Annotation of mandoc/read.c, Revision 1.120
1.120 ! schwarze 1: /* $Id: read.c,v 1.119 2015/01/28 17:32:07 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.107 schwarze 4: * Copyright (c) 2010-2015 Ingo Schwarze <schwarze@openbsd.org>
1.42 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 kristaps 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: */
1.11 kristaps 19: #include "config.h"
20:
1.80 schwarze 21: #include <sys/types.h>
1.81 schwarze 22: #if HAVE_MMAP
1.82 schwarze 23: #include <sys/mman.h>
1.80 schwarze 24: #include <sys/stat.h>
1.15 kristaps 25: #endif
1.82 schwarze 26: #include <sys/wait.h>
1.1 kristaps 27:
28: #include <assert.h>
29: #include <ctype.h>
1.40 schwarze 30: #include <errno.h>
1.1 kristaps 31: #include <fcntl.h>
1.3 kristaps 32: #include <stdarg.h>
1.28 joerg 33: #include <stdint.h>
1.1 kristaps 34: #include <stdio.h>
35: #include <stdlib.h>
36: #include <string.h>
37: #include <unistd.h>
38:
39: #include "mandoc.h"
1.46 schwarze 40: #include "mandoc_aux.h"
1.3 kristaps 41: #include "libmandoc.h"
1.1 kristaps 42: #include "mdoc.h"
43: #include "man.h"
44:
45: #define REPARSE_LIMIT 1000
46:
47: struct mparse {
48: struct man *pman; /* persistent man parser */
49: struct mdoc *pmdoc; /* persistent mdoc parser */
50: struct man *man; /* man parser */
51: struct mdoc *mdoc; /* mdoc parser */
52: struct roff *roff; /* roff parser (!NULL) */
1.94 schwarze 53: const struct mchars *mchars; /* character table */
1.45 schwarze 54: char *sodest; /* filename pointed to by .so */
1.83 schwarze 55: const char *file; /* filename of current input file */
56: struct buf *primary; /* buffer currently being parsed */
57: struct buf *secondary; /* preprocessed copy of input */
58: const char *defos; /* default operating system */
59: mandocmsg mmsg; /* warning/error message handler */
60: enum mandoclevel file_status; /* status of current parse */
61: enum mandoclevel wlevel; /* ignore messages below this */
62: int options; /* parser options */
1.93 schwarze 63: int filenc; /* encoding of the current file */
1.1 kristaps 64: int reparse_count; /* finite interp. stack */
1.83 schwarze 65: int line; /* line number in the file */
1.97 schwarze 66: pid_t child; /* the gunzip(1) process */
1.1 kristaps 67: };
68:
1.84 schwarze 69: static void choose_parser(struct mparse *);
1.1 kristaps 70: static void resize_buf(struct buf *, size_t);
1.95 schwarze 71: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 72: static int read_whole_file(struct mparse *, const char *, int,
73: struct buf *, int *);
1.1 kristaps 74: static void mparse_end(struct mparse *);
1.37 schwarze 75: static void mparse_parse_buffer(struct mparse *, struct buf,
76: const char *);
1.1 kristaps 77:
1.3 kristaps 78: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
79: MANDOCERR_OK,
80: MANDOCERR_WARNING,
81: MANDOCERR_WARNING,
82: MANDOCERR_ERROR,
1.112 schwarze 83: MANDOCERR_UNSUPP,
1.3 kristaps 84: MANDOCERR_MAX,
85: MANDOCERR_MAX
86: };
87:
1.7 kristaps 88: static const char * const mandocerrs[MANDOCERR_MAX] = {
89: "ok",
90:
91: "generic warning",
92:
93: /* related to the prologue */
1.79 schwarze 94: "missing manual title, using UNTITLED",
95: "missing manual title, using \"\"",
1.54 schwarze 96: "lower case character in document title",
1.79 schwarze 97: "missing manual section, using \"\"",
1.7 kristaps 98: "unknown manual section",
1.54 schwarze 99: "missing date, using today's date",
1.7 kristaps 100: "cannot parse date, using it verbatim",
1.79 schwarze 101: "missing Os macro, using \"\"",
102: "duplicate prologue macro",
103: "late prologue macro",
104: "skipping late title macro",
1.7 kristaps 105: "prologue macros out of order",
106:
107: /* related to document structure */
108: ".so is fragile, better use ln(1)",
1.50 schwarze 109: "no document body",
1.54 schwarze 110: "content before first section header",
111: "first section is not \"NAME\"",
1.7 kristaps 112: "bad NAME section contents",
113: "sections out of conventional order",
1.54 schwarze 114: "duplicate section title",
115: "unexpected section",
1.87 schwarze 116: "unusual Xr order",
