Annotation of src/usr.bin/mandoc/read.c, Revision 1.93
1.93 ! schwarze 1: /* $OpenBSD: read.c,v 1.92 2015/01/26 13:02:53 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",
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:
1.93 ! schwarze 169: /* related to tables */
! 170: "tbl line starts with span",
! 171: "tbl column starts with span",
! 172: "skipping vertical bar in tbl layout",
! 173:
1.1 schwarze 174: "generic error",
175:
1.3 schwarze 176: /* related to equations */
177: "unexpected equation scope closure",
178: "equation scope open on exit",
179: "overlapping equation scopes",
180: "unexpected end of equation",
181:
1.1 schwarze 182: /* related to tables */
1.91 schwarze 183: "non-alphabetic character in tbl options",
184: "skipping unknown tbl option",
185: "missing tbl option argument",
186: "wrong tbl option argument size",
1.93 ! schwarze 187: "empty tbl layout",
! 188: "invalid character in tbl layout",
! 189: "unmatched parenthesis in tbl layout",
1.1 schwarze 190: "no table data cells specified",
191: "ignore data in cell",
192: "data block still open",
193: "ignoring extra data cells",
194:
1.46 schwarze 195: /* related to document structure and macros */
1.83 schwarze 196: NULL,
1.1 schwarze 197: "input stack limit exceeded, infinite loop?",
198: "skipping bad character",
1.46 schwarze 199: "skipping unknown macro",
1.87 schwarze 200: "skipping insecure request",
1.51 schwarze 201: "skipping item outside list",
1.46 schwarze 202: "skipping column outside column list",
203: "skipping end of block that is not open",
1.90 schwarze 204: "fewer RS blocks open, skipping",
1.46 schwarze 205: "inserting missing end of block",
206: "appending missing end of block",
207:
208: /* related to request and macro arguments */
1.1 schwarze 209: "escaped character not allowed in a name",
210: "argument count wrong",
1.75 schwarze 211: "NOT IMPLEMENTED: Bd -file",
1.49 schwarze 212: "missing list type, using -item",
1.48 schwarze 213: "missing manual name, using \"\"",
1.49 schwarze 214: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 215: "unknown standard specifier",
1.49 schwarze 216: "skipping request without numeric argument",
1.84 schwarze 217: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 218: ".so request failed",
1.39 schwarze 219: "skipping all arguments",
220: "skipping excess arguments",
1.68 schwarze 221: "divide by zero",
1.87 schwarze 222:
223: "unsupported feature",
224: "input too large",
1.89 schwarze 225: "unsupported control character",
1.87 schwarze 226: "unsupported roff request",
1.93 ! schwarze 227: "unsupported tbl layout modifier",
1.87 schwarze 228: "ignoring macro in table",
1.92 schwarze 229: "eqn in tbl",
1.1 schwarze 230: };
231:
232: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
233: "SUCCESS",
234: "RESERVED",
235: "WARNING",
236: "ERROR",
1.87 schwarze 237: "UNSUPP",
1.1 schwarze 238: "BADARG",
239: "SYSERR"
240: };
241:
1.25 schwarze 242:
1.1 schwarze 243: static void
244: resize_buf(struct buf *buf, size_t initial)
245: {
246:
247: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
248: buf->buf = mandoc_realloc(buf->buf, buf->sz);
249: }
250:
251: static void
1.60 schwarze 252: choose_parser(struct mparse *curp)
1.1 schwarze 253: {
1.59 schwarze 254: char *cp, *ep;
255: int format;
1.1 schwarze 256:
1.59 schwarze 257: /*
258: * If neither command line arguments -mdoc or -man select
259: * a parser nor the roff parser found a .Dd or .TH macro
260: * yet, look ahead in the main input buffer.
