| 1230 |
leisch |
1 |
/*==============================================================================
|
|
|
2 |
|
|
|
3 |
Project: Simple JAVA Search Engine for Keyword Search
|
|
|
4 |
|
|
|
5 |
JAVA Source file for the class SearchEngine
|
|
|
6 |
|
|
|
7 |
COPYRIGHT (C), 1998, Thomas Baier
|
|
|
8 |
ALL RIGHTS RESERVED.
|
|
|
9 |
|
|
|
10 |
$Source: /scratch/CVS-ARCHIVE/R/doc/html/search/SearchEngine.java,v $
|
|
|
11 |
|
|
|
12 |
$Revision: 1.1 $
|
|
|
13 |
|
|
|
14 |
$Date: 1998/05/15 10:38:10 $
|
|
|
15 |
|
|
|
16 |
$Author: leisch $
|
|
|
17 |
|
|
|
18 |
==============================================================================*/
|
|
|
19 |
|
|
|
20 |
|
|
|
21 |
/* -------------------------------- Imports --------------------------------- */
|
|
|
22 |
|
|
|
23 |
import java.applet.*;
|
|
|
24 |
import java.awt.*;
|
|
|
25 |
import java.net.*;
|
|
|
26 |
import java.io.*;
|
|
|
27 |
import java.util.*;
|
|
|
28 |
|
|
|
29 |
|
|
|
30 |
/*==============================================================================
|
|
|
31 |
Interface of class SearchEngine
|
|
|
32 |
==============================================================================*/
|
|
|
33 |
|
|
|
34 |
/*------------------------------------------------------------------------------
|
|
|
35 |
CLASS: SearchEngine
|
|
|
36 |
SUPER: Applet
|
|
|
37 |
CONF. TO:
|
|
|
38 |
PURPOSE:
|
|
|
39 |
NOTES:
|
|
|
40 |
|
|
|
41 |
HISTORY: 98-04-26: created
|
|
|
42 |
------------------------------------------------------------------------------*/
|
|
|
43 |
public class SearchEngine extends Applet
|
|
|
44 |
{
|
|
|
45 |
/*============================================================================
|
|
|
46 |
Public methods
|
|
|
47 |
============================================================================*/
|
|
|
48 |
|
|
|
49 |
/*----------------------------------------------------------------------------
|
|
|
50 |
INTERFACE:
|
|
|
51 |
PURPOSE: default constructor
|
|
|
52 |
|
|
|
53 |
NOTES:
|
|
|
54 |
|
|
|
55 |
PARAMS:
|
|
|
56 |
THROWS:
|
|
|
57 |
RETURNS: void
|
|
|
58 |
|
|
|
59 |
HISTORY: 98-04-26: created
|
|
|
60 |
98-05-03: now reads the index file
|
|
|
61 |
----------------------------------------------------------------------------*/
|
|
|
62 |
public SearchEngine ()
|
|
|
63 |
{
|
|
|
64 |
iIndexTable = null;
|
|
|
65 |
iSearchTerm = null;
|
|
|
66 |
|
|
|
67 |
Tracer.write ("SearchEngine initializing\n");
|
|
|
68 |
|
|
|
69 |
return;
|
|
|
70 |
}
|
|
|
71 |
|
|
|
72 |
|
|
|
73 |
/*----------------------------------------------------------------------------
|
|
|
74 |
INTERFACE:
|
|
|
75 |
PURPOSE: return general information about the applet
|
|
|
76 |
|
|
|
77 |
NOTES:
|
|
|
78 |
|
|
|
79 |
PARAMS:
|
|
|
80 |
THROWS:
|
|
|
81 |
RETURNS: String: the information
|
|
|
82 |
|
|
|
83 |
HISTORY: 98-04-26: created
|
|
|
84 |
----------------------------------------------------------------------------*/