117: "unusual Xr punctuation",
1.86 schwarze 118: "AUTHORS section without An macro",
1.7 kristaps 119:
120: /* related to macros and nesting */
1.55 schwarze 121: "obsolete macro",
1.102 schwarze 122: "macro neither callable nor escaped",
1.7 kristaps 123: "skipping paragraph macro",
1.31 schwarze 124: "moving paragraph macro out of list",
1.7 kristaps 125: "skipping no-space macro",
126: "blocks badly nested",
127: "nested displays are not portable",
1.57 schwarze 128: "moving content out of list",
129: ".Vt block has child macro",
1.78 schwarze 130: "fill mode already enabled, skipping",
131: "fill mode already disabled, skipping",
1.7 kristaps 132: "line scope broken",
133:
134: /* related to missing macro arguments */
1.58 schwarze 135: "skipping empty request",
136: "conditional request controls empty scope",
1.7 kristaps 137: "skipping empty macro",
1.62 schwarze 138: "empty argument, using 0n",
1.7 kristaps 139: "argument count wrong",
1.60 schwarze 140: "missing display type, using -ragged",
141: "list type is not the first argument",
142: "missing -width in -tag list, using 8n",
1.78 schwarze 143: "missing utility name, using \"\"",
1.60 schwarze 144: "empty head in list item",
145: "empty list item",
1.61 schwarze 146: "missing font type, using \\fR",
147: "unknown font type, using \\fR",
1.103 schwarze 148: "nothing follows prefix",
1.60 schwarze 149: "missing -std argument, adding it",
1.90 schwarze 150: "missing eqn box, using \"\"",
1.7 kristaps 151:
152: /* related to bad macro arguments */
1.64 schwarze 153: "unterminated quoted argument",
1.7 kristaps 154: "duplicate argument",
1.76 schwarze 155: "skipping duplicate argument",
1.63 schwarze 156: "skipping duplicate display type",
157: "skipping duplicate list type",
1.76 schwarze 158: "skipping -width argument",
1.7 kristaps 159: "unknown AT&T UNIX version",
1.88 schwarze 160: "comma in function argument",
1.89 schwarze 161: "parenthesis in function name",
1.67 schwarze 162: "invalid content in Rs block",
1.63 schwarze 163: "invalid Boolean argument",
164: "unknown font, skipping request",
1.7 kristaps 165:
166: /* related to plain text */
1.64 schwarze 167: "blank line in fill mode, using .sp",
168: "tab in filled text",
169: "whitespace at end of input line",
1.7 kristaps 170: "bad comment style",
1.64 schwarze 171: "invalid escape sequence",
172: "undefined string, using \"\"",
1.16 kristaps 173:
1.118 schwarze 174: /* related to tables */
175: "tbl line starts with span",
176: "tbl column starts with span",
177: "skipping vertical bar in tbl layout",
178:
1.7 kristaps 179: "generic error",
180:
181: /* related to tables */
1.116 schwarze 182: "non-alphabetic character in tbl options",
183: "skipping unknown tbl option",
184: "missing tbl option argument",
185: "wrong tbl option argument size",
1.118 schwarze 186: "empty tbl layout",
187: "invalid character in tbl layout",
188: "unmatched parenthesis in tbl layout",
1.119 schwarze 189: "tbl without any data cells",
190: "ignoring data in spanned tbl cell",
191: "ignoring extra tbl data cells",
192: "data block open at end of tbl",
1.7 kristaps 193:
1.68 schwarze 194: /* related to document structure and macros */
1.108 schwarze 195: NULL,
1.7 kristaps 196: "input stack limit exceeded, infinite loop?",
197: "skipping bad character",
1.68 schwarze 198: "skipping unknown macro",
1.112 schwarze 199: "skipping insecure request",
1.73 schwarze 200: "skipping item outside list",
1.68 schwarze 201: "skipping column outside column list",
202: "skipping end of block that is not open",
1.115 schwarze 203: "fewer RS blocks open, skipping",
1.68 schwarze 204: "inserting missing end of block",
205: "appending missing end of block",
206:
207: /* related to request and macro arguments */
1.7 kristaps 208: "escaped character not allowed in a name",
209: "argument count wrong",
1.99 schwarze 210: "NOT IMPLEMENTED: Bd -file",
1.71 schwarze 211: "missing list type, using -item",
1.70 schwarze 212: "missing manual name, using \"\"",
1.71 schwarze 213: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 214: "unknown standard specifier",
