|
1
27 package org.htmlparser.tests.lexerTests;
28
29 import org.htmlparser.Node;
30 import org.htmlparser.Parser;
31 import org.htmlparser.PrototypicalNodeFactory;
32 import org.htmlparser.Tag;
33 import org.htmlparser.tags.LinkTag;
34 import org.htmlparser.tags.MetaTag;
35 import org.htmlparser.tests.ParserTestCase;
36 import org.htmlparser.util.ParserException;
37
38 public class TagTests extends ParserTestCase {
39 static
40 {
41 System.setProperty ("org.htmlparser.tests.lexerTests.TagTests", "TagTests");
42 }
43
44 private static final String  TEST_HTML = "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.0 Transitional//EN\">" +
45 "<!-- Server: sf-web2 -->\n" +
46 "<html lang=\"en\">\n" +
47 " <head><link rel=\"stylesheet\" type=\"text/css\" HREF=\"http://sourceforge.net/cssdef.php\">\n" +
48 " <meta http-equiv=\"Content-Type\" content=\"text/html; charset=iso-8859-1\">\n" +
49 " <TITLE>SourceForge.net: Modify: 711073 - HTMLTagParser not threadsafe as a static variable in Tag</TITLE>\n" +
50 " <SCRIPT language=\"JavaScript\" type=\"text/javascript\">\n" +
51 " <!--\n" +
52 " function help_window(helpurl) {\n" +
53 " HelpWin = window.open( 'http://sourceforge.net' + helpurl,'HelpWindow','scrollbars=yes,resizable=yes,toolbar=no,height=400,width=400');\n" +
54 " }\n" +
55 " // -->\n" +
56 " </SCRIPT>\n" +
57 " <link rel=\"SHORTCUT ICON\" HREF=\"/images/favicon.ico\">\n" +
58 "<!-- This is temp javascript for the jump button. If we could actually have a jump script on the server side that would be ideal -->\n" +
59 "<script language=\"JavaScript\" type=\"text/javascript\">\n" +
60 "<!--\n" +
61 " function jump(targ,selObj,restore){ //v3.0\n" +
62 " if (selObj.options[selObj.selectedIndex].value)\n" +
63 " eval(targ+\".location='\"+selObj.options[selObj.selectedIndex].value+\"'\");\n" +
64 " if (restore) selObj.selectedIndex=0;\n" +
65 " }\n" +
66 " //-->\n" +
67 "</script>\n" +
68 "<a HREF=\"http://normallink.com/sometext.html\">\n" +
69 "<style type=\"text/css\">\n" +
70 "<!--\n" +
71 "A:link { text-decoration:none }\n" +
72 "A:visited { text-decoration:none }\n" +
73 "A:active { text-decoration:none }\n" +
74 "A:hover { text-decoration:underline; color:#0066FF; }\n" +
75 "-->\n" +
76 "</style>\n" +
77 "</head>\n" +
78 "<body bgcolor=\"#FFFFFF\" text=\"#000000\" leftmargin=\"0\" topmargin=\"0\" marginwidth=\"0\" marginheight=\"0\" link=\"#003399\" vlink=\"#003399\" alink=\"#003399\">\n";
79 private int testProgress;
80
81 public TagTests (String name) {
82 super(name);
83 }
84
85 public void testTagWithQuotes() throws Exception {
86 String testHtml =
87 "<img SRC=\"http://g-images.amazon.com/images/G/01/merchants/logos/marshall-fields-logo-20.gif\" width=87 height=20 border=0 alt=\"Marshall Field's\">";
88
89 createParser(testHtml);
90 parser.setNodeFactory (new PrototypicalNodeFactory (true));
91 parseAndAssertNodeCount(1);
92 assertType("should be Tag",Tag.class,node[0]);
93 Tag tag = (Tag)node[0];
94 assertStringEquals("alt","Marshall Field's",tag.getAttribute("ALT"));
95 assertStringEquals(
96 "html",
97 testHtml,
98 tag.toHtml()
99 );
100 }
101
102 public void testEmptyTag() throws Exception
103 {
104 String html = "<custom/>";
105 createParser(html);
106 parser.setNodeFactory (new PrototypicalNodeFactory (true));
