author | aefimov |
Mon, 31 Mar 2014 19:03:41 +0400 | |
changeset 23777 | ce87cedb71cf |
parent 12457 | c348e06f0e82 |
child 25834 | aba3efbf4ec5 |
permissions | -rw-r--r-- |
6 | 1 |
/* |
2 |
* reserved comment block |
|
3 |
* DO NOT REMOVE OR ALTER! |
|
4 |
*/ |
|
5 |
/* |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
6 |
* Licensed to the Apache Software Foundation (ASF) under one or more |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
7 |
* contributor license agreements. See the NOTICE file distributed with |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
8 |
* this work for additional information regarding copyright ownership. |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
9 |
* The ASF licenses this file to You under the Apache License, Version 2.0 |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
10 |
* (the "License"); you may not use this file except in compliance with |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
11 |
* the License. You may obtain a copy of the License at |
6 | 12 |
* |
13 |
* http://www.apache.org/licenses/LICENSE-2.0 |
|
14 |
* |
|
15 |
* Unless required by applicable law or agreed to in writing, software |
|
16 |
* distributed under the License is distributed on an "AS IS" BASIS, |
|
17 |
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
|
18 |
* See the License for the specific language governing permissions and |
|
19 |
* limitations under the License. |
|
20 |
*/ |
|
21 |
||
22 |
// Sep 14, 2000: |
|
23 |
// Fixed problem with namespace handling. Contributed by |
|
24 |
// David Blondeau <blondeau@intalio.com> |
|
25 |
// Sep 14, 2000: |
|
26 |
// Fixed serializer to report IO exception directly, instead at |
|
27 |
// the end of document processing. |
|
28 |
// Reported by Patrick Higgins <phiggins@transzap.com> |
|
29 |
// Aug 21, 2000: |
|
30 |
// Fixed bug in startDocument not calling prepare. |
|
31 |
// Reported by Mikael Staldal <d96-mst-ingen-reklam@d.kth.se> |
|
32 |
// Aug 21, 2000: |
|
33 |
// Added ability to omit DOCTYPE declaration. |
|
34 |
||
35 |
package com.sun.org.apache.xml.internal.serialize; |
|
36 |
||
37 |
import java.io.IOException; |
|
38 |
import java.io.OutputStream; |
|
39 |
import java.io.Writer; |
|
40 |
||
41 |
import com.sun.org.apache.xerces.internal.dom.DOMMessageFormatter; |
|
42 |
import com.sun.org.apache.xerces.internal.util.NamespaceSupport; |
|
43 |
import com.sun.org.apache.xerces.internal.util.SymbolTable; |
|
44 |
import com.sun.org.apache.xerces.internal.util.XML11Char; |
|
45 |
import com.sun.org.apache.xerces.internal.util.XMLChar; |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
46 |
import org.w3c.dom.DOMError; |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
47 |
import org.w3c.dom.Document; |
6 | 48 |
import org.xml.sax.SAXException; |
49 |
||
50 |
/** |
|
51 |
* Implements an XML serializer supporting both DOM and SAX pretty |
|
52 |
* serializing. For usage instructions see {@link Serializer}. |
|
53 |
* <p> |
|
54 |
* If an output stream is used, the encoding is taken from the |
|
55 |
* output format (defaults to <tt>UTF-8</tt>). If a writer is |
|
56 |
* used, make sure the writer uses the same encoding (if applies) |
|
57 |
* as specified in the output format. |
|
58 |
* <p> |
|
59 |
* The serializer supports both DOM and SAX. SAX serializing is done by firing |
|
60 |
* SAX events and using the serializer as a document handler. DOM serializing is done |
|
61 |
* by calling {@link #serialize(Document)} or by using DOM Level 3 |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
62 |
* {@link org.w3c.dom.ls.LSSerializer} and |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
63 |
* serializing with {@link org.w3c.dom.ls.LSSerializer#write}, |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
64 |
* {@link org.w3c.dom.ls.LSSerializer#writeToString}. |
6 | 65 |
* <p> |
66 |
* If an I/O exception occurs while serializing, the serializer |
|
67 |
* will not throw an exception directly, but only throw it |
|
68 |
* at the end of serializing (either DOM or SAX's {@link |
|
69 |
* org.xml.sax.DocumentHandler#endDocument}. |
|
70 |
* <p> |
|
71 |
* For elements that are not specified as whitespace preserving, |
|
72 |
* the serializer will potentially break long text lines at space |