1.71 schwarze 215: "skipping request without numeric argument",
1.109 schwarze 216: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 217: ".so request failed",
1.61 schwarze 218: "skipping all arguments",
219: "skipping excess arguments",
1.92 kristaps 220: "divide by zero",
1.112 schwarze 221:
222: "unsupported feature",
223: "input too large",
1.114 schwarze 224: "unsupported control character",
1.112 schwarze 225: "unsupported roff request",
1.119 schwarze 226: "eqn delim option in tbl",
1.118 schwarze 227: "unsupported tbl layout modifier",
1.112 schwarze 228: "ignoring macro in table",
1.7 kristaps 229: };
230:
231: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
232: "SUCCESS",
233: "RESERVED",
234: "WARNING",
235: "ERROR",
1.112 schwarze 236: "UNSUPP",
1.7 kristaps 237: "BADARG",
238: "SYSERR"
239: };
240:
1.47 schwarze 241:
1.1 kristaps 242: static void
243: resize_buf(struct buf *buf, size_t initial)
244: {
245:
246: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
247: buf->buf = mandoc_realloc(buf->buf, buf->sz);
248: }
249:
250: static void
1.84 schwarze 251: choose_parser(struct mparse *curp)
1.1 kristaps 252: {
1.83 schwarze 253: char *cp, *ep;
254: int format;
1.1 kristaps 255:
1.83 schwarze 256: /*
257: * If neither command line arguments -mdoc or -man select
258: * a parser nor the roff parser found a .Dd or .TH macro
259: * yet, look ahead in the main input buffer.
260: */
261:
262: if ((format = roff_getformat(curp->roff)) == 0) {
263: cp = curp->primary->buf;
264: ep = cp + curp->primary->sz;
265: while (cp < ep) {
1.85 schwarze 266: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 267: cp++;
268: if (cp[0] == 'D' && cp[1] == 'd') {
269: format = MPARSE_MDOC;
270: break;
271: }
272: if (cp[0] == 'T' && cp[1] == 'H') {
273: format = MPARSE_MAN;
274: break;
275: }
276: }
277: cp = memchr(cp, '\n', ep - cp);
278: if (cp == NULL)
279: break;
280: cp++;
281: }
1.1 kristaps 282: }
283:
1.83 schwarze 284: if (format == MPARSE_MDOC) {
1.47 schwarze 285: if (NULL == curp->pmdoc)
1.44 schwarze 286: curp->pmdoc = mdoc_alloc(
287: curp->roff, curp, curp->defos,
288: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 289: assert(curp->pmdoc);
290: curp->mdoc = curp->pmdoc;
291: return;
1.47 schwarze 292: }
1.1 kristaps 293:
1.83 schwarze 294: /* Fall back to man(7) as a last resort. */
295:
1.47 schwarze 296: if (NULL == curp->pman)
1.106 schwarze 297: curp->pman = man_alloc(
298: curp->roff, curp, curp->defos,
1.44 schwarze 299: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 300: assert(curp->pman);
301: curp->man = curp->pman;
302: }
303:
304: /*
1.95 schwarze 305: * Main parse routine for a buffer.
306: * It assumes encoding and line numbering are already set up.
307: * It can recurse directly (for invocations of user-defined
308: * macros, inline equations, and input line traps)
309: * and indirectly (for .so file inclusion).
1.1 kristaps 310: */
311: static void
1.95 schwarze 312: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 313: {
314: const struct tbl_span *span;
315: struct buf ln;
1.113 schwarze 316: const char *save_file;
1.110 schwarze 317: char *cp;
1.95 schwarze 318: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 319: enum rofferr rr;
1.100 schwarze 320: int of;
1.1 kristaps 321: int lnn; /* line number in the real file */
1.110 schwarze 322: int fd;
323: pid_t save_child;
1.1 kristaps 324: unsigned char c;
325:
1.95 schwarze 326: memset(&ln, 0, sizeof(ln));
1.1 kristaps 327:
1.47 schwarze 328: lnn = curp->line;
329: pos = 0;
1.1 kristaps 330:
1.95 schwarze 331: while (i < blk.sz) {
1.1 kristaps 332: if (0 == pos && '\0' == blk.buf[i])
333: break;
334:
335: if (start) {
336: curp->line = lnn;
337: curp->reparse_count = 0;
1.93 schwarze 338:
339: if (lnn < 3 &&
340: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 341: curp->filenc & MPARSE_LATIN1)
342: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 343: }
344:
1.95 schwarze 345: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 346:
347: /*
348: * When finding an unescaped newline character,
349: * leave the character loop to process the line.
