01: /*
02: * @(#)StreamIn.java 1.11 2000/08/16
03: *
04: */
05:
06: package org.w3c.tidy;
07:
08: /**
09: *
10: * Input Stream
11: *
12: * (c) 1998-2000 (W3C) MIT, INRIA, Keio University
13: * Derived from <a href="http://www.w3.org/People/Raggett/tidy">
14: * HTML Tidy Release 4 Aug 2000</a>
15: *
16: * @author Dave Raggett <dsr@w3.org>
17: * @author Andy Quick <ac.quick@sympatico.ca> (translation to Java)
18: * @version 1.0, 1999/05/22
19: * @version 1.0.1, 1999/05/29
20: * @version 1.1, 1999/06/18 Java Bean
21: * @version 1.2, 1999/07/10 Tidy Release 7 Jul 1999
22: * @version 1.3, 1999/07/30 Tidy Release 26 Jul 1999
23: * @version 1.4, 1999/09/04 DOM support
24: * @version 1.5, 1999/10/23 Tidy Release 27 Sep 1999
25: * @version 1.6, 1999/11/01 Tidy Release 22 Oct 1999
26: * @version 1.7, 1999/12/06 Tidy Release 30 Nov 1999
27: * @version 1.8, 2000/01/22 Tidy Release 13 Jan 2000
28: * @version 1.9, 2000/06/03 Tidy Release 30 Apr 2000
29: * @version 1.10, 2000/07/22 Tidy Release 8 Jul 2000
30: * @version 1.11, 2000/08/16 Tidy Release 4 Aug 2000
31: */
32:
33: import java.io.InputStream;
34:
35: public abstract class StreamIn {
36:
37: public static final int EndOfStream = -1; // EOF
38:
39: /* states for ISO 2022
40:
41: A document in ISO-2022 based encoding uses some ESC sequences called
42: "designator" to switch character sets. The designators defined and
43: used in ISO-2022-JP are:
44:
45: "ESC" + "(" + ? for ISO646 variants
46:
47: "ESC" + "$" + ? and
48: "ESC" + "$" + "(" + ? for multibyte character sets
49: */
50:
51: public static final int FSM_ASCII = 0;
52: public static final int FSM_ESC = 1;
53: public static final int FSM_ESCD = 2;
54: public static final int FSM_ESCDP = 3;
55: public static final int FSM_ESCP = 4;
56: public static final int FSM_NONASCII = 5;
57:
58: /* non-raw input is cleaned up*/
59: public int state; /* FSM for ISO2022 */
60: public boolean pushed;
61: public int c;
62: public int tabs;
63: public int tabsize;
64: public int lastcol;
65: public int curcol;
66: public int curline;
67: public int encoding;
68: public InputStream stream;
69: public boolean endOfStream;
70: public Object lexer; /* needed for error reporting */
71:
72: /* read char from stream */
73: public abstract int readCharFromStream();
74:
75: public abstract int readChar();
76:
77: public abstract void ungetChar(int c);
78:
79: public abstract boolean isEndOfStream();
80:
81: }
|