|
73 |
* boundaries, indent lines, and serialize elements on separate |
|
74 |
* lines. Line terminators will be regarded as spaces, and |
|
75 |
* spaces at beginning of line will be stripped. |
|
76 |
* @author <a href="mailto:arkin@intalio.com">Assaf Arkin</a> |
|
77 |
* @author <a href="mailto:rahul.srivastava@sun.com">Rahul Srivastava</a> |
|
78 |
* @author Elena Litani IBM |
|
79 |
* @see Serializer |
|
80 |
*/ |
|
81 |
public class XML11Serializer |
|
82 |
extends XMLSerializer { |
|
83 |
||
84 |
// |
|
85 |
// constants |
|
86 |
// |
|
87 |
||
88 |
protected static final boolean DEBUG = false; |
|
89 |
||
90 |
// |
|
91 |
// data |
|
92 |
// |
|
93 |
||
94 |
// |
|
95 |
// DOM Level 3 implementation: variables intialized in DOMSerializerImpl |
|
96 |
// |
|
97 |
||
98 |
/** stores namespaces in scope */ |
|
99 |
protected NamespaceSupport fNSBinder; |
|
100 |
||
101 |
/** stores all namespace bindings on the current element */ |
|
102 |
protected NamespaceSupport fLocalNSBinder; |
|
103 |
||
104 |
/** symbol table for serialization */ |
|
105 |
protected SymbolTable fSymbolTable; |
|
106 |
||
107 |
// is node dom level 1 node? |
|
108 |
protected boolean fDOML1 = false; |
|
109 |
// counter for new prefix names |
|
110 |
protected int fNamespaceCounter = 1; |
|
111 |
protected final static String PREFIX = "NS"; |
|
112 |
||
113 |
/** |
|
114 |
* Controls whether namespace fixup should be performed during |
|
115 |
* the serialization. |
|
116 |
* NOTE: if this field is set to true the following |
|
117 |
* fields need to be initialized: fNSBinder, fLocalNSBinder, fSymbolTable, |
|
118 |
* XMLSymbols.EMPTY_STRING, fXmlSymbol, fXmlnsSymbol, fNamespaceCounter. |
|
119 |
*/ |
|
120 |
protected boolean fNamespaces = false; |
|
121 |
||
122 |
/** |
|
123 |
* Constructs a new serializer. The serializer cannot be used without |
|
124 |
* calling {@link #setOutputCharStream} or {@link #setOutputByteStream} |
|
125 |
* first. |
|
126 |
*/ |
|
127 |
public XML11Serializer() { |
|
128 |
super( ); |
|
129 |
_format.setVersion("1.1"); |
|
130 |
} |
|
131 |
||
132 |
||
133 |
/** |
|
134 |
* Constructs a new serializer. The serializer cannot be used without |
|
135 |
* calling {@link #setOutputCharStream} or {@link #setOutputByteStream} |
|
136 |
* first. |
|
137 |
*/ |
|
138 |
public XML11Serializer( OutputFormat format ) { |
|
139 |
super( format ); |
|
140 |
_format.setVersion("1.1"); |
|
141 |
} |
|
142 |
||
143 |
||
144 |
/** |
|
145 |
* Constructs a new serializer that writes to the specified writer |
|
146 |
* using the specified output format. If <tt>format</tt> is null, |
|
147 |
* will use a default output format. |
|
148 |
* |
|
149 |
* @param writer The writer to use |
|
150 |
* @param format The output format to use, null for the default |
|
151 |
*/ |
|
152 |
public XML11Serializer( Writer writer, OutputFormat format ) { |
|
153 |
super( writer, format ); |
|
154 |
_format.setVersion("1.1"); |
|
155 |
} |
|
156 |
||
157 |
||
158 |
/** |
|
159 |
* Constructs a new serializer that writes to the specified output |
|
160 |
* stream using the specified output format. If <tt>format</tt> |
|
161 |
* is null, will use a default output format. |
|
162 |
* |
|
163 |
* @param output The output stream to use |
|
164 |
* @param format The output format to use, null for the default |
|
165 |
*/ |
|
166 |
public XML11Serializer( OutputStream output, OutputFormat format ) { |
|
167 |
super( output, format != null ? format : new OutputFormat( Method.XML, null, false ) ); |
|
168 |
_format.setVersion("1.1"); |
|
169 |
} |
|
170 |
||
171 |
//-----------------------------------------// |
|
172 |
// SAX content handler serializing methods // |
|
173 |
//-----------------------------------------// |
|
174 |
||
175 |
||
176 |
public void characters( char[] chars, int start, int length ) |
|
177 |
throws SAXException |
|
178 |
{ |
|
179 |
ElementState state; |
|
180 |
||
181 |
try { |
|
182 |
state = content(); |
|
183 |
||
184 |
// Check if text should be print as CDATA section or unescaped |
|
185 |
// based on elements listed in the output format (the element |
|
186 |
// state) or whether we are inside a CDATA section or entity. |
|
187 |
||
188 |
if ( state.inCData || state.doCData ) { |
|
189 |
