.79 schwarze 143: "nothing follows prefix",
1.38 schwarze 144: "missing -std argument, adding it",
1.66 schwarze 145: "missing eqn box, using \"\"",
1.1 schwarze 146:
147: /* related to bad macro arguments */
1.42 schwarze 148: "unterminated quoted argument",
1.1 schwarze 149: "duplicate argument",
1.54 schwarze 150: "skipping duplicate argument",
1.41 schwarze 151: "skipping duplicate display type",
152: "skipping duplicate list type",
1.54 schwarze 153: "skipping -width argument",
1.1 schwarze 154: "unknown AT&T UNIX version",
1.64 schwarze 155: "comma in function argument",
1.65 schwarze 156: "parenthesis in function name",
1.45 schwarze 157: "invalid content in Rs block",
1.41 schwarze 158: "invalid Boolean argument",
159: "unknown font, skipping request",
1.1 schwarze 160:
161: /* related to plain text */
1.42 schwarze 162: "blank line in fill mode, using .sp",
163: "tab in filled text",
164: "whitespace at end of input line",
1.1 schwarze 165: "bad comment style",
1.42 schwarze 166: "invalid escape sequence",
167: "undefined string, using \"\"",
1.3 schwarze 168:
1.1 schwarze 169: "generic error",
170:
1.3 schwarze 171: /* related to equations */
172: "unexpected equation scope closure",
173: "equation scope open on exit",
174: "overlapping equation scopes",
175: "unexpected end of equation",
176:
1.1 schwarze 177: /* related to tables */
178: "bad table syntax",
179: "bad table option",
180: "bad table layout",
181: "no table layout cells specified",
182: "no table data cells specified",
183: "ignore data in cell",
184: "data block still open",
185: "ignoring extra data cells",
1.80 schwarze 186: "ignoring macro in table",
1.1 schwarze 187:
1.46 schwarze 188: /* related to document structure and macros */
1.83 schwarze 189: NULL,
1.86 ! schwarze 190: "input too large",
1.1 schwarze 191: "input stack limit exceeded, infinite loop?",
192: "skipping bad character",
1.46 schwarze 193: "skipping unknown macro",
1.51 schwarze 194: "skipping item outside list",
1.46 schwarze 195: "skipping column outside column list",
196: "skipping end of block that is not open",
197: "inserting missing end of block",
198: "appending missing end of block",
199:
200: /* related to request and macro arguments */
1.1 schwarze 201: "escaped character not allowed in a name",
202: "argument count wrong",
1.75 schwarze 203: "NOT IMPLEMENTED: Bd -file",
1.49 schwarze 204: "missing list type, using -item",
1.48 schwarze 205: "missing manual name, using \"\"",
1.49 schwarze 206: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 207: "unknown standard specifier",
1.49 schwarze 208: "skipping request without numeric argument",
1.84 schwarze 209: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 210: ".so request failed",
1.39 schwarze 211: "skipping all arguments",
212: "skipping excess arguments",
1.68 schwarze 213: "divide by zero",
1.1 schwarze 214: };
215:
216: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
217: "SUCCESS",
218: "RESERVED",
219: "WARNING",
220: "ERROR",
221: "FATAL",
222: "BADARG",
223: "SYSERR"
224: };
225:
1.25 schwarze 226:
1.1 schwarze 227: static void
228: resize_buf(struct buf *buf, size_t initial)
229: {
230:
231: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
232: buf->buf = mandoc_realloc(buf->buf, buf->sz);
233: }
234:
235: static void
1.60 schwarze 236: choose_parser(struct mparse *curp)
1.1 schwarze 237: {
1.59 schwarze 238: char *cp, *ep;
239: int format;
1.1 schwarze 240:
1.59 schwarze 241: /*
242: * If neither command line arguments -mdoc or -man select
243: * a parser nor the roff parser found a .Dd or .TH macro
244: * yet, look ahead in the main input buffer.