261: */
262:
263: if ((format = roff_getformat(curp->roff)) == 0) {
264: cp = curp->primary->buf;
265: ep = cp + curp->primary->sz;
266: while (cp < ep) {
1.61 schwarze 267: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 268: cp++;
269: if (cp[0] == 'D' && cp[1] == 'd') {
270: format = MPARSE_MDOC;
271: break;
272: }
273: if (cp[0] == 'T' && cp[1] == 'H') {
274: format = MPARSE_MAN;
275: break;
276: }
277: }
278: cp = memchr(cp, '\n', ep - cp);
279: if (cp == NULL)
280: break;
281: cp++;
282: }
1.1 schwarze 283: }
284:
1.59 schwarze 285: if (format == MPARSE_MDOC) {
1.25 schwarze 286: if (NULL == curp->pmdoc)
1.22 schwarze 287: curp->pmdoc = mdoc_alloc(
288: curp->roff, curp, curp->defos,
289: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 schwarze 290: assert(curp->pmdoc);
291: curp->mdoc = curp->pmdoc;
292: return;
1.25 schwarze 293: }
1.1 schwarze 294:
1.59 schwarze 295: /* Fall back to man(7) as a last resort. */
296:
1.25 schwarze 297: if (NULL == curp->pman)
1.81 schwarze 298: curp->pman = man_alloc(
299: curp->roff, curp, curp->defos,
1.22 schwarze 300: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 schwarze 301: assert(curp->pman);
302: curp->man = curp->pman;
303: }
304:
305: /*
1.71 schwarze 306: * Main parse routine for a buffer.
307: * It assumes encoding and line numbering are already set up.
308: * It can recurse directly (for invocations of user-defined
309: * macros, inline equations, and input line traps)
310: * and indirectly (for .so file inclusion).
1.1 schwarze 311: */
312: static void
1.71 schwarze 313: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 314: {
315: const struct tbl_span *span;
316: struct buf ln;
1.88 schwarze 317: const char *save_file;
1.85 schwarze 318: char *cp;
1.71 schwarze 319: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 320: enum rofferr rr;
1.76 schwarze 321: int of;
1.1 schwarze 322: int lnn; /* line number in the real file */
1.85 schwarze 323: int fd;
324: pid_t save_child;
1.1 schwarze 325: unsigned char c;
326:
1.71 schwarze 327: memset(&ln, 0, sizeof(ln));
1.1 schwarze 328:
1.25 schwarze 329: lnn = curp->line;
330: pos = 0;
1.1 schwarze 331:
1.71 schwarze 332: while (i < blk.sz) {
1.1 schwarze 333: if (0 == pos && '\0' == blk.buf[i])
334: break;
335:
336: if (start) {
337: curp->line = lnn;
338: curp->reparse_count = 0;
1.70 schwarze 339:
340: if (lnn < 3 &&
341: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 342: curp->filenc & MPARSE_LATIN1)
343: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 344: }
345:
1.71 schwarze 346: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 347:
348: /*
349: * When finding an unescaped newline character,
350: * leave the character loop to process the line.
351: * Skip a preceding carriage return, if any.
352: */
353:
1.71 schwarze 354: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 355: '\n' == blk.buf[i + 1])
356: ++i;
357: if ('\n' == blk.buf[i]) {
358: ++i;
359: ++lnn;
360: break;
361: }
362:
1.13 schwarze 363: /*
1.70 schwarze 364: * Make sure we have space for the worst
365: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 366: */
367:
1.71 schwarze 368: if (pos + 11 > ln.sz)
1.13 schwarze 369: resize_buf(&ln, 256);
370:
1.25 schwarze 371: /*
1.70 schwarze 372: * Encode 8-bit input.
1.1 schwarze 373: */
374:
1.70 schwarze 375: c = blk.buf[i];
376: if (c & 0x80) {
1.71 schwarze 377: if ( ! (curp->filenc && preconv_encode(
378: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 379: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
380: curp->line, pos, "0x%x", c);
1.70 schwarze 381: ln.buf[pos++] = '?';
382: i++;
383: }
384: continue;
385: }
386:
387: /*
388: * Exclude control characters.
