Annotation of src/usr.bin/mandoc/read.c, Revision 1.99
1.99 ! schwarze 1: /* $OpenBSD: read.c,v 1.98 2015/02/06 03:31:11 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,
1.87 schwarze 78: MANDOCERR_UNSUPP,
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",
1.96 schwarze 108: "missing description line, using \"\"",
1.1 schwarze 109: "sections out of conventional order",
1.32 schwarze 110: "duplicate section title",
111: "unexpected section",
1.63 schwarze 112: "unusual Xr order",
113: "unusual Xr punctuation",
1.62 schwarze 114: "AUTHORS section without An macro",
1.1 schwarze 115:
116: /* related to macros and nesting */
1.33 schwarze 117: "obsolete macro",
1.78 schwarze 118: "macro neither callable nor escaped",
1.1 schwarze 119: "skipping paragraph macro",
1.10 schwarze 120: "moving paragraph macro out of list",
1.1 schwarze 121: "skipping no-space macro",
122: "blocks badly nested",
123: "nested displays are not portable",
1.35 schwarze 124: "moving content out of list",
125: ".Vt block has child macro",
1.56 schwarze 126: "fill mode already enabled, skipping",
127: "fill mode already disabled, skipping",
1.1 schwarze 128: "line scope broken",
129:
130: /* related to missing macro arguments */
1.36 schwarze 131: "skipping empty request",
132: "conditional request controls empty scope",
1.1 schwarze 133: "skipping empty macro",
1.99 ! schwarze 134: "empty block",
1.40 schwarze 135: "empty argument, using 0n",
1.38 schwarze 136: "missing display type, using -ragged",
137: "list type is not the first argument",
138: "missing -width in -tag list, using 8n",
1.56 schwarze 139: "missing utility name, using \"\"",
1.98 schwarze 140: "missing function name, using \"\"",
1.38 schwarze 141: "empty head in list item",
142: "empty list item",
1.39 schwarze 143: "missing font type, using \\fR",
144: "unknown font type, using \\fR",
1.79 schwarze 145: "nothing follows prefix",
1.97 schwarze 146: "empty reference block",
1.38 schwarze 147: "missing -std argument, adding it",
1.66 schwarze 148: "missing eqn box, using \"\"",
1.1 schwarze 149:
150: /* related to bad macro arguments */
1.42 schwarze 151: "unterminated quoted argument",
1.1 schwarze 152: "duplicate argument",
1.54 schwarze 153: "skipping duplicate argument",
1.41 schwarze 154: "skipping duplicate display type",
155: "skipping duplicate list type",
1.54 schwarze 156: "skipping -width argument",
1.1 schwarze 157: "unknown AT&T UNIX version",
1.64 schwarze 158: "comma in function argument",
1.65 schwarze 159: "parenthesis in function name",
1.45 schwarze 160: "invalid content in Rs block",
1.41 schwarze 161: "invalid Boolean argument",
162: "unknown font, skipping request",
1.1 schwarze 163:
164: /* related to plain text */
1.42 schwarze 165: "blank line in fill mode, using .sp",
166: "tab in filled text",
167: "whitespace at end of input line",
1.1 schwarze 168: "bad comment style",
1.42 schwarze 169: "invalid escape sequence",
170: "undefined string, using \"\"",
1.3 schwarze 171:
1.93 schwarze 172: /* related to tables */
173: "tbl line starts with span",
174: "tbl column starts with span",
175: "skipping vertical bar in tbl layout",
176:
1.1 schwarze 177: "generic error",
1.3 schwarze 178:
1.1 schwarze 179: /* related to tables */
1.91 schwarze 180: "non-alphabetic character in tbl options",
181: "skipping unknown tbl option",
182: "missing tbl option argument",
183: "wrong tbl option argument size",
1.93 schwarze 184: "empty tbl layout",
185: "invalid character in tbl layout",
186: "unmatched parenthesis in tbl layout",
1.94 schwarze 187: "tbl without any data cells",
188: "ignoring data in spanned tbl cell",
189: "ignoring extra tbl data cells",
190: "data block open at end of tbl",
1.1 schwarze 191:
1.46 schwarze 192: /* related to document structure and macros */
1.83 schwarze 193: NULL,
