annotate src/pagerank/LinkConvertGraph.java @ 5:140272228818 draft

fix LinkToVertex.java
author one
date Wed, 05 Sep 2012 15:38:57 +0900
parents dcd59917a2dd
children c7b139ff27e2
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2
1744340f8be6 add some java files
one
parents:
diff changeset
1 package pagerank;
1744340f8be6 add some java files
one
parents:
diff changeset
2
1744340f8be6 add some java files
one
parents:
diff changeset
3
1744340f8be6 add some java files
one
parents:
diff changeset
4 import java.io.File;
1744340f8be6 add some java files
one
parents:
diff changeset
5 import java.io.FileInputStream;
1744340f8be6 add some java files
one
parents:
diff changeset
6 import java.io.FileNotFoundException;
1744340f8be6 add some java files
one
parents:
diff changeset
7 import java.io.FileOutputStream;
1744340f8be6 add some java files
one
parents:
diff changeset
8 import java.io.IOException;
1744340f8be6 add some java files
one
parents:
diff changeset
9 import java.util.HashMap;
1744340f8be6 add some java files
one
parents:
diff changeset
10 import java.util.HashSet;
1744340f8be6 add some java files
one
parents:
diff changeset
11 import java.util.Map;
1744340f8be6 add some java files
one
parents:
diff changeset
12
1744340f8be6 add some java files
one
parents:
diff changeset
13 import javax.xml.parsers.ParserConfigurationException;
1744340f8be6 add some java files
one
parents:
diff changeset
14 import javax.xml.parsers.SAXParser;
1744340f8be6 add some java files
one
parents:
diff changeset
15 import javax.xml.parsers.SAXParserFactory;
1744340f8be6 add some java files
one
parents:
diff changeset
16
1744340f8be6 add some java files
one
parents:
diff changeset
17 import org.xml.sax.SAXException;
1744340f8be6 add some java files
one
parents:
diff changeset
18
1744340f8be6 add some java files
one
parents:
diff changeset
19 import com.tinkerpop.blueprints.Graph;
1744340f8be6 add some java files
one
parents:
diff changeset
20 import com.tinkerpop.blueprints.Vertex;
1744340f8be6 add some java files
one
parents:
diff changeset
21 import com.tinkerpop.blueprints.impls.tg.TinkerGraph;
1744340f8be6 add some java files
one
parents:
diff changeset
22 import com.tinkerpop.blueprints.util.io.graphml.GraphMLWriter;
1744340f8be6 add some java files
one
parents:
diff changeset
23
1744340f8be6 add some java files
one
parents:
diff changeset
24 import xmlParser.TextTagParser;
1744340f8be6 add some java files
one
parents:
diff changeset
25
1744340f8be6 add some java files
one
parents:
diff changeset
26 public class LinkConvertGraph {
1744340f8be6 add some java files
one
parents:
diff changeset
27
1744340f8be6 add some java files
one
parents:
diff changeset
28 private String filename;
1744340f8be6 add some java files
one
parents:
diff changeset
29 private FileInputStream fis;
1744340f8be6 add some java files
one
parents:
diff changeset
30 private SAXParserFactory factory;
1744340f8be6 add some java files
one
parents:
diff changeset
31 private SAXParser parser;
1744340f8be6 add some java files
one
parents:
diff changeset
32 private TextTagParser xmlParser;
1744340f8be6 add some java files
one
parents:
diff changeset
33
1744340f8be6 add some java files
one
parents:
diff changeset
34 private HashMap<String,HashSet<String>> hash;
1744340f8be6 add some java files
one
parents:
diff changeset
35
1744340f8be6 add some java files
one
parents:
diff changeset
36
1744340f8be6 add some java files
one
parents:
diff changeset
37 LinkConvertGraph() throws ParserConfigurationException, SAXException {
1744340f8be6 add some java files
one
parents:
diff changeset
38 xmlParser = new TextTagParser();
