|
1
19
20 package org.netbeans.modules.xml.xdm.nodes;
21
22 import java.io.ByteArrayInputStream  ;
23 import java.io.CharConversionException ;
24 import java.io.EOFException ;
25 import java.io.IOException ;
26 import java.io.InputStream ;
27 import java.io.InputStreamReader ;
28 import java.io.Reader ;
29 import java.io.UnsupportedEncodingException ;
30 import java.util.List ;
31 import java.util.ArrayList ;
32 import java.util.Stack ;
33 import javax.swing.text.BadLocationException ;
34 import org.netbeans.editor.BaseDocument;
35 import org.netbeans.editor.TokenID;
36 import org.netbeans.editor.TokenItem;
37 import org.netbeans.editor.ext.ExtSyntaxSupport;
38 import org.netbeans.modules.xml.text.syntax.XMLTokenIDs;
39
40 public class XMLSyntaxParser {
41
42 public Document parse(BaseDocument basedoc)
43 throws IOException , BadLocationException {
44
45 ExtSyntaxSupport sup = (ExtSyntaxSupport)basedoc.getSyntaxSupport();
47 TokenItem token = sup.getTokenChain(0, basedoc.getLength());
48
49 Stack <NodeImpl> stack = new Stack <NodeImpl>();
51 Document doc = new Document();
52 stack.push(doc);
53 NodeImpl currentNode = doc;
54 List <Token> currentTokens = new ArrayList <Token>();
55 if(isValid(token) && token.getTokenID().getNumericID() == XMLTokenIDs.TEXT_ID) {
57 currentTokens.add(Token.create(token.getImage(),TokenType.TOKEN_CHARACTER_DATA));
58 token = token.getNext();
59 if(isValid(token) && token.getTokenID().getNumericID() != XMLTokenIDs.PI_START_ID) {
61 currentNode.setTokens(new ArrayList <Token>(currentTokens));
62 currentTokens.clear();
63 }
64 }
65
66 while (token != null) {
67 isValid(token);
68 TokenID tokenId = token.getTokenID();
69 int numericId = tokenId.getNumericID();
70 String image = token.getImage();
71 TokenType coreTokenId = TokenType.TOKEN_WHITESPACE;
72 switch(numericId) {
73 case XMLTokenIDs.PI_START_ID:
74 {
75 coreTokenId = TokenType.TOKEN_PI_START_TAG;
76 currentTokens.add(Token.create(image,coreTokenId));
77 break;
78 }
79 case XMLTokenIDs.PI_END_ID:
80 {
81 coreTokenId = TokenType.TOKEN_PI_END_TAG;
82 currentTokens.add(Token.create(image,coreTokenId));
83 if(currentNode instanceof Document) {
84 if(token.getNext().getTokenID().getNumericID() == XMLTokenIDs.TEXT_ID) {
85 token = token.getNext();
86 currentTokens.add(Token.create(token.getImage(),TokenType.TOKEN_CHARACTER_DATA));
87 }
88 stack.push(currentNode);
89 }
90 List <Token> list = new ArrayList <Token>(currentNode.getTokens());
91 list.addAll(currentTokens);
92 currentNode.setTokens(list);
93 currentTokens.clear();
94 break;
95 }
96 case XMLTokenIDs.TAG_ID:
97 {
98 int len = image.length();
99 if (image.charAt(len-1) == '>') {
100 Token endToken =
101 Token.create(image,TokenType.TOKEN_ELEMENT_END_TAG);
102 if(len == 2) {
103 currentNode = stack.pop();
104 endToken =
105 Token.create(image,TokenType.TOKEN_ELEMENT_END_TAG);
106 } else if(!(currentNode instanceof Element)) {
107 currentNode = stack.peek();
108 }
109 currentTokens.add(endToken);
110 currentNode.getTokensForWrite().addAll(currentTokens);
111 currentTokens.clear();
112 } else {
113 coreTokenId = TokenType.TOKEN_ELEMENT_START_TAG;
114 if(image.startsWith("</")) {
115 currentNode = stack.pop();
116 if(!currentNode.getTokens().get(0).getValue().substring(1).