1.1 schwarze 194: "input stack limit exceeded, infinite loop?",
195: "skipping bad character",
1.46 schwarze 196: "skipping unknown macro",
1.87 schwarze 197: "skipping insecure request",
1.51 schwarze 198: "skipping item outside list",
1.46 schwarze 199: "skipping column outside column list",
200: "skipping end of block that is not open",
1.90 schwarze 201: "fewer RS blocks open, skipping",
1.46 schwarze 202: "inserting missing end of block",
203: "appending missing end of block",
204:
205: /* related to request and macro arguments */
1.1 schwarze 206: "escaped character not allowed in a name",
207: "argument count wrong",
1.75 schwarze 208: "NOT IMPLEMENTED: Bd -file",
1.49 schwarze 209: "missing list type, using -item",
1.48 schwarze 210: "missing manual name, using \"\"",
1.49 schwarze 211: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 212: "unknown standard specifier",
1.49 schwarze 213: "skipping request without numeric argument",
1.84 schwarze 214: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 215: ".so request failed",
1.39 schwarze 216: "skipping all arguments",
217: "skipping excess arguments",
1.68 schwarze 218: "divide by zero",
1.87 schwarze 219:
220: "unsupported feature",
221: "input too large",
1.89 schwarze 222: "unsupported control character",
1.87 schwarze 223: "unsupported roff request",
1.94 schwarze 224: "eqn delim option in tbl",
1.93 schwarze 225: "unsupported tbl layout modifier",
1.87 schwarze 226: "ignoring macro in table",
1.1 schwarze 227: };
228:
229: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
230: "SUCCESS",
231: "RESERVED",
232: "WARNING",
233: "ERROR",
1.87 schwarze 234: "UNSUPP",
1.1 schwarze 235: "BADARG",
236: "SYSERR"
237: };
238:
1.25 schwarze 239:
1.1 schwarze 240: static void
241: resize_buf(struct buf *buf, size_t initial)
242: {
243:
244: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
245: buf->buf = mandoc_realloc(buf->buf, buf->sz);
246: }
247:
248: static void
1.60 schwarze 249: choose_parser(struct mparse *curp)
1.1 schwarze 250: {
1.59 schwarze 251: char *cp, *ep;
252: int format;
1.1 schwarze 253:
1.59 schwarze 254: /*
255: * If neither command line arguments -mdoc or -man select
256: * a parser nor the roff parser found a .Dd or .TH macro
257: * yet, look ahead in the main input buffer.
258: */
259:
260: if ((format = roff_getformat(curp->roff)) == 0) {
261: cp = curp->primary->buf;
262: ep = cp + curp->primary->sz;
263: while (cp < ep) {
1.61 schwarze 264: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 265: cp++;
266: if (cp[0] == 'D' && cp[1] == 'd') {
267: format = MPARSE_MDOC;
268: break;
269: }
270: if (cp[0] == 'T' && cp[1] == 'H') {
271: format = MPARSE_MAN;
272: break;
273: }
274: }
275: cp = memchr(cp, '\n', ep - cp);
276: if (cp == NULL)
277: break;
278: cp++;
279: }
1.1 schwarze 280: }
281:
1.59 schwarze 282: if (format == MPARSE_MDOC) {
1.25 schwarze 283: if (NULL == curp->pmdoc)
1.22 schwarze 284: curp->pmdoc = mdoc_alloc(
285: curp->roff, curp, curp->defos,
286: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 schwarze 287: assert(curp->pmdoc);
288: curp->mdoc = curp->pmdoc;
289: return;
1.25 schwarze 290: }
1.1 schwarze 291:
1.59 schwarze 292: /* Fall back to man(7) as a last resort. */
293:
1.25 schwarze 294: if (NULL == curp->pman)
1.81 schwarze 295: curp->pman = man_alloc(
296: curp->roff, curp, curp->defos,
1.22 schwarze 297: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 schwarze 298: assert(curp->pman);
299: curp->man = curp->pman;
300: }
301:
302: /*
1.71 schwarze 303: * Main parse routine for a buffer.
304: * It assumes encoding and line numbering are already set up.
305: * It can recurse directly (for invocations of user-defined
306: * macros, inline equations, and input line traps)
307: * and indirectly (for .so file inclusion).
