|
1 package org.apache.ecs.html2ecs;
2
3
60
61 import java.util.Hashtable  ;
62 import java.io.InputStream ;
63 import java.io.IOException ;
64 import java.io.FileInputStream ;
65 import java.io.FileNotFoundException ;
66 import java.lang.reflect.Method ;
67 import org.apache.xerces.parsers.DOMParser;
68 import org.xml.sax.InputSource ;
69 import org.w3c.dom.Node ;
70 import org.w3c.dom.NodeList ;
71 import org.w3c.dom.NamedNodeMap ;
72 import org.w3c.dom.Attr ;
73 import org.apache.ecs.filter.CharacterFilter;
74 import org.apache.ecs.Entities;
75 import org.xml.sax.SAXException ;
76
77
89 public class Html2Ecs extends Hashtable
90 {
91 private static org.w3c.dom.Document doc;
92 private static DOMParser parser;
93 private static InputStream input;
94
95 {
98 put("A","org.apache.ecs.html.A");
99 put("ABBR","org.apache.ecs.html.Abbr");
100 put("ACRONYM","org.apache.ecs.html.Acronym");
101 put("ADDRESS","org.apache.ecs.html.Address");
102 put("APPLET","org.apache.ecs.html.Applet");
103 put("AREA","org.apache.ecs.html.Area");
104 put("B","org.apache.ecs.B");
105 put("BASE","org.apache.ecs.html.Base");
106 put("BASEFONT","org.apache.ecs.html.BaseFont");
107 put("BDO","org.apache.ecs.html.Bdo");
108 put("BIG","org.apache.ecs.html.Big");
109 put("BLINK","org.apache.ecs.html.Blink");
110 put("BLOCKQUOTE","org.apache.ecs.html.BlockQuote");
111 put("BODY","org.apache.ecs.html.Body");
112 put("BR","org.apache.ecs.html.BR");
113 put("BUTTON","org.apache.ecs.html.Button");
114 put("CAPTION","org.apache.ecs.html.Caption");
115 put("CENTER","org.apache.ecs.html.Center");
116 put("CITE","org.apache.ecs.html.Cite");
117 put("CODE","org.apache.ecs.html.Code");
118 put("COL","org.apache.ecs.html.Col");
119 put("COLGROUP","org.apache.ecs.html.ColGroup");
120 put("COMMENT","org.apache.ecs.html.Comment");
121 put("DD","org.apache.ecs.html.DD");
122 put("DEL","org.apache.ecs.html.Del");
123 put("DFN","org.apache.ecs.html.Dfn");
124 put("DIV","org.apache.ecs.html.Div");
125 put("DL","org.apache.ecs.html.DL");
126 put("EM","org.apache.ecs.html.EM");
127 put("FILEDSET","org.apache.ecs.html.FieldSet");
128 put("FONT","org.apache.ecs.html.Font");
129 put("FORM","org.apache.ecs.html.Form");
130 put("FRAME","org.apache.ecs.html.Frame");
131 put("FRAMESET","org.apache.ecs.html.FrameSet");
132 put("H1","org.apache.ecs.html.H1");
133 put("H2","org.apache.ecs.html.H2");
134 put("H3","org.apache.ecs.html.H3");
135 put("H4","org.apache.ecs.html.H4");
136 put("H5","org.apache.ecs.html.H5");
137 put("H6","org.apache.ecs.html.H6");
138 put("HEAD","org.apache.ecs.html.Head");
139 put("HR","org.apache.ecs.html.HR");
140 put("HTML","org.apache.ecs.html.Html");
141 put("I","org.apache.ecs.html.I");
142 put("IFRAME","org.apache.ecs.html.IFrame");
143 put("IMG","org.apache.ecs.html.IMG");
144 put("INPUT","org.apache.ecs.html.Input");
145 put("INS","org.apache.ecs.html.Ins");
146 put("KBD","org.apache.ecs.html.Kbd");
147 put("LABEL","org.apache.ecs.html.Label");
148 put("LEGEND","org.apache.ecs.html.Legend");
149 put("LI","org.apache.ecs.html.LI");
150 put("LINK","org.apache.ecs.html.Link");
151 put("MAP","org.apache.ecs.html.Map");
152 put("META","org.apache.ecs.html.Meta");
153 put("NOBR","org.apache.ecs.html.NOBR");
154 put("NOFRAMES","org.apache.ecs.html.NoFrames");
155 put("NOSCRIPT","org.apache.ecs.html.NoScript");
156 put("OBJECT","org.apache.ecs.html.ObjectElement");
157 put("OL","org.apache.ecs.html.OL");
158 put("OPTGROUP","org.apache.ecs.html.OptGroup");
159 put("OPTION","org.apache.ecs.html.Option");
160 put("P","org.apache.ecs.html.P");
161 put("PARAM","org.apache.ecs.html.Param");