int saveIndent; |
|
190 |
||
191 |
// Print a CDATA section. The text is not escaped, but ']]>' |
|
192 |
// appearing in the code must be identified and dealt with. |
|
193 |
// The contents of a text node is considered space preserving. |
|
194 |
if ( ! state.inCData ) { |
|
195 |
_printer.printText( "<![CDATA[" ); |
|
196 |
state.inCData = true; |
|
197 |
} |
|
198 |
saveIndent = _printer.getNextIndent(); |
|
199 |
_printer.setNextIndent( 0 ); |
|
200 |
char ch; |
|
201 |
final int end = start + length; |
|
202 |
for ( int index = start; index < end; ++index ) { |
|
203 |
ch = chars[index]; |
|
204 |
if ( ch == ']' && index + 2 < end && |
|
205 |
chars[ index + 1 ] == ']' && chars[ index + 2 ] == '>' ) { |
|
206 |
_printer.printText("]]]]><![CDATA[>"); |
|
207 |
index +=2; |
|
208 |
continue; |
|
209 |
} |
|
210 |
if (!XML11Char.isXML11Valid(ch)) { |
|
211 |
// check if it is surrogate |
|
212 |
if (++index < end) { |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
213 |
surrogates(ch, chars[index], true); |
6 | 214 |
} |
215 |
else { |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
216 |
fatalError("The character '"+ch+"' is an invalid XML character"); |
6 | 217 |
} |
218 |
continue; |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
219 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
220 |
if ( _encodingInfo.isPrintable(ch) && XML11Char.isXML11ValidLiteral(ch)) { |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
221 |
_printer.printText(ch); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
222 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
223 |
else { |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
224 |
// The character is not printable -- split CDATA section |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
225 |
_printer.printText("]]>&#x"); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
226 |
_printer.printText(Integer.toHexString(ch)); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
227 |
_printer.printText(";<![CDATA["); |
6 | 228 |
} |
229 |
} |
|
230 |
_printer.setNextIndent( saveIndent ); |
|
231 |
||
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
232 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
233 |
else { |
6 | 234 |
|
235 |
int saveIndent; |
|
236 |
||
237 |
if ( state.preserveSpace ) { |
|
238 |
// If preserving space then hold of indentation so no |
|
239 |
// excessive spaces are printed at line breaks, escape |
|
240 |
// the text content without replacing spaces and print |
|
241 |
// the text breaking only at line breaks. |
|
242 |
saveIndent = _printer.getNextIndent(); |
|
243 |
_printer.setNextIndent( 0 ); |
|
244 |
printText( chars, start, length, true, state.unescaped ); |
|
245 |
_printer.setNextIndent( saveIndent ); |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
246 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
247 |
else { |
6 | 248 |
printText( chars, start, length, false, state.unescaped ); |
249 |
} |
|
250 |
} |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
251 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
252 |
catch ( IOException except ) { |
6 | 253 |
throw new SAXException( except ); |
254 |
} |
|
255 |
} |
|
256 |
||
257 |
// |
|
258 |
// overwrite printing functions to make sure serializer prints out valid XML |
|
259 |
// |
|
260 |
protected void printEscaped( String source ) throws IOException { |
|
261 |
int length = source.length(); |
|
262 |
for ( int i = 0 ; i < length ; ++i ) { |
|
263 |
int ch = source.charAt(i); |
|
264 |
if (!XML11Char.isXML11Valid(ch)) { |
|
265 |
if (++i <length) { |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
266 |
surrogates(ch, source.charAt(i), false); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
267 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
268 |
else { |
6 | 269 |
fatalError("The character '"+(char)ch+"' is an invalid XML character"); |
270 |
} |
|
271 |
continue; |
|
272 |
} |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
273 |
if (ch == '\n' || ch == '\r' || ch == '\t' || ch == 0x0085 || ch == 0x2028) { |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
274 |
printHex(ch); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
275 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
276 |
