|
1
57
58 package org.enhydra.apache.xerces.readers;
59
60 import java.io.IOException  ;
61 import java.io.InputStreamReader ;
62 import java.io.UnsupportedEncodingException ;
63
64 import org.enhydra.apache.xerces.framework.XMLErrorReporter;
65 import org.enhydra.apache.xerces.utils.ChunkyByteArray;
66 import org.enhydra.apache.xerces.utils.QName;
67 import org.enhydra.apache.xerces.utils.StringPool;
68
69
73 final class UTF8Recognizer extends XMLDeclRecognizer {
74 private byte[] fUTF8BOM = {(byte)0xEF, (byte)0xBB, (byte)0xBF};
75 public XMLEntityHandler.EntityReader recognize(XMLEntityReaderFactory readerFactory,
79 XMLEntityHandler entityHandler,
80 XMLErrorReporter errorReporter,
81 boolean sendCharDataAsCharArray,
82 StringPool stringPool,
83 ChunkyByteArray data,
84 boolean xmlDecl,
85 boolean allowJavaEncodingName) throws Exception {
86 XMLEntityHandler.EntityReader reader = null;
87
88 boolean seeBOM = false;
90 byte bom0 = data.byteAt(0);
91 if (bom0 == fUTF8BOM[0]) {
92 byte bom1 = data.byteAt(1);
93 if (bom1 == fUTF8BOM[1]) {
94 byte bom2 = data.byteAt(2);
95 if (bom2 == fUTF8BOM[2]) {
96 seeBOM = true;
97 }
98 }
99 }
100 if (seeBOM) {
101 data.read(fUTF8BOM, 0, 3);
103 }
104
105 byte b0 = data.byteAt(0);
106 boolean debug = false;
107
108 if (b0 == '<') {
109 int b1 = data.byteAt(1);
110 if (b1 == '?') {
111 if (data.byteAt(2) == 'x' && data.byteAt(3) == 'm' && data.byteAt(4) == 'l') {
112 int b5 = data.byteAt(5);
113 if (b5 == 0x20 || b5 == 0x09 || b5 == 0x0a || b5 == 0x0d) {
114 XMLEntityHandler.EntityReader declReader = new XMLDeclReader(entityHandler, errorReporter, sendCharDataAsCharArray, data, stringPool);
115 int encoding = prescanXMLDeclOrTextDecl(declReader, xmlDecl);
116 if (encoding != -1) {
117 String encname = stringPool.orphanString(encoding);
118 String enc = encname.toUpperCase();
119 if ("ISO-10646-UCS-2".equals(enc)) throw new UnsupportedEncodingException (encname);
120 if ("ISO-10646-UCS-4".equals(enc)) throw new UnsupportedEncodingException (encname);
121 if ("UTF-16".equals(enc)) throw new UnsupportedEncodingException (encname);
122
123 String javaencname = MIME2Java.convert(enc);
124 if (null == javaencname) {
125 if (allowJavaEncodingName) {
127 javaencname = encname;
128 } else {
129 throw new UnsupportedEncodingException (encname);
130 }
131 }
132 try {
133 data.rewind();
134 if ("UTF-8".equalsIgnoreCase(javaencname) || "UTF8".equalsIgnoreCase(javaencname)) {
135 reader = readerFactory.createUTF8Reader(entityHandler, errorReporter, sendCharDataAsCharArray, data, stringPool);
136 } else {
137 reader = readerFactory.createCharReader(entityHandler, errorReporter, sendCharDataAsCharArray,
138 new InputStreamReader (data, javaencname), stringPool);
139 }
140 } catch (UnsupportedEncodingException e) {
141 throw new UnsupportedEncodingException (encname);
142 } catch (Exception e) {
143 if( debug == true )
144 e.printStackTrace(); }
146 } else {
147 data.rewind();
148 reader = readerFactory.createUTF8Reader(entityHandler, errorReporter, sendCharDataAsCharArray, data, stringPool);
149 }
150 }
151 }
152 }
153 }
154 return reader;
155 }
156
157 final class XMLDeclReader extends XMLEntityReader {
158 private StringPool fStringPool = null;
162 private ChunkyByteArray fData = null;
163 XMLDeclReader(XMLEntityHandler entityHandler, XMLErrorReporter errorReporter, boolean sendCharDataAsCharArray, ChunkyByteArray data, StringPool stringPool) {
167 super(entityHandler, errorReporter, sendCharDataAsCharArray);
168 fStringPool = stringPool;
169 fData = data;
170 }
171 public boolean lookingAtChar(char ch, boolean skipPastChar) throws IOException {
175 if (fData.byteAt(fCurrentOffset) != ch)
