Annotation of src/usr.bin/mandoc/read.c, Revision 1.97
1.97 ! schwarze 1: /* $OpenBSD: read.c,v 1.96 2015/02/04 16:38:31 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",
1.40 schwarze 134: "empty argument, using 0n",
1.1 schwarze 135: "argument count wrong",
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 \"\"",
1.38 schwarze 140: "empty head in list item",
141: "empty list item",
1.39 schwarze 142: "missing font type, using \\fR",
143: "unknown font type, using \\fR",
1.79 schwarze 144: "nothing follows prefix",
1.97 ! schwarze 145: "empty reference block",
1.38 schwarze 146: "missing -std argument, adding it",
1.66 schwarze 147: "missing eqn box, using \"\"",
1.1 schwarze 148:
149: /* related to bad macro arguments */
1.42 schwarze 150: "unterminated quoted argument",
1.1 schwarze 151: "duplicate argument",
1.54 schwarze 152: "skipping duplicate argument",
1.41 schwarze 153: "skipping duplicate display type",
154: "skipping duplicate list type",
1.54 schwarze 155: "skipping -width argument",
1.1 schwarze 156: "unknown AT&T UNIX version",
1.64 schwarze 157: "comma in function argument",
1.65 schwarze 158: "parenthesis in function name",
1.45 schwarze 159: "invalid content in Rs block",
1.41 schwarze 160: "invalid Boolean argument",
161: "unknown font, skipping request",
1.1 schwarze 162:
163: /* related to plain text */
1.42 schwarze 164: "blank line in fill mode, using .sp",
165: "tab in filled text",
166: "whitespace at end of input line",
1.1 schwarze 167: "bad comment style",
1.42 schwarze 168: "invalid escape sequence",
169: "undefined string, using \"\"",
1.3 schwarze 170:
1.93 schwarze 171: /* related to tables */
172: "tbl line starts with span",
173: "tbl column starts with span",
174: "skipping vertical bar in tbl layout",
175:
1.1 schwarze 176: "generic error",
1.3 schwarze 177:
1.1 schwarze 178: /* related to tables */
1.91 schwarze 179: "non-alphabetic character in tbl options",
180: "skipping unknown tbl option",
181: "missing tbl option argument",
182: "wrong tbl option argument size",
1.93 schwarze 183: "empty tbl layout",
184: "invalid character in tbl layout",
185: "unmatched parenthesis in tbl layout",
1.94 schwarze 186: "tbl without any data cells",
187: "ignoring data in spanned tbl cell",
188: "ignoring extra tbl data cells",
189: "data block open at end of tbl",
1.1 schwarze 190:
1.46 schwarze 191: /* related to document structure and macros */
1.83 schwarze 192: NULL,
1.1 schwarze 193: "input stack limit exceeded, infinite loop?",
194: "skipping bad character",
1.46 schwarze 195: "skipping unknown macro",
1.87 schwarze 196: "skipping insecure request",
1.51 schwarze 197: "skipping item outside list",
1.46 schwarze 198: "skipping column outside column list",
199: "skipping end of block that is not open",
1.90 schwarze 200: "fewer RS blocks open, skipping",
1.46 schwarze 201: "inserting missing end of block",
202: "appending missing end of block",
203:
204: /* related to request and macro arguments */
1.1 schwarze 205: "escaped character not allowed in a name",
206: "argument count wrong",
1.75 schwarze 207: "NOT IMPLEMENTED: Bd -file",
1.49 schwarze 208: "missing list type, using -item",
1.48 schwarze 209: "missing manual name, using \"\"",
1.49 schwarze 210: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 211: "unknown standard specifier",
1.49 schwarze 212: "skipping request without numeric argument",
1.84 schwarze 213: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 214: ".so request failed",
1.39 schwarze 215: "skipping all arguments",
216: "skipping excess arguments",
1.68 schwarze 217: "divide by zero",
1.87 schwarze 218:
219: "unsupported feature",
220: "input too large",
1.89 schwarze 221: "unsupported control character",
1.87 schwarze 222: "unsupported roff request",
1.94 schwarze 223: "eqn delim option in tbl",
1.93 schwarze 224: "unsupported tbl layout modifier",
1.87 schwarze 225: "ignoring macro in table",
1.1 schwarze 226: };
227:
228: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
229: "SUCCESS",
230: "RESERVED",
231: "WARNING",
232: "ERROR",
1.87 schwarze 233: "UNSUPP",
1.1 schwarze 234: "BADARG",
235: "SYSERR"
236: };
237:
1.25 schwarze 238:
1.1 schwarze 239: static void
240: resize_buf(struct buf *buf, size_t initial)
241: {
242:
243: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
244: buf->buf = mandoc_realloc(buf->buf, buf->sz);
245: }
246:
247: static void
1.60 schwarze 248: choose_parser(struct mparse *curp)
1.1 schwarze 249: {
1.59 schwarze 250: char *cp, *ep;
251: int format;
1.1 schwarze 252:
1.59 schwarze 253: /*
254: * If neither command line arguments -mdoc or -man select
255: * a parser nor the roff parser found a .Dd or .TH macro
256: * yet, look ahead in the main input buffer.
