|
1
16 package org.apache.cocoon.components.sax;
17
18 import java.util.ArrayList  ;
19
20 import org.apache.avalon.excalibur.pool.Recyclable;
21 import org.apache.cocoon.xml.AbstractXMLProducer;
22 import org.xml.sax.SAXException ;
23 import org.xml.sax.helpers.AttributesImpl ;
24
25
33
34 public final class XMLByteStreamInterpreter extends AbstractXMLProducer implements XMLDeserializer, Recyclable {
35
36 private static final int START_DOCUMENT = 0;
37 private static final int END_DOCUMENT = 1;
38 private static final int START_PREFIX_MAPPING = 2;
39 private static final int END_PREFIX_MAPPING = 3;
40 private static final int START_ELEMENT = 4;
41 private static final int END_ELEMENT = 5;
42 private static final int CHARACTERS = 6;
43 private static final int IGNORABLE_WHITESPACE = 7;
44 private static final int PROCESSING_INSTRUCTION = 8;
45 private static final int COMMENT = 9;
46 private static final int LOCATOR = 10;
47 private static final int START_DTD = 11;
48 private static final int END_DTD = 12;
49 private static final int START_CDATA = 13;
50 private static final int END_CDATA = 14;
51 private static final int SKIPPED_ENTITY = 15;
52 private static final int START_ENTITY = 16;
53 private static final int END_ENTITY = 17;
54
55 private ArrayList list = new ArrayList ();
56 private byte[] input;
57 private int currentPos;
58
59 public void recycle() {
60 super.recycle();
61 this.list.clear();
62 this.input = null;
63 }
64
65 public void deserialize(Object saxFragment) throws SAXException {
66 if (!(saxFragment instanceof byte[])) {
67 throw new SAXException ("XMLDeserializer needs byte array for deserialization.");
68 }
69 this.list.clear();
70 this.input = (byte[])saxFragment;
71 this.currentPos = 0;
72 this.checkProlog();
73 this.parse();
74 }
75
76 private void parse() throws SAXException {
77 while ( currentPos < input.length) {
78 switch (this.readEvent()) {
79 case START_DOCUMENT:
80 contentHandler.startDocument();
81 break;
82 case END_DOCUMENT:
83 contentHandler.endDocument();
84 break;
85 case START_PREFIX_MAPPING:
86 contentHandler.startPrefixMapping(this.readString(), this.readString());
87 break;
88 case END_PREFIX_MAPPING:
89 contentHandler.endPrefixMapping(this.readString());
90 break;
91 case START_ELEMENT:
92 int attributes = this.readAttributes();
93 AttributesImpl atts = new AttributesImpl ();
94 for (int i = 0; i < attributes; i++) {
95 atts.addAttribute(this.readString(), this.readString(), this.readString(), this.readString(), this.readString());
96 }
97 contentHandler.startElement(this.readString(), this.readString(), this.readString(), atts);
98 break;
99 case END_ELEMENT:
100 contentHandler.endElement(this.readString(), this.readString(), this.readString());
101 break;
102 case CHARACTERS:
103 char[] chars = this.readChars();
104 int len = chars.length;
105 while (len > 0 && chars[len-1]==0) len--;
106 if (len > 0) contentHandler.characters(chars, 0, len);
107 break;
108 case IGNORABLE_WHITESPACE:
109 char[] spaces = this.readChars();
110 len = spaces.length;
111 while (len > 0 && spaces[len-1]==0) len--;
112 if (len > 0) contentHandler.characters(spaces, 0, len);
113 break;
114 case PROCESSING_INSTRUCTION:
115 contentHandler.processingInstruction(this.readString(), this.readString());
116 break;
117 case COMMENT:
118 chars = this.readChars();
119 if (this.lexicalHandler != null) {
120 len = chars.length;
121 while (len > 0 && chars[len-1]==0) len--;
122 if (len > 0) lexicalHandler.comment(chars, 0, len);
123 }
124 break;
125 case LOCATOR:
126 {
127 String publicId = this.readString();
128 String systemId = this.readString();
129 int lineNumber = this.read();
130 int columnNumber = this.read();
131 org.xml.sax.helpers.LocatorImpl locator = new org.xml.sax.helpers.LocatorImpl ();
132 locator.setPublicId(publicId);
133 locator.setSystemId(systemId);
134 locator.setLineNumber(lineNumber);
135 locator.setColumnNumber(columnNumber);