350: * Skip a preceding carriage return, if any.
351: */
352:
1.95 schwarze 353: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 354: '\n' == blk.buf[i + 1])
355: ++i;
356: if ('\n' == blk.buf[i]) {
357: ++i;
358: ++lnn;
359: break;
360: }
361:
1.35 schwarze 362: /*
1.93 schwarze 363: * Make sure we have space for the worst
364: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 365: */
366:
1.95 schwarze 367: if (pos + 11 > ln.sz)
1.35 schwarze 368: resize_buf(&ln, 256);
369:
1.47 schwarze 370: /*
1.93 schwarze 371: * Encode 8-bit input.
1.1 kristaps 372: */
373:
1.93 schwarze 374: c = blk.buf[i];
375: if (c & 0x80) {
1.95 schwarze 376: if ( ! (curp->filenc && preconv_encode(
377: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 378: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
379: curp->line, pos, "0x%x", c);
1.93 schwarze 380: ln.buf[pos++] = '?';
381: i++;
382: }
383: continue;
384: }
385:
386: /*
387: * Exclude control characters.
388: */
1.1 kristaps 389:
1.93 schwarze 390: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 391: mandoc_vmsg(c == 0x00 || c == 0x04 ||
392: c > 0x0a ? MANDOCERR_CHAR_BAD :
393: MANDOCERR_CHAR_UNSUPP,
394: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 395: i++;
1.27 joerg 396: ln.buf[pos++] = '?';
1.1 kristaps 397: continue;
398: }
399:
400: /* Trailing backslash = a plain char. */
401:
1.95 schwarze 402: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 403: ln.buf[pos++] = blk.buf[i++];
404: continue;
405: }
406:
407: /*
408: * Found escape and at least one other character.
409: * When it's a newline character, skip it.
410: * When there is a carriage return in between,
411: * skip that one as well.
412: */
413:
1.95 schwarze 414: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 415: '\n' == blk.buf[i + 2])
416: ++i;
417: if ('\n' == blk.buf[i + 1]) {
418: i += 2;
419: ++lnn;
420: continue;
421: }
422:
1.13 kristaps 423: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 424: i += 2;
425: /* Comment, skip to end of line */
1.95 schwarze 426: for (; i < blk.sz; ++i) {
1.1 kristaps 427: if ('\n' == blk.buf[i]) {
428: ++i;
429: ++lnn;
430: break;
431: }
432: }
433:
434: /* Backout trailing whitespaces */
435: for (; pos > 0; --pos) {
436: if (ln.buf[pos - 1] != ' ')
437: break;
438: if (pos > 2 && ln.buf[pos - 2] == '\\')
439: break;
440: }
441: break;
442: }
443:
1.35 schwarze 444: /* Catch escaped bogus characters. */
445:
446: c = (unsigned char) blk.buf[i+1];
447:
1.47 schwarze 448: if ( ! (isascii(c) &&
449: (isgraph(c) || isblank(c)))) {
1.114 schwarze 450: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
1.78 schwarze 451: curp->line, pos, "0x%x", c);
1.35 schwarze 452: i += 2;
453: ln.buf[pos++] = '?';
454: continue;
455: }
456:
1.1 kristaps 457: /* Some other escape sequence, copy & cont. */
458:
459: ln.buf[pos++] = blk.buf[i++];
460: ln.buf[pos++] = blk.buf[i++];
461: }
462:
1.95 schwarze 463: if (pos >= ln.sz)
1.1 kristaps 464: resize_buf(&ln, 256);
465:
466: ln.buf[pos] = '\0';
467:
468: /*
469: * A significant amount of complexity is contained by
470: * the roff preprocessor. It's line-oriented but can be
471: * expressed on one line, so we need at times to
472: * readjust our starting point and re-run it. The roff
473: * preprocessor can also readjust the buffers with new
474: * data, so we pass them in wholesale.