389: */
1.1 schwarze 390:
1.70 schwarze 391: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 392: mandoc_vmsg(c == 0x00 || c == 0x04 ||
393: c > 0x0a ? MANDOCERR_CHAR_BAD :
394: MANDOCERR_CHAR_UNSUPP,
395: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 396: i++;
1.6 schwarze 397: ln.buf[pos++] = '?';
1.1 schwarze 398: continue;
399: }
400:
401: /* Trailing backslash = a plain char. */
402:
1.71 schwarze 403: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 schwarze 404: ln.buf[pos++] = blk.buf[i++];
405: continue;
406: }
407:
408: /*
409: * Found escape and at least one other character.
410: * When it's a newline character, skip it.
411: * When there is a carriage return in between,
412: * skip that one as well.
413: */
414:
1.71 schwarze 415: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 schwarze 416: '\n' == blk.buf[i + 2])
417: ++i;
418: if ('\n' == blk.buf[i + 1]) {
419: i += 2;
420: ++lnn;
421: continue;
422: }
423:
424: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
425: i += 2;
426: /* Comment, skip to end of line */
1.71 schwarze 427: for (; i < blk.sz; ++i) {
1.1 schwarze 428: if ('\n' == blk.buf[i]) {
429: ++i;
430: ++lnn;
431: break;
432: }
433: }
434:
435: /* Backout trailing whitespaces */
436: for (; pos > 0; --pos) {
437: if (ln.buf[pos - 1] != ' ')
438: break;
439: if (pos > 2 && ln.buf[pos - 2] == '\\')
440: break;
441: }
442: break;
443: }
444:
1.13 schwarze 445: /* Catch escaped bogus characters. */
446:
447: c = (unsigned char) blk.buf[i+1];
448:
1.25 schwarze 449: if ( ! (isascii(c) &&
450: (isgraph(c) || isblank(c)))) {
1.89 schwarze 451: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.56 schwarze 452: curp->line, pos, "0x%x", c);
1.13 schwarze 453: i += 2;
454: ln.buf[pos++] = '?';
455: continue;
456: }
457:
1.1 schwarze 458: /* Some other escape sequence, copy & cont. */
459:
460: ln.buf[pos++] = blk.buf[i++];
461: ln.buf[pos++] = blk.buf[i++];
462: }
463:
1.71 schwarze 464: if (pos >= ln.sz)
1.1 schwarze 465: resize_buf(&ln, 256);
466:
467: ln.buf[pos] = '\0';
468:
469: /*
470: * A significant amount of complexity is contained by
471: * the roff preprocessor. It's line-oriented but can be
472: * expressed on one line, so we need at times to
473: * readjust our starting point and re-run it. The roff
474: * preprocessor can also readjust the buffers with new
475: * data, so we pass them in wholesale.
476: */
477:
478: of = 0;
479:
1.4 schwarze 480: /*
481: * Maintain a lookaside buffer of all parsed lines. We
482: * only do this if mparse_keep() has been invoked (the
483: * buffer may be accessed with mparse_getkeep()).
484: */
485:
486: if (curp->secondary) {
1.25 schwarze 487: curp->secondary->buf = mandoc_realloc(
488: curp->secondary->buf,
489: curp->secondary->sz + pos + 2);
490: memcpy(curp->secondary->buf +
491: curp->secondary->sz,
492: ln.buf, pos);
1.4 schwarze 493: curp->secondary->sz += pos;
494: curp->secondary->buf
495: [curp->secondary->sz] = '\n';
496: curp->secondary->sz++;
497: curp->secondary->buf
498: [curp->secondary->sz] = '\0';
499: }
1.1 schwarze 500: rerun:
1.72 schwarze 501: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 502:
503: switch (rr) {
1.25 schwarze 504: case ROFF_REPARSE:
1.1 schwarze 505: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.71 schwarze 506: mparse_buf_r(curp, ln, of, 0);
1.1 schwarze 507: else
508: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 509: curp->line, pos, NULL);
1.1 schwarze 510: pos = 0;
511: continue;
1.25 schwarze 512: case ROFF_APPEND:
1.71 schwarze 513: pos = strlen(ln.buf);
1.1 schwarze 514: continue;
1.25 schwarze 515: case ROFF_RERUN:
1.1 schwarze 516: goto rerun;
1.25 schwarze 517: case ROFF_IGN:
1.1 schwarze 518: pos = 0;
519: continue;
1.25 schwarze 520: case ROFF_SO:
1.71 schwarze 521: if ( ! (curp->options & MPARSE_SO) &&
522: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 523: curp->sodest = mandoc_strdup(ln.buf + of);
524: free(ln.buf);
525: return;
526: }
1.4 schwarze 527: /*
528: * We remove `so' clauses from our lookaside
529: * buffer because we're going to descend into
530: * the file recursively.