176 return false;
177 if (skipPastChar)
178 fCurrentOffset++;
179 return true;
180 }
181 public boolean lookingAtSpace(boolean skipPastChar) throws IOException {
182 int ch = fData.byteAt(fCurrentOffset) & 0xff;
183 if (ch != 0x20 && ch != 0x09 && ch != 0x0A && ch != 0x0D)
184 return false;
185 if (skipPastChar)
186 fCurrentOffset++;
187 return true;
188 }
189 public void skipPastSpaces() throws IOException {
190 while (true) {
191 int ch = fData.byteAt(fCurrentOffset) & 0xff;
192 if (ch != 0x20 && ch != 0x09 && ch != 0x0A && ch != 0x0D)
193 return;
194 fCurrentOffset++;
195 }
196 }
197 public boolean skippedString(char[] s) throws IOException {
198 int offset = fCurrentOffset;
199 for (int i = 0; i < s.length; i++) {
200 if (fData.byteAt(offset) != s[i])
201 return false;
202 offset++;
203 }
204 fCurrentOffset = offset;
205 return true;
206 }
207 public int scanStringLiteral() throws Exception {
208 boolean single;
209 if (!(single = lookingAtChar('\'', true)) && !lookingAtChar('\"', true)) {
210 return XMLEntityHandler.STRINGLIT_RESULT_QUOTE_REQUIRED;
211 }
212 int offset = fCurrentOffset;
213 char qchar = single ? '\'' : '\"';
214 while (true) {
215 byte b = fData.byteAt(fCurrentOffset);
216 if (b == qchar)
217 break;
218 if (b == -1)
219 return XMLEntityHandler.STRINGLIT_RESULT_QUOTE_REQUIRED;
220 fCurrentOffset++;
221 }
222 int length = fCurrentOffset - offset;
223 StringBuffer str = new StringBuffer (length);
224 for (int i = 0; i < length; i++) {
225 str.append((char)fData.byteAt(offset + i));
226 }
227 int stringIndex = fStringPool.addString(str.toString());
228 fCurrentOffset++; return stringIndex;
230 }
231 public void append(XMLEntityHandler.CharBuffer charBuffer, int offset, int length) {
235 throw new RuntimeException ("RDR002 cannot happen");
236 }
237 public int addString(int offset, int length) {
238 throw new RuntimeException ("RDR002 cannot happen");
239 }
240 public int addSymbol(int offset, int length) {
241 throw new RuntimeException ("RDR002 cannot happen");
242 }
243 public void skipToChar(char ch) throws IOException {
244 throw new IOException ("RDR002 cannot happen");
245 }
246 public void skipPastName(char fastcheck) throws IOException {
247 throw new IOException ("RDR002 cannot happen");
248 }
249 public void skipPastNmtoken(char fastcheck) throws IOException {
250 throw new IOException ("RDR002 cannot happen");
251 }
252 public boolean lookingAtValidChar(boolean skipPastChar) throws IOException {
253 throw new IOException ("RDR002 cannot happen");
254 }
255 public int scanInvalidChar() throws IOException {
256 throw new IOException ("RDR002 cannot happen");
257 }
258 public int scanCharRef(boolean hex) throws IOException {
259 throw new IOException ("RDR002 cannot happen");
260 }
261 public int scanAttValue(char qchar, boolean asSymbol) throws IOException {
262 throw new IOException ("RDR002 cannot happen");
263 }
264 public int scanEntityValue(int qchar, boolean createString) throws IOException {
265 throw new IOException ("RDR002 cannot happen");
266 }
267 public boolean scanExpectedName(char fastcheck, StringPool.CharArrayRange expectedName) throws IOException {
268 throw new IOException ("RDR002 cannot happen");
269 }
270 public void scanQName(char fastcheck, QName qname) throws IOException {
271 throw new IOException ("RDR002 cannot happen");
272 }
273 public int scanName(char fastcheck) throws IOException {
274 throw new IOException ("RDR002 cannot happen");
275 }
276 public int scanContent(QName element) throws IOException {
277 throw new IOException ("RDR002 cannot happen");
278 }
279 }
280 }
281 |
Popular Tags |
Java API By Example, From Geeks To Geeks.