1.1 schwarze 308: */
309: static void
1.71 schwarze 310: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 311: {
312: const struct tbl_span *span;
313: struct buf ln;
1.88 schwarze 314: const char *save_file;
1.85 schwarze 315: char *cp;
1.71 schwarze 316: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 317: enum rofferr rr;
1.76 schwarze 318: int of;
1.1 schwarze 319: int lnn; /* line number in the real file */
1.85 schwarze 320: int fd;
321: pid_t save_child;
1.1 schwarze 322: unsigned char c;
323:
1.71 schwarze 324: memset(&ln, 0, sizeof(ln));
1.1 schwarze 325:
1.25 schwarze 326: lnn = curp->line;
327: pos = 0;
1.1 schwarze 328:
1.71 schwarze 329: while (i < blk.sz) {
1.1 schwarze 330: if (0 == pos && '\0' == blk.buf[i])
331: break;
332:
333: if (start) {
334: curp->line = lnn;
335: curp->reparse_count = 0;
1.70 schwarze 336:
337: if (lnn < 3 &&
338: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 339: curp->filenc & MPARSE_LATIN1)
340: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 341: }
342:
1.71 schwarze 343: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 344:
345: /*
346: * When finding an unescaped newline character,
347: * leave the character loop to process the line.
348: * Skip a preceding carriage return, if any.
349: */
350:
1.71 schwarze 351: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 352: '\n' == blk.buf[i + 1])
353: ++i;
354: if ('\n' == blk.buf[i]) {
355: ++i;
356: ++lnn;
357: break;
358: }
359:
1.13 schwarze 360: /*
1.70 schwarze 361: * Make sure we have space for the worst
362: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 363: */
364:
1.71 schwarze 365: if (pos + 11 > ln.sz)
1.13 schwarze 366: resize_buf(&ln, 256);
367:
1.25 schwarze 368: /*
1.70 schwarze 369: * Encode 8-bit input.
1.1 schwarze 370: */
371:
1.70 schwarze 372: c = blk.buf[i];
373: if (c & 0x80) {
1.71 schwarze 374: if ( ! (curp->filenc && preconv_encode(
375: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 376: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
377: curp->line, pos, "0x%x", c);
1.70 schwarze 378: ln.buf[pos++] = '?';
379: i++;
380: }
381: continue;
382: }
383:
384: /*
385: * Exclude control characters.
386: */
1.1 schwarze 387:
1.70 schwarze 388: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 389: mandoc_vmsg(c == 0x00 || c == 0x04 ||
390: c > 0x0a ? MANDOCERR_CHAR_BAD :
391: MANDOCERR_CHAR_UNSUPP,
392: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 393: i++;
1.6 schwarze 394: ln.buf[pos++] = '?';
1.1 schwarze 395: continue;
396: }
397:
398: /* Trailing backslash = a plain char. */
399:
1.71 schwarze 400: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 schwarze 401: ln.buf[pos++] = blk.buf[i++];
402: continue;
403: }
404:
405: /*
406: * Found escape and at least one other character.
407: * When it's a newline character, skip it.
408: * When there is a carriage return in between,
409: * skip that one as well.
410: */
411:
1.71 schwarze 412: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 schwarze 413: '\n' == blk.buf[i + 2])
414: ++i;
415: if ('\n' == blk.buf[i + 1]) {
416: i += 2;
417: ++lnn;
418: continue;
419: }
420:
421: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
422: i += 2;
423: /* Comment, skip to end of line */
1.71 schwarze 424: for (; i < blk.sz; ++i) {
1.1 schwarze 425: if ('\n' == blk.buf[i]) {
426: ++i;
427: ++lnn;
428: break;
429: }
430: }
431:
432: /* Backout trailing whitespaces */
433: for (; pos > 0; --pos) {
434: if (ln.buf[pos - 1] != ' ')
435: break;
436: if (pos > 2 && ln.buf[pos - 2] == '\\')
437: break;
438: }
439: break;
440: }
441:
1.13 schwarze 442: /* Catch escaped bogus characters. */
443:
444: c = (unsigned char) blk.buf[i+1];
445:
1.25 schwarze 446: if ( ! (isascii(c) &&
447: (isgraph(c) || isblank(c)))) {
1.89 schwarze 448: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.56 schwarze 449: curp->line, pos, "0x%x", c);
1.13 schwarze 450: i += 2;
451: ln.buf[pos++] = '?';
452: continue;
453: }
454:
1.1 schwarze 455: /* Some other escape sequence, copy & cont. */
456:
457: ln.buf[pos++] = blk.buf[i++];
458: ln.buf[pos++] = blk.buf[i++];
459: }
460:
1.71 schwarze 461: if (pos >= ln.sz)
1.1 schwarze 462: resize_buf(&ln, 256);
463:
464: ln.buf[pos] = '\0';
465:
466: /*
467: * A significant amount of complexity is contained by
468: * the roff preprocessor. It's line-oriented but can be
469: * expressed on one line, so we need at times to
470: * readjust our starting point and re-run it. The roff
471: * preprocessor can also readjust the buffers with new
472: * data, so we pass them in wholesale.