|
|
|
85 |
public String getAppletInfo ()
|
|
|
86 |
{
|
|
|
87 |
return "Name: SearchEngine\r\n" +
|
|
|
88 |
"Author: Thomas Baier\r\n" +
|
|
|
89 |
"(C) 1998 Thomas Baier, ALL RIGHTS RESERVED";
|
|
|
90 |
}
|
|
|
91 |
|
|
|
92 |
|
|
|
93 |
/*----------------------------------------------------------------------------
|
|
|
94 |
INTERFACE:
|
|
|
95 |
PURPOSE: perform initialization
|
|
|
96 |
|
|
|
97 |
NOTES: creates the controls
|
|
|
98 |
|
|
|
99 |
PARAMS:
|
|
|
100 |
THROWS:
|
|
|
101 |
RETURNS: void
|
|
|
102 |
|
|
|
103 |
HISTORY: 98-04-26: created
|
|
|
104 |
98-05-03: now a do-nothing
|
|
|
105 |
----------------------------------------------------------------------------*/
|
|
|
106 |
public void init ()
|
|
|
107 |
{
|
|
|
108 |
resize(640, 240);
|
|
|
109 |
|
|
|
110 |
// get the name of the index file
|
|
|
111 |
String indexName = getParameter (cIndexKeyword);
|
|
|
112 |
String searchTerm = getParameter (cSearchKeyword);
|
|
|
113 |
|
|
|
114 |
Tracer.write ("Index file is \"" + indexName + "\"\n");
|
|
|
115 |
Tracer.write ("Search term is \"" + searchTerm + "\"\n");
|
|
|
116 |
|
|
|
117 |
// use a default index file if none specified
|
|
|
118 |
if (indexName == null) {
|
|
|
119 |
indexName = cIndexFile;
|
|
|
120 |
}
|
|
|
121 |
|
|
|
122 |
iSearchTerm = searchTerm;
|
|
|
123 |
|
|
|
124 |
/*
|
|
|
125 |
* examine the URL to get the search term...
|
|
|
126 |
*
|
|
|
127 |
* if the URL ends with ?SEARCHTERM=xxxxx we know, xxxxx is the search term
|
|
|
128 |
*/
|
|
|
129 |
{
|
|
|
130 |
URL url = getDocumentBase ();
|
|
|
131 |
String urlString = url.toString ();
|
|
|
132 |
int index = urlString.indexOf ("?" + cSearchKeyword + "=");
|
|
|
133 |
|
|
|
134 |
Tracer.write ("URL is \"" + urlString + "\"\n");
|
|
|
135 |
|
|
|
136 |
// if found, take the rest as the search string
|
|
|
137 |
if (index >= 0) {
|
|
|
138 |
iSearchTerm =
|
|
|
139 |
urlString.substring (index + 2 + cSearchKeyword.length ());
|
|
|
140 |
Tracer.write ("found search term \"" + iSearchTerm + "\" in URL\n");
|
|
|
141 |
}
|
|
|
142 |
}
|
|
|
143 |
|
|
|
144 |
readIndexFile (indexName);
|
|
|
145 |
|
|
|
146 |
return;
|
|
|
147 |
}
|
|
|
148 |
|
|
|
149 |
|
|
|
150 |
/*----------------------------------------------------------------------------
|
|
|
151 |
INTERFACE:
|
|
|
152 |
PURPOSE: perform "destructor" code
|
|
|
153 |
|
|
|
154 |
NOTES: not required here
|
|
|
155 |
|
|
|
156 |
PARAMS:
|
|
|
157 |
THROWS:
|
|
|
158 |
RETURNS: void
|
|
|
159 |
|
|
|
160 |
HISTORY: 98-04-26: created
|
|
|
161 |
----------------------------------------------------------------------------*/
|
|
|
162 |
public void destroy ()
|
|
|
163 |
{
|
|
|
164 |
return;
|
|
|
165 |
}
|
|
|
166 |
|
|
|
167 |
|
|
|
168 |
/*----------------------------------------------------------------------------
|
|
|
169 |
INTERFACE:
|
|
|
170 |
PURPOSE: perform painting of the applet
|
|
|
171 |
|
|
|
172 |
