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 // Spanish has no initial headings, tried to also detect {{ES as such
43 // with "^(\\{\\{ES|(=+)[^=]).*$" but that broke English.
44 static final Pattern headingStart = Pattern.compile("^(=+)[^=].*$", Pattern.MULTILINE);
46 final Map<String,List<Selector>> pathToSelectors = new LinkedHashMap<String, List<Selector>>();
47 List<Selector> currentSelectors = null;
49 StringBuilder titleBuilder;
50 StringBuilder textBuilder;
51 StringBuilder currentBuilder = null;
53 public static void main(final String[] args) throws Exception {
54 final WiktionarySplitter wiktionarySplitter = new WiktionarySplitter();
55 wiktionarySplitter.go();
58 private WiktionarySplitter() {
59 List<Selector> selectors;
60 for (final String code : WiktionaryLangs.wikiCodeToIsoCodeToWikiName.keySet()) {
61 //if (!code.equals("fr")) {continue;}
62 selectors = new ArrayList<WiktionarySplitter.Selector>();
63 pathToSelectors.put(String.format("data/inputs/%swiktionary-pages-articles.xml", code), selectors);
64 for (final Map.Entry<String, String> entry : WiktionaryLangs.wikiCodeToIsoCodeToWikiName.get(code).entrySet()) {
65 final String dir = String.format("data/inputs/wikiSplit/%s", code);
66 new File(dir).mkdirs();
67 selectors.add(new Selector(String.format("%s/%s.data", dir, entry.getKey()), entry.getValue()));
72 private void go() throws Exception {
73 final SAXParser parser = SAXParserFactoryImpl.newInstance().newSAXParser();
76 for (final Map.Entry<String, List<Selector>> pathToSelectorsEntry : pathToSelectors.entrySet()) {
78 currentSelectors = pathToSelectorsEntry.getValue();
80 for (final Selector selector : currentSelectors) {
81 selector.out = new DataOutputStream(new BufferedOutputStream(new FileOutputStream(selector.outFilename)));
86 parser.parse(new File(pathToSelectorsEntry.getKey()), this);
87 } catch (Exception e) {
88 System.err.println("Exception during parse, lastPageTitle=" + lastPageTitle + ", titleBuilder=" + titleBuilder.toString());
93 for (final Selector selector : currentSelectors) {
100 String lastPageTitle = null;
102 private void endPage() {
103 final String title = titleBuilder.toString();
104 lastPageTitle = title;
105 if (++pageCount % 1000 == 0) {
106 System.out.println("endPage: " + title + ", count=" + pageCount);
108 if (title.startsWith("Wiktionary:") ||
109 title.startsWith("Appendix:") ||
110 title.startsWith("Help:") ||
111 title.startsWith("Index:") ||
112 title.startsWith("MediaWiki:") ||
113 title.startsWith("Citations:") ||
114 title.startsWith("Concordance:") ||
115 title.startsWith("Glossary:") ||
116 title.startsWith("Rhymes:") ||
117 title.startsWith("Category:") ||
118 title.startsWith("Wikisaurus:") ||
119 title.startsWith("Unsupported titles/") ||
120 title.startsWith("Transwiki:") ||
121 title.startsWith("File:") ||
122 title.startsWith("Thread:") ||
123 title.startsWith("Template:") ||
124 title.startsWith("Summary:") ||
125 title.startsWith("Module:") ||
127 title.startsWith("Datei:") ||
128 title.startsWith("Verzeichnis:") ||
129 title.startsWith("Vorlage:") ||
130 title.startsWith("Thesaurus:") ||
131 title.startsWith("Kategorie:") ||
132 title.startsWith("Hilfe:") ||
133 title.startsWith("Reim:") ||
135 title.startsWith("Annexe:") ||
136 title.startsWith("Catégori:") ||
137 title.startsWith("Modèle:") ||
138 title.startsWith("Thésaurus:") ||
139 title.startsWith("Projet:") ||
140 title.startsWith("Aide:") ||
141 title.startsWith("Fichier:") ||
142 title.startsWith("Wiktionnaire:") ||
143 title.startsWith("Catégorie:") ||
144 title.startsWith("Portail:") ||
145 title.startsWith("utiliusateur:") ||
146 title.startsWith("Kategorio:") ||
148 title.startsWith("Wikizionario:") ||
149 title.startsWith("Appendice:") ||
150 title.startsWith("Categoria:") ||
151 title.startsWith("Aiuto:") ||
152 title.startsWith("Portail:") ||
154 title.startsWith("Apéndice:") ||
155 title.startsWith("Archivo:") ||
156 title.startsWith("Ayuda:") ||
157 title.startsWith("Categoría:") ||