257: */
258:
259: if ((format = roff_getformat(curp->roff)) == 0) {
260: cp = curp->primary->buf;
261: ep = cp + curp->primary->sz;
262: while (cp < ep) {
1.61 schwarze 263: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 264: cp++;
265: if (cp[0] == 'D' && cp[1] == 'd') {
266: format = MPARSE_MDOC;
267: break;
268: }
269: if (cp[0] == 'T' && cp[1] == 'H') {
270: format = MPARSE_MAN;
271: break;
272: }
273: }
274: cp = memchr(cp, '\n', ep - cp);
275: if (cp == NULL)
276: break;
277: cp++;
278: }
1.1 schwarze 279: }
280:
1.59 schwarze 281: if (format == MPARSE_MDOC) {
1.25 schwarze 282: if (NULL == curp->pmdoc)
1.22 schwarze 283: curp->pmdoc = mdoc_alloc(
284: curp->roff, curp, curp->defos,
285: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 schwarze 286: assert(curp->pmdoc);
287: curp->mdoc = curp->pmdoc;
288: return;
1.25 schwarze 289: }
1.1 schwarze 290:
1.59 schwarze 291: /* Fall back to man(7) as a last resort. */
292:
1.25 schwarze 293: if (NULL == curp->pman)
1.81 schwarze 294: curp->pman = man_alloc(
295: curp->roff, curp, curp->defos,
1.22 schwarze 296: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 schwarze 297: assert(curp->pman);
298: curp->man = curp->pman;
299: }
300:
301: /*
1.71 schwarze 302: * Main parse routine for a buffer.
303: * It assumes encoding and line numbering are already set up.
304: * It can recurse directly (for invocations of user-defined
305: * macros, inline equations, and input line traps)
306: * and indirectly (for .so file inclusion).
1.1 schwarze 307: */
308: static void
1.71 schwarze 309: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 310: {
311: const struct tbl_span *span;
312: struct buf ln;
1.88 schwarze 313: const char *save_file;
1.85 schwarze 314: char *cp;
1.71 schwarze 315: size_t pos; /* byte number in the ln buffer */
1.1 schwarze 316: enum rofferr rr;
1.76 schwarze 317: int of;
1.1 schwarze 318: int lnn; /* line number in the real file */
1.85 schwarze 319: int fd;
320: pid_t save_child;
1.1 schwarze 321: unsigned char c;
322:
1.71 schwarze 323: memset(&ln, 0, sizeof(ln));
1.1 schwarze 324:
1.25 schwarze 325: lnn = curp->line;
326: pos = 0;
1.1 schwarze 327:
1.71 schwarze 328: while (i < blk.sz) {
1.1 schwarze 329: if (0 == pos && '\0' == blk.buf[i])
330: break;
331:
332: if (start) {
333: curp->line = lnn;
334: curp->reparse_count = 0;
1.70 schwarze 335:
336: if (lnn < 3 &&
337: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 338: curp->filenc & MPARSE_LATIN1)
339: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 340: }
341:
1.71 schwarze 342: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 343:
344: /*
345: * When finding an unescaped newline character,
346: * leave the character loop to process the line.
347: * Skip a preceding carriage return, if any.