107 parseAndAssertNodeCount(1);
108 assertType("should be Tag",Tag.class,node[0]);
109 Tag tag = (Tag)node[0];
110 assertStringEquals("tag name","CUSTOM",tag.getTagName());
111 assertTrue("empty tag",tag.isEmptyXmlTag());
112 assertStringEquals(
113 "html",
114 html,
115 tag.toHtml()
116 );
117 }
118
119 public void testTagWithCloseTagSymbolInAttribute() throws ParserException {
120 createParser("<tag att=\"a>b\">");
121 parser.setNodeFactory (new PrototypicalNodeFactory (true));
122 parseAndAssertNodeCount(1);
123 assertType("should be Tag",Tag.class,node[0]);
124 Tag tag = (Tag)node[0];
125 assertStringEquals("attribute","a>b",tag.getAttribute("att"));
126 }
127
128 public void testTagWithOpenTagSymbolInAttribute() throws ParserException {
129 createParser("<tag att=\"a<b\">");
130 parser.setNodeFactory (new PrototypicalNodeFactory (true));
131 parseAndAssertNodeCount(1);
132 assertType("should be Tag",Tag.class,node[0]);
133 Tag tag = (Tag)node[0];
134 assertStringEquals("attribute","a<b",tag.getAttribute("att"));
135 }
136
137 public void testTagWithSingleQuote() throws ParserException {
138 String html = "<tag att=\'a<b\'>";
139 createParser(html);
140 parser.setNodeFactory (new PrototypicalNodeFactory (true));
141 parseAndAssertNodeCount(1);
142 assertType("should be Tag",Tag.class,node[0]);
143 Tag tag = (Tag)node[0];
144 assertStringEquals("html",html,tag.toHtml());
145 assertStringEquals("attribute","a<b",tag.getAttribute("att"));
146 }
147
148
153 public void testMultiLine1 () throws ParserException
154 {
155 String html = "<meta name=\"foo\" content=\"foo<bar>\">";
156 createParser(html);
157 parseAndAssertNodeCount (1);
158 assertType ("should be MetaTag", MetaTag.class, node[0]);
159 Tag tag = (Tag)node[0];
160 assertStringEquals ("html",html, tag.toHtml ());
161 String attribute1 = tag.getAttribute ("NAME");
162 assertStringEquals ("attribute 1","foo", attribute1);
163 String attribute2 = tag.getAttribute ("CONTENT");
164 assertStringEquals ("attribute 2","foo<bar>", attribute2);
165 }
166
167 public void testMultiLine2 () throws ParserException
168 {
169 String html = "<meta name=\"foo\" content=\"foo<bar\">";
170 createParser(html);
171 parseAndAssertNodeCount (1);
172 assertType ("should be MetaTag", MetaTag.class, node[0]);
173 Tag tag = (Tag)node[0];
174 assertStringEquals ("html",html, tag.toHtml ());
175 String attribute1 = tag.getAttribute ("NAME");
176 assertStringEquals ("attribute 1","foo", attribute1);
177 String attribute2 = tag.getAttribute ("CONTENT");
178 assertStringEquals ("attribute 2","foo<bar", attribute2);
179 }
180
181 public void testMultiLine3 () throws ParserException
182 {
183 String html = "<meta name=\"foo\" content=\"foobar>\">";
184 createParser(html);
185 parseAndAssertNodeCount (1);
186 assertType ("should be MetaTag", MetaTag.class, node[0]);
187 Tag tag = (Tag)node[0];
188 assertStringEquals ("html",html, tag.toHtml ());
189 String attribute1 = tag.getAttribute ("NAME");
190 assertStringEquals ("attribute 1","foo", attribute1);
191 String attribute2 = tag.getAttribute ("CONTENT");
192 assertStringEquals ("attribute 2","foobar>", attribute2);
193 }
194
195 public void testMultiLine4 () throws ParserException
196 {
197 String html = "<meta name=\"foo\" content=\"foo\nbar>\">";
198 createParser(html);