473: */
474:
475: of = 0;
476:
1.4 schwarze 477: /*
478: * Maintain a lookaside buffer of all parsed lines. We
479: * only do this if mparse_keep() has been invoked (the
480: * buffer may be accessed with mparse_getkeep()).
481: */
482:
483: if (curp->secondary) {
1.25 schwarze 484: curp->secondary->buf = mandoc_realloc(
485: curp->secondary->buf,
486: curp->secondary->sz + pos + 2);
487: memcpy(curp->secondary->buf +
488: curp->secondary->sz,
489: ln.buf, pos);
1.4 schwarze 490: curp->secondary->sz += pos;
491: curp->secondary->buf
492: [curp->secondary->sz] = '\n';
493: curp->secondary->sz++;
494: curp->secondary->buf
495: [curp->secondary->sz] = '\0';
496: }
1.1 schwarze 497: rerun:
1.72 schwarze 498: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 499:
500: switch (rr) {
1.25 schwarze 501: case ROFF_REPARSE:
1.1 schwarze 502: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.71 schwarze 503: mparse_buf_r(curp, ln, of, 0);
1.1 schwarze 504: else
505: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 506: curp->line, pos, NULL);
1.1 schwarze 507: pos = 0;
508: continue;
1.25 schwarze 509: case ROFF_APPEND:
1.71 schwarze 510: pos = strlen(ln.buf);
1.1 schwarze 511: continue;
1.25 schwarze 512: case ROFF_RERUN:
1.1 schwarze 513: goto rerun;
1.25 schwarze 514: case ROFF_IGN:
1.1 schwarze 515: pos = 0;
516: continue;
1.25 schwarze 517: case ROFF_SO:
1.71 schwarze 518: if ( ! (curp->options & MPARSE_SO) &&
519: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 520: curp->sodest = mandoc_strdup(ln.buf + of);
521: free(ln.buf);
522: return;
523: }
1.4 schwarze 524: /*
525: * We remove `so' clauses from our lookaside
526: * buffer because we're going to descend into
527: * the file recursively.
528: */
1.25 schwarze 529: if (curp->secondary)
1.4 schwarze 530: curp->secondary->sz -= pos + 1;
1.88 schwarze 531: save_file = curp->file;
1.85 schwarze 532: save_child = curp->child;
533: if (mparse_open(curp, &fd, ln.buf + of) ==
1.88 schwarze 534: MANDOCLEVEL_OK) {
1.85 schwarze 535: mparse_readfd(curp, fd, ln.buf + of);
1.88 schwarze 536: curp->file = save_file;
537: } else {
538: curp->file = save_file;
1.30 schwarze 539: mandoc_vmsg(MANDOCERR_SO_FAIL,
540: curp, curp->line, pos,
541: ".so %s", ln.buf + of);
1.85 schwarze 542: ln.sz = mandoc_asprintf(&cp,
543: ".sp\nSee the file %s.\n.sp",
544: ln.buf + of);
545: free(ln.buf);
546: ln.buf = cp;
547: of = 0;
548: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 549: }
1.85 schwarze 550: curp->child = save_child;
1.1 schwarze 551: pos = 0;
552: continue;
553: default:
554: break;
555: }
556:
557: /*
558: * If input parsers have not been allocated, do so now.
1.2 schwarze 559: * We keep these instanced between parsers, but set them
1.1 schwarze 560: * locally per parse routine since we can use different
561: * parsers with each one.
562: */
563:
564: if ( ! (curp->man || curp->mdoc))
1.60 schwarze 565: choose_parser(curp);
1.1 schwarze 566:
1.25 schwarze 567: /*
1.60 schwarze 568: * Lastly, push down into the parsers themselves.
1.1 schwarze 569: * If libroff returns ROFF_TBL, then add it to the
570: * currently open parse. Since we only get here if
571: * there does exist data (see tbl_data.c), we're
572: * guaranteed that something's been allocated.
573: * Do the same for ROFF_EQN.