348: */
349:
1.71 schwarze 350: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 351: '\n' == blk.buf[i + 1])
352: ++i;
353: if ('\n' == blk.buf[i]) {
354: ++i;
355: ++lnn;
356: break;
357: }
358:
1.13 schwarze 359: /*
1.70 schwarze 360: * Make sure we have space for the worst
361: * case of 11 bytes: "\\[u10ffff]\0"
1.13 schwarze 362: */
363:
1.71 schwarze 364: if (pos + 11 > ln.sz)
1.13 schwarze 365: resize_buf(&ln, 256);
366:
1.25 schwarze 367: /*
1.70 schwarze 368: * Encode 8-bit input.
1.1 schwarze 369: */
370:
1.70 schwarze 371: c = blk.buf[i];
372: if (c & 0x80) {
1.71 schwarze 373: if ( ! (curp->filenc && preconv_encode(
374: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 375: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
376: curp->line, pos, "0x%x", c);
1.70 schwarze 377: ln.buf[pos++] = '?';
378: i++;
379: }
380: continue;
381: }
382:
383: /*
384: * Exclude control characters.
385: */
1.1 schwarze 386:
1.70 schwarze 387: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 388: mandoc_vmsg(c == 0x00 || c == 0x04 ||
389: c > 0x0a ? MANDOCERR_CHAR_BAD :
390: MANDOCERR_CHAR_UNSUPP,
391: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 392: i++;
1.6 schwarze 393: ln.buf[pos++] = '?';
1.1 schwarze 394: continue;
395: }
396:
397: /* Trailing backslash = a plain char. */
398:
1.71 schwarze 399: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 schwarze 400: ln.buf[pos++] = blk.buf[i++];
401: continue;
402: }
403:
404: /*
405: * Found escape and at least one other character.
406: * When it's a newline character, skip it.
407: * When there is a carriage return in between,
408: * skip that one as well.
409: */
410:
1.71 schwarze 411: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 schwarze 412: '\n' == blk.buf[i + 2])
413: ++i;
414: if ('\n' == blk.buf[i + 1]) {
415: i += 2;
416: ++lnn;
417: continue;
418: }
419:
420: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
421: i += 2;
422: /* Comment, skip to end of line */
1.71 schwarze 423: for (; i < blk.sz; ++i) {
1.1 schwarze 424: if ('\n' == blk.buf[i]) {
425: ++i;
426: ++lnn;
427: break;
428: }
429: }
430:
431: /* Backout trailing whitespaces */
432: for (; pos > 0; --pos) {
433: if (ln.buf[pos - 1] != ' ')
434: break;
435: if (pos > 2 && ln.buf[pos - 2] == '\\')
436: break;
437: }
438: break;
439: }
440:
1.13 schwarze 441: /* Catch escaped bogus characters. */
442:
443: c = (unsigned char) blk.buf[i+1];
444:
1.25 schwarze 445: if ( ! (isascii(c) &&
446: (isgraph(c) || isblank(c)))) {
1.89 schwarze 447: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.56 schwarze 448: curp->line, pos, "0x%x", c);
1.13 schwarze 449: i += 2;
450: ln.buf[pos++] = '?';
451: continue;
452: }
453:
1.1 schwarze 454: /* Some other escape sequence, copy & cont. */
455:
456: ln.buf[pos++] = blk.buf[i++];
457: ln.buf[pos++] = blk.buf[i++];
458: }
459:
1.71 schwarze 460: if (pos >= ln.sz)
1.1 schwarze 461: resize_buf(&ln, 256);
462:
463: ln.buf[pos] = '\0';
464:
465: /*
466: * A significant amount of complexity is contained by
467: * the roff preprocessor. It's line-oriented but can be
468: * expressed on one line, so we need at times to
469: * readjust our starting point and re-run it. The roff
470: * preprocessor can also readjust the buffers with new
471: * data, so we pass them in wholesale.
472: */
473:
474: of = 0;
475:
1.4 schwarze 476: /*
477: * Maintain a lookaside buffer of all parsed lines. We
478: * only do this if mparse_keep() has been invoked (the
479: * buffer may be accessed with mparse_getkeep()).