245: */
246:
247: if ((format = roff_getformat(curp->roff)) == 0) {
248: cp = curp->primary->buf;
249: ep = cp + curp->primary->sz;
250: while (cp < ep) {
1.61 schwarze 251: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 252: cp++;
253: if (cp[0] == 'D' && cp[1] == 'd') {
254: format = MPARSE_MDOC;
255: break;
256: }
257: if (cp[0] == 'T' && cp[1] == 'H') {
258: format = MPARSE_MAN;
259: break;
260: }
261: }
262: cp = memchr(cp, '\n', ep - cp);
263: if (cp == NULL)
264: break;
265: cp++;
266: }
1.1 schwarze 267: }
268:
1.59 schwarze 269: if (format == MPARSE_MDOC) {
1.25 schwarze 270: if (NULL == curp->pmdoc)
1.22 schwarze 271: curp->pmdoc = mdoc_alloc(
272: curp->roff, curp, curp->defos,
273: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 schwarze 274: assert(curp->pmdoc);
275: curp->mdoc = curp->pmdoc;
276: return;
1.25 schwarze 277: }
1.1 schwarze 278:
1.59 schwarze 279: /* Fall back to man(7) as a last resort. */
280:
1.25 schwarze 281: if (NULL == curp->pman)
1.81 schwarze 282: curp->pman = man_alloc(
283: curp->roff, curp, curp->defos,
1.22 schwarze 284: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 schwarze 285: assert(curp->pman);
286: curp->man = curp->pman;
287: }
288:
289: /*
1.71 schwarze 290: * Main parse routine for a buffer.
291: * It assumes encoding and line numbering are already set up.
292: * It can recurse directly (for invocations of user-defined
293: * macros, inline equations, and input line traps)
294: * and indirectly (for .so file inclusion).
1.1 schwarze 295: */
296: static void
1.71 schwarze 297: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 298: {
299: const struct tbl_span *span;
300: struct buf ln;
1.85 schwarze 301: char *cp;
1.71 schwarze 302: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 303: enum rofferr rr;
1.76 schwarze 304: int of;
1.1 schwarze 305: int lnn; /* line number in the real file */
1.85 schwarze 306: int fd;
307: pid_t save_child;
1.1 schwarze 308: unsigned char c;
309:
1.71 schwarze 310: memset(&ln, 0, sizeof(ln));
1.1 schwarze 311:
1.25 schwarze 312: lnn = curp->line;
313: pos = 0;
1.1 schwarze 314:
1.71 schwarze 315: while (i < blk.sz) {
1.1 schwarze 316: if (0 == pos && '\0' == blk.buf[i])
317: break;
318:
319: if (start) {
320: curp->line = lnn;
321: curp->reparse_count = 0;
1.70 schwarze 322:
323: if (lnn < 3 &&
324: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 325: curp->filenc & MPARSE_LATIN1)
326: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 327: }
328:
1.71 schwarze 329: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 330:
331: /*
332: * When finding an unescaped newline character,
333: * leave the character loop to process the line.
334: * Skip a preceding carriage return, if any.
335: */
336:
1.71 schwarze 337: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 338: '\n' == blk.buf[i + 1])
339: ++i;
340: if ('\n' == blk.buf[i]) {
341: ++i;
342: ++lnn;
343: break;
344: }
345:
1.13 schwarze 346: /*
1.70 schwarze 347: * Make sure we have space for the worst
348: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 349: */
350:
1.71 schwarze 351: if (pos + 11 > ln.sz)
1.13 schwarze 352: resize_buf(&ln, 256);
353:
1.25 schwarze 354: /*
1.70 schwarze 355: * Encode 8-bit input.
1.1 schwarze 356: */
357:
1.70 schwarze 358: c = blk.buf[i];
359: if (c & 0x80) {
1.71 schwarze 360: if ( ! (curp->filenc && preconv_encode(
361: &blk, &i, &ln, &pos, &curp->filenc))) {
1.70 schwarze 362: mandoc_vmsg(MANDOCERR_BADCHAR,
363: curp, curp->line, pos,
364: "0x%x", c);
365: ln.buf[pos++] = '?';
366: i++;
367: }
368: continue;
369: }
370:
371: /*
372: * Exclude control characters.