else if (ch == '<') { |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
277 |
_printer.printText("<"); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
278 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
279 |
else if (ch == '&') { |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
280 |
_printer.printText("&"); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
281 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
282 |
else if (ch == '"') { |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
283 |
_printer.printText("""); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
284 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
285 |
else if ((ch >= ' ' && _encodingInfo.isPrintable((char) ch))) { |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
286 |
_printer.printText((char) ch); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
287 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
288 |
else { |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
289 |
printHex(ch); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
290 |
} |
6 | 291 |
} |
292 |
} |
|
293 |
||
294 |
protected final void printCDATAText(String text) throws IOException { |
|
295 |
int length = text.length(); |
|
296 |
char ch; |
|
297 |
||
298 |
for (int index = 0; index < length; ++index) { |
|
299 |
ch = text.charAt(index); |
|
300 |
||
301 |
if (ch == ']' |
|
302 |
&& index + 2 < length |
|
303 |
&& text.charAt(index + 1) == ']' |
|
304 |
&& text.charAt(index + 2) == '>') { // check for ']]>' |
|
305 |
if (fDOMErrorHandler != null){ |
|
306 |
// REVISIT: this means that if DOM Error handler is not registered we don't report any |
|
307 |
// fatal errors and might serialize not wellformed document |
|
308 |
if ((features & DOMSerializerImpl.SPLITCDATA) == 0 |
|
309 |
&& (features & DOMSerializerImpl.WELLFORMED) == 0) { |
|
310 |
// issue fatal error |
|
311 |
String msg = |
|
312 |
DOMMessageFormatter.formatMessage( |
|
313 |
DOMMessageFormatter.SERIALIZER_DOMAIN, |
|
314 |
"EndingCDATA", |
|
315 |
null); |
|
316 |
modifyDOMError( |
|
317 |
msg, |
|
318 |
DOMError.SEVERITY_FATAL_ERROR, |
|
319 |
null, fCurrentNode); |
|
320 |
boolean continueProcess = |
|
321 |
fDOMErrorHandler.handleError(fDOMError); |
|
322 |
if (!continueProcess) { |
|
323 |
throw new IOException(); |
|
324 |
} |
|
325 |
} else { |
|
326 |
// issue warning |
|
327 |
String msg = |
|
328 |
DOMMessageFormatter.formatMessage( |
|
329 |
DOMMessageFormatter.SERIALIZER_DOMAIN, |
|
330 |
"SplittingCDATA", |
|
331 |
null); |
|
332 |
modifyDOMError( |
|
333 |
msg, |
|
334 |
DOMError.SEVERITY_WARNING, |
|
335 |
null, fCurrentNode); |
|
336 |
fDOMErrorHandler.handleError(fDOMError); |
|
337 |
} |
|
338 |
} |
|
339 |
// split CDATA section |
|
340 |
_printer.printText("]]]]><![CDATA[>"); |
|
341 |
index += 2; |
|
342 |
continue; |
|
343 |
} |
|
344 |
||
345 |
if (!XML11Char.isXML11Valid(ch)) { |
|
346 |
// check if it is surrogate |
|
347 |
if (++index < length) { |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
348 |
surrogates(ch, text.charAt(index), true); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
349 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
350 |
else { |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
351 |
fatalError("The character '" + ch + "' is an invalid XML character"); |
6 | 352 |
} |
353 |
continue; |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
354 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
355 |
if (_encodingInfo.isPrintable(ch) |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
356 |
&& XML11Char.isXML11ValidLiteral(ch)) { |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
357 |
_printer.printText(ch); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
358 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
359 |
else { |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
360 |
// The character is not printable -- split CDATA section |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
361 |
_printer.printText("]]>&#x"); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
362 |
_printer.printText(Integer.toHexString(ch)); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
363 |
_printer.printText(";<![CDATA["); |
6 | 364 |
} |
365 |
} |
|
366 |
} |