162 put("PRE","org.apache.ecs.html.PRE");
163 put("Q","org.apache.ecs.html.Q");
164 put("S","org.apache.ecs.html.S");
165 put("SAMP","org.apache.ecs.html.Samp");
166 put("SCRIPT","org.apache.ecs.html.Script");
167 put("SELECT","org.apache.ecs.html.Select");
168 put("SMALL","org.apache.ecs.html.Small");
169 put("SPAN","org.apache.ecs.html.Span");
170 put("STRIKE","org.apache.ecs.html.Strike");
171 put("STRONG","org.apache.ecs.html.Strong");
172 put("STYLE","org.apache.ecs.html.Style");
173 put("SUB","org.apache.ecs.html.Sub");
174 put("SUP","org.apache.ecs.html.Sup");
175 put("TABLE","org.apache.ecs.html.Table");
176 put("TBODY","org.apache.ecs.html.TBody");
177 put("TD","org.apache.ecs.html.TD");
178 put("THEAD","org.apache.ecs.html.THead");
179 put("TITLE","org.apache.ecs.html.Title");
180 put("TR","org.apache.ecs.html.TR");
181 put("TT","org.apache.ecs.html.TT");
182 put("U","org.apache.ecs.html.U");
183 put("UL","org.apache.ecs.html.UL");
184 put("VAR","org.apache.ecs.html.Var");
185
186 }
187
188 public Html2Ecs()
189 {
190 super(4);
191 }
192
193 private void process( Node node )
194 {
195 while( node != null)
196 {
197 constructElement(node);
198
199 if ( node.hasChildNodes() &&
201 !node.getNodeName().equals("#document") &&
202 get(node.getNodeName()) == null )
203 {
204 System.out.print(".addElement(");
205 process ( node.getFirstChild() );
206 System.out.println(")");
207 }
208 else
209 {
210 process ( node.getFirstChild() );
211 }
212 node = node.getNextSibling();
214 System.out.print("\n");
215 }
216 }
217
218 private void constructElement( Node node )
219 {
220 if( get(node.getNodeName().toUpperCase()) != null )
221 {
222 System.out.print("new ");
223 System.out.print(get(node.getNodeName().toUpperCase()));
224 System.out.print("()");
225 }
226 constructAttributes( node );
227 constructValue( node );
228 }
229
230 private void constructValue( Node node )
231 {
232 CharacterFilter cf = new CharacterFilter();
233 cf.addAttribute("\"",Entities.QUOT);
234
235 if( node.getNodeName().equals("#text")
236 && node.getNodeValue().length() > 1 )
237 {
238 System.out.print(".addElement(\"");
239 System.out.print(node.getNodeValue());
240 System.out.print("\")");
241 }
242 }
243
244 private void constructAttributes( Node node )
245 {
246 NamedNodeMap attrList = node.getAttributes();
247 Attr attr;
248 Class c = null;
249
250 try
251 {
252 c = Class.forName( (String ) get(node.getNodeName().toUpperCase()) );
253 }
254 catch(Exception e)
255 {
256 }
257
258 if( attrList != null && c != null)
259 {
260 for(int x = 0; x < attrList.getLength(); x++)
261 {
262 attr = (Attr ) attrList.item(x);
263
264 Method [] m = c.getMethods();
265 for(int y = 0; y < m.length; ++y)
266 {
267 if(m[y].getName().toLowerCase().endsWith(attr.getName().toLowerCase()))
268 {
269 System.out.print(".");
270 System.out.print(m[y].getName());
271 System.out.print("(\"");
272 System.out.print(attr.getValue());
273 System.out.print("\")");
274 break;
275 }
276 }
277 }
278 }
279 }
280
281 public static void main(String [] args)
282 {
283 Html2Ecs html2ecs = new Html2Ecs();
284 try
285 {
286 input = new FileInputStream ( args[0] );
287 parser = new DOMParser();
288 parser.parse(new InputSource (input));
289 doc = parser.getDocument();
290 }
291 catch(SAXException se)
292 {
293 System.out.println(se.toString());
294 }
295 catch(FileNotFoundException fnfe)
296 {
297 System.out.println(fnfe.toString());
298 }
299 catch(IOException ioe)
300 {
301 System.out.println(ioe.toString());
302 }
303 html2ecs.process( doc );
304 }
305 }
306 |
Popular Tags |
Java API By Example, From Geeks To Geeks.