1 // Copyright 2011 Google Inc. All Rights Reserved.
3 // Licensed under the Apache License, Version 2.0 (the "License");
4 // you may not use this file except in compliance with the License.
5 // You may obtain a copy of the License at
7 // http://www.apache.org/licenses/LICENSE-2.0
9 // Unless required by applicable law or agreed to in writing, software
10 // distributed under the License is distributed on an "AS IS" BASIS,
11 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12 // See the License for the specific language governing permissions and
13 // limitations under the License.
15 package com.hughes.android.dictionary.engine;
17 import java.io.BufferedOutputStream;
18 import java.io.DataOutputStream;
20 import java.io.FileOutputStream;
21 import java.io.IOException;
22 import java.util.ArrayList;
23 import java.util.Arrays;
24 import java.util.List;
25 import java.util.regex.Matcher;
26 import java.util.regex.Pattern;
28 import javax.xml.parsers.ParserConfigurationException;
29 import javax.xml.parsers.SAXParser;
30 import javax.xml.parsers.SAXParserFactory;
32 import org.xml.sax.Attributes;
33 import org.xml.sax.SAXException;
35 public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler {
37 static class Section implements java.io.Serializable {
38 private static final long serialVersionUID = -7676549898325856822L;
44 public Section(final String title, final String heading, final String text) {
46 this.heading = heading;
49 //System.out.printf("TITLE:%s\nHEADING:%s\nTEXT:%s\n\n\n\n\n\n", title, heading, text);
53 static class Selector {
57 public Selector(final String filename, final String pattern) throws IOException {
58 this.out = new DataOutputStream(new BufferedOutputStream(new FileOutputStream(filename)));
59 this.pattern = Pattern.compile(pattern);
63 final List<Selector> selectors = new ArrayList<Selector>();
64 StringBuilder titleBuilder;
65 StringBuilder textBuilder;
66 StringBuilder currentBuilder = null;
68 public static void main(final String[] args) throws SAXException, IOException, ParserConfigurationException {
69 final SAXParser parser = SAXParserFactory.newInstance().newSAXParser();
70 final WiktionarySplitter wiktionarySplitter = new WiktionarySplitter();
73 final File file = new File(args[0]);
74 final List<Selector> selectors = wiktionarySplitter.selectors;
75 for (int i = 1; i < args.length; i += 2) {
76 final Selector selector = new Selector(args[i], args[i+1]);
77 selectors.add(selector);
80 if (selectors.isEmpty()) {
81 selectors.addAll(Arrays.asList(
82 new Selector("../DictionaryData/inputs/enWikiSplit/arabic.data", ".*[Ar]rabic.*"),
83 new Selector("../DictionaryData/inputs/enWikiSplit/croation.data", ".*[Cc]roation.*"),
84 new Selector("../DictionaryData/inputs/enWikiSplit/czech.data", ".*[Cc]zech.*"),
85 new Selector("../DictionaryData/inputs/enWikiSplit/mandarin.data", ".*[Mm]andarin|[Cc]hinese.*"),
86 new Selector("../DictionaryData/inputs/enWikiSplit/dutch.data", ".*[Du]utch.*"),
87 new Selector("../DictionaryData/inputs/enWikiSplit/english.data", ".*[Ee]nglish.*"),
88 new Selector("../DictionaryData/inputs/enWikiSplit/french.data", ".*[Ff]rench.*"),
89 new Selector("../DictionaryData/inputs/enWikiSplit/german.data", ".*[Gg]erman.*"),
90 new Selector("../DictionaryData/inputs/enWikiSplit/greek.data", ".*[Gg]reek.*"),
91 new Selector("../DictionaryData/inputs/enWikiSplit/hindi.data", ".*[Hh]indi.*"),
92 new Selector("../DictionaryData/inputs/enWikiSplit/italian.data", ".*[Ii]talian.*"),
93 new Selector("../DictionaryData/inputs/enWikiSplit/japanese.data", ".*[Jj]apanese.*"),
94 new Selector("../DictionaryData/inputs/enWikiSplit/korean.data", ".*[Kk]orean.*"),
95 new Selector("../DictionaryData/inputs/enWikiSplit/persian.data", ".*[Pp]ersian.*"),
96 new Selector("../DictionaryData/inputs/enWikiSplit/portuguese.data", ".*[Pp]ortuguese.*"),
97 new Selector("../DictionaryData/inputs/enWikiSplit/romanian.data", ".*[Rr]omanian.*"),
98 new Selector("../DictionaryData/inputs/enWikiSplit/russian.data", ".*[Rr]ussian.*"),
99 new Selector("../DictionaryData/inputs/enWikiSplit/spanish.data", ".*[Ss]panish.*"),
100 new Selector("../DictionaryData/inputs/enWikiSplit/swedish.data", ".*[Ss]wedish.*"),
101 new Selector("../DictionaryData/inputs/enWikiSplit/thai.data", ".*[Tt]hai.*"),
102 new Selector("../DictionaryData/inputs/enWikiSplit/vietnamese.data", ".*[Vv]ietnamese.*")
107 parser.parse(file, wiktionarySplitter);
110 for (final Selector selector : selectors) {
111 selector.out.close();
115 static final Pattern headingStart = Pattern.compile("^(=+)[^=]+=+", Pattern.MULTILINE);
117 private void endPage() {
118 final String title = titleBuilder.toString();
119 System.out.println("endPage: " + title);
121 String text = textBuilder.toString();
123 while (text.length() > 0) {
125 final Matcher startMatcher = headingStart.matcher(text);
126 if (!startMatcher.find()) {
129 text = text.substring(startMatcher.end());
131 final String heading = startMatcher.group();
132 for (final Selector selector : selectors) {
133 if (selector.pattern.matcher(heading).find()) {
136 final int depth = startMatcher.group(1).length();
137 final Pattern endPattern = Pattern.compile(String.format("^={1,%d}[^=]+=+", depth), Pattern.MULTILINE);
139 final Matcher endMatcher = endPattern.matcher(text);
141 if (endMatcher.find()) {
142 end = endMatcher.start();
147 final String sectionText = text.substring(0, end);
148 final Section section = new Section(title, heading, sectionText);
151 selector.out.writeUTF(section.title);
152 selector.out.writeUTF(section.heading);
153 final byte[] bytes = section.text.getBytes("UTF8");
154 selector.out.writeInt(bytes.length);
155 selector.out.write(bytes);
156 } catch (IOException e) {
157 throw new RuntimeException(e);
160 text = text.substring(end);
167 // -----------------------------------------------------------------------
170 public void startElement(String uri, String localName, String qName,
171 Attributes attributes) {
172 currentBuilder = null;
173 if ("page".equals(qName)) {
174 titleBuilder = new StringBuilder();
176 // Start with "\n" to better match certain strings.
177 textBuilder = new StringBuilder("\n");
178 } else if ("title".equals(qName)) {
179 currentBuilder = titleBuilder;
180 } else if ("text".equals(qName)) {
181 currentBuilder = textBuilder;
186 public void characters(char[] ch, int start, int length) throws SAXException {
187 if (currentBuilder != null) {
188 currentBuilder.append(ch, start, length);
193 public void endElement(String uri, String localName, String qName)
194 throws SAXException {
195 currentBuilder = null;
196 if ("page".equals(qName)) {
202 public void parse(final File file) throws ParserConfigurationException,
203 SAXException, IOException {
204 final SAXParser parser = SAXParserFactory.newInstance().newSAXParser();
205 parser.parse(file, this);