annotate src/pagerank/LinkConvertGraph.java @ 2:1744340f8be6 draft

add some java files
author one
date Wed, 05 Sep 2012 11:56:21 +0900
parents
children b44abb9aa09f
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) {
1744340f8be6 add some java files
one
parents:
diff changeset
93 // final String filename = "./resource/article.xml";
1744340f8be6 add some java files
one
parents:
diff changeset
94 final String filename = "/Users/aotokage/testProgram/wiki/ja-pages_current.xml";
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();
1744340f8be6 add some java files
one
parents:
diff changeset
104
1744340f8be6 add some java files
one
parents:
diff changeset
105 FileOutputStream fos = new FileOutputStream("./resource/wikiLink.log");
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
1744340f8be6 add some java files
one
parents:
diff changeset
111 final String filenameD = "./resource/tinkerpopDB";
1744340f8be6 add some java files
one
parents:
diff changeset
112
1744340f8be6 add some java files
one
parents:
diff changeset
113 Graph graph = new TinkerGraph();
1744340f8be6 add some java files
one
parents:
diff changeset
114 FileOutputStream out = new FileOutputStream(new File(filename));
1744340f8be6 add some java files
one
parents:
diff changeset
115 LinkToVertex ltn = new LinkToVertex(graph);
1744340f8be6 add some java files
one
parents:
diff changeset
116
1744340f8be6 add some java files
one
parents:
diff changeset
117 for (Map.Entry<String, HashSet<String>> map : hash.entrySet()) {
1744340f8be6 add some java files
one
parents:
diff changeset
118 String pageTitle = map.getKey();
1744340f8be6 add some java files
one
parents:
diff changeset
119
1744340f8be6 add some java files
one
parents:
diff changeset
120 Vertex v;// = graph.addVertex(null);
1744340f8be6 add some java files
one
parents:
diff changeset
121
1744340f8be6 add some java files
one
parents:
diff changeset
122 if ( ltn.getId(pageTitle) == null ) {
1744340f8be6 add some java files
one
parents:
diff changeset
123 v = ltn.createVertexWithPageTitle(pageTitle);
1744340f8be6 add some java files
one
parents:
diff changeset
124
1744340f8be6 add some java files
one
parents:
diff changeset
125 } else {
1744340f8be6 add some java files
one
parents:
diff changeset
126 v = ltn.getVertex(pageTitle);
1744340f8be6 add some java files
one
parents:
diff changeset
127 }
1744340f8be6 add some java files
one
parents:
diff changeset
128
1744340f8be6 add some java files
one
parents:
diff changeset
129 for (String linkPageTitle : map.getValue()) {
1744340f8be6 add some java files
one
parents:
diff changeset
130 Vertex linkV;
1744340f8be6 add some java files
one
parents:
diff changeset
131 if ( ltn.getId(linkPageTitle) == null) {
1744340f8be6 add some java files
one
parents:
diff changeset
132 linkV = ltn.createVertexWithPageTitle(linkPageTitle);
1744340f8be6 add some java files
one
parents:
diff changeset
133 ltn.setPageRank(linkV, (Double)0.0);
1744340f8be6 add some java files
one
parents:
diff changeset
134 } else {
1744340f8be6 add some java files
one
parents:
diff changeset
135 linkV = ltn.getVertex(linkPageTitle);
1744340f8be6 add some java files
one
parents:
diff changeset
136 }
1744340f8be6 add some java files
one
parents:
diff changeset
137 ltn.setHasLink(v, linkV);
1744340f8be6 add some java files
one
parents:
diff changeset
138 }
1744340f8be6 add some java files
one
parents:
diff changeset
139
1744340f8be6 add some java files
one
parents:
diff changeset
140 }
1744340f8be6 add some java files
one
parents:
diff changeset
141
1744340f8be6 add some java files
one
parents:
diff changeset
142 GraphMLWriter.outputGraph(graph, out);
1744340f8be6 add some java files
one
parents:
diff changeset
143
1744340f8be6 add some java files
one
parents:
diff changeset
144 } catch (FileNotFoundException e) {
1744340f8be6 add some java files
one
parents:
diff changeset
145 e.printStackTrace();
1744340f8be6 add some java files
one
parents:
diff changeset
146 } catch (SAXException e) {
1744340f8be6 add some java files
one
parents:
diff changeset
147 e.printStackTrace();
1744340f8be6 add some java files
one
parents:
diff changeset
148 } catch (ParserConfigurationException e) {
1744340f8be6 add some java files
one
parents:
diff changeset
149 e.printStackTrace();
1744340f8be6 add some java files
one
parents:
diff changeset
150 } catch (IOException e) {
1744340f8be6 add some java files
one
parents:
diff changeset
151 System.err.println("Failed to parse xml");
1744340f8be6 add some java files
one
parents:
diff changeset
152 e.printStackTrace();
1744340f8be6 add some java files
one
parents:
diff changeset
153 }
1744340f8be6 add some java files
one
parents:
diff changeset
154
1744340f8be6 add some java files
one
parents:
diff changeset
155
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 }