475: */
476:
477: of = 0;
478:
1.24 kristaps 479: /*
480: * Maintain a lookaside buffer of all parsed lines. We
481: * only do this if mparse_keep() has been invoked (the
482: * buffer may be accessed with mparse_getkeep()).
483: */
484:
485: if (curp->secondary) {
1.47 schwarze 486: curp->secondary->buf = mandoc_realloc(
487: curp->secondary->buf,
488: curp->secondary->sz + pos + 2);
489: memcpy(curp->secondary->buf +
490: curp->secondary->sz,
491: ln.buf, pos);
1.24 kristaps 492: curp->secondary->sz += pos;
493: curp->secondary->buf
494: [curp->secondary->sz] = '\n';
495: curp->secondary->sz++;
496: curp->secondary->buf
497: [curp->secondary->sz] = '\0';
498: }
1.1 kristaps 499: rerun:
1.96 schwarze 500: rr = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 501:
502: switch (rr) {
1.47 schwarze 503: case ROFF_REPARSE:
1.1 kristaps 504: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 schwarze 505: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 506: else
1.3 kristaps 507: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 508: curp->line, pos, NULL);
1.1 kristaps 509: pos = 0;
510: continue;
1.47 schwarze 511: case ROFF_APPEND:
1.95 schwarze 512: pos = strlen(ln.buf);
1.1 kristaps 513: continue;
1.47 schwarze 514: case ROFF_RERUN:
1.1 kristaps 515: goto rerun;
1.47 schwarze 516: case ROFF_IGN:
1.1 kristaps 517: pos = 0;
518: continue;
1.47 schwarze 519: case ROFF_SO:
1.95 schwarze 520: if ( ! (curp->options & MPARSE_SO) &&
521: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 522: curp->sodest = mandoc_strdup(ln.buf + of);
523: free(ln.buf);
524: return;
525: }
1.24 kristaps 526: /*
527: * We remove `so' clauses from our lookaside
528: * buffer because we're going to descend into
529: * the file recursively.
530: */
1.47 schwarze 531: if (curp->secondary)
1.25 kristaps 532: curp->secondary->sz -= pos + 1;
1.113 schwarze 533: save_file = curp->file;
1.110 schwarze 534: save_child = curp->child;
535: if (mparse_open(curp, &fd, ln.buf + of) ==
1.113 schwarze 536: MANDOCLEVEL_OK) {
1.110 schwarze 537: mparse_readfd(curp, fd, ln.buf + of);
1.113 schwarze 538: curp->file = save_file;
539: } else {
540: curp->file = save_file;
1.52 schwarze 541: mandoc_vmsg(MANDOCERR_SO_FAIL,
542: curp, curp->line, pos,
543: ".so %s", ln.buf + of);
1.110 schwarze 544: ln.sz = mandoc_asprintf(&cp,
545: ".sp\nSee the file %s.\n.sp",
546: ln.buf + of);
547: free(ln.buf);
548: ln.buf = cp;
549: of = 0;
550: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 551: }
1.110 schwarze 552: curp->child = save_child;
1.1 kristaps 553: pos = 0;
554: continue;
555: default:
556: break;
557: }
558:
559: /*
560: * If input parsers have not been allocated, do so now.
1.14 kristaps 561: * We keep these instanced between parsers, but set them
1.1 kristaps 562: * locally per parse routine since we can use different
563: * parsers with each one.
564: */
565:
566: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 567: choose_parser(curp);
1.1 kristaps 568:
1.47 schwarze 569: /*
1.84 schwarze 570: * Lastly, push down into the parsers themselves.
1.1 kristaps 571: * If libroff returns ROFF_TBL, then add it to the
572: * currently open parse. Since we only get here if
573: * there does exist data (see tbl_data.c), we're
574: * guaranteed that something's been allocated.
575: * Do the same for ROFF_EQN.
576: */
577:
1.100 schwarze 578: if (rr == ROFF_TBL) {
579: while ((span = roff_span(curp->roff)) != NULL)
580: if (curp->man == NULL)
581: mdoc_addspan(curp->mdoc, span);
582: else
583: man_addspan(curp->man, span);
584: } else if (rr == ROFF_EQN) {
585: if (curp->man == NULL)
586: mdoc_addeqn(curp->mdoc, roff_eqn(curp->roff));
587: else
588: man_addeqn(curp->man, roff_eqn(curp->roff));
589: } else if ((curp->man == NULL ?
