|
1
17
18 package org.ajaxanywhere.parser;
19
20 import javax.swing.text.MutableAttributeSet  ;
21 import javax.swing.text.html.HTML ;
22 import javax.swing.text.html.HTMLEditorKit ;
23 import javax.swing.text.html.parser.ParserDelegator ;
24 import java.io.IOException ;
25 import java.io.Reader ;
26 import java.io.StringReader ;
27 import java.util.*;
28 import java.util.regex.Pattern ;
29
30 public class ResponseParserHTML implements ResponseParser {
31 private static ResponseParser ourInstance = new ResponseParserHTML();
32 public static final Pattern SCRIPT_START_PATTERN = Pattern.compile("<script",Pattern.CASE_INSENSITIVE);
33 public static final Pattern SCRIPT_END_PATTERN = Pattern.compile("</script",Pattern.CASE_INSENSITIVE);
34
35 public static ResponseParser getInstance() {
36 return ourInstance;
37 }
38
39 private ResponseParserHTML() {
40 }
41
42 public ResponseBean parse(String html) {
43
47 String dummy = findDummy(html);
48 html = html.replaceAll("<>", "<" + dummy + ">").replaceAll("</>", "</" + dummy + ">");
49
50 html = SCRIPT_START_PATTERN.matcher(html).replaceAll("<DIR");
53 html = SCRIPT_END_PATTERN.matcher(html).replaceAll("</DIR");
54
55 ResponseBean responseBean = doParse(html);
56 String htmlContent = responseBean.getHtmlContent();
57 if (htmlContent != null)
58 responseBean.setHtmlContent(htmlContent.replaceAll("<" + dummy + ">", "<>").replaceAll("</" + dummy + ">", "</>"));
59 for (int i = 0; i < responseBean.getScriptContents().size(); i++) {
60 String s = (String ) responseBean.getScriptContents().get(i);
61 responseBean.getScriptContents().set(i, s.replaceAll("<" + dummy + ">", "<>").replaceAll("</" + dummy + ">", "</>"));
62 }
63 return responseBean;
64 }
65
66 private String findDummy(String html) {
67 String dummy;
68 do {
69 dummy = Double.toString(Math.random());
70 } while (html.indexOf(dummy) != -1);
71
72 return dummy;
73 }
74
75 public ResponseBean doParse(final String html) {
76
77 ResponseBean res = new ResponseBean();
78 try {
79 final StringBuffer contentHTML = new StringBuffer ();
80 final List scripts = res.getScriptContents();
81 final Set images = res.getImages();
82
83 HTMLEditorKit.ParserCallback callback = new HTMLEditorKit.ParserCallback () {
84 private boolean insideScript;
85 private StringBuffer scriptContent = new StringBuffer ();
86 private int lastStop;
87
88 private void append(String str) {
89 if (insideScript) {
90 scriptContent.append(str);
91 } else {
92 contentHTML.append(str);
93 }
94 }
95
96 public void appendSinceLastStop(int newPos) {
97 if (lastStop > newPos)
98 return;
99 append(html.substring(lastStop, newPos));
100 lastStop = newPos;
101 }
102
103 private void flushScript() {
104 int posScriptEnd = scriptContent.indexOf(">");
105 if (posScriptEnd == -1)
106 posScriptEnd = 0;
107 int posC1 = scriptContent.indexOf("<!--", posScriptEnd);
108 int posC11 = scriptContent.indexOf("<![CDATA[", posScriptEnd);
109 int posQ1 = scriptContent.indexOf("'", posScriptEnd);
110 int posQ2 = scriptContent.indexOf("\"", posScriptEnd);
111
112 if ((posC1 != -1) && (posQ2 == -1 || posC1 < posQ2) && (posQ1 == -1 || posC1 < posQ1))
113 scriptContent.delete(posC1, posC1 + 4);
114
115 if ((posC11 != -1) && (posQ2 == -1 || posC11 < posQ2) && (posQ1 == -1 || posC11 < posQ1))
116 scriptContent.delete(posC11, posC11 + 9);
117
118
119 posQ1 = scriptContent.lastIndexOf("'", posScriptEnd);
120 posQ2 = scriptContent.lastIndexOf("\"", posScriptEnd);
121 int posC2 = scriptContent.indexOf("-->", posScriptEnd);
122 int posC22 = scriptContent.indexOf("//]]>", posScriptEnd);
123
124 if ((posC2 != -1) && (posQ2 == -1 || posC2 > posQ2) && (posQ1 == -1 || posC2 > posQ1))
125 scriptContent.delete(posC2, posC2 + 3);
126 if ((posC22 != -1) && (posQ2 == -1 || posC22 > posQ2) && (posQ1 == -1 || posC22 > posQ1))
127 scriptContent.delete(posC22, posC22 + 5);
128
129 int len = scriptContent.length();
130 if (len >0 && scriptContent.charAt(len -1) =='>'){
131 int lastEndTagPos = scriptContent.lastIndexOf("</");
132 if (lastEndTagPos!=-1)
133 scriptContent.setLength(lastEndTagPos);
134 }
135 scripts.add(scriptContent.toString());
136 scriptContent.setLength(0);
137 }
138
139 public void handleText(char[] data, int pos) {
140 appendSinceLastStop(pos);
141 }
142
143 public void handleComment(char[] data, int pos) {
144 appendSinceLastStop(pos);
145 if (data == null)
146 flushScript();
147 }
148
149 public void handleError(String errorMsg, int pos) {
150
151 }
152
153 public void handleEndTag(HTML.Tag tag, int pos) {
154 if (pos == -1)
155 return;
156 if (lastStop > pos)
157 return;
158 appendSinceLastStop(pos);
159
160 if (tag == HTML.Tag.DIR && insideScript) {
161 int posScriptEnd = html.indexOf('>', pos);
162
163 if (posScriptEnd != -1)
164 lastStop = posScriptEnd + 1;
165
166 insideScript = false;
167 flushScript();
168 }
169 }
170
171 public void handleSimpleTag(HTML.Tag tag, MutableAttributeSet attributes, int pos) {
172 handleStartTag(tag, attributes, pos);
173 }
174
175 public void handleStartTag(HTML.Tag tag, MutableAttributeSet attributes, int pos) {
176 appendSinceLastStop(pos);
177 if (tag == HTML.Tag.DIR) {
178 insideScript = true;
179 } else if (tag == HTML.Tag.IMG) {
180 images.add(attributes.getAttribute(HTML.Attribute.SRC));
181 }
182 }
183
184 };
185
186 Reader reader = new StringReader (html);
187 new ParserDelegator ().parse(reader, callback, false);
188 callback.handleComment(null, html.length());
189
190 res.setHtmlContent(contentHTML.toString());
191 return res;
192
193 } catch (IOException e) {
194 throw new RuntimeException (e.toString()); }
196 }
197
198 }
199 |
Popular Tags |
Java API By Example, From Geeks To Geeks.