574: */
575:
1.76 schwarze 576: if (rr == ROFF_TBL) {
577: while ((span = roff_span(curp->roff)) != NULL)
578: if (curp->man == NULL)
579: mdoc_addspan(curp->mdoc, span);
580: else
581: man_addspan(curp->man, span);
582: } else if (rr == ROFF_EQN) {
583: if (curp->man == NULL)
584: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
585: else
586: man_addeqn(curp->man, roff_eqn(curp->roff));
587: } else if ((curp->man == NULL ?
588: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
589: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
590: break;
1.1 schwarze 591:
592: /* Temporary buffers typically are not full. */
593:
594: if (0 == start && '\0' == blk.buf[i])
595: break;
596:
597: /* Start the next input line. */
598:
599: pos = 0;
600: }
601:
602: free(ln.buf);
603: }
604:
605: static int
1.18 schwarze 606: read_whole_file(struct mparse *curp, const char *file, int fd,
607: struct buf *fb, int *with_mmap)
1.1 schwarze 608: {
609: struct stat st;
610: size_t off;
611: ssize_t ssz;
612:
613: if (-1 == fstat(fd, &st)) {
1.82 schwarze 614: perror(file);
615: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 616: }
617:
618: /*
619: * If we're a regular file, try just reading in the whole entry
620: * via mmap(). This is faster than reading it into blocks, and
621: * since each file is only a few bytes to begin with, I'm not
622: * concerned that this is going to tank any machines.
623: */
624:
625: if (S_ISREG(st.st_mode)) {
626: if (st.st_size >= (1U << 31)) {
1.86 schwarze 627: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 schwarze 628: return(0);
629: }
630: *with_mmap = 1;
631: fb->sz = (size_t)st.st_size;
1.15 schwarze 632: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 633: if (fb->buf != MAP_FAILED)
634: return(1);
635: }
636:
637: /*
638: * If this isn't a regular file (like, say, stdin), then we must
639: * go the old way and just read things in bit by bit.
640: */
641:
642: *with_mmap = 0;
643: off = 0;
644: fb->sz = 0;
645: fb->buf = NULL;
646: for (;;) {
647: if (off == fb->sz) {
648: if (fb->sz == (1U << 31)) {
1.86 schwarze 649: mandoc_msg(MANDOCERR_TOOLARGE, curp,
650: 0, 0, NULL);
1.1 schwarze 651: break;
652: }
653: resize_buf(fb, 65536);
654: }
655: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
656: if (ssz == 0) {
657: fb->sz = off;
658: return(1);
659: }
660: if (ssz == -1) {
1.82 schwarze 661: perror(file);
662: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 663: }
664: off += (size_t)ssz;
665: }
666:
667: free(fb->buf);
668: fb->buf = NULL;
669: return(0);
670: }
671:
672: static void
673: mparse_end(struct mparse *curp)
674: {
675:
1.50 schwarze 676: if (curp->mdoc == NULL &&
677: curp->man == NULL &&
678: curp->sodest == NULL) {
679: if (curp->options & MPARSE_MDOC)
680: curp->mdoc = curp->pmdoc;
681: else {
682: if (curp->pman == NULL)
1.81 schwarze 683: curp->pman = man_alloc(
684: curp->roff, curp, curp->defos,
1.50 schwarze 685: curp->options & MPARSE_QUICK ? 1 : 0);
686: curp->man = curp->pman;
687: }
688: }
1.86 schwarze 689: if (curp->mdoc)
690: mdoc_endparse(curp->mdoc);
691: if (curp->man)
692: man_endparse(curp->man);
1.1 schwarze 693: roff_endparse(curp->roff);
694: }
695:
1.15 schwarze 696: static void
697: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 698: {
1.61 schwarze 699: struct buf *svprimary;
1.1 schwarze 700: const char *svfile;
1.71 schwarze 701: size_t offset;
1.14 schwarze 702: static int recursion_depth;
703:
704: if (64 < recursion_depth) {
705: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 706: return;
1.14 schwarze 707: }
1.1 schwarze 708:
1.15 schwarze 709: /* Line number is per-file. */
710: svfile = curp->file;
711: curp->file = file;
1.61 schwarze 712: svprimary = curp->primary;
1.59 schwarze 713: curp->primary = &blk;
1.15 schwarze 714: curp->line = 1;
715: recursion_depth++;
716:
1.70 schwarze 717: /* Skip an UTF-8 byte order mark. */
718: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
719: (unsigned char)blk.buf[0] == 0xef &&
720: (unsigned char)blk.buf[1] == 0xbb &&
721: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 722: offset = 3;
1.70 schwarze 723: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 724: } else
725: offset = 0;
1.70 schwarze 726:
1.71 schwarze 727: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 728:
1.86 schwarze 729: if (--recursion_depth == 0)
1.15 schwarze 730: mparse_end(curp);
731:
1.61 schwarze 732: curp->primary = svprimary;
1.15 schwarze 733: curp->file = svfile;
734: }
735:
1.74 schwarze 736: /*
737: * Read the whole file into memory and call the parsers.
