1 // Copyright 2012 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.parser.wiktionary;
17 import java.io.BufferedInputStream;
18 import java.io.DataInputStream;
19 import java.io.EOFException;
21 import java.io.FileInputStream;
22 import java.io.InputStream;
23 import java.io.IOException;
24 import java.util.LinkedHashMap;
25 import java.util.LinkedHashSet;
26 import java.util.List;
29 import java.util.SortedMap;
30 import java.util.TreeMap;
31 import java.util.concurrent.atomic.AtomicInteger;
32 import java.util.logging.Level;
33 import java.util.logging.Logger;
34 import java.util.regex.Pattern;
36 import org.apache.commons.compress.compressors.CompressorStreamFactory;
37 import org.apache.commons.compress.compressors.CompressorException;
39 import com.hughes.android.dictionary.engine.EntrySource;
40 import com.hughes.android.dictionary.engine.EntryTypeName;
41 import com.hughes.android.dictionary.engine.IndexBuilder;
42 import com.hughes.android.dictionary.engine.IndexedEntry;
43 import com.hughes.android.dictionary.engine.ReadAheadBuffer;
44 import com.hughes.android.dictionary.parser.Parser;
45 import com.hughes.android.dictionary.parser.WikiTokenizer;
46 import com.hughes.util.EnumUtil;
48 public abstract class AbstractWiktionaryParser implements Parser {
50 static final Logger LOG = Logger.getLogger("WiktionaryParser");
52 final SortedMap<String, AtomicInteger> counters = new TreeMap<String, AtomicInteger>();
53 final Set<String> pairsAdded = new LinkedHashSet<String>();
55 public EntrySource entrySource;
59 abstract void parseSection(final String heading, final String text);
61 abstract void removeUselessArgs(final Map<String, String> namedArgs);
64 public void parse(final File file, final EntrySource entrySource, final int pageLimit) throws IOException {
65 this.entrySource = entrySource;
67 File input = new File(file.getPath() + ".bz2");
68 if (!input.exists()) input = new File(file.getPath() + ".gz");
69 if (!input.exists()) input = new File(file.getPath() + ".xz");
71 if (!input.exists()) {
72 // Fallback to uncompressed file
73 dis = new DataInputStream(new BufferedInputStream(new FileInputStream(file)));
75 InputStream compressedIn = new BufferedInputStream(new FileInputStream(input));
77 InputStream in = new CompressorStreamFactory().createCompressorInputStream(compressedIn);
78 in = new ReadAheadBuffer(in, 20 * 1024 * 1024);
79 dis = new DataInputStream(in);
80 } catch (CompressorException e) {
81 throw new IOException(e);
86 if (pageLimit >= 0 && pageCount >= pageLimit) {
91 title = dis.readUTF();
92 } catch (EOFException e) {
93 LOG.log(Level.INFO, "EOF reading split.");
97 final String heading = dis.readUTF();
98 final int bytesLength = dis.readInt();
99 final byte[] bytes = new byte[bytesLength];
100 dis.readFully(bytes);
101 final String text = new String(bytes, "UTF8");
103 parseSection(heading, text);
106 if (pageCount % 1000 == 0) {
107 LOG.info("pageCount=" + pageCount);
112 LOG.info("***COUNTERS***");
113 for (final Map.Entry<String, AtomicInteger> entry : counters.entrySet()) {
114 LOG.info(entry.getKey() + ": " + entry.getValue());
119 static final Pattern whitespace = Pattern.compile("\\s+");
120 static String trim(final String s) {
121 return whitespace.matcher(s).replaceAll(" ").trim();
124 public void incrementCount(final String string) {
125 AtomicInteger counter = counters.get(string);
126 if (counter == null) {
127 counter = new AtomicInteger();
128 counters.put(string, counter);
130 counter.incrementAndGet();
133 public void addLinkToCurrentEntry(final String token, final String lang, final EntryTypeName entryTypeName) {
134 assert false : token + ", title=" + title;
138 // -------------------------------------------------------------------------
140 static class AppendAndIndexWikiCallback<T extends AbstractWiktionaryParser> implements WikiTokenizer.Callback {
143 StringBuilder builder;
144 IndexedEntry indexedEntry;
145 IndexBuilder indexBuilder;
146 final Map<String,FunctionCallback<T>> functionCallbacks = new LinkedHashMap<String, FunctionCallback<T>>();
148 boolean entryTypeNameSticks = false;
149 EntryTypeName entryTypeName = null;
151 final Map<String,AtomicInteger> langCodeToTCount = new LinkedHashMap<String, AtomicInteger>();
153 final NameAndArgs<T> nameAndArgs = new NameAndArgs<T>();
155 public AppendAndIndexWikiCallback(final T parser) {
156 this.parser = parser;
159 public void reset(final StringBuilder builder, final IndexedEntry indexedEntry) {
160 this.builder = builder;
161 this.indexedEntry = indexedEntry;
162 this.indexBuilder = null;
163 entryTypeName = null;
164 entryTypeNameSticks = false;
167 public void dispatch(final String wikiText, final IndexBuilder indexBuilder, final EntryTypeName entryTypeName) {
168 final IndexBuilder oldIndexBuilder = this.indexBuilder;
169 final EntryTypeName oldEntryTypeName = this.entryTypeName;
170 this.indexBuilder = indexBuilder;
171 if (!entryTypeNameSticks) {
172 this.entryTypeName = EnumUtil.min(entryTypeName, this.entryTypeName);
174 if (entryTypeName == null) this.entryTypeName = null;
175 WikiTokenizer.dispatch(wikiText, false, this);
176 this.indexBuilder = oldIndexBuilder;
177 this.entryTypeName = oldEntryTypeName;
180 public String dispatch(final String wikiText, final EntryTypeName entryTypeName) {
181 final int start = builder.length();
182 dispatch(wikiText, this.indexBuilder, entryTypeName);
183 return builder.substring(start);
187 public void onPlainText(final String plainText) {
188 // The only non-recursive callback. Just appends to the builder, and indexes.