531: */
1.25 schwarze 532: if (curp->secondary)
1.4 schwarze 533: curp->secondary->sz -= pos + 1;
1.88 schwarze 534: save_file = curp->file;
1.85 schwarze 535: save_child = curp->child;
536: if (mparse_open(curp, &fd, ln.buf + of) ==
1.88 schwarze 537: MANDOCLEVEL_OK) {
1.85 schwarze 538: mparse_readfd(curp, fd, ln.buf + of);
1.88 schwarze 539: curp->file = save_file;
540: } else {
541: curp->file = save_file;
1.30 schwarze 542: mandoc_vmsg(MANDOCERR_SO_FAIL,
543: curp, curp->line, pos,
544: ".so %s", ln.buf + of);
1.85 schwarze 545: ln.sz = mandoc_asprintf(&cp,
546: ".sp\nSee the file %s.\n.sp",
547: ln.buf + of);
548: free(ln.buf);
549: ln.buf = cp;
550: of = 0;
551: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 552: }
1.85 schwarze 553: curp->child = save_child;
1.1 schwarze 554: pos = 0;
555: continue;
556: default:
557: break;
558: }
559:
560: /*
561: * If input parsers have not been allocated, do so now.
1.2 schwarze 562: * We keep these instanced between parsers, but set them
1.1 schwarze 563: * locally per parse routine since we can use different
564: * parsers with each one.
565: */
566:
567: if ( ! (curp->man || curp->mdoc))
1.60 schwarze 568: choose_parser(curp);
1.1 schwarze 569:
1.25 schwarze 570: /*
1.60 schwarze 571: * Lastly, push down into the parsers themselves.
1.1 schwarze 572: * If libroff returns ROFF_TBL, then add it to the
573: * currently open parse. Since we only get here if
574: * there does exist data (see tbl_data.c), we're
575: * guaranteed that something's been allocated.
576: * Do the same for ROFF_EQN.
577: */
578:
1.76 schwarze 579: if (rr == ROFF_TBL) {
580: while ((span = roff_span(curp->roff)) != NULL)
581: if (curp->man == NULL)
582: mdoc_addspan(curp->mdoc, span);
583: else
584: man_addspan(curp->man, span);
585: } else if (rr == ROFF_EQN) {
586: if (curp->man == NULL)
587: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
588: else
589: man_addeqn(curp->man, roff_eqn(curp->roff));
590: } else if ((curp->man == NULL ?
591: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
592: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
593: break;
1.1 schwarze 594:
595: /* Temporary buffers typically are not full. */
596:
597: if (0 == start && '\0' == blk.buf[i])
598: break;
599:
600: /* Start the next input line. */
601:
602: pos = 0;
603: }
604:
605: free(ln.buf);
606: }
607:
608: static int
1.18 schwarze 609: read_whole_file(struct mparse *curp, const char *file, int fd,
610: struct buf *fb, int *with_mmap)
1.1 schwarze 611: {
612: struct stat st;
613: size_t off;
614: ssize_t ssz;
615:
616: if (-1 == fstat(fd, &st)) {
1.82 schwarze 617: perror(file);
618: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 619: }
620:
621: /*
622: * If we're a regular file, try just reading in the whole entry
623: * via mmap(). This is faster than reading it into blocks, and
624: * since each file is only a few bytes to begin with, I'm not
625: * concerned that this is going to tank any machines.