199 parseAndAssertNodeCount (1);
200 assertType ("should be MetaTag", MetaTag.class, node[0]);
201 Tag tag = (Tag)node[0];
202 assertStringEquals ("html",html, tag.toHtml ());
203 String attribute1 = tag.getAttribute ("NAME");
204 assertStringEquals ("attribute 1","foo", attribute1);
205 String attribute2 = tag.getAttribute ("CONTENT");
206 assertStringEquals ("attribute 2","foo\nbar>", attribute2);
207 }
208
209
213 public void testMultiLine5 () throws ParserException
214 {
215 String html = "<meta name=\"foo\" content=\"<foo>\nbar\">";
218 createParser(html);
219 parseAndAssertNodeCount (1);
220 assertType ("should be MetaTag", MetaTag.class, node[0]);
221 Tag tag = (Tag)node[0];
222 assertStringEquals ("html",html, tag.toHtml ());
223 String attribute1 = tag.getAttribute ("NAME");
224 assertStringEquals ("attribute 1","foo", attribute1);
225 String attribute2 = tag.getAttribute ("CONTENT");
226 assertStringEquals ("attribute 2","<foo>\nbar", attribute2);
227 }
228
229
233 public void testMultiLine6 () throws ParserException
234 {
235 String html = "<meta name=\"foo\" content=\"foo>\nbar\">";
238 createParser(html);
239 parseAndAssertNodeCount (1);
240 assertType ("should be MetaTag", MetaTag.class, node[0]);
241 Tag tag = (Tag)node[0];
242 assertStringEquals ("html",html, tag.toHtml ());
243 String attribute1 = tag.getAttribute ("NAME");
244 assertStringEquals ("attribute 1","foo", attribute1);
245 String attribute2 = tag.getAttribute ("CONTENT");
246 assertStringEquals ("attribute 2","foo>\nbar", attribute2);
247 }
248
249
253 public void testMultiLine7 () throws ParserException
254 {
255 String html = "<meta name=\"foo\" content=\"<foo\nbar\"";
258 createParser(html);
259 parseAndAssertNodeCount (1);
260 assertType ("should be MetaTag", MetaTag.class, node[0]);
261 Tag tag = (Tag)node[0];
262 assertStringEquals ("html",html + ">", tag.toHtml ());
263 String attribute1 = tag.getAttribute ("NAME");
264 assertStringEquals ("attribute 1","foo", attribute1);
265 String attribute2 = tag.getAttribute ("CONTENT");
266 assertStringEquals ("attribute 2","<foo\nbar", attribute2);
267 }
268
269
272
273
277 public void testThreadSafety() throws Exception
278 {
279 createParser("<html></html>");
280 parser.setNodeFactory (new PrototypicalNodeFactory (true));
281 String testHtml1 = "<a HREF=\"/cgi-bin/view_search?query_text=postdate>20020701&txt_clr=White&bg_clr=Red&url=http://localhost/Testing/Report1.html\">20020702 Report 1</A>" +
282 TEST_HTML;
283
284 String testHtml2 = "<a HREF=\"http://normallink.com/sometext.html\">" +
285 TEST_HTML;
286 ParsingThread parsingThread [] =
287 new ParsingThread[100];
288 testProgress = 0;
289 for (int i=0;i<parsingThread.length;i++) {
290 if (i<parsingThread.length/2)
291 parsingThread[i] =
292 new ParsingThread(i,testHtml1,parsingThread.length);
293 else
294 parsingThread[i] =
295 new ParsingThread(i,testHtml2,parsingThread.length);
296
297 Thread thread = new Thread (parsingThread[i]);
298 thread.start();
299 }
300
301 int completionValue = computeCompletionValue(parsingThread.length);
302
303 do {
304 try {
305 Thread.sleep(500);
306 }
307 catch (InterruptedException e) {
308 }
309 }
310 while (testProgress!=completionValue);
311 for (int i=0;i<parsingThread.length;i++)
312 {