373: */
1.1 schwarze 374:
1.70 schwarze 375: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.56 schwarze 376: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
377: curp->line, pos, "0x%x", c);
1.1 schwarze 378: i++;
1.6 schwarze 379: ln.buf[pos++] = '?';
1.1 schwarze 380: continue;
381: }
382:
383: /* Trailing backslash = a plain char. */
384:
1.71 schwarze 385: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 schwarze 386: ln.buf[pos++] = blk.buf[i++];
387: continue;
388: }
389:
390: /*
391: * Found escape and at least one other character.
392: * When it's a newline character, skip it.
393: * When there is a carriage return in between,
394: * skip that one as well.
395: */
396:
1.71 schwarze 397: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 schwarze 398: '\n' == blk.buf[i + 2])
399: ++i;
400: if ('\n' == blk.buf[i + 1]) {
401: i += 2;
402: ++lnn;
403: continue;
404: }
405:
406: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
407: i += 2;
408: /* Comment, skip to end of line */
1.71 schwarze 409: for (; i < blk.sz; ++i) {
1.1 schwarze 410: if ('\n' == blk.buf[i]) {
411: ++i;
412: ++lnn;
413: break;
414: }
415: }
416:
417: /* Backout trailing whitespaces */
418: for (; pos > 0; --pos) {
419: if (ln.buf[pos - 1] != ' ')
420: break;
421: if (pos > 2 && ln.buf[pos - 2] == '\\')
422: break;
423: }
424: break;
425: }
426:
1.13 schwarze 427: /* Catch escaped bogus characters. */
428:
429: c = (unsigned char) blk.buf[i+1];
430:
1.25 schwarze 431: if ( ! (isascii(c) &&
432: (isgraph(c) || isblank(c)))) {
1.56 schwarze 433: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
434: curp->line, pos, "0x%x", c);
1.13 schwarze 435: i += 2;
436: ln.buf[pos++] = '?';
437: continue;
438: }
439:
1.1 schwarze 440: /* Some other escape sequence, copy & cont. */
441:
442: ln.buf[pos++] = blk.buf[i++];
443: ln.buf[pos++] = blk.buf[i++];
444: }
445:
1.71 schwarze 446: if (pos >= ln.sz)
1.1 schwarze 447: resize_buf(&ln, 256);
448:
449: ln.buf[pos] = '\0';
450:
451: /*
452: * A significant amount of complexity is contained by
453: * the roff preprocessor. It's line-oriented but can be
454: * expressed on one line, so we need at times to
455: * readjust our starting point and re-run it. The roff
456: * preprocessor can also readjust the buffers with new
457: * data, so we pass them in wholesale.
458: */
459:
460: of = 0;
461:
1.4 schwarze 462: /*
463: * Maintain a lookaside buffer of all parsed lines. We
464: * only do this if mparse_keep() has been invoked (the
465: * buffer may be accessed with mparse_getkeep()).
466: */
467:
468: if (curp->secondary) {
1.25 schwarze 469: curp->secondary->buf = mandoc_realloc(
470: curp->secondary->buf,
471: curp->secondary->sz + pos + 2);
472: memcpy(curp->secondary->buf +
473: curp->secondary->sz,
474: ln.buf, pos);
1.4 schwarze 475: curp->secondary->sz += pos;
476: curp->secondary->buf
477: [curp->secondary->sz] = '\n';
478: curp->secondary->sz++;
479: curp->secondary->buf
480: [curp->secondary->sz] = '\0';
481: }
1.1 schwarze 482: rerun:
1.72 schwarze 483: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 484:
485: switch (rr) {
1.25 schwarze 486: case ROFF_REPARSE:
1.1 schwarze 487: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.71 schwarze 488: mparse_buf_r(curp, ln, of, 0);
1.1 schwarze 489: else
490: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 491: curp->line, pos, NULL);
1.1 schwarze 492: pos = 0;
493: continue;
1.25 schwarze 494: case ROFF_APPEND:
1.71 schwarze 495: pos = strlen(ln.buf);
1.1 schwarze 496: continue;
1.25 schwarze 497: case ROFF_RERUN:
1.1 schwarze 498: goto rerun;
1.25 schwarze 499: case ROFF_IGN:
1.1 schwarze 500: pos = 0;
501: continue;
1.25 schwarze 502: case ROFF_SO:
1.71 schwarze 503: if ( ! (curp->options & MPARSE_SO) &&
504: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 505: curp->sodest = mandoc_strdup(ln.buf + of);
506: free(ln.buf);
507: return;
508: }
1.4 schwarze 509: /*
510: * We remove `so' clauses from our lookaside
511: * buffer because we're going to descend into
512: * the file recursively.
