1 | import org.greenstone.gsdl3.util.GSEntityResolver;
|
---|
2 | import org.greenstone.LuceneWrapper4.GSLuceneUtil;
|
---|
3 | import org.greenstone.LuceneWrapper4.GSLuceneConstants;
|
---|
4 |
|
---|
5 | import org.xml.sax.Attributes;
|
---|
6 | import org.xml.sax.helpers.DefaultHandler;
|
---|
7 | import org.xml.sax.InputSource;
|
---|
8 | import org.xml.sax.SAXException;
|
---|
9 | //import org.xml.sax.SAXParseException;
|
---|
10 | import javax.xml.parsers.SAXParser;
|
---|
11 | import javax.xml.parsers.SAXParserFactory;
|
---|
12 |
|
---|
13 |
|
---|
14 | import org.apache.lucene.analysis.Analyzer;
|
---|
15 | import org.apache.lucene.analysis.miscellaneous.LimitTokenCountAnalyzer;
|
---|
16 | import org.apache.lucene.index.IndexWriter;
|
---|
17 | import org.apache.lucene.index.IndexWriterConfig;
|
---|
18 | import org.apache.lucene.store.FSDirectory;
|
---|
19 | import org.apache.lucene.util.Version;
|
---|
20 |
|
---|
21 | import org.apache.lucene.document.Document;
|
---|
22 | import org.apache.lucene.document.Field;
|
---|
23 | import org.apache.lucene.document.StoredField;
|
---|
24 | import org.apache.lucene.document.TextField;
|
---|
25 | //import org.apache.lucene.document.DateField;
|
---|
26 | import org.apache.lucene.index.IndexWriter;
|
---|
27 | import org.apache.lucene.analysis.standard.StandardAnalyzer;
|
---|
28 | import org.apache.lucene.util.Version;
|
---|
29 | //import org.apache.lucene.analysis.SimpleAnalyzer;
|
---|
30 |
|
---|
31 | import java.util.Stack;
|
---|
32 | import java.io.FileInputStream;
|
---|
33 | import java.io.File;
|
---|
34 | import java.net.URL;
|
---|
35 |
|
---|
36 |
|
---|
37 |
|
---|
38 | public class Indexer extends DefaultHandler {
|
---|
39 | IndexWriter writer = null;
|
---|
40 | SAXParser sax_parser = null;
|
---|
41 | Stack stack = null;
|
---|
42 | String path = "";
|
---|
43 | String current_node = "";
|
---|
44 | String current_contents = "";
|
---|
45 | Document current_doc = null;
|
---|
46 | String scope = "";
|
---|
47 | protected String file_id = null;
|
---|
48 | private String base_path = null;
|
---|
49 | /** pass in true if want to create a new index, false if want to use the existing one */
|
---|
50 | public Indexer (File index_dir, boolean create) {
|
---|
51 | try {
|
---|
52 | stack = new Stack();
|
---|
53 | SAXParserFactory sax_factory = SAXParserFactory.newInstance();
|
---|
54 | sax_parser = sax_factory.newSAXParser();
|
---|
55 | Analyzer analyzer = new StandardAnalyzer(Version.LUCENE_47);
|
---|
56 | Analyzer ltcAn = new LimitTokenCountAnalyzer(analyzer,Integer.MAX_VALUE);
|
---|
57 |
|
---|
58 | IndexWriterConfig.OpenMode open_mode;
|
---|
59 | if (create) {
|
---|
60 | open_mode = IndexWriterConfig.OpenMode.CREATE;
|
---|
61 | } else {
|
---|
62 | open_mode = IndexWriterConfig.OpenMode.APPEND;
|
---|
63 | }
|
---|
64 | IndexWriterConfig indexWriterConfig = new IndexWriterConfig(Version.LUCENE_47, ltcAn);
|
---|
65 | indexWriterConfig.setOpenMode(open_mode);
|
---|
66 |
|
---|
67 | FSDirectory index_fs_dir = FSDirectory.open(index_dir);
|
---|
68 | writer = new IndexWriter(index_fs_dir, indexWriterConfig);
|
---|
69 |
|
---|
70 | } catch (Exception e) {
|
---|
71 |
|
---|
72 | }
|
---|
73 | }
|
---|
74 |
|
---|
75 | /** index one document */
|
---|
76 | public void index (String file_id, File file) {
|
---|
77 | this.file_id = file_id;
|
---|
78 | this.path = "";
|
---|
79 | this.base_path = file.getPath();
|
---|
80 | this.base_path = this.base_path.substring(0, this.base_path.lastIndexOf(File.separatorChar));
|
---|
81 | try {
|
---|
82 | sax_parser.parse(new InputSource(new FileInputStream(file)), this);
|
---|
83 | }
|
---|
84 | catch (Exception e) {
|
---|
85 | println("parse error:");
|
---|
86 | e.printStackTrace();
|
---|
87 | }
|
---|
88 | }
|
---|
89 |
|
---|
90 | /** optimise the index */
|
---|
91 | public void finish() {
|
---|
92 | try {
|
---|
93 | writer.close();
|
---|
94 | } catch (Exception e) {}
|
---|
95 | }
|
---|
96 |
|
---|
97 | protected void println(String s) { System.out.println(s); }
|
---|
98 |
|
---|
99 | public void startDocument() throws SAXException {
|
---|
100 | println("Starting to index " + file_id);
|
---|
101 | }
|
---|
102 | public void endDocument() throws SAXException {
|
---|
103 | println("... indexing finished.");
|
---|
104 | }
|
---|