NOTES: not required for our applet, controls do everything
|
|
|
173 |
|
|
|
174 |
PARAMS: Graphics g: the graphics context to draw on
|
|
|
175 |
THROWS:
|
|
|
176 |
RETURNS: void
|
|
|
177 |
|
|
|
178 |
HISTORY: 98-04-26: created
|
|
|
179 |
----------------------------------------------------------------------------*/
|
|
|
180 |
public void paint (Graphics g)
|
|
|
181 |
{
|
|
|
182 |
|
|
|
183 |
return;
|
|
|
184 |
}
|
|
|
185 |
|
|
|
186 |
|
|
|
187 |
/*----------------------------------------------------------------------------
|
|
|
188 |
INTERFACE:
|
|
|
189 |
PURPOSE: perform startup code everytime visiting the applet
|
|
|
190 |
|
|
|
191 |
NOTES:
|
|
|
192 |
|
|
|
193 |
PARAMS:
|
|
|
194 |
THROWS:
|
|
|
195 |
RETURNS: void
|
|
|
196 |
|
|
|
197 |
HISTORY: 98-04-26: created
|
|
|
198 |
98-05-10: start the tracer
|
|
|
199 |
----------------------------------------------------------------------------*/
|
|
|
200 |
public void start ()
|
|
|
201 |
{
|
|
|
202 |
Tracer.start ();
|
|
|
203 |
return;
|
|
|
204 |
}
|
|
|
205 |
|
|
|
206 |
|
|
|
207 |
/*----------------------------------------------------------------------------
|
|
|
208 |
INTERFACE:
|
|
|
209 |
PURPOSE: perform cleanup evertime the applet "loses" the focus
|
|
|
210 |
|
|
|
211 |
NOTES:
|
|
|
212 |
|
|
|
213 |
PARAMS:
|
|
|
214 |
THROWS:
|
|
|
215 |
RETURNS: void
|
|
|
216 |
|
|
|
217 |
HISTORY: 98-04-26: created
|
|
|
218 |
98-05-10: stop the tracer
|
|
|
219 |
----------------------------------------------------------------------------*/
|
|
|
220 |
public void stop ()
|
|
|
221 |
{
|
|
|
222 |
Tracer.stop ();
|
|
|
223 |
return;
|
|
|
224 |
}
|
|
|
225 |
|
|
|
226 |
|
|
|
227 |
/*----------------------------------------------------------------------------
|
|
|
228 |
INTERFACE:
|
|
|
229 |
PURPOSE: perform the search and return the search results as a string
|
|
|
230 |
|
|
|
231 |
NOTES:
|
|
|
232 |
|
|
|
233 |
PARAMS:
|
|
|
234 |
THROWS:
|
|
|
235 |
RETURNS: void
|
|
|
236 |
|
|
|
237 |
HISTORY: 98-05-03: created
|
|
|
238 |
98-05-08: new format for output
|
|
|
239 |
98-05-09: added trace
|
|
|
240 |
98-05-10: now a front-end for search()
|
|
|
241 |
----------------------------------------------------------------------------*/
|
|
|
242 |
public String search (String key)
|
|
|
243 |
{
|
|
|
244 |
iSearchTerm = key;
|
|
|
245 |
|
|
|
246 |
return internalSearch ();
|
|
|
247 |
}
|
|
|
248 |
|
|
|
249 |
|
|
|
250 |
/*----------------------------------------------------------------------------
|
|
|
251 |
INTERFACE:
|
|
|
252 |
PURPOSE: perform the search (back-end)
|
|
|
253 |
|
|
|
254 |
NOTES:
|
|
|
255 |
|
|
|
256 |
PARAMS:
|
|
|
257 |
THROWS:
|
|
|
258 |
RETURNS: void
|
|
|
259 |
|
|
|
260 |
HISTORY: 98-05-10: created
|
|
|
261 |
----------------------------------------------------------------------------*/
|
|
|
262 |
public String internalSearch ()
|
|
|
263 |
{
|
|
|
264 |
Tracer.write ("Search for \"" + iSearchTerm + "\" started");