626: */
627:
628: if (S_ISREG(st.st_mode)) {
629: if (st.st_size >= (1U << 31)) {
1.86 schwarze 630: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 schwarze 631: return(0);
632: }
633: *with_mmap = 1;
634: fb->sz = (size_t)st.st_size;
1.15 schwarze 635: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 636: if (fb->buf != MAP_FAILED)
637: return(1);
638: }
639:
640: /*
641: * If this isn't a regular file (like, say, stdin), then we must
642: * go the old way and just read things in bit by bit.
643: */
644:
645: *with_mmap = 0;
646: off = 0;
647: fb->sz = 0;
648: fb->buf = NULL;
649: for (;;) {
650: if (off == fb->sz) {
651: if (fb->sz == (1U << 31)) {
1.86 schwarze 652: mandoc_msg(MANDOCERR_TOOLARGE, curp,
653: 0, 0, NULL);
1.1 schwarze 654: break;
655: }
656: resize_buf(fb, 65536);
657: }
658: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
659: if (ssz == 0) {
660: fb->sz = off;
661: return(1);
662: }
663: if (ssz == -1) {
1.82 schwarze 664: perror(file);
665: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 666: }
667: off += (size_t)ssz;
668: }
669:
670: free(fb->buf);
671: fb->buf = NULL;
672: return(0);
673: }
674:
675: static void
676: mparse_end(struct mparse *curp)
677: {
678:
1.50 schwarze 679: if (curp->mdoc == NULL &&
680: curp->man == NULL &&
681: curp->sodest == NULL) {
682: if (curp->options & MPARSE_MDOC)
683: curp->mdoc = curp->pmdoc;
684: else {
685: if (curp->pman == NULL)
1.81 schwarze 686: curp->pman = man_alloc(
687: curp->roff, curp, curp->defos,
1.50 schwarze 688: curp->options & MPARSE_QUICK ? 1 : 0);
689: curp->man = curp->pman;
690: }
691: }
1.86 schwarze 692: if (curp->mdoc)
693: mdoc_endparse(curp->mdoc);
694: if (curp->man)
695: man_endparse(curp->man);
1.1 schwarze 696: roff_endparse(curp->roff);
697: }
698:
1.15 schwarze 699: static void
700: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 701: {
1.61 schwarze 702: struct buf *svprimary;
1.1 schwarze 703: const char *svfile;
1.71 schwarze 704: size_t offset;
1.14 schwarze 705: static int recursion_depth;
706:
707: if (64 < recursion_depth) {
708: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 709: return;
1.14 schwarze 710: }
1.1 schwarze 711:
1.15 schwarze 712: /* Line number is per-file. */
713: svfile = curp->file;
714: curp->file = file;
1.61 schwarze 715: svprimary = curp->primary;
1.59 schwarze 716: curp->primary = &blk;
1.15 schwarze 717: curp->line = 1;
718: recursion_depth++;
719:
1.70 schwarze 720: /* Skip an UTF-8 byte order mark. */
721: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
722: (unsigned char)blk.buf[0] == 0xef &&
723: (unsigned char)blk.buf[1] == 0xbb &&
724: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 725: offset = 3;
1.70 schwarze 726: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 727: } else
728: offset = 0;
1.70 schwarze 729:
1.71 schwarze 730: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 731:
1.86 schwarze 732: if (--recursion_depth == 0)
1.15 schwarze 733: mparse_end(curp);
734:
1.61 schwarze 735: curp->primary = svprimary;
1.15 schwarze 736: curp->file = svfile;
737: }
738:
1.74 schwarze 739: /*
740: * Read the whole file into memory and call the parsers.
741: * Called recursively when an .so request is encountered.
