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;
32 import org.apache.xerces.jaxp.SAXParserFactoryImpl;
33 import org.xml.sax.Attributes;
34 import org.xml.sax.SAXException;
36 import com.hughes.android.dictionary.parser.wiktionary.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 Exception {
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("en") || code.equals("de") || 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 Exception {
71 final SAXParser parser = SAXParserFactoryImpl.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)));
84 parser.parse(new File(pathToSelectorsEntry.getKey()), this);
85 } catch (Exception e) {
86 System.err.println("Exception during parse, lastPageTitle=" + lastPageTitle + ", titleBuilder=" + titleBuilder.toString());
91 for (final Selector selector : currentSelectors) {
98 String lastPageTitle = null;
100 private void endPage() {
101 final String title = titleBuilder.toString();
102 lastPageTitle = title;
103 if (++pageCount % 1000 == 0) {
104 System.out.println("endPage: " + title + ", count=" + pageCount);
106 if (title.startsWith("Wiktionary:") ||
107 title.startsWith("Appendix:") ||
108 title.startsWith("Help:") ||
109 title.startsWith("Index:") ||
110 title.startsWith("MediaWiki:") ||
111 title.startsWith("Citations:") ||
112 title.startsWith("Concordance:") ||
113 title.startsWith("Glossary:") ||
114 title.startsWith("Rhymes:") ||
115 title.startsWith("Category:") ||
116 title.startsWith("Wikisaurus:") ||
117 title.startsWith("Unsupported titles/") ||
118 title.startsWith("Transwiki:") ||
119 title.startsWith("File:") ||
120 title.startsWith("Thread:") ||
121 title.startsWith("Template:") ||
122 title.startsWith("Summary:") ||
124 title.startsWith("Datei:") ||
125 title.startsWith("Verzeichnis:") ||
126 title.startsWith("Vorlage:") ||
127 title.startsWith("Thesaurus:") ||
128 title.startsWith("Kategorie:") ||
129 title.startsWith("Hilfe:") ||
131 title.startsWith("Annexe:") ||
132 title.startsWith("Catégori:") ||
133 title.startsWith("Modèle:") ||
134 title.startsWith("Thésaurus:") ||
135 title.startsWith("Projet:") ||
136 title.startsWith("Aide:") ||
137 title.startsWith("Fichier:") ||
138 title.startsWith("Wiktionnaire:") ||
139 title.startsWith("Catégorie:") ||
140 title.startsWith("Portail:") ||
141 title.startsWith("utiliusateur:") ||
142 title.startsWith("Kategorio:") ||
144 title.startsWith("Wikizionario:") ||
145 title.startsWith("Appendice:") ||
146 title.startsWith("Categoria:") ||
147 title.startsWith("Aiuto:") ||
148 title.startsWith("Portail:") ||
155 if (title.contains(":")) {
156 if (!title.startsWith("Sign gloss:")) {
157 System.err.println("title with colon: " + title);
161 String text = textBuilder.toString();
163 while (text.length() > 0) {
165 final Matcher startMatcher = headingStart.matcher(text);
166 if (!startMatcher.find()) {
169 text = text.substring(startMatcher.end());
171 final String heading = startMatcher.group();
172 for (final Selector selector : currentSelectors) {
173 if (selector.pattern.matcher(heading).find()) {
176 final int depth = startMatcher.group(1).length();
177 final Pattern endPattern = Pattern.compile(String.format("^={1,%d}[^=].*$", depth), Pattern.MULTILINE);
179 final Matcher endMatcher = endPattern.matcher(text);
181 if (endMatcher.find()) {
182 end = endMatcher.start();
187 final String sectionText = text.substring(0, end);
188 final Section section = new Section(title, heading, sectionText);
191 selector.out.writeUTF(section.title);
192 selector.out.writeUTF(section.heading);
193 final byte[] bytes = section.text.getBytes("UTF8");
194 selector.out.writeInt(bytes.length);
195 selector.out.write(bytes);
196 } catch (IOException e) {
197 throw new RuntimeException(e);
200 text = text.substring(end);
207 // -----------------------------------------------------------------------
209 static class Section implements java.io.Serializable {
210 private static final long serialVersionUID = -7676549898325856822L;
213 final String heading;
216 public Section(final String title, final String heading, final String text) {
218 this.heading = heading;
221 //System.out.printf("TITLE:%s\nHEADING:%s\nTEXT:%s\n\n\n\n\n\n", title, heading, text);
225 static class Selector {
226 final String outFilename;
227 final Pattern pattern;
229 DataOutputStream out;
231 public Selector(final String filename, final String pattern) {
232 this.outFilename = filename;
233 this.pattern = Pattern.compile(pattern, Pattern.CASE_INSENSITIVE);
237 // -----------------------------------------------------------------------
240 public void startElement(String uri, String localName, String qName,
241 Attributes attributes) {
242 currentBuilder = null;
243 if ("page".equals(qName)) {
244 titleBuilder = new StringBuilder();
246 // Start with "\n" to better match certain strings.
247 textBuilder = new StringBuilder("\n");
248 } else if ("title".equals(qName)) {
249 currentBuilder = titleBuilder;
250 } else if ("text".equals(qName)) {
251 currentBuilder = textBuilder;
256 public void characters(char[] ch, int start, int length) throws SAXException {
257 if (currentBuilder != null) {
258 currentBuilder.append(ch, start, length);
263 public void endElement(String uri, String localName, String qName)
264 throws SAXException {
265 currentBuilder = null;
266 if ("page".equals(qName)) {
271 public void parse(final File file) throws ParserConfigurationException,
272 SAXException, IOException {
273 final SAXParser parser = SAXParserFactoryImpl.newInstance().newSAXParser();
274 parser.parse(file, this);