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.LinkedHashMap;
24 import java.util.List;
26 import java.util.regex.Matcher;
27 import java.util.regex.Pattern;
29 import javax.xml.parsers.ParserConfigurationException;
30 import javax.xml.parsers.SAXParser;
31 import javax.xml.parsers.SAXParserFactory;
33 import org.xml.sax.Attributes;
34 import org.xml.sax.SAXException;
36 import com.hughes.android.dictionary.parser.enwiktionary.WiktionaryLangs;
38 public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler {
40 // The matches the whole line, otherwise regexes don't work well on French:
42 static final Pattern headingStart = Pattern.compile("^(=+)[^=].*$", Pattern.MULTILINE);
44 final Map<String,List<Selector>> pathToSelectors = new LinkedHashMap<String, List<Selector>>();
45 List<Selector> currentSelectors = null;
47 StringBuilder titleBuilder;
48 StringBuilder textBuilder;
49 StringBuilder currentBuilder = null;
51 public static void main(final String[] args) throws SAXException, IOException, ParserConfigurationException {
52 final WiktionarySplitter wiktionarySplitter = new WiktionarySplitter();
53 wiktionarySplitter.go();
56 private WiktionarySplitter() {
57 List<Selector> selectors;
58 for (final String code : WiktionaryLangs.wikiCodeToIsoCodeToWikiName.keySet()) {
59 //if (!code.equals("fr")) {continue;}
60 selectors = new ArrayList<WiktionarySplitter.Selector>();
61 pathToSelectors.put(String.format("data/inputs/%swiktionary-pages-articles.xml", code), selectors);
62 for (final Map.Entry<String, String> entry : WiktionaryLangs.wikiCodeToIsoCodeToWikiName.get(code).entrySet()) {
63 final String dir = String.format("data/inputs/wikiSplit/%s", code);
64 new File(dir).mkdirs();
65 selectors.add(new Selector(String.format("%s/%s.data", dir, entry.getKey()), entry.getValue()));
70 private void go() throws ParserConfigurationException, SAXException, IOException {
71 final SAXParser parser = SAXParserFactory.newInstance().newSAXParser();
74 for (final Map.Entry<String, List<Selector>> pathToSelectorsEntry : pathToSelectors.entrySet()) {
76 currentSelectors = pathToSelectorsEntry.getValue();
78 for (final Selector selector : currentSelectors) {
79 selector.out = new DataOutputStream(new BufferedOutputStream(new FileOutputStream(selector.outFilename)));
83 parser.parse(new File(pathToSelectorsEntry.getKey()), this);
86 for (final Selector selector : currentSelectors) {
94 private void endPage() {
95 final String title = titleBuilder.toString();
96 if (++pageCount % 1000 == 0) {
97 System.out.println("endPage: " + title + ", count=" + pageCount);
100 String text = textBuilder.toString();
102 while (text.length() > 0) {
104 final Matcher startMatcher = headingStart.matcher(text);
105 if (!startMatcher.find()) {
108 text = text.substring(startMatcher.end());
110 final String heading = startMatcher.group();
111 for (final Selector selector : currentSelectors) {
112 if (selector.pattern.matcher(heading).find()) {
115 final int depth = startMatcher.group(1).length();
116 final Pattern endPattern = Pattern.compile(String.format("^={1,%d}[^=].*$", depth), Pattern.MULTILINE);
118 final Matcher endMatcher = endPattern.matcher(text);
120 if (endMatcher.find()) {
121 end = endMatcher.start();
126 final String sectionText = text.substring(0, end);
127 final Section section = new Section(title, heading, sectionText);
130 selector.out.writeUTF(section.title);
131 selector.out.writeUTF(section.heading);
132 final byte[] bytes = section.text.getBytes("UTF8");
133 selector.out.writeInt(bytes.length);
134 selector.out.write(bytes);
135 } catch (IOException e) {
136 throw new RuntimeException(e);
139 text = text.substring(end);
146 // -----------------------------------------------------------------------
148 static class Section implements java.io.Serializable {
149 private static final long serialVersionUID = -7676549898325856822L;
152 final String heading;
155 public Section(final String title, final String heading, final String text) {
157 this.heading = heading;
160 //System.out.printf("TITLE:%s\nHEADING:%s\nTEXT:%s\n\n\n\n\n\n", title, heading, text);
164 static class Selector {
165 final String outFilename;
166 final Pattern pattern;
168 DataOutputStream out;
170 public Selector(final String filename, final String pattern) {
171 this.outFilename = filename;
172 this.pattern = Pattern.compile(pattern, Pattern.CASE_INSENSITIVE);
176 // -----------------------------------------------------------------------
179 public void startElement(String uri, String localName, String qName,
180 Attributes attributes) {
181 currentBuilder = null;
182 if ("page".equals(qName)) {
183 titleBuilder = new StringBuilder();
185 // Start with "\n" to better match certain strings.
186 textBuilder = new StringBuilder("\n");
187 } else if ("title".equals(qName)) {
188 currentBuilder = titleBuilder;
189 } else if ("text".equals(qName)) {
190 currentBuilder = textBuilder;
195 public void characters(char[] ch, int start, int length) throws SAXException {
196 if (currentBuilder != null) {
197 currentBuilder.append(ch, start, length);
202 public void endElement(String uri, String localName, String qName)
203 throws SAXException {
204 currentBuilder = null;
205 if ("page".equals(qName)) {
211 public void parse(final File file) throws ParserConfigurationException,
212 SAXException, IOException {
213 final SAXParser parser = SAXParserFactory.newInstance().newSAXParser();
214 parser.parse(file, this);