738: * Called recursively when an .so request is encountered.
739: */
1.15 schwarze 740: enum mandoclevel
741: mparse_readfd(struct mparse *curp, int fd, const char *file)
742: {
743: struct buf blk;
744: int with_mmap;
1.70 schwarze 745: int save_filenc;
1.1 schwarze 746:
1.67 schwarze 747: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 748: save_filenc = curp->filenc;
749: curp->filenc = curp->options &
750: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 751: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 752: curp->filenc = save_filenc;
1.67 schwarze 753: if (with_mmap)
754: munmap(blk.buf, blk.sz);
755: else
756: free(blk.buf);
757: }
1.1 schwarze 758:
1.74 schwarze 759: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 schwarze 760: perror(file);
1.67 schwarze 761:
1.74 schwarze 762: mparse_wait(curp);
1.1 schwarze 763: return(curp->file_status);
1.58 schwarze 764: }
765:
766: enum mandoclevel
1.73 schwarze 767: mparse_open(struct mparse *curp, int *fd, const char *file)
1.58 schwarze 768: {
769: int pfd[2];
1.74 schwarze 770: int save_errno;
1.58 schwarze 771: char *cp;
772:
773: curp->file = file;
1.74 schwarze 774:
775: /* Unless zipped, try to just open the file. */
776:
1.58 schwarze 777: if ((cp = strrchr(file, '.')) == NULL ||
778: strcmp(cp + 1, "gz")) {
1.73 schwarze 779: curp->child = 0;
1.74 schwarze 780: if ((*fd = open(file, O_RDONLY)) != -1)
781: return(MANDOCLEVEL_OK);
782:
783: /* Open failed; try to append ".gz". */
784:
785: mandoc_asprintf(&cp, "%s.gz", file);
786: file = cp;
787: } else
788: cp = NULL;
789:
790: /* Before forking, make sure the file can be read. */
791:
792: save_errno = errno;
793: if (access(file, R_OK) == -1) {
794: if (cp != NULL)
795: errno = save_errno;
1.83 schwarze 796: free(cp);
797: *fd = -1;
798: curp->child = 0;
799: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
800: return(MANDOCLEVEL_ERROR);
1.58 schwarze 801: }
802:
1.74 schwarze 803: /* Run gunzip(1). */
804:
1.58 schwarze 805: if (pipe(pfd) == -1) {
1.82 schwarze 806: perror("pipe");
807: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 808: }
809:
1.73 schwarze 810: switch (curp->child = fork()) {
1.58 schwarze 811: case -1:
1.82 schwarze 812: perror("fork");
813: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 814: case 0:
815: close(pfd[0]);
816: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.82 schwarze 817: perror("dup");
818: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 819: }
820: execlp("gunzip", "gunzip", "-c", file, NULL);
1.82 schwarze 821: perror("exec");
822: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 823: default:
824: close(pfd[1]);
825: *fd = pfd[0];
826: return(MANDOCLEVEL_OK);
827: }
828: }
829:
830: enum mandoclevel
1.73 schwarze 831: mparse_wait(struct mparse *curp)
1.58 schwarze 832: {
833: int status;
834:
1.73 schwarze 835: if (curp->child == 0)
836: return(MANDOCLEVEL_OK);
837:
838: if (waitpid(curp->child, &status, 0) == -1) {
1.82 schwarze 839: perror("wait");
840: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 841: }
842: if (WIFSIGNALED(status)) {
1.83 schwarze 843: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
844: "gunzip died from signal %d", WTERMSIG(status));
845: return(MANDOCLEVEL_ERROR);
1.58 schwarze 846: }
847: if (WEXITSTATUS(status)) {
1.83 schwarze 848: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
849: "gunzip failed with code %d", WEXITSTATUS(status));
850: return(MANDOCLEVEL_ERROR);
1.58 schwarze 851: }
852: return(MANDOCLEVEL_OK);
1.1 schwarze 853: }
854:
855: struct mparse *