1744340f8be6 add some java files
one
parents:
diff changeset
39 factory = SAXParserFactory.newInstance();
1744340f8be6 add some java files
one
parents:
diff changeset
40 parser = factory.newSAXParser();
1744340f8be6 add some java files
one
parents:
diff changeset
41 }
1744340f8be6 add some java files
one
parents:
diff changeset
42
1744340f8be6 add some java files
one
parents:
diff changeset
43 LinkConvertGraph(final String filename) throws FileNotFoundException, ParserConfigurationException, SAXException {
1744340f8be6 add some java files
one
parents:
diff changeset
44 this.filename = filename;
1744340f8be6 add some java files
one
parents:
diff changeset
45 fis = new FileInputStream(filename);
1744340f8be6 add some java files
one
parents:
diff changeset
46 xmlParser = new TextTagParser();
1744340f8be6 add some java files
one
parents:
diff changeset
47 factory = SAXParserFactory.newInstance();
1744340f8be6 add some java files
one
parents:
diff changeset
48 parser = factory.newSAXParser();
1744340f8be6 add some java files
one
parents:
diff changeset
49 }
1744340f8be6 add some java files
one
parents:
diff changeset
50
1744340f8be6 add some java files
one
parents:
diff changeset
51 public void setFilename(final String filename) throws FileNotFoundException {
1744340f8be6 add some java files
one
parents:
diff changeset
52 this.filename = filename;
1744340f8be6 add some java files
one
parents:
diff changeset
53 this.fis = new FileInputStream(filename);
1744340f8be6 add some java files
one
parents:
diff changeset
54 }
1744340f8be6 add some java files
one
parents:
diff changeset
55
1744340f8be6 add some java files
one
parents:
diff changeset
56 private void parseXml() throws SAXException, IOException {
1744340f8be6 add some java files
one
parents:
diff changeset
57 parser.parse(this.fis, this.xmlParser);
1744340f8be6 add some java files
one
parents:
diff changeset
58 hash = xmlParser.getHash();
1744340f8be6 add some java files
one
parents:
diff changeset
59 }
1744340f8be6 add some java files
one
parents:
diff changeset
60
1744340f8be6 add some java files
one
parents:
diff changeset
61 private HashMap<String,HashSet<String>> getHash() {
1744340f8be6 add some java files
one
parents:
diff changeset
62 return hash;
1744340f8be6 add some java files
one
parents:
diff changeset
63 }
1744340f8be6 add some java files
one
parents:
diff changeset
64
1744340f8be6 add some java files
one
parents:
diff changeset
65 public void printHash() {
1744340f8be6 add some java files
one
parents:
diff changeset
66 for (Map.Entry<String,HashSet<String>> entry : hash.entrySet()) {
1744340f8be6 add some java files
one
parents:
diff changeset
67 String title = entry.getKey();
1744340f8be6 add some java files
one
parents:
diff changeset
68 System.out.println("title: " + title);
1744340f8be6 add some java files
one
parents:
diff changeset
69 for (String link : entry.getValue()) {
1744340f8be6 add some java files
one
parents:
diff changeset
70 System.out.println("\t"+link);
1744340f8be6 add some java files
one
parents:
diff changeset
71 }
1744340f8be6 add some java files
one
parents:
diff changeset
72 System.out.println();
1744340f8be6 add some java files
one
parents:
diff changeset
73 }
1744340f8be6 add some java files
one
parents:
diff changeset
74 }
1744340f8be6 add some java files
one
parents:
diff changeset
75
1744340f8be6 add some java files
one
parents:
diff changeset
76 private void printHash(FileOutputStream os) throws IOException {
1744340f8be6 add some java files
one
parents:
diff changeset
77 for (Map.Entry<String,HashSet<String>> entry : hash.entrySet()) {
1744340f8be6 add some java files