742: */
1.15 schwarze 743: enum mandoclevel
744: mparse_readfd(struct mparse *curp, int fd, const char *file)
745: {
746: struct buf blk;
747: int with_mmap;
1.70 schwarze 748: int save_filenc;
1.1 schwarze 749:
1.67 schwarze 750: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 751: save_filenc = curp->filenc;
752: curp->filenc = curp->options &
753: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 754: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 755: curp->filenc = save_filenc;
1.67 schwarze 756: if (with_mmap)
757: munmap(blk.buf, blk.sz);
758: else
759: free(blk.buf);
760: }
1.1 schwarze 761:
1.74 schwarze 762: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 schwarze 763: perror(file);
1.67 schwarze 764:
1.74 schwarze 765: mparse_wait(curp);
1.1 schwarze 766: return(curp->file_status);
1.58 schwarze 767: }
768:
769: enum mandoclevel
1.73 schwarze 770: mparse_open(struct mparse *curp, int *fd, const char *file)
1.58 schwarze 771: {
772: int pfd[2];
1.74 schwarze 773: int save_errno;
1.58 schwarze 774: char *cp;
775:
776: curp->file = file;
1.74 schwarze 777:
778: /* Unless zipped, try to just open the file. */
779:
1.58 schwarze 780: if ((cp = strrchr(file, '.')) == NULL ||
781: strcmp(cp + 1, "gz")) {
1.73 schwarze 782: curp->child = 0;
1.74 schwarze 783: if ((*fd = open(file, O_RDONLY)) != -1)
784: return(MANDOCLEVEL_OK);
785:
786: /* Open failed; try to append ".gz". */
787:
788: mandoc_asprintf(&cp, "%s.gz", file);
789: file = cp;
790: } else
791: cp = NULL;
792:
793: /* Before forking, make sure the file can be read. */
794:
795: save_errno = errno;
796: if (access(file, R_OK) == -1) {
797: if (cp != NULL)
798: errno = save_errno;
1.83 schwarze 799: free(cp);
800: *fd = -1;
801: curp->child = 0;
802: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
803: return(MANDOCLEVEL_ERROR);
1.58 schwarze 804: }
805:
1.74 schwarze 806: /* Run gunzip(1). */
807:
1.58 schwarze 808: if (pipe(pfd) == -1) {
1.82 schwarze 809: perror("pipe");
810: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 811: }
812:
1.73 schwarze 813: switch (curp->child = fork()) {
1.58 schwarze 814: case -1:
1.82 schwarze 815: perror("fork");
816: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 817: case 0:
818: close(pfd[0]);
819: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.82 schwarze 820: perror("dup");
821: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 822: }
823: execlp("gunzip", "gunzip", "-c", file, NULL);
1.82 schwarze 824: perror("exec");
825: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 826: default:
827: close(pfd[1]);
828: *fd = pfd[0];
829: return(MANDOCLEVEL_OK);
830: }
831: }
832:
833: enum mandoclevel
1.73 schwarze 834: mparse_wait(struct mparse *curp)
1.58 schwarze 835: {
836: int status;
837:
1.73 schwarze 838: if (curp->child == 0)
839: return(MANDOCLEVEL_OK);
840:
841: if (waitpid(curp->child, &status, 0) == -1) {
1.82 schwarze 842: perror("wait");
843: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 844: }
845: if (WIFSIGNALED(status)) {
1.83 schwarze 846: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
847: "gunzip died from signal %d", WTERMSIG(status));
848: return(MANDOCLEVEL_ERROR);
1.58 schwarze 849: }
850: if (WEXITSTATUS(status)) {
1.83 schwarze 851: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
852: "gunzip failed with code %d", WEXITSTATUS(status));
853: return(MANDOCLEVEL_ERROR);
1.58 schwarze 854: }
855: return(MANDOCLEVEL_OK);
1.1 schwarze 856: }
857:
858: struct mparse *
1.69 schwarze 859: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
