KickJava   Java API By Example, From Geeks To Geeks.

Java > Open Source Codes > org > w3c > tidy > StreamInImpl


1 /*
2  * @(#)StreamInImpl.java 1.11 2000/08/16
3  *
4  */
5
6 package org.w3c.tidy;
7
8 /**
9  *
10  * Input Stream Implementation
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 JavaDoc;
34 import java.io.IOException JavaDoc;
35
36 public class StreamInImpl extends StreamIn {
37
38     /* Mapping for Windows Western character set (128-159) to Unicode */
39     private static int[] Win2Unicode =
40     {
41     0x20AC, 0x0000, 0x201A, 0x0192, 0x201E, 0x2026, 0x2020, 0x2021,
42     0x02C6, 0x2030, 0x0160, 0x2039, 0x0152, 0x0000, 0x017D, 0x0000,
43     0x0000, 0x2018, 0x2019, 0x201C, 0x201D, 0x2022, 0x2013, 0x2014,
44     0x02DC, 0x2122, 0x0161, 0x203A, 0x0153, 0x0000, 0x017E, 0x0178
45     };
46
47     /*
48     John Love-Jensen contributed this table for mapping MacRoman
49     character set to Unicode
50     */
51
52     private static int[] Mac2Unicode =
53     {
54
55     0x0000, 0x0001, 0x0002, 0x0003, 0x0004, 0x0005, 0x0006, 0x0007,
56     0x0008, 0x0009, 0x000A, 0x000B, 0x000C, 0x000D, 0x000E, 0x000F,
57
58     0x0010, 0x0011, 0x0012, 0x0013, 0x0014, 0x0015, 0x0016, 0x0017,
59     0x0018, 0x0019, 0x001A, 0x001B, 0x001C, 0x001D, 0x001E, 0x001F,
60
61     0x0020, 0x0021, 0x0022, 0x0023, 0x0024, 0x0025, 0x0026, 0x0027,
62     0x0028, 0x0029, 0x002A, 0x002B, 0x002C, 0x002D, 0x002E, 0x002F,
63
64     0x0030, 0x0031, 0x0032, 0x0033, 0x0034, 0x0035, 0x0036, 0x0037,
65     0x0038, 0x0039, 0x003A, 0x003B, 0x003C, 0x003D, 0x003E, 0x003F,
66
67     0x0040, 0x0041, 0x0042, 0x0043, 0x0044, 0x0045, 0x0046, 0x0047,
68     0x0048, 0x0049, 0x004A, 0x004B, 0x004C, 0x004D, 0x004E, 0x004F,
69
70     0x0050, 0x0051, 0x0052, 0x0053, 0x0054, 0x0055, 0x0056, 0x0057,
71     0x0058, 0x0059, 0x005A, 0x005B, 0x005C, 0x005D, 0x005E, 0x005F,
72
73     0x0060, 0x0061, 0x0062, 0x0063, 0x0064, 0x0065, 0x0066, 0x0067,
74     0x0068, 0x0069, 0x006A, 0x006B, 0x006C, 0x006D, 0x006E, 0x006F,
75
76     0x0070, 0x0071, 0x0072, 0x0073, 0x0074, 0x0075, 0x0076, 0x0077,
77     0x0078, 0x0079, 0x007A, 0x007B, 0x007C, 0x007D, 0x007E, 0x007F,
78     /* x7F = DEL */
79     0x00C4, 0x00C5, 0x00C7, 0x00C9, 0x00D1, 0x00D6, 0x00DC, 0x00E1,
80     0x00E0, 0x00E2, 0x00E4, 0x00E3, 0x00E5, 0x00E7, 0x00E9, 0x00E8,
81
82     0x00EA, 0x00EB, 0x00ED, 0x00EC, 0x00EE, 0x00EF, 0x00F1, 0x00F3,
83     0x00F2, 0x00F4, 0x00F6, 0x00F5, 0x00FA, 0x00F9, 0x00FB, 0x00FC,
84