480: */
481:
482: if (curp->secondary) {
1.25 schwarze 483: curp->secondary->buf = mandoc_realloc(
484: curp->secondary->buf,
485: curp->secondary->sz + pos + 2);
486: memcpy(curp->secondary->buf +
487: curp->secondary->sz,
488: ln.buf, pos);
1.4 schwarze 489: curp->secondary->sz += pos;
490: curp->secondary->buf
491: [curp->secondary->sz] = '\n';
492: curp->secondary->sz++;
493: curp->secondary->buf
494: [curp->secondary->sz] = '\0';
495: }
1.1 schwarze 496: rerun:
1.72 schwarze 497: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 498:
499: switch (rr) {
1.25 schwarze 500: case ROFF_REPARSE:
1.1 schwarze 501: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.71 schwarze 502: mparse_buf_r(curp, ln, of, 0);
1.1 schwarze 503: else
504: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 505: curp->line, pos, NULL);
1.1 schwarze 506: pos = 0;
507: continue;
1.25 schwarze 508: case ROFF_APPEND:
1.71 schwarze 509: pos = strlen(ln.buf);
1.1 schwarze 510: continue;
1.25 schwarze 511: case ROFF_RERUN:
1.1 schwarze 512: goto rerun;
1.25 schwarze 513: case ROFF_IGN:
1.1 schwarze 514: pos = 0;
515: continue;
1.25 schwarze 516: case ROFF_SO:
1.71 schwarze 517: if ( ! (curp->options & MPARSE_SO) &&
518: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 519: curp->sodest = mandoc_strdup(ln.buf + of);
520: free(ln.buf);
521: return;
522: }
1.4 schwarze 523: /*
524: * We remove `so' clauses from our lookaside
525: * buffer because we're going to descend into
526: * the file recursively.
527: */
1.25 schwarze 528: if (curp->secondary)
1.4 schwarze 529: curp->secondary->sz -= pos + 1;
1.88 schwarze 530: save_file = curp->file;
1.85 schwarze 531: save_child = curp->child;
532: if (mparse_open(curp, &fd, ln.buf + of) ==
1.88 schwarze 533: MANDOCLEVEL_OK) {
1.85 schwarze 534: mparse_readfd(curp, fd, ln.buf + of);
1.88 schwarze 535: curp->file = save_file;
536: } else {
537: curp->file = save_file;
1.30 schwarze 538: mandoc_vmsg(MANDOCERR_SO_FAIL,
539: curp, curp->line, pos,
540: ".so %s", ln.buf + of);
1.85 schwarze 541: ln.sz = mandoc_asprintf(&cp,
542: ".sp\nSee the file %s.\n.sp",
543: ln.buf + of);
544: free(ln.buf);
545: ln.buf = cp;
546: of = 0;
547: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 548: }
1.85 schwarze 549: curp->child = save_child;
1.1 schwarze 550: pos = 0;
551: continue;
552: default:
553: break;
554: }
555:
556: /*
557: * If input parsers have not been allocated, do so now.
1.2 schwarze 558: * We keep these instanced between parsers, but set them
1.1 schwarze 559: * locally per parse routine since we can use different
560: * parsers with each one.
561: */
562:
563: if ( ! (curp->man || curp->mdoc))
1.60 schwarze 564: choose_parser(curp);
1.1 schwarze 565:
1.25 schwarze 566: /*
1.60 schwarze 567: * Lastly, push down into the parsers themselves.
1.1 schwarze 568: * If libroff returns ROFF_TBL, then add it to the
569: * currently open parse. Since we only get here if
570: * there does exist data (see tbl_data.c), we're
571: * guaranteed that something's been allocated.
572: * Do the same for ROFF_EQN.
573: */
574:
1.76 schwarze 575: if (rr == ROFF_TBL) {
576: while ((span = roff_span(curp->roff)) != NULL)
577: if (curp->man == NULL)
578: mdoc_addspan(curp->mdoc, span);
579: else
580: man_addspan(curp->man, span);
581: } else if (rr == ROFF_EQN) {
582: if (curp->man == NULL)
583: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
584: else
585: man_addeqn(curp->man, roff_eqn(curp->roff));
586: } else if ((curp->man == NULL ?