860: const struct mchars *mchars, const char *defos)
1.1 schwarze 861: {
862: struct mparse *curp;
863:
864: curp = mandoc_calloc(1, sizeof(struct mparse));
865:
1.22 schwarze 866: curp->options = options;
1.1 schwarze 867: curp->wlevel = wlevel;
868: curp->mmsg = mmsg;
1.7 schwarze 869: curp->defos = defos;
1.1 schwarze 870:
1.69 schwarze 871: curp->mchars = mchars;
872: curp->roff = roff_alloc(curp, curp->mchars, options);
1.50 schwarze 873: if (curp->options & MPARSE_MDOC)
874: curp->pmdoc = mdoc_alloc(
875: curp->roff, curp, curp->defos,
876: curp->options & MPARSE_QUICK ? 1 : 0);
877: if (curp->options & MPARSE_MAN)
1.81 schwarze 878: curp->pman = man_alloc(
879: curp->roff, curp, curp->defos,
1.50 schwarze 880: curp->options & MPARSE_QUICK ? 1 : 0);
881:
1.1 schwarze 882: return(curp);
883: }
884:
885: void
886: mparse_reset(struct mparse *curp)
887: {
888:
889: roff_reset(curp->roff);
890:
891: if (curp->mdoc)
892: mdoc_reset(curp->mdoc);
893: if (curp->man)
894: man_reset(curp->man);
1.4 schwarze 895: if (curp->secondary)
896: curp->secondary->sz = 0;
1.1 schwarze 897:
898: curp->file_status = MANDOCLEVEL_OK;
899: curp->mdoc = NULL;
900: curp->man = NULL;
1.23 schwarze 901:
902: free(curp->sodest);
903: curp->sodest = NULL;
1.1 schwarze 904: }
905:
906: void
907: mparse_free(struct mparse *curp)
908: {
909:
910: if (curp->pmdoc)
911: mdoc_free(curp->pmdoc);
912: if (curp->pman)
913: man_free(curp->pman);
914: if (curp->roff)
915: roff_free(curp->roff);
1.4 schwarze 916: if (curp->secondary)
917: free(curp->secondary->buf);
1.1 schwarze 918:
1.4 schwarze 919: free(curp->secondary);
1.23 schwarze 920: free(curp->sodest);
1.1 schwarze 921: free(curp);
922: }
923:
924: void
1.23 schwarze 925: mparse_result(struct mparse *curp,
926: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 schwarze 927: {
928:
1.23 schwarze 929: if (sodest && NULL != (*sodest = curp->sodest)) {
930: *mdoc = NULL;
931: *man = NULL;
932: return;
933: }
1.1 schwarze 934: if (mdoc)
935: *mdoc = curp->mdoc;
936: if (man)
937: *man = curp->man;
938: }
939:
940: void
941: mandoc_vmsg(enum mandocerr t, struct mparse *m,
942: int ln, int pos, const char *fmt, ...)
943: {
944: char buf[256];
945: va_list ap;
946:
947: va_start(ap, fmt);
1.26 schwarze 948: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 949: va_end(ap);
950:
951: mandoc_msg(t, m, ln, pos, buf);
952: }
953:
954: void
1.25 schwarze 955: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 956: int ln, int col, const char *msg)
957: {
958: enum mandoclevel level;
959:
1.87 schwarze 960: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 961: while (er < mandoclimits[level])
962: level--;
963:
1.83 schwarze 964: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 965: return;
966:
967: if (m->mmsg)
968: (*m->mmsg)(er, level, m->file, ln, col, msg);
969:
970: if (m->file_status < level)
971: m->file_status = level;
972: }
973:
974: const char *
975: mparse_strerror(enum mandocerr er)
976: {
977:
978: return(mandocerrs[er]);
979: }
980:
981: const char *
982: mparse_strlevel(enum mandoclevel lvl)
983: {
984: return(mandoclevels[lvl]);
1.4 schwarze 985: }
986:
987: void
988: mparse_keep(struct mparse *p)
989: {
990:
991: assert(NULL == p->secondary);
992: p->secondary = mandoc_calloc(1, sizeof(struct buf));
993: }
994:
995: const char *
996: mparse_getkeep(const struct mparse *p)
997: {
998:
999: assert(p->secondary);
1000: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 schwarze 1001: }
![[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