|
|
|
265 |
|
|
|
266 |
Vector foundItems = null;
|
|
|
267 |
|
|
|
268 |
if (iSearchTerm != null) {
|
|
|
269 |
foundItems = iIndexTable.search (iSearchTerm);
|
|
|
270 |
} else {
|
|
|
271 |
foundItems = null;
|
|
|
272 |
}
|
|
|
273 |
|
|
|
274 |
String result = null;
|
|
|
275 |
|
|
|
276 |
// if nothing found, return a special string
|
|
|
277 |
if (foundItems == null) {
|
|
|
278 |
result = "No matches for <b>\"" +
|
|
|
279 |
iSearchTerm +
|
|
|
280 |
"\"</b> have been found!<hr>";
|
|
|
281 |
} else {
|
|
|
282 |
Enumeration cursor = foundItems.elements ();
|
|
|
283 |
|
|
|
284 |
result =
|
|
|
285 |
"The search string was <b>\"" +
|
|
|
286 |
iSearchTerm +
|
|
|
287 |
"<b>\"" +
|
|
|
288 |
"<hr>" +
|
|
|
289 |
"<dl>";
|
|
|
290 |
|
|
|
291 |
while (cursor.hasMoreElements ()) {
|
|
|
292 |
IndexEntry entry = (IndexEntry) cursor.nextElement ();
|
|
|
293 |
|
|
|
294 |
/*
|
|
|
295 |
* the format for every entry is
|
|
|
296 |
*
|
|
|
297 |
* title
|
|
|
298 |
* description
|
|
|
299 |
*/
|
|
|
300 |
|
|
|
301 |
result +=
|
|
|
302 |
"<dt><a href=\"" +
|
|
|
303 |
entry.getURL () +
|
|
|
304 |
"\">" +
|
|
|
305 |
entry.getTitle () +
|
|
|
306 |
"</a></dt>\n";
|
|
|
307 |
result += "<dd>" + entry.getDescription () + "</dd>\n";
|
|
|
308 |
}
|
|
|
309 |
|
|
|
310 |
result += "</dl>";
|
|
|
311 |
}
|
|
|
312 |
|
|
|
313 |
return result;
|
|
|
314 |
}
|
|
|
315 |
|
|
|
316 |
|
|
|
317 |
/*============================================================================
|
|
|
318 |
Protected methods
|
|
|
319 |
============================================================================*/
|
|
|
320 |
|
|
|
321 |
/*============================================================================
|
|
|
322 |
Private methods
|
|
|
323 |
============================================================================*/
|
|
|
324 |
|
|
|
325 |
/*----------------------------------------------------------------------------
|
|
|
326 |
INTERFACE:
|
|
|
327 |
PURPOSE: read the index file
|
|
|
328 |
|
|
|
329 |
NOTES:
|
|
|
330 |
|
|
|
331 |
PARAMS:
|
|
|
332 |
THROWS:
|
|
|
333 |
RETURNS: void
|
|
|
334 |
|
|
|
335 |
HISTORY: 98-04-26: created
|
|
|
336 |
98-05-08: now use an IndexStream
|
|
|
337 |
98-05-10: also use prefix and suffix, build URL from first key
|
|
|
338 |
----------------------------------------------------------------------------*/
|
|
|
339 |
private void readIndexFile (String idxFile)
|
|
|
340 |
{
|
|
|
341 |
// create the index table
|
|
|
342 |
iIndexTable = new IndexTable ();
|
|
|
343 |
|
|
|
344 |
URL baseURL = getCodeBase ();
|
|
|
345 |
|
|
|
346 |
// get the index file and parse its contents
|
|
|
347 |
try {
|
|
|
348 |
URL idxFileURL = new URL (baseURL,idxFile);
|
|
|
349 |
|
|
|
350 |
// get an IndexStream object for ease of parsing
|
|
|
351 |
IndexStream idxStream = new IndexStream (idxFileURL);
|
|
|
352 |
|
|
|
353 |
// now start parsing...