513: */
1.25 schwarze 514: if (curp->secondary)
1.4 schwarze 515: curp->secondary->sz -= pos + 1;
1.85 schwarze 516: save_child = curp->child;
517: if (mparse_open(curp, &fd, ln.buf + of) ==
518: MANDOCLEVEL_OK)
519: mparse_readfd(curp, fd, ln.buf + of);
520: else {
1.30 schwarze 521: mandoc_vmsg(MANDOCERR_SO_FAIL,
522: curp, curp->line, pos,
523: ".so %s", ln.buf + of);
1.85 schwarze 524: ln.sz = mandoc_asprintf(&cp,
525: ".sp\nSee the file %s.\n.sp",
526: ln.buf + of);
527: free(ln.buf);
528: ln.buf = cp;
529: of = 0;
530: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 531: }
1.85 schwarze 532: curp->child = save_child;
1.1 schwarze 533: pos = 0;
534: continue;
535: default:
536: break;
537: }
538:
539: /*
540: * If input parsers have not been allocated, do so now.
1.2 schwarze 541: * We keep these instanced between parsers, but set them
1.1 schwarze 542: * locally per parse routine since we can use different
543: * parsers with each one.
544: */
545:
546: if ( ! (curp->man || curp->mdoc))
1.60 schwarze 547: choose_parser(curp);
1.1 schwarze 548:
1.25 schwarze 549: /*
1.60 schwarze 550: * Lastly, push down into the parsers themselves.
1.1 schwarze 551: * If libroff returns ROFF_TBL, then add it to the
552: * currently open parse. Since we only get here if
553: * there does exist data (see tbl_data.c), we're
554: * guaranteed that something's been allocated.
555: * Do the same for ROFF_EQN.
556: */
557:
1.76 schwarze 558: if (rr == ROFF_TBL) {
559: while ((span = roff_span(curp->roff)) != NULL)
560: if (curp->man == NULL)
561: mdoc_addspan(curp->mdoc, span);
562: else
563: man_addspan(curp->man, span);
564: } else if (rr == ROFF_EQN) {
565: if (curp->man == NULL)
566: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
567: else
568: man_addeqn(curp->man, roff_eqn(curp->roff));
569: } else if ((curp->man == NULL ?
570: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
571: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
572: break;
1.1 schwarze 573:
574: /* Temporary buffers typically are not full. */
575:
576: if (0 == start && '\0' == blk.buf[i])
577: break;
578:
579: /* Start the next input line. */
580:
581: pos = 0;
582: }
583:
584: free(ln.buf);
585: }
586:
587: static int
1.18 schwarze 588: read_whole_file(struct mparse *curp, const char *file, int fd,
589: struct buf *fb, int *with_mmap)
1.1 schwarze 590: {
591: struct stat st;
592: size_t off;
593: ssize_t ssz;
594:
595: if (-1 == fstat(fd, &st)) {
1.82 schwarze 596: perror(file);
597: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 598: }
599:
600: /*
601: * If we're a regular file, try just reading in the whole entry
602: * via mmap(). This is faster than reading it into blocks, and
603: * since each file is only a few bytes to begin with, I'm not
604: * concerned that this is going to tank any machines.