1.69 schwarze 856: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
857: const struct mchars *mchars, const char *defos)
1.1 schwarze 858: {
859: struct mparse *curp;
860:
861: curp = mandoc_calloc(1, sizeof(struct mparse));
862:
1.22 schwarze 863: curp->options = options;
1.1 schwarze 864: curp->wlevel = wlevel;
865: curp->mmsg = mmsg;
1.7 schwarze 866: curp->defos = defos;
1.1 schwarze 867:
1.69 schwarze 868: curp->mchars = mchars;
869: curp->roff = roff_alloc(curp, curp->mchars, options);
1.50 schwarze 870: if (curp->options & MPARSE_MDOC)
871: curp->pmdoc = mdoc_alloc(
872: curp->roff, curp, curp->defos,
873: curp->options & MPARSE_QUICK ? 1 : 0);
874: if (curp->options & MPARSE_MAN)
1.81 schwarze 875: curp->pman = man_alloc(
876: curp->roff, curp, curp->defos,
1.50 schwarze 877: curp->options & MPARSE_QUICK ? 1 : 0);
878:
1.1 schwarze 879: return(curp);
880: }
881:
882: void
883: mparse_reset(struct mparse *curp)
884: {
885:
886: roff_reset(curp->roff);
887:
888: if (curp->mdoc)
889: mdoc_reset(curp->mdoc);
890: if (curp->man)
891: man_reset(curp->man);
1.4 schwarze 892: if (curp->secondary)
893: curp->secondary->sz = 0;
1.1 schwarze 894:
895: curp->file_status = MANDOCLEVEL_OK;
896: curp->mdoc = NULL;
897: curp->man = NULL;
1.23 schwarze 898:
899: free(curp->sodest);
900: curp->sodest = NULL;
1.1 schwarze 901: }
902:
903: void
904: mparse_free(struct mparse *curp)
905: {
906:
907: if (curp->pmdoc)
908: mdoc_free(curp->pmdoc);
909: if (curp->pman)
910: man_free(curp->pman);
911: if (curp->roff)
912: roff_free(curp->roff);
1.4 schwarze 913: if (curp->secondary)
914: free(curp->secondary->buf);
1.1 schwarze 915:
1.4 schwarze 916: free(curp->secondary);
1.23 schwarze 917: free(curp->sodest);
1.1 schwarze 918: free(curp);
919: }
920:
921: void
1.23 schwarze 922: mparse_result(struct mparse *curp,
923: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 schwarze 924: {
925:
1.23 schwarze 926: if (sodest && NULL != (*sodest = curp->sodest)) {
927: *mdoc = NULL;
928: *man = NULL;
929: return;
930: }
1.1 schwarze 931: if (mdoc)
932: *mdoc = curp->mdoc;
933: if (man)
934: *man = curp->man;
935: }
936:
937: void
938: mandoc_vmsg(enum mandocerr t, struct mparse *m,
939: int ln, int pos, const char *fmt, ...)
940: {
941: char buf[256];
942: va_list ap;
943:
944: va_start(ap, fmt);
1.26 schwarze 945: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 946: va_end(ap);
947:
948: mandoc_msg(t, m, ln, pos, buf);
949: }
950:
951: void
1.25 schwarze 952: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 953: int ln, int col, const char *msg)
954: {
955: enum mandoclevel level;
956:
1.87 schwarze 957: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 958: while (er < mandoclimits[level])
959: level--;
960:
1.83 schwarze 961: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 962: return;
963:
964: if (m->mmsg)
965: (*m->mmsg)(er, level, m->file, ln, col, msg);
966:
967: if (m->file_status < level)
968: m->file_status = level;
969: }
970:
971: const char *
972: mparse_strerror(enum mandocerr er)
973: {
974:
975: return(mandocerrs[er]);
976: }
977:
978: const char *
979: mparse_strlevel(enum mandoclevel lvl)
980: {
981: return(mandoclevels[lvl]);
1.4 schwarze 982: }
983:
984: void
985: mparse_keep(struct mparse *p)
986: {
987:
988: assert(NULL == p->secondary);
989: p->secondary = mandoc_calloc(1, sizeof(struct buf));
990: }
991:
992: const char *
993: mparse_getkeep(const struct mparse *p)
994: {
995:
996: assert(p->secondary);
997: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 schwarze 998: }
![[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