85     0x2020, 0x00B0, 0x00A2, 0x00A3, 0x00A7, 0x2022, 0x00B6, 0x00DF,
86     0x00AE, 0x00A9, 0x2122, 0x00B4, 0x00A8, 0x2260, 0x00C6, 0x00D8,
87
88     0x221E, 0x00B1, 0x2264, 0x2265, 0x00A5, 0x00B5, 0x2202, 0x2211,
89     0x220F, 0x03C0, 0x222B, 0x00AA, 0x00BA, 0x03A9, 0x00E6, 0x00F8,
90
91     0x00BF, 0x00A1, 0x00AC, 0x221A, 0x0192, 0x2248, 0x2206, 0x00AB,
92     0x00BB, 0x2026, 0x00A0, 0x00C0, 0x00C3, 0x00D5, 0x0152, 0x0153,
93
94     0x2013, 0x2014, 0x201C, 0x201D, 0x2018, 0x2019, 0x00F7, 0x25CA,
95     0x00FF, 0x0178, 0x2044, 0x20AC, 0x2039, 0x203A, 0xFB01, 0xFB02,
96
97     0x2021, 0x00B7, 0x201A, 0x201E, 0x2030, 0x00C2, 0x00CA, 0x00C1,
98     0x00CB, 0x00C8, 0x00CD, 0x00CE, 0x00CF, 0x00CC, 0x00D3, 0x00D4,
99     /* xF0 = Apple Logo */
100     0xF8FF, 0x00D2, 0x00DA, 0x00DB, 0x00D9, 0x0131, 0x02C6, 0x02DC,
101     0x00AF, 0x02D8, 0x02D9, 0x02DA, 0x00B8, 0x02DD, 0x02DB, 0x02C7
102     };
103
104     public StreamInImpl(InputStream JavaDoc stream, int encoding, int tabsize)
105     {
106         this.stream = stream;
107         this.pushed = false;
108         this.c = (int)'\0';
109         this.tabs = 0;
110         this.tabsize = tabsize;
111         this.curline = 1;
112         this.curcol = 1;
113         this.encoding = encoding;
114         this.state = FSM_ASCII;
115         this.endOfStream = false;
116     }
117
118     /* read char from stream */
119     public int readCharFromStream()
120     {
121         int n, c, i, count;
122
123         try {
124             c = this.stream.read();
125
126             if (c == EndOfStream) {
127                 this.endOfStream = true;
128                 return c;
129             }
130
131             /*
132                A document in ISO-2022 based encoding uses some ESC sequences
133                called "designator" to switch character sets. The designators
134                defined and used in ISO-2022-JP are:
135
136                 "ESC" + "(" + ? for ISO646 variants
137
138                 "ESC" + "$" + ? and
139                 "ESC" + "$" + "(" + ? for multibyte character sets
140
141                Where ? stands for a single character used to indicate the
142                character set for multibyte characters.
143
144                Tidy handles this by preserving the escape sequence and
145                setting the top bit of each byte for non-ascii chars. This
146                bit is then cleared on output. The input stream keeps track
147                of the state to determine when to set/clear the bit.