189 builder.append(plainText);
190 if (indexBuilder != null && entryTypeName != null && indexedEntry != null) {
191 indexBuilder.addEntryWithString(indexedEntry, plainText, entryTypeName);
196 public void onWikiLink(WikiTokenizer wikiTokenizer) {
197 final String text = wikiTokenizer.wikiLinkText();
198 @SuppressWarnings("unused")
199 final String link = wikiTokenizer.wikiLinkDest();
200 dispatch(text, entryTypeName);
204 public void onFunction(
205 final WikiTokenizer wikiTokenizer,
207 final List<String> args,
208 final Map<String, String> namedArgs) {
210 FunctionCallback<T> functionCallback = functionCallbacks.get(name);
211 if (functionCallback == null || !functionCallback.onWikiFunction(wikiTokenizer, name, args, namedArgs, parser, this)) {
212 // Default function handling:
213 parser.removeUselessArgs(namedArgs);
214 final boolean single = args.isEmpty() && namedArgs.isEmpty();
215 builder.append(single ? "{" : "{{");
217 final IndexBuilder oldIndexBuilder = indexBuilder;
219 nameAndArgs.onWikiFunction(wikiTokenizer, name, args, namedArgs, parser, this);
220 indexBuilder = oldIndexBuilder;
222 builder.append(single ? "}" : "}}");
227 public void onHtml(WikiTokenizer wikiTokenizer) {
228 if (wikiTokenizer.token().startsWith("<ref>")) {
232 // Unindexed for now.
233 builder.append(wikiTokenizer.token());
237 public void onMarkup(WikiTokenizer wikiTokenizer) {
242 public final void onComment(WikiTokenizer wikiTokenizer) {
247 public void onNewline(WikiTokenizer wikiTokenizer) {
252 public void onHeading(WikiTokenizer wikiTokenizer) {
257 public void onListItem(WikiTokenizer wikiTokenizer) {
263 // --------------------------------------------------------------------
265 static final class NameAndArgs<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
267 public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
268 final Map<String, String> namedArgs, final T parser,
269 final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
272 appendAndIndexWikiCallback.dispatch(name, null);
274 for (int i = 0; i < args.size(); ++i) {
275 if (args.get(i).length() > 0) {
276 appendAndIndexWikiCallback.builder.append("|");
277 appendAndIndexWikiCallback.dispatch(args.get(i), null, null);
280 appendNamedArgs(namedArgs, appendAndIndexWikiCallback);
284 static NameAndArgs<AbstractWiktionaryParser> NAME_AND_ARGS = new NameAndArgs<AbstractWiktionaryParser>();
286 static void appendNamedArgs(final Map<String, String> namedArgs,
287 final AppendAndIndexWikiCallback<?> appendAndIndexWikiCallback) {
288 for (final Map.Entry<String, String> entry : namedArgs.entrySet()) {
289 appendAndIndexWikiCallback.builder.append("|");
290 appendAndIndexWikiCallback.dispatch(entry.getKey(), null, null);
291 appendAndIndexWikiCallback.builder.append("=");
292 EntryTypeName entryTypeName = null;
293 IndexBuilder indexBuilder = null;
294 // This doesn't work: we'd need to add to word-forms.
295 // System.out.println(entry.getKey());
296 // if (entry.getKey().equals("tr")) {
297 // entryTypeName = EntryTypeName.WIKTIONARY_TRANSLITERATION;
298 // indexBuilder = appendAndIndexWikiCallback.parser.foreignIndexBuilder;
300 appendAndIndexWikiCallback.dispatch(entry.getValue(), indexBuilder, entryTypeName);