587: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
588: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
589: break;
1.1 schwarze 590:
591: /* Temporary buffers typically are not full. */
592:
593: if (0 == start && '\0' == blk.buf[i])
594: break;
595:
596: /* Start the next input line. */
597:
598: pos = 0;
599: }
600:
601: free(ln.buf);
602: }
603:
604: static int
1.18 schwarze 605: read_whole_file(struct mparse *curp, const char *file, int fd,
606: struct buf *fb, int *with_mmap)
1.1 schwarze 607: {
608: struct stat st;
609: size_t off;
610: ssize_t ssz;
611:
612: if (-1 == fstat(fd, &st)) {
1.82 schwarze 613: perror(file);
614: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 615: }
616:
617: /*
618: * If we're a regular file, try just reading in the whole entry
619: * via mmap(). This is faster than reading it into blocks, and
620: * since each file is only a few bytes to begin with, I'm not
621: * concerned that this is going to tank any machines.
622: */
623:
624: if (S_ISREG(st.st_mode)) {
625: if (st.st_size >= (1U << 31)) {
1.86 schwarze 626: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 schwarze 627: return(0);
628: }
629: *with_mmap = 1;
630: fb->sz = (size_t)st.st_size;
1.15 schwarze 631: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 632: if (fb->buf != MAP_FAILED)
633: return(1);
634: }
635:
636: /*
637: * If this isn't a regular file (like, say, stdin), then we must
638: * go the old way and just read things in bit by bit.
639: */
640:
641: *with_mmap = 0;
642: off = 0;
643: fb->sz = 0;
644: fb->buf = NULL;
645: for (;;) {
646: if (off == fb->sz) {
647: if (fb->sz == (1U << 31)) {
1.86 schwarze 648: mandoc_msg(MANDOCERR_TOOLARGE, curp,
649: 0, 0, NULL);
1.1 schwarze 650: break;
651: }
652: resize_buf(fb, 65536);
653: }
654: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
655: if (ssz == 0) {
656: fb->sz = off;
657: return(1);
658: }
659: if (ssz == -1) {
1.82 schwarze 660: perror(file);
661: exit((int)MANDOCLEVEL_SYSERR);
1.1 schwarze 662: }
663: off += (size_t)ssz;
664: }
665:
666: free(fb->buf);
667: fb->buf = NULL;
668: return(0);
669: }
670:
671: static void
672: mparse_end(struct mparse *curp)
673: {
674:
1.50 schwarze 675: if (curp->mdoc == NULL &&
676: curp->man == NULL &&
677: curp->sodest == NULL) {
678: if (curp->options & MPARSE_MDOC)
679: curp->mdoc = curp->pmdoc;
680: else {
681: if (curp->pman == NULL)
1.81 schwarze 682: curp->pman = man_alloc(
683: curp->roff, curp, curp->defos,
1.50 schwarze 684: curp->options & MPARSE_QUICK ? 1 : 0);
685: curp->man = curp->pman;
686: }
687: }
1.86 schwarze 688: if (curp->mdoc)
689: mdoc_endparse(curp->mdoc);
690: if (curp->man)
691: man_endparse(curp->man);
1.1 schwarze 692: roff_endparse(curp->roff);
693: }
694:
1.15 schwarze 695: static void
696: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 697: {
1.61 schwarze 698: struct buf *svprimary;
1.1 schwarze 699: const char *svfile;
1.71 schwarze 700: size_t offset;
1.14 schwarze 701: static int recursion_depth;
702:
703: if (64 < recursion_depth) {
704: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 705: return;
1.14 schwarze 706: }
1.1 schwarze 707:
1.15 schwarze 708: /* Line number is per-file. */
709: svfile = curp->file;
710: curp->file = file;
1.61 schwarze 711: svprimary = curp->primary;
1.59 schwarze 712: curp->primary = &blk;
1.15 schwarze 713: curp->line = 1;
714: recursion_depth++;
715:
1.70 schwarze 716: /* Skip an UTF-8 byte order mark. */
717: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
718: (unsigned char)blk.buf[0] == 0xef &&
719: (unsigned char)blk.buf[1] == 0xbb &&
720: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 721: offset = 3;
1.70 schwarze 722: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 723: } else
724: offset = 0;
1.70 schwarze 725:
1.71 schwarze 726: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 727:
1.86 schwarze 728: if (--recursion_depth == 0)
1.15 schwarze 729: mparse_end(curp);
730:
1.61 schwarze 731: curp->primary = svprimary;
1.15 schwarze 732: curp->file = svfile;
733: }
734:
1.74 schwarze 735: /*
736: * Read the whole file into memory and call the parsers.