148             */
149
150             if (this.encoding == Configuration.ISO2022)
151             {
152                 if (c == 0x1b) /* ESC */
153                 {
154                     this.state = FSM_ESC;
155                     return c;
156                 }
157
158                 switch (this.state)
159                 {
160                 case FSM_ESC:
161                     if (c == '$')
162                         this.state = FSM_ESCD;
163                     else if (c == '(')
164                         this.state = FSM_ESCP;
165                     else
166                         this.state = FSM_ASCII;
167                     break;
168
169                 case FSM_ESCD:
170                     if (c == '(')
171                         this.state = FSM_ESCDP;
172                     else
173                         this.state = FSM_NONASCII;
174                     break;
175
176                 case FSM_ESCDP:
177                     this.state = FSM_NONASCII;
178                     break;
179
180                 case FSM_ESCP:
181                     this.state = FSM_ASCII;
182                     break;
183
184                 case FSM_NONASCII:
185                     c |= 0x80;
186                     break;
187                 }
188
189                 return c;
190             }
191
192             if (this.encoding != Configuration.UTF8)
193                 return c;
194
195             /* deal with UTF-8 encoded char */
196
197             if ((c & 0xE0) == 0xC0) /* 110X XXXX two bytes */
198             {
199                 n = c & 31;
200                 count = 1;
201             }
202             else if ((c & 0xF0) == 0xE0) /* 1110 XXXX three bytes */
203             {
204                 n = c & 15;
205                 count = 2;
206             }
207             else if ((c & 0xF8) == 0xF0) /* 1111 0XXX four bytes */
208             {
209                 n = c & 7;
210                 count = 3;
211             }
212             else if ((c & 0xFC) == 0xF8) /* 1111 10XX five bytes */
213             {
214                 n = c & 3;
215                 count = 4;
216             }
217             else if ((c & 0xFE) == 0xFC) /* 1111 110X six bytes */
218             {
219                 n = c & 1;
220                 count = 5;
221             }
222             else /* 0XXX XXXX one byte */
223                 return c;
224
225             /* successor bytes should have the form 10XX XXXX */
226             for (i = 1; i <= count; ++i)
227             {
228                 c = this.stream.read();
229
230                 if (c == EndOfStream) {
231                     this.endOfStream = true;
232                     return c;
233                 }
234
235                 n = (n << 6) | (c & 0x3F);
236             }
237         }
238         catch (IOException JavaDoc e) {
239             System.err.println("StreamInImpl.readCharFromStream: " + e.toString());
240             n = EndOfStream;
241         }
242
243         return n;
244     }
245
246     public int readChar()
247     {
248         int c;
249
250         if (this.pushed)
251         {
252             this.pushed = false;
253             c = this.c;
254
255             if (c == '\n')
256             {
257                 this.curcol = 1;
258                 this.curline++;
259                 return c;
260             }
261
262             this.curcol++;
263             return c;
264         }
265
266         this.lastcol = this.curcol;
267
268         if (this.tabs > 0)
269         {
270             this.curcol++;
271             this.tabs--;
272             return ' ';
273         }
274     
275         for (;;)
276         {
277             c = readCharFromStream();
278
279             if (c < 0)
280                 return EndOfStream;
281
282             if (c == '\n')
283             {
284                 this.curcol = 1;
285                 this.curline++;
286                 break;
287             }
288
289             if (c == '\r')
290             {
291                 c = readCharFromStream();
292                 if (c != '\n')
293                 {
294                     ungetChar(c);
295                     c = '\n';
296                 }
297                 this.curcol = 1;
298                 this.curline++;
299                 break;
300             }
301
302             if (c == '\t')
303             {
304                 this.tabs = this.tabsize - ((this.curcol - 1) % this.tabsize) - 1;
305                 this.curcol++;
306                 c = ' ';
307                 break;
308             }
309
310             /* strip control characters, except for Esc */
311
312             if (c == '\033')
313                 break;
314
315             if (0 < c && c < 32)
316                 continue;
317
318             /* watch out for IS02022 */
319
320             if (this.encoding == Configuration.RAW ||
321                 this.encoding == Configuration.ISO2022)
322             {
323                 this.curcol++;
324                 break;
325             }
326
327             if (this.encoding == Configuration.MACROMAN)
328                 c = Mac2Unicode[c];
329
330             /* produced e.g. as a side-effect of smart quotes in Word */
331
332             if (127 < c && c < 160)
333             {
334                 Report.encodingError((Lexer)this.lexer, Report.WINDOWS_CHARS, c);
335
336                 c = Win2Unicode[c - 128];
337
338                 if (c == 0)
339                     continue;
340             }
341
342             this.curcol++;
343             break;
344         }
345
346         return c;
347     }
348
349     public void ungetChar(int c)
350     {
351         this.pushed = true;
352         this.c = c;
353
354         if (c == '\n')
355         {
356             --this.curline;
357         }
358
359         this.curcol = this.lastcol;
360     }
361
362     public boolean isEndOfStream()
363     {
364         return this.endOfStream;
365     }
366
367 }
368
Popular Tags