158 title.startsWith("Plantilla:") ||
159 title.startsWith("Wikcionario:") ||
166 if (title.contains(":")) {
167 if (!title.startsWith("Sign gloss:")) {
168 System.err.println("title with colon: " + title);
172 String text = textBuilder.toString();
173 String translingual = "";
175 while (text.length() > 0) {
177 final Matcher startMatcher = headingStart.matcher(text);
178 if (!startMatcher.find()) {
181 text = text.substring(startMatcher.end());
183 final String heading = startMatcher.group();
184 for (final Selector selector : currentSelectors) {
185 if (heading.indexOf("Translingual") != -1) {
187 final int depth = startMatcher.group(1).length();
188 final Pattern endPattern = Pattern.compile(String.format("^={1,%d}[^=].*$", depth), Pattern.MULTILINE);
190 final Matcher endMatcher = endPattern.matcher(text);
191 if (endMatcher.find()) {
192 int end = endMatcher.start();
193 translingual = text.substring(0, endMatcher.start());
194 text = text.substring(end);
198 if (selector.pattern.matcher(heading).find()) {
201 final int depth = startMatcher.group(1).length();
202 final Pattern endPattern = Pattern.compile(String.format("^={1,%d}[^=].*$", depth), Pattern.MULTILINE);
204 final Matcher endMatcher = endPattern.matcher(text);
206 if (endMatcher.find()) {
207 end = endMatcher.start();
212 String sectionText = text.substring(0, end);
213 // Hack to remove empty dummy section from French
214 if (sectionText.startsWith("\n=== {{S|étymologie}} ===\n: {{ébauche-étym"))
216 int dummy_end = sectionText.indexOf("}}", 41) + 2;
217 while (dummy_end + 1 < sectionText.length() &&
218 sectionText.charAt(dummy_end) == '\n' &&
219 sectionText.charAt(dummy_end + 1) == '\n') ++dummy_end;
220 sectionText = sectionText.substring(dummy_end);
222 if (heading.indexOf("Japanese") == -1) sectionText += translingual;
223 final Section section = new Section(title, heading, sectionText);
226 selector.out.writeUTF(section.title);
227 selector.out.writeUTF(section.heading);
228 final byte[] bytes = section.text.getBytes("UTF8");
229 selector.out.writeInt(bytes.length);
230 selector.out.write(bytes);
231 } catch (IOException e) {
232 throw new RuntimeException(e);
235 text = text.substring(end);
243 // -----------------------------------------------------------------------
245 static class Section implements java.io.Serializable {
246 private static final long serialVersionUID = -7676549898325856822L;
249 final String heading;
252 public Section(final String title, final String heading, final String text) {
254 this.heading = heading;
257 //System.out.printf("TITLE:%s\nHEADING:%s\nTEXT:%s\n\n\n\n\n\n", title, heading, text);
261 static class Selector {
262 final String outFilename;
263 final Pattern pattern;
265 DataOutputStream out;
267 public Selector(final String filename, final String pattern) {
268 this.outFilename = filename;
269 this.pattern = Pattern.compile(pattern, Pattern.CASE_INSENSITIVE);
273 // -----------------------------------------------------------------------
276 public void startElement(String uri, String localName, String qName,
277 Attributes attributes) {
278 currentBuilder = null;
279 if ("page".equals(qName)) {
280 titleBuilder = new StringBuilder();
282 // Start with "\n" to better match certain strings.
283 textBuilder = new StringBuilder("\n");
284 } else if ("title".equals(qName)) {
285 currentBuilder = titleBuilder;
286 } else if ("text".equals(qName)) {
287 currentBuilder = textBuilder;
292 public void characters(char[] ch, int start, int length) throws SAXException {
293 if (currentBuilder != null) {
294 currentBuilder.append(ch, start, length);
299 public void endElement(String uri, String localName, String qName)
300 throws SAXException {
301 currentBuilder = null;
302 if ("page".equals(qName)) {
307 public void parse(final File file) throws ParserConfigurationException,
308 SAXException, IOException {
309 final SAXParser parser = SAXParserFactoryImpl.newInstance().newSAXParser();
310 parser.parse(file, this);