136 contentHandler.setDocumentLocator(locator);
137 }
138 break;
139 case START_DTD:
140 lexicalHandler.startDTD(this.readString(),
141 this.readString(),
142 this.readString());
143 break;
144 case END_DTD:
145 lexicalHandler.endDTD();
146 break;
147 case START_CDATA:
148 lexicalHandler.startCDATA();
149 break;
150 case END_CDATA:
151 lexicalHandler.endCDATA();
152 break;
153 case SKIPPED_ENTITY:
154 contentHandler.skippedEntity( this.readString() );
155 break;
156 case START_ENTITY:
157 lexicalHandler.startEntity( this.readString() );
158 break;
159 case END_ENTITY:
160 lexicalHandler.endEntity( this.readString() );
161 break;
162 default:
163 throw new SAXException ("parsing error: event not supported.");
164 }
165 }
166 }
167
168 private void checkProlog() throws SAXException {
169 int valid = 0;
170 if (this.read() == 'C') valid++;
171 if (this.read() == 'X') valid++;
172 if (this.read() == 'M') valid++;
173 if (this.read() == 'L') valid++;
174 if (this.read() == 1) valid++;
175 if (this.read() == 0) valid++;
176 if (valid != 6) throw new SAXException ("Unrecognized file format.");
177 }
178
179 protected int readEvent() throws SAXException {
180 return this.read();
181 }
182
183 private int readAttributes() throws SAXException {
184 int ch1 = this.read();
185 int ch2 = this.read();
186 return ((ch1 << 8) + (ch2 << 0));
187 }
188
189 private String readString() throws SAXException {
190 int length = this.readWord();
191 int index = length & 0x00007FFF;
192 if (length >= 0x00008000) {
193 return (String ) list.get(index);
194 }
195 else {
196 if (length == 0x00007FFF) {
197 length = this.readLong();
198 }
199 char[] chars = this.readChars(length);
200 int len = chars.length;
201 if (len > 0) {
202 while (chars[len-1]==0) len--;
203 }
204 String str;
205 if (len == 0) {
206 str = "";
207 } else {
208 str = new String (chars, 0, len);
209 }
210 list.add(str);
211 return str;
212 }
213 }
214
215
219 private char[] readChars() throws SAXException {
220 int length = this.readWord();
221 if (length == 0x00007FFF) {
222 length = this.readLong();
223 }
224 return this.readChars(length);
225 }
226
227 private int read() throws SAXException {
228 if (currentPos >= input.length)
229 throw new SAXException ("Reached end of input.");
230 return input[currentPos++] & 0xff;
231 }
232
233
237 private char[] readChars(int len) throws SAXException {
238 char[] str = new char[len];
239 byte[] bytearr = new byte[len];
240 int c, char2, char3;
241 int count = 0;
242 int i = 0;
243
244 this.readBytes(bytearr);
245
246 while (count < len) {
247 c = bytearr[count] & 0xff;
248 switch (c >> 4) {
249 case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
250 count++;
252 str[i++] = (char) c;
253 break;
254 case 12: case 13:
255 count += 2;
257 char2 = bytearr[count-1];
258 str[i++] = (char) (((c & 0x1F) << 6) | (char2 & 0x3F));
259 break;
260 case 14:
261 count += 3;
263 char2 = bytearr[count-2];
264 char3 = bytearr[count-1];
265 str[i++] = ((char)(((c & 0x0F) << 12) | ((char2 & 0x3F) << 6) | ((char3 & 0x3F) << 0)));
266 break;
267 default:
268 throw new SAXException ("UTFDataFormatException");
270 }
271 }
272
273 return str;
274 }
275
276 private void readBytes(byte[] b) throws SAXException {
277 if (this.currentPos + b.length > this.input.length) {
278 throw new SAXException ("End of input reached.");
287 }
288 System.arraycopy(this.input, this.currentPos, b, 0, b.length);
289 this.currentPos += b.length;
290 }
291
292 private int readWord() throws SAXException {
293 int ch1 = this.read();
294 int ch2 = this.read();
295 return ((ch1 << 8) + (ch2 << 0));
296 }
297
298 private int readLong() throws SAXException {
299 int ch1 = this.read();
300 int ch2 = this.read();
301 int ch3 = this.read();
302 int ch4 = this.read();
303 return ((ch1 << 24) + (ch2 << 16) + (ch3 << 8) + (ch4 << 0));
304 }
305 }
306 |
Popular Tags |
Java API By Example, From Geeks To Geeks.