|
1
32
33 package websphinx.searchengine;
34
35 import websphinx.*;
36 import java.util.Vector  ;
37 import java.util.Enumeration ;
38 import java.util.NoSuchElementException ;
39
40 public class Search extends Crawler implements Enumeration {
41
42 int maxResults;
43 int walkedResults;
45 Vector results = new Vector (); int nextResult = 0; int approxCount = -1; boolean crawling = false;
49
50 public Search () {
51 this (Integer.MAX_VALUE);
52 }
53
54 public Search (int maxResults) {
55 this.maxResults = maxResults;
56 setDepthFirst (false);
57 setMaxDepth (Integer.MAX_VALUE);
58 EventLog.monitor (this); }
60
61 public Search (SearchEngine engine, String keywords, int maxResults) {
62 this (maxResults);
63 addQuery (engine, keywords);
64 search ();
65 }
66
67 public Search (SearchEngine engine, String keywords) {
68 this (engine, keywords, Integer.MAX_VALUE);
69 }
70
71 public void addQuery (SearchEngine engine, String keywords) {
72 addRoot (new Link (engine.makeQuery (keywords)));
73 addClassifier (engine);
74 walkedResults += engine.getResultsPerPage ();
75 }
76
77 public void search () {
78 crawling = true;
79 Thread thread = new Thread (this, "Search");
80 thread.setDaemon (true);
81 thread.start ();
82 }
83
84 public int count () {
85 synchronized (results) {
86 try {
88 while (approxCount == -1 && crawling)
89 results.wait ();
90 } catch (InterruptedException e) {}
91 return approxCount;
92 }
93 }
94
95 public boolean hasMoreElements () {
96 synchronized (results) {
97 try {
98 while (nextResult >= results.size() && crawling)
99 results.wait ();
100 } catch (InterruptedException e) {}
101
102 return nextResult < results.size ();
103 }
104 }
105
106 public Object nextElement () {
107 return nextResult ();
108 }
109
110 public SearchEngineResult nextResult () {
111 if (!hasMoreElements ())
112 throw new NoSuchElementException ();
113 synchronized (results) {
114 SearchEngineResult result = (SearchEngineResult)results.elementAt (nextResult++);
115 if (result.rank == 0)
116 result.rank = nextResult;
117 return result;
118 }
119 }
120
121 public void run () {
122 super.run ();
123 synchronized (results) {
124 if (approxCount == -1)
125 approxCount = 0;
126 crawling = false;
127 results.notify ();
128 }
129 }
130
131 public void visit (Page page) {
132 synchronized (results) {
133 if (approxCount == -1)
134 approxCount = page.getNumericLabel ("searchengine.count", new Integer (0)).intValue();
135
136 Region[] ser = page.getFields ("searchengine.results");
137 for (int i=0; i<ser.length; ++i) {
138 if (results.size() == maxResults) {
139 stop ();
140 return;
141 }
142 results.addElement (ser[i]);
143 }
144 results.notify ();
145 }
146 }
147
148 public boolean shouldVisit (Link link) {
149 if (walkedResults >= maxResults
150 || !link.hasLabel ("searchengine.more-results"))
151 return false;
152 SearchEngine engine = (SearchEngine)link.getSource().getObjectLabel("searchengine.source");
153 walkedResults += engine.getResultsPerPage ();
154 return true;
155 }
156
157
158 public static void main (String [] args) throws Exception {
159 if (args.length == 0) {
160 System.err.println ("Search <search engine classname> [-max n] <keywords>*");
161 return;
162 }
163
164 SearchEngine engine = (SearchEngine) Class.forName (args[0]).newInstance ();
165
166 int max = Integer.MAX_VALUE;
167 int firstKeyword = 1;
168 if (args[1].equals ("-max")) {
169 max = Integer.parseInt (args[2]);
170 firstKeyword = 3;
171 }
172
173 Search ms = new Search (max);
174 ms.addQuery (engine, concat (args, firstKeyword));
175 ms.search ();
176 while (ms.hasMoreElements ())
177 System.out.println (ms.nextResult ());
178 }
179
180 static String concat (String [] args, int start) {
181 StringBuffer buf = new StringBuffer ();
182 for (int i=start; i<args.length; ++i) {
183 if (buf.length() > 0)
184 buf.append (' ');
185 buf.append (args[i]);
186 }
187 return buf.toString ();
188 }
189
190 }
191
|
Popular Tags
|
Java API By Example, From Geeks To Geeks.