one
parents:
diff changeset
78 String title = entry.getKey();
1744340f8be6 add some java files
one
parents:
diff changeset
79 os.write( ("title: " + title + "\n").getBytes());
1744340f8be6 add some java files
one
parents:
diff changeset
80 for (String link : entry.getValue()) {
1744340f8be6 add some java files
one
parents:
diff changeset
81 os.write( ("\t"+link+"\n").getBytes());
1744340f8be6 add some java files
one
parents:
diff changeset
82 }
1744340f8be6 add some java files
one
parents:
diff changeset
83 os.write( ("\n").getBytes());
1744340f8be6 add some java files
one
parents:
diff changeset
84 os.flush();
1744340f8be6 add some java files
one
parents:
diff changeset
85 }
1744340f8be6 add some java files
one
parents:
diff changeset
86 }
1744340f8be6 add some java files
one
parents:
diff changeset
87
1744340f8be6 add some java files
one
parents:
diff changeset
88
1744340f8be6 add some java files
one
parents:
diff changeset
89
1744340f8be6 add some java files
one
parents:
diff changeset
90
1744340f8be6 add some java files
one
parents:
diff changeset
91
1744340f8be6 add some java files
one
parents:
diff changeset
92 public static void main(String[] args) {
5
140272228818 fix LinkToVertex.java
one
parents: 4
diff changeset
93 final String filename = "./resources/article.xml";
140272228818 fix LinkToVertex.java
one
parents: 4
diff changeset
94 // final String filename = "/Users/aotokage/testProgram/wiki/ja-pages_current.xml";
2
1744340f8be6 add some java files
one
parents:
diff changeset
95
1744340f8be6 add some java files
one
parents:
diff changeset
96 LinkConvertGraph lcg;
1744340f8be6 add some java files
one
parents:
diff changeset
97
1744340f8be6 add some java files
one
parents:
diff changeset
98 try {
1744340f8be6 add some java files
one
parents:
diff changeset
99
1744340f8be6 add some java files
one
parents:
diff changeset
100 lcg = new LinkConvertGraph(filename);
1744340f8be6 add some java files
one
parents:
diff changeset
101
1744340f8be6 add some java files
one
parents:
diff changeset
102 lcg.parseXml();
1744340f8be6 add some java files
one
parents:
diff changeset
103 // lcg.printHash();
4
dcd59917a2dd fix LinkToVertex
one
parents: 3
diff changeset
104
3
b44abb9aa09f add resources/article.xml
one
parents: 2
diff changeset
105 FileOutputStream fos = new FileOutputStream("./resources/wikiLink.log");
2
1744340f8be6 add some java files
one
parents:
diff changeset
106 lcg.printHash(fos);
1744340f8be6 add some java files
one
parents:
diff changeset
107
1744340f8be6 add some java files
one
parents:
diff changeset
108 HashMap<String,HashSet<String>> hash = lcg.getHash();
1744340f8be6 add some java files
one
parents:
diff changeset
109
1744340f8be6 add some java files
one
parents:
diff changeset
110
4
dcd59917a2dd fix LinkToVertex
one
parents: 3
diff changeset
111 final String fileDB = "./resources/tinkerpopDB";
2
1744340f8be6 add some java files
one
parents:
diff changeset
112
1744340f8be6 add some java files
one
parents:
diff changeset
113 Graph graph = new TinkerGraph();
4
dcd59917a2dd fix LinkToVertex
one
parents: 3
diff changeset
114 FileOutputStream out = new FileOutputStream(new File(fileDB));
5
140272228818 fix LinkToVertex.java
one
parents: 4
diff changeset
115 LinkToVertex ltv = new LinkToVertex(graph);
140272228818 fix LinkToVertex.java
one
parents: 4
diff changeset
116
140272228818 fix LinkToVertex.java
one
parents: 4
diff changeset
117 long countId = 1;
2
1744340f8be6 add some java files
one
parents:
diff changeset
118 for (Map.Entry<String, HashSet<String>> map : hash.entrySet()) {
1744340f8be6 add some java files
one
parents:
diff changeset