|
367 |
||
368 |
// note that this "int" should, in all cases, be a char. |
|
369 |
// REVISIT: make it a char... |
|
370 |
protected final void printXMLChar( int ch ) throws IOException { |
|
371 |
||
372 |
if (ch == '\r' || ch == 0x0085 || ch == 0x2028) { |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
373 |
printHex(ch); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
374 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
375 |
else if ( ch == '<') { |
6 | 376 |
_printer.printText("<"); |
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
377 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
378 |
else if (ch == '&') { |
6 | 379 |
_printer.printText("&"); |
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
380 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
381 |
else if (ch == '>'){ |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
382 |
// character sequence "]]>" can't appear in content, therefore |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
383 |
// we should escape '>' |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
384 |
_printer.printText(">"); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
385 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
386 |
else if ( _encodingInfo.isPrintable((char)ch) && XML11Char.isXML11ValidLiteral(ch)) { |
6 | 387 |
_printer.printText((char)ch); |
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
388 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
389 |
else { |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
390 |
printHex(ch); |
6 | 391 |
} |
392 |
} |
|
393 |
||
394 |
||
395 |
||
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
396 |
protected final void surrogates(int high, int low, boolean inContent) throws IOException{ |
6 | 397 |
if (XMLChar.isHighSurrogate(high)) { |
398 |
if (!XMLChar.isLowSurrogate(low)) { |
|
399 |
//Invalid XML |
|
400 |
fatalError("The character '"+(char)low+"' is an invalid XML character"); |
|
401 |
} |
|
402 |
else { |
|
403 |
int supplemental = XMLChar.supplemental((char)high, (char)low); |
|
404 |
if (!XML11Char.isXML11Valid(supplemental)) { |
|
405 |
//Invalid XML |
|
406 |
fatalError("The character '"+(char)supplemental+"' is an invalid XML character"); |
|
407 |
} |
|
408 |
else { |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
409 |
if (inContent && content().inCData) { |
6 | 410 |
_printer.printText("]]>&#x"); |
411 |
_printer.printText(Integer.toHexString(supplemental)); |
|
412 |
_printer.printText(";<![CDATA["); |
|
413 |
} |
|
414 |
else { |
|
415 |
printHex(supplemental); |
|
416 |
} |
|
417 |
} |
|
418 |
} |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
419 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
420 |
else { |
6 | 421 |
fatalError("The character '"+(char)high+"' is an invalid XML character"); |
422 |
} |
|
423 |
||
424 |
} |
|
425 |
||
426 |
||
427 |
protected void printText( String text, boolean preserveSpace, boolean unescaped ) |
|
428 |
throws IOException { |
|
429 |
int index; |
|
430 |
char ch; |
|
431 |
int length = text.length(); |
|
432 |
if ( preserveSpace ) { |
|
433 |
// Preserving spaces: the text must print exactly as it is, |
|
434 |
// without breaking when spaces appear in the text and without |
|
435 |
// consolidating spaces. If a line terminator is used, a line |
|
436 |
// break will occur. |
|
437 |
for ( index = 0 ; index < length ; ++index ) { |
|
438 |
ch = text.charAt( index ); |
|
439 |
if (!XML11Char.isXML11Valid(ch)) { |
|
440 |
// check if it is surrogate |
|
441 |
if (++index <length) { |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
442 |
surrogates(ch, text.charAt(index), true); |
6 | 443 |
} else { |
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
444 |
fatalError("The character '"+ch+"' is an invalid XML character"); |
6 | 445 |
} |
446 |
continue; |
|
447 |
} |
|
448 |
if ( unescaped && XML11Char.isXML11ValidLiteral(ch)) { |
|
449 |
_printer.printText( ch ); |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
450 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
451 |
else { |
6 | 452 |
printXMLChar( ch ); |
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
453 |
} |
6 | 454 |
} |
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