117 equals(image.substring(2))) {
118 throw new IOException ("Invalid token '" + image +
119 "' found in document: " +
120 "Please use the text editor to resolve the issues...");
121 } else { String saveTokenImage = image;
123 currentTokens.add(Token.create(image,coreTokenId));
124 token = token.getNext();
125 while(token != null) {
126 int nextNumericId = token.getTokenID().getNumericID();
127 if(nextNumericId != XMLTokenIDs.WS_ID)
128 break;
129 coreTokenId = TokenType.TOKEN_WHITESPACE;
130 currentTokens.add(Token.create(token.getImage(), coreTokenId));
131 token = token.getNext();
132 }
133 if(token == null || !token.getImage().equals(">"))
134 throw new IOException ("Invalid token '" + saveTokenImage +
135 "' does not end with '>': Please use the " +
136 "text editor to resolve the issues...");
137 continue;
138 }
139 } else {
140 currentNode = new Element();
141 Node parent = stack.peek();
142 parent.appendChild(currentNode);
143 stack.push(currentNode);
144 currentTokens.add(Token.create(image,coreTokenId));
145 currentNode.setTokens(new ArrayList <Token>(currentTokens));
146 currentTokens.clear();
147 }
148 }
149 break;
150 }
151 case XMLTokenIDs.ARGUMENT_ID:
152 {
153 coreTokenId = TokenType.TOKEN_ATTR_NAME;
154 currentNode = new Attribute();
155 Element parent = (Element)stack.peek();
156 parent.appendAttribute((Attribute)currentNode);
157 currentTokens.add(Token.create(image,coreTokenId));
158 break;
159 }
160 case XMLTokenIDs.VALUE_ID:
161 {
162 TokenItem nextToken = token.getNext();
163 isValid(nextToken);
164 int nextNumericId = nextToken.getTokenID().getNumericID();
165 while(nextNumericId == XMLTokenIDs.VALUE_ID || nextNumericId == XMLTokenIDs.CHARACTER_ID) {
166 token = token.getNext();
167 image = image.concat(token.getImage());
168 nextNumericId = token.getNext().getTokenID().getNumericID();
169 }
170 coreTokenId = TokenType.TOKEN_ATTR_VAL;
171 currentTokens.add(Token.create(image,coreTokenId));
172 currentNode.setTokens(new ArrayList <Token>(currentTokens));
173 currentTokens.clear();
174 break;
175 }
176 case XMLTokenIDs.BLOCK_COMMENT_ID:
177 {
178 Node parent = stack.peek();
179 currentTokens.add(Token.create(image, coreTokenId));
180 if (image.endsWith(Token.COMMENT_END.getValue())) {
181 String combinedString = combineString(currentTokens);
182 Comment comment = new Comment(combinedString);
183 if (parent instanceof Element) {
184 ((Element)parent).appendChild(comment, false);
185 } else { if(numericId != XMLTokenIDs.BLOCK_COMMENT_ID &&
187 token.getImage().trim().length() > 0) {
188 throw new IOException ("Invalid token '" + token.getImage() +
189 "' found in document: " +
190 "Please use the text editor to resolve the issues...");
191 }
192 parent.appendChild(comment);
193 }
194 currentTokens.clear();
195 }
196 break;
197 }
198 case XMLTokenIDs.TEXT_ID:
199 case XMLTokenIDs.CHARACTER_ID:
200 {
201 coreTokenId = TokenType.TOKEN_CHARACTER_DATA;
202 currentNode = new Text();
203 currentTokens.add(Token.create(image,coreTokenId));
204 if(numericId == XMLTokenIDs.TEXT_ID) {
205 while(token.getNext() != null) {
206 int nextNumericId = token.getNext().getTokenID().getNumericID();
207 if(nextNumericId != XMLTokenIDs.TEXT_ID && nextNumericId != XMLTokenIDs.CHARACTER_ID)