119 String pageTitle = map.getKey();
1744340f8be6 add some java files
one
parents:
diff changeset
120
1744340f8be6 add some java files
one
parents:
diff changeset
121 Vertex v;// = graph.addVertex(null);
1744340f8be6 add some java files
one
parents:
diff changeset
122
5
140272228818 fix LinkToVertex.java
one
parents: 4
diff changeset
123 if ( ltv.getId(pageTitle) == null ) {
140272228818 fix LinkToVertex.java
one
parents: 4
diff changeset
124 v = ltv.createVertexWithPageTitle(pageTitle,countId);
140272228818 fix LinkToVertex.java
one
parents: 4
diff changeset
125 ltv.setPageRank(v, (Double)0.0);
140272228818 fix LinkToVertex.java
one
parents: 4
diff changeset
126 countId++;
2
1744340f8be6 add some java files
one
parents:
diff changeset
127 } else {
5
140272228818 fix LinkToVertex.java
one
parents: 4
diff changeset
128 v = ltv.getVertex(pageTitle);
2
1744340f8be6 add some java files
one
parents:
diff changeset
129 }
1744340f8be6 add some java files
one
parents:
diff changeset
130
1744340f8be6 add some java files
one
parents:
diff changeset
131 for (String linkPageTitle : map.getValue()) {
1744340f8be6 add some java files
one
parents:
diff changeset
132 Vertex linkV;
5
140272228818 fix LinkToVertex.java
one
parents: 4
diff changeset
133 if ( ltv.getId(linkPageTitle) == null) {
140272228818 fix LinkToVertex.java
one
parents: 4
diff changeset
134 linkV = ltv.createVertexWithPageTitle(linkPageTitle,countId);
140272228818 fix LinkToVertex.java
one
parents: 4
diff changeset
135 countId++;
140272228818 fix LinkToVertex.java
one
parents: 4
diff changeset
136 ltv.setPageRank(linkV, (Double)0.0);
2
1744340f8be6 add some java files
one
parents:
diff changeset
137 } else {
5
140272228818 fix LinkToVertex.java
one
parents: 4
diff changeset
138 linkV = ltv.getVertex(linkPageTitle);
2
1744340f8be6 add some java files
one
parents:
diff changeset
139 }
5
140272228818 fix LinkToVertex.java
one
parents: 4
diff changeset
140 ltv.setHasLink(v, linkV);
2
1744340f8be6 add some java files
one
parents:
diff changeset
141 }
1744340f8be6 add some java files
one
parents:
diff changeset
142
1744340f8be6 add some java files
one
parents:
diff changeset
143 }
1744340f8be6 add some java files
one
parents:
diff changeset
144
1744340f8be6 add some java files
one
parents:
diff changeset
145 GraphMLWriter.outputGraph(graph, out);
1744340f8be6 add some java files
one
parents:
diff changeset
146
1744340f8be6 add some java files
one
parents:
diff changeset
147 } catch (FileNotFoundException e) {
1744340f8be6 add some java files
one
parents:
diff changeset
148 e.printStackTrace();
1744340f8be6 add some java files
one
parents:
diff changeset
149 } catch (SAXException e) {
1744340f8be6 add some java files
one
parents:
diff changeset
150 e.printStackTrace();
1744340f8be6 add some java files
one
parents:
diff changeset
151 } catch (ParserConfigurationException e) {
1744340f8be6 add some java files
one
parents:
diff changeset
152 e.printStackTrace();
1744340f8be6 add some java files
one
parents:
diff changeset
153 } catch (IOException e) {
1744340f8be6 add some java files
one
parents:
diff changeset
154 System.err.println("Failed to parse xml");
1744340f8be6 add some java files
one
parents:
diff changeset
155 e.printStackTrace();
1744340f8be6 add some java files
one
parents:
diff changeset
156 }
1744340f8be6 add some java files
one
parents:
diff changeset
157
1744340f8be6 add some java files
one
parents:
diff changeset
158
1744340f8be6 add some java files
one
parents:
diff changeset
159
1744340f8be6 add some java files
one
parents:
diff changeset
160 }
1744340f8be6 add some java files
one
parents:
diff changeset
161 }