|
|
|
354 |
|
|
|
355 |
/*
|
|
|
356 |
* An entry consists of a title, keywords, an URL and a description.
|
|
|
357 |
* everything else is ignored. Every entry starts with the keyword
|
|
|
358 |
* "Entry" (case is ignored)
|
|
|
359 |
*
|
|
|
360 |
* must-have entries are "Entry" and "Keywords"
|
|
|
361 |
*/
|
|
|
362 |
String entry = null;
|
|
|
363 |
String keywords = null;
|
|
|
364 |
String url = null;
|
|
|
365 |
String description = "";
|
|
|
366 |
String prefix = "";
|
|
|
367 |
String suffix = "";
|
|
|
368 |
|
|
|
369 |
Value value = idxStream.popEntry ();
|
|
|
370 |
|
|
|
371 |
while (value != null) {
|
|
|
372 |
// parse the value now
|
|
|
373 |
if (value.getKey ().equalsIgnoreCase ("entry")) {
|
|
|
374 |
// if a new entry is about to start, add the current one
|
|
|
375 |
addEntry (entry,keywords,description,url,prefix,suffix);
|
|
|
376 |
|
|
|
377 |
entry = value.getValue ();
|
|
|
378 |
keywords = null;
|
|
|
379 |
url = null;
|
|
|
380 |
description = "";
|
|
|
381 |
} else if (value.getKey ().equalsIgnoreCase ("keywords")) {
|
|
|
382 |
keywords = value.getValue ();
|
|
|
383 |
} else if (value.getKey ().equalsIgnoreCase ("url")) {
|
|
|
384 |
// use prefix and suffix
|
|
|
385 |
url = prefix + value.getValue () + suffix;
|
|
|
386 |
} else if (value.getKey ().equalsIgnoreCase ("description")) {
|
|
|
387 |
description = value.getValue ();
|
|
|
388 |
} else if (value.getKey ().equalsIgnoreCase ("prefix")) {
|
|
|
389 |
prefix = value.getValue ();
|
|
|
390 |
Tracer.write ("using new URL prefix \"" + prefix + "\"\n");
|
|
|
391 |
} else if (value.getKey ().equalsIgnoreCase ("suffix")) {
|
|
|
392 |
suffix = value.getValue ();
|
|
|
393 |
Tracer.write ("using new URL suffix \"" + suffix + "\"\n");
|
|
|
394 |
}
|
|
|
395 |
value = idxStream.popEntry ();
|
|
|
396 |
}
|
|
|
397 |
|
|
|
398 |
// the final entry just read
|
|
|
399 |
addEntry (entry,keywords,description,url,prefix,suffix);
|
|
|
400 |
} catch (MalformedURLException exc) {
|
|
|
401 |
// an error occured while reading...