605: */
606:
607: if (S_ISREG(st.st_mode)) {
608: if (st.st_size >= (1U << 31)) {
1.86 ! schwarze 609: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 schwarze 610: return(0);
611: }
612: *with_mmap = 1;
613: fb->sz = (size_t)st.st_size;
1.15 schwarze 614: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 615: if (fb->buf != MAP_FAILED)
616: return(1);
617: }
618:
619: /*
620: * If this isn't a regular file (like, say, stdin), then we must
621: * go the old way and just read things in bit by bit.
622: */
623:
624: *with_mmap = 0;
625: off = 0;
626: fb->sz = 0;
627: fb->buf = NULL;
628: for (;;) {
629: if (off == fb->sz) {
630: if (fb->sz == (1U << 31)) {
1.86 ! schwarze 631: mandoc_msg(MANDOCERR_TOOLARGE, curp,
! 632: 0, 0, NULL);
1.1 schwarze 633: break;
634: }
635: resize_buf(fb, 65536);
636: }
637: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
638: if (ssz == 0) {
639: fb->sz = off;
640: return(1);
641: }
642: if (ssz == -1) {
1.82 schwarze 643: perror(file);
644: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 645: }
646: off += (size_t)ssz;
647: }
648:
649: free(fb->buf);
650: fb->buf = NULL;
651: return(0);
652: }
653:
654: static void
655: mparse_end(struct mparse *curp)
656: {
657:
1.50 schwarze 658: if (curp->mdoc == NULL &&
659: curp->man == NULL &&
660: curp->sodest == NULL) {
661: if (curp->options & MPARSE_MDOC)
662: curp->mdoc = curp->pmdoc;
663: else {
664: if (curp->pman == NULL)
1.81 schwarze 665: curp->pman = man_alloc(
666: curp->roff, curp, curp->defos,
1.50 schwarze 667: curp->options & MPARSE_QUICK ? 1 : 0);
668: curp->man = curp->pman;
669: }
670: }
1.86 ! schwarze 671: if (curp->mdoc)
! 672: mdoc_endparse(curp->mdoc);
! 673: if (curp->man)
! 674: man_endparse(curp->man);
1.1 schwarze 675: roff_endparse(curp->roff);
676: }
677:
1.15 schwarze 678: static void
679: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 680: {
1.61 schwarze 681: struct buf *svprimary;
1.1 schwarze 682: const char *svfile;
1.71 schwarze 683: size_t offset;
1.14 schwarze 684: static int recursion_depth;
685:
686: if (64 < recursion_depth) {
687: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 688: return;
1.14 schwarze 689: }
1.1 schwarze 690:
1.15 schwarze 691: /* Line number is per-file. */
692: svfile = curp->file;
693: curp->file = file;
1.61 schwarze 694: svprimary = curp->primary;
1.59 schwarze 695: curp->primary = &blk;
1.15 schwarze 696: curp->line = 1;
697: recursion_depth++;
698:
1.70 schwarze 699: /* Skip an UTF-8 byte order mark. */
700: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
701: (unsigned char)blk.buf[0] == 0xef &&
702: (unsigned char)blk.buf[1] == 0xbb &&
703: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 704: offset = 3;
1.70 schwarze 705: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 706: } else
707: offset = 0;
1.70 schwarze 708:
1.71 schwarze 709: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 710:
1.86 ! schwarze 711: if (--recursion_depth == 0)
1.15 schwarze 712: mparse_end(curp);
713:
1.61 schwarze 714: curp->primary = svprimary;
1.15 schwarze 715: curp->file = svfile;
716: }
717:
1.74 schwarze 718: /*
719: * Read the whole file into memory and call the parsers.
720: * Called recursively when an .so request is encountered.