737: * Called recursively when an .so request is encountered.
738: */
1.15 schwarze 739: enum mandoclevel
740: mparse_readfd(struct mparse *curp, int fd, const char *file)
741: {
742: struct buf blk;
743: int with_mmap;
1.70 schwarze 744: int save_filenc;
1.1 schwarze 745:
1.67 schwarze 746: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 747: save_filenc = curp->filenc;
748: curp->filenc = curp->options &
749: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 750: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 751: curp->filenc = save_filenc;
1.67 schwarze 752: if (with_mmap)
753: munmap(blk.buf, blk.sz);
754: else
755: free(blk.buf);
756: }
1.1 schwarze 757:
1.74 schwarze 758: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 schwarze 759: perror(file);
1.67 schwarze 760:
1.74 schwarze 761: mparse_wait(curp);
1.1 schwarze 762: return(curp->file_status);
1.58 schwarze 763: }
764:
765: enum mandoclevel
1.73 schwarze 766: mparse_open(struct mparse *curp, int *fd, const char *file)
1.58 schwarze 767: {
768: int pfd[2];
1.74 schwarze 769: int save_errno;
1.58 schwarze 770: char *cp;
771:
772: curp->file = file;
1.74 schwarze 773:
774: /* Unless zipped, try to just open the file. */
775:
1.58 schwarze 776: if ((cp = strrchr(file, '.')) == NULL ||
777: strcmp(cp + 1, "gz")) {
1.73 schwarze 778: curp->child = 0;
1.74 schwarze 779: if ((*fd = open(file, O_RDONLY)) != -1)
780: return(MANDOCLEVEL_OK);
781:
782: /* Open failed; try to append ".gz". */
783:
784: mandoc_asprintf(&cp, "%s.gz", file);
785: file = cp;
786: } else
787: cp = NULL;
788:
789: /* Before forking, make sure the file can be read. */
790:
791: save_errno = errno;
792: if (access(file, R_OK) == -1) {
793: if (cp != NULL)
794: errno = save_errno;
1.83 schwarze 795: free(cp);
796: *fd = -1;
797: curp->child = 0;
798: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
799: return(MANDOCLEVEL_ERROR);
1.58 schwarze 800: }
801:
1.74 schwarze 802: /* Run gunzip(1). */
803:
1.58 schwarze 804: if (pipe(pfd) == -1) {
1.82 schwarze 805: perror("pipe");
806: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 807: }
808:
1.73 schwarze 809: switch (curp->child = fork()) {
1.58 schwarze 810: case -1:
1.82 schwarze 811: perror("fork");
812: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 813: case 0:
814: close(pfd[0]);
815: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.82 schwarze 816: perror("dup");
817: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 818: }
819: execlp("gunzip", "gunzip", "-c", file, NULL);
1.82 schwarze 820: perror("exec");
821: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 822: default:
823: close(pfd[1]);
824: *fd = pfd[0];
825: return(MANDOCLEVEL_OK);
826: }
827: }
828:
829: enum mandoclevel
1.73 schwarze 830: mparse_wait(struct mparse *curp)
1.58 schwarze 831: {
832: int status;
833:
1.73 schwarze 834: if (curp->child == 0)
835: return(MANDOCLEVEL_OK);
836:
837: if (waitpid(curp->child, &status, 0) == -1) {
1.82 schwarze 838: perror("wait");
839: exit((int)MANDOCLEVEL_SYSERR);
1.58 schwarze 840: }
841: if (WIFSIGNALED(status)) {
1.83 schwarze 842: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
843: "gunzip died from signal %d", WTERMSIG(status));
844: return(MANDOCLEVEL_ERROR);
1.58 schwarze 845: }
846: if (WEXITSTATUS(status)) {
1.83 schwarze 847: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
848: "gunzip failed with code %d", WEXITSTATUS(status));
849: return(MANDOCLEVEL_ERROR);
1.58 schwarze 850: }
851: return(MANDOCLEVEL_OK);
1.1 schwarze 852: }
853:
854: struct mparse *