590: mdoc_parseln(curp->mdoc, curp->line, ln.buf, of) :
591: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
592: break;
1.1 kristaps 593:
594: /* Temporary buffers typically are not full. */
595:
596: if (0 == start && '\0' == blk.buf[i])
597: break;
598:
599: /* Start the next input line. */
600:
601: pos = 0;
602: }
603:
604: free(ln.buf);
605: }
606:
607: static int
1.40 schwarze 608: read_whole_file(struct mparse *curp, const char *file, int fd,
609: struct buf *fb, int *with_mmap)
1.1 kristaps 610: {
611: size_t off;
612: ssize_t ssz;
613:
1.81 schwarze 614: #if HAVE_MMAP
1.15 kristaps 615: struct stat st;
1.1 kristaps 616: if (-1 == fstat(fd, &st)) {
1.107 schwarze 617: perror(file);
618: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 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)) {
1.111 schwarze 630: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.1 kristaps 631: return(0);
632: }
633: *with_mmap = 1;
634: fb->sz = (size_t)st.st_size;
1.37 schwarze 635: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 636: if (fb->buf != MAP_FAILED)
637: return(1);
638: }
1.15 kristaps 639: #endif
1.1 kristaps 640:
641: /*
642: * If this isn't a regular file (like, say, stdin), then we must
643: * go the old way and just read things in bit by bit.
644: */
645:
646: *with_mmap = 0;
647: off = 0;
648: fb->sz = 0;
649: fb->buf = NULL;
650: for (;;) {
651: if (off == fb->sz) {
652: if (fb->sz == (1U << 31)) {
1.111 schwarze 653: mandoc_msg(MANDOCERR_TOOLARGE, curp,
654: 0, 0, NULL);
1.1 kristaps 655: break;
656: }
657: resize_buf(fb, 65536);
658: }
659: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
660: if (ssz == 0) {
661: fb->sz = off;
662: return(1);
663: }
664: if (ssz == -1) {
1.107 schwarze 665: perror(file);
666: exit((int)MANDOCLEVEL_SYSERR);
1.1 kristaps 667: }
668: off += (size_t)ssz;
669: }
670:
671: free(fb->buf);
672: fb->buf = NULL;
673: return(0);
674: }
675:
676: static void
677: mparse_end(struct mparse *curp)
678: {
679:
1.72 schwarze 680: if (curp->mdoc == NULL &&
681: curp->man == NULL &&
682: curp->sodest == NULL) {
683: if (curp->options & MPARSE_MDOC)
684: curp->mdoc = curp->pmdoc;
685: else {
686: if (curp->pman == NULL)
1.106 schwarze 687: curp->pman = man_alloc(
688: curp->roff, curp, curp->defos,
1.72 schwarze 689: curp->options & MPARSE_QUICK ? 1 : 0);
690: curp->man = curp->pman;
691: }
692: }
1.111 schwarze 693: if (curp->mdoc)
694: mdoc_endparse(curp->mdoc);
695: if (curp->man)
696: man_endparse(curp->man);
1.1 kristaps 697: roff_endparse(curp->roff);
698: }
699:
700: static void
1.36 schwarze 701: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 702: {
1.85 schwarze 703: struct buf *svprimary;
1.28 joerg 704: const char *svfile;
1.95 schwarze 705: size_t offset;
1.36 schwarze 706: static int recursion_depth;
707:
708: if (64 < recursion_depth) {
709: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
710: return;
711: }
1.28 joerg 712:
713: /* Line number is per-file. */
714: svfile = curp->file;
715: curp->file = file;
1.85 schwarze 716: svprimary = curp->primary;
1.83 schwarze 717: curp->primary = &blk;
1.28 joerg 718: curp->line = 1;
1.36 schwarze 719: recursion_depth++;
1.28 joerg 720:
1.93 schwarze 721: /* Skip an UTF-8 byte order mark. */
722: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
723: (unsigned char)blk.buf[0] == 0xef &&
724: (unsigned char)blk.buf[1] == 0xbb &&
725: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 726: offset = 3;
1.93 schwarze 727: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 728: } else
729: offset = 0;
1.93 schwarze 730:
1.95 schwarze 731: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 732:
1.111 schwarze 733: if (--recursion_depth == 0)
1.28 joerg 734: mparse_end(curp);
735:
1.85 schwarze 736: curp->primary = svprimary;
1.28 joerg 737: curp->file = svfile;
738: }
739:
740: enum mandoclevel
1.104 schwarze 741: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 742: const char *file)
743: {
744: struct buf blk;
745:
1.104 schwarze 746: blk.buf = buf;
1.28 joerg 747: blk.sz = len;
748:
1.36 schwarze 749: mparse_parse_buffer(curp, blk, file);
1.28 joerg 750: return(curp->file_status);
751: }
752:
1.98 schwarze 753: /*
754: * Read the whole file into memory and call the parsers.