313 if (!parsingThread[i].passed())
314 {
315 assertNotNull("Thread "+i+" link 1",parsingThread[i].getLink1());
316 assertNotNull("Thread "+i+" link 2",parsingThread[i].getLink2());
317 if (i<parsingThread.length/2) {
318 assertStringEquals(
319 "Thread "+i+", link 1:",
320 "/cgi-bin/view_search?query_text=postdate>20020701&txt_clr=White&bg_clr=Red&url=http://localhost/Testing/Report1.html",
321 parsingThread[i].getLink1().getLink()
322 );
323 assertStringEquals(
324 "Thread "+i+", link 2:",
325 "http://normallink.com/sometext.html",
326 parsingThread[i].getLink2().getLink()
327 );
328 } else {
329 assertStringEquals(
330 "Thread "+i+", link 1:",
331 "http://normallink.com/sometext.html",
332 parsingThread[i].getLink1().getLink()
333 );
334 assertNotNull("Thread "+i+" link 2",parsingThread[i].getLink2());
335 assertStringEquals(
336 "Thread "+i+", link 2:",
337 "/cgi-bin/view_search?query_text=postdate>20020701&txt_clr=White&bg_clr=Red&url=http://localhost/Testing/Report1.html",
338 parsingThread[i].getLink2().getLink()
339 );
340 }
341 }
342 }
343 }
344
345 private int computeCompletionValue(int numThreads) {
346 return numThreads * (numThreads - 1) / 2;
347 }
348
349 class ParsingThread implements Runnable {
350 Parser mParser;
351 int mId;
352 LinkTag mLink1;
353 LinkTag mLink2;
354 boolean mResult;
355 int mMax;
356
357 ParsingThread(int id, String testHtml, int max) {
358 mId = id;
359 mMax = max;
360 mParser = Parser.createParser(testHtml, null);
361 }
362
363 public void run() {
364 try {
365 mResult = false;
366 Node linkTag [] = mParser.extractAllNodesThatAre(LinkTag.class);
367 mLink1 = (LinkTag)linkTag[0];
368 mLink2 = (LinkTag)linkTag[1];
369 if (mId < mMax / 2) {
370 if (mLink1.getLink().equals("/cgi-bin/view_search?query_text=postdate>20020701&txt_clr=White&bg_clr=Red&url=http://localhost/Testing/Report1.html") &&
371 mLink2.getLink().equals("http://normallink.com/sometext.html"))
372 mResult = true;
373 } else {
374 if (mLink1.getLink().equals("http://normallink.com/sometext.html") &&
375 mLink2.getLink().equals("http://normallink.com/sometext.html"))
376 mResult = true;
377 }
378 }
379 catch (ParserException e) {
380 System.err.println("Parser Exception");
381 e.printStackTrace();
382 }
383 finally {
384 testProgress += mId;
385 }
386 }
387
388 public LinkTag getLink1() {
389 return (mLink1);
390 }
391
392 public LinkTag getLink2() {
393 return (mLink2);
394 }
395
396 public boolean passed() {
397 return (mResult);
398 }
399 }
400
401
404 public void testStandAloneToHTML () throws ParserException
405 {
406 String html = "<input disabled>";
407 createParser(html);
408 parser.setNodeFactory (new PrototypicalNodeFactory (true));
409 parseAndAssertNodeCount (1);
410 assertType ("should be Tag", Tag.class, node[0]);
411 Tag tag = (Tag)node[0];
412 assertStringEquals ("html", html, tag.toHtml ());
413 }
414
415
418 public void testMissingValueToHTML () throws ParserException
419 {
420 String html = "<input disabled=>";
421 createParser(html);
422 parser.setNodeFactory (new PrototypicalNodeFactory (true));
423 parseAndAssertNodeCount (1);
424 assertType ("should be Tag", Tag.class, node[0]);
425 Tag tag = (Tag)node[0];
426 assertStringEquals ("html", html, tag.toHtml ());
427 }
428 } |
Popular Tags |
Java API By Example, From Geeks To Geeks.