455 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
456 |
else { |
6 | 457 |
// Not preserving spaces: print one part at a time, and |
458 |
// use spaces between parts to break them into different |
|
459 |
// lines. Spaces at beginning of line will be stripped |
|
460 |
// by printing mechanism. Line terminator is treated |
|
461 |
// no different than other text part. |
|
462 |
for ( index = 0 ; index < length ; ++index ) { |
|
463 |
ch = text.charAt( index ); |
|
464 |
if (!XML11Char.isXML11Valid(ch)) { |
|
465 |
// check if it is surrogate |
|
466 |
if (++index <length) { |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
467 |
surrogates(ch, text.charAt(index), true); |
6 | 468 |
} else { |
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
469 |
fatalError("The character '"+ch+"' is an invalid XML character"); |
6 | 470 |
} |
471 |
continue; |
|
472 |
} |
|
473 |
||
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
474 |
if ( unescaped && XML11Char.isXML11ValidLiteral(ch) ) { |
6 | 475 |
_printer.printText( ch ); |
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
476 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
477 |
else { |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
478 |
printXMLChar( ch ); |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
479 |
} |
6 | 480 |
} |
481 |
} |
|
482 |
} |
|
483 |
||
484 |
protected void printText( char[] chars, int start, int length, |
|
485 |
boolean preserveSpace, boolean unescaped ) throws IOException { |
|
486 |
||
487 |
if ( preserveSpace ) { |
|
488 |
// Preserving spaces: the text must print exactly as it is, |
|
489 |
// without breaking when spaces appear in the text and without |
|
490 |
// consolidating spaces. If a line terminator is used, a line |
|
491 |
// break will occur. |
|
492 |
while ( length-- > 0 ) { |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
493 |
char ch = chars[start++]; |
6 | 494 |
if (!XML11Char.isXML11Valid(ch)) { |
495 |
// check if it is surrogate |
|
496 |
if ( length-- > 0) { |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
497 |
surrogates(ch, chars[start++], true); |
6 | 498 |
} else { |
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
499 |
fatalError("The character '"+ch+"' is an invalid XML character"); |
6 | 500 |
} |
501 |
continue; |
|
502 |
} |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
503 |
if ( unescaped && XML11Char.isXML11ValidLiteral(ch)) { |
6 | 504 |
_printer.printText( ch ); |
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
505 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
506 |
else { |
6 | 507 |
printXMLChar( ch ); |
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
508 |
} |
6 | 509 |
} |
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
510 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
511 |
else { |
6 | 512 |
// Not preserving spaces: print one part at a time, and |
513 |
// use spaces between parts to break them into different |
|
514 |
// lines. Spaces at beginning of line will be stripped |
|
515 |
// by printing mechanism. Line terminator is treated |
|
516 |
// no different than other text part. |
|
517 |
while ( length-- > 0 ) { |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
518 |
char ch = chars[start++]; |
6 | 519 |
if (!XML11Char.isXML11Valid(ch)) { |
520 |
// check if it is surrogate |
|
521 |
if ( length-- > 0) { |
|
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
522 |
surrogates(ch, chars[start++], true); |
6 | 523 |
} else { |
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
524 |
fatalError("The character '"+ch+"' is an invalid XML character"); |
6 | 525 |
} |
526 |
continue; |
|
527 |
} |
|
528 |
||
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
529 |
if ( unescaped && XML11Char.isXML11ValidLiteral(ch)) { |
6 | 530 |
_printer.printText( ch ); |
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
531 |
} |
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
532 |
else { |
6 | 533 |
printXMLChar( ch ); |
23777
ce87cedb71cf
8035437: Xerces Update: xml/serialize/DOMSerializerImpl
aefimov
parents:
12457
diff
changeset
|
534 |
} |
6 | 535 |
} |
536 |
} |
|
537 |
} |
|
538 |
||
539 |
public boolean reset() { |
|
540 |
super.reset(); |
|
541 |
return true; |
|
542 |
} |
|
543 |
||
544 |
} |