105 | public void startElement(String uri, String localName, String qName, Attributes atts) throws SAXException {
|
---|
106 | path = appendPathLink(path, qName, atts);
|
---|
107 | if (XMLTagInfo.isScopable(qName)) {
|
---|
108 | scope = qName;
|
---|
109 | }
|
---|
110 | if (XMLTagInfo.isIndexable(qName)) {
|
---|
111 | pushOnStack();
|
---|
112 | current_node = qName;
|
---|
113 | System.out.println("going to index "+qName );
|
---|
114 | String node_id = "";
|
---|
115 | String id = "<"+qName;
|
---|
116 | for (int i=0; i<atts.getLength(); i++) {
|
---|
117 | String name = atts.getQName(i);
|
---|
118 | String value = atts.getValue(i);
|
---|
119 | if (name!=null && value != null) {
|
---|
120 | id += " "+name+"="+value;
|
---|
121 | }
|
---|
122 | if (name.equals("gs3:id")) {
|
---|
123 | node_id = value;
|
---|
124 | }
|
---|
125 | }
|
---|
126 | id += "/>";
|
---|
127 |
|
---|
128 | String value;
|
---|
129 | if (scope.equals(qName)) {
|
---|
130 | value = this.file_id+"."+qName;
|
---|
131 | } else {
|
---|
132 | value = this.file_id+"."+scope+"."+qName+"."+node_id;
|
---|
133 | }
|
---|
134 | current_doc.add(new StoredField("nodeID", value));
|
---|
135 |
|
---|
136 | }
|
---|
137 | }
|
---|
138 | public void endElement(String uri, String localName, String qName) throws SAXException {
|
---|
139 | if (XMLTagInfo.isIndexable(qName) && qName.equals(current_node)) {
|
---|
140 | current_doc.add(new TextField("content", current_contents, Field.Store.NO));
|
---|
141 | try {
|
---|
142 | writer.addDocument(current_doc);
|
---|
143 | } catch (java.io.IOException e) {
|
---|
144 | e.printStackTrace();
|
---|
145 | }
|
---|
146 | popOffStack();
|
---|
147 | }
|
---|
148 |
|
---|
149 | path = removePathLink(path);
|
---|
150 | }
|
---|
151 |
|
---|
152 | public void characters(char ch[], int start, int length) throws SAXException {
|
---|
153 | String data = new String(ch, start, length).trim();
|
---|
154 | if (data.length() > 0 ) {
|
---|
155 | current_contents += data;
|
---|
156 | }
|
---|
157 | }
|
---|
158 |
|
---|
159 | protected String appendPathLink(String path, String qName, Attributes atts) {
|
---|
160 |
|
---|
161 | path = path + "/"+qName;
|
---|
162 | if (atts.getLength()>0) {
|
---|
163 | String id = atts.getValue("gs3:id");
|
---|
164 | if (id != null) {
|
---|
165 | path += "[@gs3:id='"+id+"']";
|
---|
166 | }
|
---|
167 | }
|
---|
168 | return path;
|
---|
169 | }
|
---|
170 | protected String removePathLink(String path) {
|
---|
171 |
|
---|
172 | int i=path.lastIndexOf('/');
|
---|
173 | if (i==-1) {
|
---|
174 | path="";
|
---|
175 | } else {
|
---|
176 | path = path.substring(0, i);
|
---|
177 | }
|
---|
178 | return path;
|
---|
179 | }
|
---|
180 | /** these are what we save on the stack */
|
---|
181 | private class MyDocument {
|
---|
182 |
|
---|
183 | public Document doc = null;
|
---|
184 | public String contents = null;
|
---|
185 | public String tagname = "";
|
---|
186 |
|
---|
187 | }
|
---|
188 | protected void pushOnStack() {
|
---|
189 | if (current_doc != null) {
|
---|
190 | MyDocument save = new MyDocument();
|
---|
191 | save.doc = current_doc;
|
---|
192 | save.contents = current_contents;
|
---|
193 | save.tagname = current_node;
|
---|
194 | stack.push(save);
|
---|
195 | }
|
---|
196 | current_doc = new Document();
|
---|
197 | current_contents = "";
|
---|
198 | current_node = "";
|
---|
199 | }
|
---|
200 |
|
---|
201 | protected void popOffStack() {
|
---|
202 | if (!stack.empty()) {
|
---|
203 | MyDocument saved = (MyDocument)stack.pop();
|
---|
204 | current_doc = saved.doc;
|
---|
205 | current_contents = saved.contents;
|
---|
206 | current_node = saved.tagname;
|
---|
207 | } else {
|
---|
208 | current_doc = new Document();
|
---|
209 | current_contents = "";
|
---|
210 | current_node = "";
|
---|
211 | }
|
---|
212 | }
|
---|
213 |
|
---|
214 | public InputSource resolveEntity (String public_id, String system_id) {
|
---|
215 |
|
---|
216 | if (system_id.startsWith("file://")) {
|
---|
217 | return new InputSource(system_id);
|
---|
218 | }
|
---|
219 | if (!system_id.startsWith(File.separator)) {
|
---|
220 | system_id = base_path+File.separatorChar+system_id;
|
---|
221 | }
|
---|
222 | return new InputSource("file://"+system_id);
|
---|
223 | }
|
---|
224 |
|
---|
225 | }
|
---|
226 |
|
---|
227 |
|
---|