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.IOException;
23 import java.io.InputStream;
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.Matcher;
35 import java.util.regex.Pattern;
37 import org.apache.commons.compress.compressors.CompressorException;
38 import org.apache.commons.compress.compressors.CompressorStreamFactory;
40 import com.hughes.android.dictionary.engine.EntrySource;
41 import com.hughes.android.dictionary.engine.EntryTypeName;
42 import com.hughes.android.dictionary.engine.IndexBuilder;
43 import com.hughes.android.dictionary.engine.IndexedEntry;
44 import com.hughes.android.dictionary.engine.ReadAheadBuffer;
45 import com.hughes.android.dictionary.parser.Parser;
46 import com.hughes.android.dictionary.parser.WikiTokenizer;
47 import com.hughes.util.EnumUtil;
49 public abstract class AbstractWiktionaryParser implements Parser {
51 static final Logger LOG = Logger.getLogger("WiktionaryParser");
53 private static final Pattern SUPERSCRIPT = Pattern.compile("<sup>[0-9]*</sup>");
55 final SortedMap<String, AtomicInteger> counters = new TreeMap<String, AtomicInteger>();
56 final Set<String> pairsAdded = new LinkedHashSet<String>();
58 public EntrySource entrySource;
62 abstract void parseSection(final String heading, final String text);
64 abstract void removeUselessArgs(final Map<String, String> namedArgs);
66 private static String replaceSuperscript(String in) {
68 while ((matcher = SUPERSCRIPT.matcher(in)).find()) {
70 String orig = matcher.group();
71 for (int i = 5; i < orig.length() - 6; i++)
74 switch (orig.charAt(i)) {
75 case '0': c = '\u2070'; break;
76 case '1': c = '\u00b9'; break;
77 case '2': c = '\u00b2'; break;
78 case '3': c = '\u00b3'; break;
79 case '4': c = '\u2074'; break;
80 case '5': c = '\u2075'; break;
81 case '6': c = '\u2076'; break;
82 case '7': c = '\u2077'; break;
83 case '8': c = '\u2078'; break;
84 case '9': c = '\u2079'; break;
86 if (c == 0) throw new RuntimeException();
89 in = matcher.replaceFirst(replace);
95 public void parse(final File file, final EntrySource entrySource, final int pageLimit) throws IOException {
96 this.entrySource = entrySource;
98 File input = new File(file.getPath() + ".bz2");
99 if (!input.exists()) input = new File(file.getPath() + ".gz");
100 if (!input.exists()) input = new File(file.getPath() + ".xz");
102 if (!input.exists()) {
103 // Fallback to uncompressed file
104 dis = new DataInputStream(new BufferedInputStream(new FileInputStream(file)));
106 InputStream compressedIn = new BufferedInputStream(new FileInputStream(input));
108 InputStream in = new CompressorStreamFactory().createCompressorInputStream(compressedIn);
109 in = new ReadAheadBuffer(in, 20 * 1024 * 1024);
110 dis = new DataInputStream(in);
111 } catch (CompressorException e) {
112 throw new IOException(e);
117 if (pageLimit >= 0 && pageCount >= pageLimit) {
122 title = dis.readUTF();
123 } catch (EOFException e) {
124 LOG.log(Level.INFO, "EOF reading split.");
128 final String heading = dis.readUTF();
129 final int bytesLength = dis.readInt();
130 final byte[] bytes = new byte[bytesLength];
131 dis.readFully(bytes);
132 final String text = new String(bytes, "UTF8");
134 parseSection(heading, replaceSuperscript(text));
137 if (pageCount % 1000 == 0) {
138 LOG.info("pageCount=" + pageCount);
143 LOG.info("***COUNTERS***");
144 for (final Map.Entry<String, AtomicInteger> entry : counters.entrySet()) {
145 LOG.info(entry.getKey() + ": " + entry.getValue());
150 static final Pattern whitespace = Pattern.compile("\\s+");
151 static String trim(final String s) {
152 return whitespace.matcher(s).replaceAll(" ").trim();
155 public void incrementCount(final String string) {
156 AtomicInteger counter = counters.get(string);