755: * Called recursively when an .so request is encountered.
756: */
1.36 schwarze 757: enum mandoclevel
758: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 759: {
1.28 joerg 760: struct buf blk;
761: int with_mmap;
1.93 schwarze 762: int save_filenc;
1.1 kristaps 763:
1.91 schwarze 764: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 765: save_filenc = curp->filenc;
766: curp->filenc = curp->options &
767: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 768: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 769: curp->filenc = save_filenc;
1.81 schwarze 770: #if HAVE_MMAP
1.91 schwarze 771: if (with_mmap)
772: munmap(blk.buf, blk.sz);
773: else
1.28 joerg 774: #endif
1.91 schwarze 775: free(blk.buf);
776: }
1.1 kristaps 777:
1.98 schwarze 778: if (fd != STDIN_FILENO && close(fd) == -1)
1.1 kristaps 779: perror(file);
1.91 schwarze 780:
1.98 schwarze 781: mparse_wait(curp);
1.1 kristaps 782: return(curp->file_status);
1.82 schwarze 783: }
784:
785: enum mandoclevel
1.97 schwarze 786: mparse_open(struct mparse *curp, int *fd, const char *file)
1.82 schwarze 787: {
788: int pfd[2];
1.98 schwarze 789: int save_errno;
1.82 schwarze 790: char *cp;
791:
792: curp->file = file;
1.98 schwarze 793:
794: /* Unless zipped, try to just open the file. */
795:
1.82 schwarze 796: if ((cp = strrchr(file, '.')) == NULL ||
797: strcmp(cp + 1, "gz")) {
1.97 schwarze 798: curp->child = 0;
1.98 schwarze 799: if ((*fd = open(file, O_RDONLY)) != -1)
800: return(MANDOCLEVEL_OK);
801:
802: /* Open failed; try to append ".gz". */
803:
804: mandoc_asprintf(&cp, "%s.gz", file);
805: file = cp;
806: } else
807: cp = NULL;
808:
809: /* Before forking, make sure the file can be read. */
810:
811: save_errno = errno;
812: if (access(file, R_OK) == -1) {
813: if (cp != NULL)
814: errno = save_errno;
1.108 schwarze 815: free(cp);
816: *fd = -1;
817: curp->child = 0;
818: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
819: return(MANDOCLEVEL_ERROR);
1.82 schwarze 820: }
821:
1.98 schwarze 822: /* Run gunzip(1). */
823:
1.82 schwarze 824: if (pipe(pfd) == -1) {
1.107 schwarze 825: perror("pipe");
826: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 827: }
828:
1.97 schwarze 829: switch (curp->child = fork()) {
1.82 schwarze 830: case -1:
1.107 schwarze 831: perror("fork");
832: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 833: case 0:
834: close(pfd[0]);
835: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
1.107 schwarze 836: perror("dup");
837: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 838: }
839: execlp("gunzip", "gunzip", "-c", file, NULL);
1.107 schwarze 840: perror("exec");
841: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 842: default:
843: close(pfd[1]);
844: *fd = pfd[0];
845: return(MANDOCLEVEL_OK);
846: }
847: }
848:
849: enum mandoclevel
1.97 schwarze 850: mparse_wait(struct mparse *curp)
1.82 schwarze 851: {
852: int status;
853:
1.97 schwarze 854: if (curp->child == 0)
855: return(MANDOCLEVEL_OK);
856:
857: if (waitpid(curp->child, &status, 0) == -1) {
1.107 schwarze 858: perror("wait");
859: exit((int)MANDOCLEVEL_SYSERR);
1.82 schwarze 860: }
861: if (WIFSIGNALED(status)) {
1.108 schwarze 862: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
863: "gunzip died from signal %d", WTERMSIG(status));
864: return(MANDOCLEVEL_ERROR);
1.82 schwarze 865: }
866: if (WEXITSTATUS(status)) {
1.108 schwarze 867: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
868: "gunzip failed with code %d", WEXITSTATUS(status));
869: return(MANDOCLEVEL_ERROR);
1.82 schwarze 870: }
871: return(MANDOCLEVEL_OK);
1.1 kristaps 872: }
873:
874: struct mparse *