721: */
1.15 schwarze 722: enum mandoclevel
723: mparse_readfd(struct mparse *curp, int fd, const char *file)
724: {
725: struct buf blk;
726: int with_mmap;
1.70 schwarze 727: int save_filenc;
1.1 schwarze 728:
1.67 schwarze 729: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 730: save_filenc = curp->filenc;
731: curp->filenc = curp->options &
732: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 733: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 734: curp->filenc = save_filenc;
1.67 schwarze 735: if (with_mmap)
736: munmap(blk.buf, blk.sz);
737: else
738: free(blk.buf);
739: }
1.1 schwarze 740:
1.74 schwarze 741: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 schwarze 742: perror(file);
1.67 schwarze 743:
1.74 schwarze 744: mparse_wait(curp);
1.1 schwarze 745: return(curp->file_status);
1.58 schwarze 746: }
747:
748: enum mandoclevel
1.73 schwarze 749: mparse_open(struct mparse *curp, int *fd, const char *file)
1.58 schwarze 750: {
751: int pfd[2];
1.74 schwarze 752: int save_errno;
1.58 schwarze 753: char *cp;
754:
755: curp->file = file;
1.74 schwarze 756:
757: /* Unless zipped, try to just open the file. */
758:
1.58 schwarze 759: if ((cp = strrchr(file, '.')) == NULL ||
760: strcmp(cp + 1, "gz")) {
1.73 schwarze 761: curp->child = 0;
1.74 schwarze 762: if ((*fd = open(file, O_RDONLY)) != -1)
763: return(MANDOCLEVEL_OK);
764:
765: /* Open failed; try to append ".gz". */
766:
767: mandoc_asprintf(&cp, "%s.gz", file);
768: file = cp;
769: } else
770: cp = NULL;
771:
772: /* Before forking, make sure the file can be read. */
773:
774: save_errno = errno;
775: if (access(file, R_OK) == -1) {
776: if (cp != NULL)
777: errno = save_errno;
1.83 schwarze 778: free(cp);
779: *fd = -1;
780: curp->child = 0;
781: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
782: return(MANDOCLEVEL_ERROR);
1.58 schwarze 783: }
784:
1.74 schwarze 785: /* Run gunzip(1). */
786:
1.58 schwarze 787: if (pipe(pfd) == -1) {
1.82 schwarze 788: perror("pipe");
789: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 790: }
791:
1.73 schwarze 792: switch (curp->child = fork()) {
1.58 schwarze 793: case -1:
1.82 schwarze 794: perror("fork");
795: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 796: case 0:
797: close(pfd[0]);
798: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.82 schwarze 799: perror("dup");
800: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 801: }
802: execlp("gunzip", "gunzip", "-c", file, NULL);
1.82 schwarze 803: perror("exec");
804: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 805: default:
806: close(pfd[1]);
807: *fd = pfd[0];
808: return(MANDOCLEVEL_OK);
809: }
810: }
811:
812: enum mandoclevel
1.73 schwarze 813: mparse_wait(struct mparse *curp)
1.58 schwarze 814: {
815: int status;
816:
1.73 schwarze 817: if (curp->child == 0)
818: return(MANDOCLEVEL_OK);
819:
820: if (waitpid(curp->child, &status, 0) == -1) {
1.82 schwarze 821: perror("wait");
822: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 823: }
824: if (WIFSIGNALED(status)) {
1.83 schwarze 825: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
826: "gunzip died from signal %d", WTERMSIG(status));
827: return(MANDOCLEVEL_ERROR);
1.58 schwarze 828: }
829: if (WEXITSTATUS(status)) {
1.83 schwarze 830: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
831: "gunzip failed with code %d", WEXITSTATUS(status));
832: return(MANDOCLEVEL_ERROR);
1.58 schwarze 833: }
834: return(MANDOCLEVEL_OK);
1.1 schwarze 835: }
836:
837: struct mparse *
1.69 schwarze 838: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