208 break;
209 token = token.getNext();
210 currentTokens.add(Token.create(token.getImage(),coreTokenId));
211 }
212 }
213 currentNode.setTokens(new ArrayList <Token>(currentTokens));
214 Node parent = stack.peek();
215 if (parent instanceof Element) {
216 ((Element)parent).appendChild(currentNode, false);
217 } else { if(numericId != XMLTokenIDs.BLOCK_COMMENT_ID &&
219 token.getImage().trim().length() > 0) {
220 throw new IOException ("Invalid token '" + token.getImage() +
221 "' found in document: " +
222 "Please use the text editor to resolve the issues...");
223 }
224 parent.appendChild(currentNode);
225 }
226 currentTokens.clear();
227 break;
228 }
229 case XMLTokenIDs.WS_ID:
230 {
231 coreTokenId = TokenType.TOKEN_WHITESPACE;
232 currentTokens.add(Token.create(image, coreTokenId));
233 break;
234 }
235 case XMLTokenIDs.OPERATOR_ID:
236 {
237 coreTokenId = TokenType.TOKEN_ATTR_EQUAL;
238 currentTokens.add(Token.create(image,coreTokenId));
239 break;
240 }
241 case XMLTokenIDs.DECLARATION_ID:
242 {
243 coreTokenId = TokenType.TOKEN_DTD_VAL;
244 currentTokens.add(Token.create(image, coreTokenId));
245 while(token.getNext() != null) {
246 int nextNumericId = token.getNext().getTokenID().getNumericID();
247 if(nextNumericId != XMLTokenIDs.DECLARATION_ID && nextNumericId != XMLTokenIDs.VALUE_ID)
248 break;
249 token = token.getNext();
250 currentTokens.add(Token.create(token.getImage(),coreTokenId));
251 }
252 break;
253 }
254 case XMLTokenIDs.PI_CONTENT_ID:
255 {
256 coreTokenId = TokenType.TOKEN_PI_VAL;
257 currentTokens.add(Token.create(image, coreTokenId));
258 break;
259 }
260 case XMLTokenIDs.PI_TARGET_ID:
261 {
262 coreTokenId = TokenType.TOKEN_PI_NAME;
263 currentTokens.add(Token.create(image, coreTokenId));
264 break;
265 }
266 case XMLTokenIDs.CDATA_SECTION_ID:
267 {
268 Node parent = stack.peek();
269 CData cdata = new CData(image);
270 if (parent instanceof Element) {
271 ((Element)parent).appendChild(cdata, false);
272 } else { throw new IOException ("CDATA is not valid as direct child of document" +
274 "Please use the text editor to resolve the issues...");
275 }
276 coreTokenId = TokenType.TOKEN_CDATA_VAL;
277 break;
278 }
279 case XMLTokenIDs.ERROR_ID:
280 case XMLTokenIDs.EOL_ID:
281 default:
282 throw new IOException ("Invalid token '" + token.getImage() + "' found in document: " +
284 "Please use the text editor to resolve the issues...");
285 }
286 token = token.getNext();
287 }
288 Node result = stack.pop();
289 if(result instanceof Document) {
290 return (Document)result;
291 }
292 else
293 throw new IOException ("Document not well formed/Invalid: " +
295 "Please use the text editor to resolve the issues...");
296 }
297
298 private boolean isValid(TokenItem token) throws IOException {
299 if(token!=null && token.getTokenID()!=null)
300 return true;
301 else
302 throw new IOException ("Document parsed is invalid: Please use the text " +
303 "editor to resolve the issues...");
304 }
305
306 private String combineString(List <Token> tokens) {
307 StringBuilder sb = new StringBuilder ();
308 for (Token t: tokens) {
309 sb.append(t.getValue());
310 }
311 return sb.toString();
312 }
313 }
314 |
Popular Tags |
Java API By Example, From Geeks To Geeks.