1.69 schwarze 855: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
856: const struct mchars *mchars, const char *defos)
1.1 schwarze 857: {
858: struct mparse *curp;
859:
860: curp = mandoc_calloc(1, sizeof(struct mparse));
861:
1.22 schwarze 862: curp->options = options;
1.1 schwarze 863: curp->wlevel = wlevel;
864: curp->mmsg = mmsg;
1.7 schwarze 865: curp->defos = defos;
1.1 schwarze 866:
1.69 schwarze 867: curp->mchars = mchars;
868: curp->roff = roff_alloc(curp, curp->mchars, options);
1.50 schwarze 869: if (curp->options & MPARSE_MDOC)
870: curp->pmdoc = mdoc_alloc(
871: curp->roff, curp, curp->defos,
872: curp->options & MPARSE_QUICK ? 1 : 0);
873: if (curp->options & MPARSE_MAN)
1.81 schwarze 874: curp->pman = man_alloc(
875: curp->roff, curp, curp->defos,
1.50 schwarze 876: curp->options & MPARSE_QUICK ? 1 : 0);
877:
1.1 schwarze 878: return(curp);
879: }
880:
881: void
882: mparse_reset(struct mparse *curp)
883: {
884:
885: roff_reset(curp->roff);
886:
887: if (curp->mdoc)
888: mdoc_reset(curp->mdoc);
889: if (curp->man)
890: man_reset(curp->man);
1.4 schwarze 891: if (curp->secondary)
892: curp->secondary->sz = 0;
1.1 schwarze 893:
894: curp->file_status = MANDOCLEVEL_OK;
895: curp->mdoc = NULL;
896: curp->man = NULL;
1.23 schwarze 897:
898: free(curp->sodest);
899: curp->sodest = NULL;
1.1 schwarze 900: }
901:
902: void
903: mparse_free(struct mparse *curp)
904: {
905:
906: if (curp->pmdoc)
907: mdoc_free(curp->pmdoc);
908: if (curp->pman)
909: man_free(curp->pman);
910: if (curp->roff)
911: roff_free(curp->roff);
1.4 schwarze 912: if (curp->secondary)
913: free(curp->secondary->buf);
1.1 schwarze 914:
1.4 schwarze 915: free(curp->secondary);
1.23 schwarze 916: free(curp->sodest);
1.1 schwarze 917: free(curp);
918: }
919:
920: void
1.23 schwarze 921: mparse_result(struct mparse *curp,
922: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 schwarze 923: {
924:
1.23 schwarze 925: if (sodest && NULL != (*sodest = curp->sodest)) {
926: *mdoc = NULL;
927: *man = NULL;
928: return;
929: }
1.1 schwarze 930: if (mdoc)
931: *mdoc = curp->mdoc;
932: if (man)
933: *man = curp->man;
934: }
935:
936: void
937: mandoc_vmsg(enum mandocerr t, struct mparse *m,
938: int ln, int pos, const char *fmt, ...)
939: {
940: char buf[256];
941: va_list ap;
942:
943: va_start(ap, fmt);
1.26 schwarze 944: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 945: va_end(ap);
946:
947: mandoc_msg(t, m, ln, pos, buf);
948: }
949:
950: void
1.25 schwarze 951: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 952: int ln, int col, const char *msg)
953: {
954: enum mandoclevel level;
955:
1.87 schwarze 956: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 957: while (er < mandoclimits[level])
958: level--;
959:
1.83 schwarze 960: if (level < m->wlevel && er != MANDOCERR_FILE)
1.1 schwarze 961: return;
962:
963: if (m->mmsg)
964: (*m->mmsg)(er, level, m->file, ln, col, msg);
965:
966: if (m->file_status < level)
967: m->file_status = level;
968: }
969:
970: const char *
971: mparse_strerror(enum mandocerr er)
972: {
973:
974: return(mandocerrs[er]);
975: }
976:
977: const char *
978: mparse_strlevel(enum mandoclevel lvl)
979: {
980: return(mandoclevels[lvl]);
1.4 schwarze 981: }
982:
983: void
984: mparse_keep(struct mparse *p)
985: {
986:
987: assert(NULL == p->secondary);
988: p->secondary = mandoc_calloc(1, sizeof(struct buf));
989: }
990:
991: const char *
992: mparse_getkeep(const struct mparse *p)
993: {
994:
995: assert(p->secondary);
996: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 schwarze 997: }
![[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