839: const struct mchars *mchars, const char *defos)
1.1 schwarze 840: {
841: struct mparse *curp;
842:
843: curp = mandoc_calloc(1, sizeof(struct mparse));
844:
1.22 schwarze 845: curp->options = options;
1.1 schwarze 846: curp->wlevel = wlevel;
847: curp->mmsg = mmsg;
1.7 schwarze 848: curp->defos = defos;
1.1 schwarze 849:
1.69 schwarze 850: curp->mchars = mchars;
851: curp->roff = roff_alloc(curp, curp->mchars, options);
1.50 schwarze 852: if (curp->options & MPARSE_MDOC)
853: curp->pmdoc = mdoc_alloc(
854: curp->roff, curp, curp->defos,
855: curp->options & MPARSE_QUICK ? 1 : 0);
856: if (curp->options & MPARSE_MAN)
1.81 schwarze 857: curp->pman = man_alloc(
858: curp->roff, curp, curp->defos,
1.50 schwarze 859: curp->options & MPARSE_QUICK ? 1 : 0);
860:
1.1 schwarze 861: return(curp);
862: }
863:
864: void
865: mparse_reset(struct mparse *curp)
866: {
867:
868: roff_reset(curp->roff);
869:
870: if (curp->mdoc)
871: mdoc_reset(curp->mdoc);
872: if (curp->man)
873: man_reset(curp->man);
1.4 schwarze 874: if (curp->secondary)
875: curp->secondary->sz = 0;
1.1 schwarze 876:
877: curp->file_status = MANDOCLEVEL_OK;
878: curp->mdoc = NULL;
879: curp->man = NULL;
1.23 schwarze 880:
881: free(curp->sodest);
882: curp->sodest = NULL;
1.1 schwarze 883: }
884:
885: void
886: mparse_free(struct mparse *curp)
887: {
888:
889: if (curp->pmdoc)
890: mdoc_free(curp->pmdoc);
891: if (curp->pman)
892: man_free(curp->pman);
893: if (curp->roff)
894: roff_free(curp->roff);
1.4 schwarze 895: if (curp->secondary)
896: free(curp->secondary->buf);
1.1 schwarze 897:
1.4 schwarze 898: free(curp->secondary);
1.23 schwarze 899: free(curp->sodest);
1.1 schwarze 900: free(curp);
901: }
902:
903: void
1.23 schwarze 904: mparse_result(struct mparse *curp,
905: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 schwarze 906: {
907:
1.23 schwarze 908: if (sodest && NULL != (*sodest = curp->sodest)) {
909: *mdoc = NULL;
910: *man = NULL;
911: return;
912: }
1.1 schwarze 913: if (mdoc)
914: *mdoc = curp->mdoc;
915: if (man)
916: *man = curp->man;
917: }
918:
919: void
920: mandoc_vmsg(enum mandocerr t, struct mparse *m,
921: int ln, int pos, const char *fmt, ...)
922: {
923: char buf[256];
924: va_list ap;
925:
926: va_start(ap, fmt);
1.26 schwarze 927: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 928: va_end(ap);
929:
930: mandoc_msg(t, m, ln, pos, buf);
931: }
932:
933: void
1.25 schwarze 934: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 935: int ln, int col, const char *msg)
936: {
937: enum mandoclevel level;
938:
1.86 ! schwarze 939: level = MANDOCLEVEL_ERROR;
1.1 schwarze 940: while (er < mandoclimits[level])
941: level--;
942:
1.83 schwarze 943: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 944: return;
945:
946: if (m->mmsg)
947: (*m->mmsg)(er, level, m->file, ln, col, msg);
948:
949: if (m->file_status < level)
950: m->file_status = level;
951: }
952:
953: const char *
954: mparse_strerror(enum mandocerr er)
955: {
956:
957: return(mandocerrs[er]);
958: }
959:
960: const char *
961: mparse_strlevel(enum mandoclevel lvl)
962: {
963: return(mandoclevels[lvl]);
1.4 schwarze 964: }
965:
966: void
967: mparse_keep(struct mparse *p)
968: {
969:
970: assert(NULL == p->secondary);
971: p->secondary = mandoc_calloc(1, sizeof(struct buf));
972: }
973:
974: const char *
975: mparse_getkeep(const struct mparse *p)
976: {
977:
978: assert(p->secondary);
979: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 schwarze 980: }
![[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