1.94 schwarze 875: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
876: const struct mchars *mchars, const char *defos)
1.1 kristaps 877: {
878: struct mparse *curp;
1.10 kristaps 879:
1.1 kristaps 880: curp = mandoc_calloc(1, sizeof(struct mparse));
881:
1.44 schwarze 882: curp->options = options;
1.3 kristaps 883: curp->wlevel = wlevel;
1.1 kristaps 884: curp->mmsg = mmsg;
1.29 schwarze 885: curp->defos = defos;
1.1 kristaps 886:
1.94 schwarze 887: curp->mchars = mchars;
888: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 889: if (curp->options & MPARSE_MDOC)
890: curp->pmdoc = mdoc_alloc(
891: curp->roff, curp, curp->defos,
892: curp->options & MPARSE_QUICK ? 1 : 0);
893: if (curp->options & MPARSE_MAN)
1.106 schwarze 894: curp->pman = man_alloc(
895: curp->roff, curp, curp->defos,
1.72 schwarze 896: curp->options & MPARSE_QUICK ? 1 : 0);
897:
1.1 kristaps 898: return(curp);
899: }
900:
901: void
902: mparse_reset(struct mparse *curp)
903: {
904:
905: roff_reset(curp->roff);
906:
907: if (curp->mdoc)
908: mdoc_reset(curp->mdoc);
909: if (curp->man)
910: man_reset(curp->man);
1.24 kristaps 911: if (curp->secondary)
912: curp->secondary->sz = 0;
1.1 kristaps 913:
914: curp->file_status = MANDOCLEVEL_OK;
915: curp->mdoc = NULL;
916: curp->man = NULL;
1.45 schwarze 917:
918: free(curp->sodest);
919: curp->sodest = NULL;
1.1 kristaps 920: }
921:
922: void
923: mparse_free(struct mparse *curp)
924: {
925:
926: if (curp->pmdoc)
927: mdoc_free(curp->pmdoc);
928: if (curp->pman)
929: man_free(curp->pman);
930: if (curp->roff)
931: roff_free(curp->roff);
1.24 kristaps 932: if (curp->secondary)
933: free(curp->secondary->buf);
1.1 kristaps 934:
1.24 kristaps 935: free(curp->secondary);
1.45 schwarze 936: free(curp->sodest);
1.1 kristaps 937: free(curp);
938: }
939:
940: void
1.45 schwarze 941: mparse_result(struct mparse *curp,
942: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 943: {
944:
1.45 schwarze 945: if (sodest && NULL != (*sodest = curp->sodest)) {
946: *mdoc = NULL;
947: *man = NULL;
948: return;
949: }
1.9 kristaps 950: if (mdoc)
951: *mdoc = curp->mdoc;
952: if (man)
953: *man = curp->man;
1.3 kristaps 954: }
955:
956: void
957: mandoc_vmsg(enum mandocerr t, struct mparse *m,
958: int ln, int pos, const char *fmt, ...)
959: {
960: char buf[256];
961: va_list ap;
962:
963: va_start(ap, fmt);
1.48 schwarze 964: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 965: va_end(ap);
966:
967: mandoc_msg(t, m, ln, pos, buf);
968: }
969:
970: void
1.47 schwarze 971: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 972: int ln, int col, const char *msg)
973: {
974: enum mandoclevel level;
975:
1.112 schwarze 976: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 977: while (er < mandoclimits[level])
978: level--;
979:
1.108 schwarze 980: if (level < m->wlevel && er != MANDOCERR_FILE)
1.3 kristaps 981: return;
982:
1.8 kristaps 983: if (m->mmsg)
984: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 985:
986: if (m->file_status < level)
987: m->file_status = level;
1.7 kristaps 988: }
989:
990: const char *
991: mparse_strerror(enum mandocerr er)
992: {
993:
994: return(mandocerrs[er]);
995: }
996:
997: const char *
998: mparse_strlevel(enum mandoclevel lvl)
999: {
1000: return(mandoclevels[lvl]);
1.24 kristaps 1001: }
1002:
1003: void
1004: mparse_keep(struct mparse *p)
1005: {
1006:
1007: assert(NULL == p->secondary);
1008: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1009: }
1010:
1011: const char *
1012: mparse_getkeep(const struct mparse *p)
1013: {
1014:
1015: assert(p->secondary);
1016: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1017: }
CVSweb