|
|
|
402 |
}
|
|
|
403 |
|
|
|
404 |
return;
|
|
|
405 |
}
|
|
|
406 |
|
|
|
407 |
|
|
|
408 |
/*----------------------------------------------------------------------------
|
|
|
409 |
INTERFACE:
|
|
|
410 |
PURPOSE:
|
|
|
411 |
|
|
|
412 |
NOTES:
|
|
|
413 |
|
|
|
414 |
PARAMS:
|
|
|
415 |
THROWS:
|
|
|
416 |
RETURNS: void
|
|
|
417 |
|
|
|
418 |
HISTORY: 98-05-10: created
|
|
|
419 |
----------------------------------------------------------------------------*/
|
|
|
420 |
private void addEntry (String entry,String keywords,
|
|
|
421 |
String description,String url,
|
|
|
422 |
String prefix,String suffix)
|
|
|
423 |
{
|
|
|
424 |
// the entry must be set
|
|
|
425 |
if (entry == null) {
|
|
|
426 |
return;
|
|
|
427 |
}
|
|
|
428 |
|
|
|
429 |
// the keywords must be set, else ignore it
|
|
|
430 |
if (keywords != null) {
|
|
|
431 |
if (url == null) {
|
|
|
432 |
// if the URL is empty, construct one following the rule:
|
|
|
433 |
// URL = prefix + first keyword + suffix
|
|
|
434 |
int endOfFirstKeyword = keywords.indexOf (" ");
|
|
|
435 |
|
|
|
436 |
// because we have trimmed the string, the first character must
|
|
|
437 |
// not be a blank
|
|
|
438 |
if (endOfFirstKeyword >= 0) {
|
|
|
439 |
url = keywords.substring (0,endOfFirstKeyword);
|
|
|
440 |
Tracer.write ("constructing URL, keywords=\"" +
|
|
|
441 |
keywords + "\"" +
|
|
|
442 |
"using \"" +
|
|
|
443 |
url +
|
|
|
444 |
"\" (results in \"" +
|
|
|
445 |
prefix + url + suffix + "\")\n");
|
|
|
446 |
} else {
|
|
|
447 |
// just a single keyword
|
|
|
448 |
url = keywords;
|
|
|
449 |
Tracer.write ("constructing URL, using keyword \"" +
|
|
|
450 |
url +
|
|
|
451 |
"\" (results in \"" +
|
|
|
452 |
prefix + url + suffix + "\")\n");
|
|
|
453 |
}
|
|
|
454 |
// add prefix and suffix
|
|
|
455 |
url = prefix + url + suffix;
|
|
|
456 |
}
|
|
|
457 |
IndexEntry idxEntry =
|
|
|
458 |
new IndexEntry (entry,keywords,description,url);
|
|
|
459 |
iIndexTable.addElement (idxEntry);
|
|
|
460 |
}
|
|
|
461 |
|
|
|
462 |
return;
|
|
|
463 |
}
|
|
|
464 |
|
|
|
465 |
|
|
|
466 |
/*============================================================================
|
|
|
467 |
Instance Variables
|
|
|
468 |
============================================================================*/
|
|
|
469 |
private IndexTable iIndexTable;
|
|
|
470 |
private String iSearchTerm;
|
|
|
471 |
|
|
|
472 |
|
|
|
473 |
/*============================================================================
|
|
|
474 |
Static Data
|
|
|
475 |
============================================================================*/
|
|
|
476 |
|
|
|
477 |
private static final String cIndexFile = "index.txt";
|
|
|
478 |
private static final String cIndexKeyword = "INDEXFILE";
|
|
|
479 |
private static final String cSearchKeyword = "SEARCHTERM";
|
|
|
480 |
}
|
|
|
481 |
|
|
|
482 |
/*==============================================================================
|
|
|
483 |
|
|
|
484 |
HISTORY:
|
|
|
485 |
|
|
|
486 |
$Log: SearchEngine.java,v $
|
|
|
487 |
Revision 1.1 1998/05/15 10:38:10 leisch
|
|
|
488 |
New: Search Engine
|
|
|
489 |
|
|
|
490 |
Revision 1.4 1998/05/10 22:56:53 baier
|
|
|
491 |
internal search function, parameter expansion
|
|
|
492 |
|
|
|
493 |
Revision 1.3 1998/05/10 02:44:32 baier
|
|
|
494 |
traces, output in HTML via JavaScript, new index generation
|
|
|
495 |
|
|
|
496 |
Revision 1.2 1998/04/26 22:36:34 baier
|
|
|
497 |
documentation changes
|
|
|
498 |
|
|
|
499 |
|
|
|
500 |
|
|
|
501 |
Revision 1.1 1998/04/26 21:32:54 baier
|
|
|
502 |
Initial revision
|
|
|
503 |
|
|
|
504 |
|
|
|
505 |
==============================================================================*/
|
|
|
506 |
|
|
|
507 |
|
|
|
508 |
// Local Variables:
|
|
|
509 |
// mode: Java
|
|
|
510 |
// mode: font-lock
|
|
|
511 |
// End:
|