157 if (counter == null) {
158 counter = new AtomicInteger();
159 counters.put(string, counter);
161 counter.incrementAndGet();
164 public void addLinkToCurrentEntry(final String token, final String lang, final EntryTypeName entryTypeName) {
165 assert false : token + ", title=" + title;
169 // -------------------------------------------------------------------------
171 static class AppendAndIndexWikiCallback<T extends AbstractWiktionaryParser> implements WikiTokenizer.Callback {
174 StringBuilder builder;
175 IndexedEntry indexedEntry;
176 IndexBuilder indexBuilder;
177 final Map<String,FunctionCallback<T>> functionCallbacks = new LinkedHashMap<String, FunctionCallback<T>>();
179 boolean entryTypeNameSticks = false;
180 EntryTypeName entryTypeName = null;
182 final Map<String,AtomicInteger> langCodeToTCount = new LinkedHashMap<String, AtomicInteger>();
184 final NameAndArgs<T> nameAndArgs = new NameAndArgs<T>();
186 public AppendAndIndexWikiCallback(final T parser) {
187 this.parser = parser;
190 public void reset(final StringBuilder builder, final IndexedEntry indexedEntry) {
191 this.builder = builder;
192 this.indexedEntry = indexedEntry;
193 this.indexBuilder = null;
194 entryTypeName = null;
195 entryTypeNameSticks = false;
198 public void dispatch(final String wikiText, final IndexBuilder indexBuilder, final EntryTypeName entryTypeName) {
199 final IndexBuilder oldIndexBuilder = this.indexBuilder;
200 final EntryTypeName oldEntryTypeName = this.entryTypeName;
201 this.indexBuilder = indexBuilder;
202 if (!entryTypeNameSticks) {
203 this.entryTypeName = EnumUtil.min(entryTypeName, this.entryTypeName);
205 if (entryTypeName == null) this.entryTypeName = null;
206 WikiTokenizer.dispatch(wikiText, false, this);
207 this.indexBuilder = oldIndexBuilder;
208 this.entryTypeName = oldEntryTypeName;
211 public String dispatch(final String wikiText, final EntryTypeName entryTypeName) {
212 final int start = builder.length();
213 dispatch(wikiText, this.indexBuilder, entryTypeName);
214 return builder.substring(start);
218 public void onPlainText(final String plainText) {
219 // The only non-recursive callback. Just appends to the builder, and indexes.
220 builder.append(plainText);
221 if (indexBuilder != null && entryTypeName != null && indexedEntry != null) {
222 indexBuilder.addEntryWithString(indexedEntry, plainText, entryTypeName);
227 public void onWikiLink(WikiTokenizer wikiTokenizer) {
228 final String text = wikiTokenizer.wikiLinkText();
229 @SuppressWarnings("unused")
230 final String link = wikiTokenizer.wikiLinkDest();
231 dispatch(text, entryTypeName);
235 public void onFunction(
236 final WikiTokenizer wikiTokenizer,
238 final List<String> args,
239 final Map<String, String> namedArgs) {
241 FunctionCallback<T> functionCallback = functionCallbacks.get(name);
242 if (functionCallback == null || !functionCallback.onWikiFunction(wikiTokenizer, name, args, namedArgs, parser, this)) {
243 // Default function handling:
244 parser.removeUselessArgs(namedArgs);
245 final boolean single = args.isEmpty() && namedArgs.isEmpty();
246 builder.append(single ? "{" : "{{");
248 final IndexBuilder oldIndexBuilder = indexBuilder;
250 nameAndArgs.onWikiFunction(wikiTokenizer, name, args, namedArgs, parser, this);
251 indexBuilder = oldIndexBuilder;
253 builder.append(single ? "}" : "}}");
258 public void onHtml(WikiTokenizer wikiTokenizer) {
259 if (wikiTokenizer.token().startsWith("<ref>")) {
263 // Unindexed for now.
264 builder.append(wikiTokenizer.token());
268 public void onMarkup(WikiTokenizer wikiTokenizer) {
273 public final void onComment(WikiTokenizer wikiTokenizer) {
278 public void onNewline(WikiTokenizer wikiTokenizer) {
283 public void onHeading(WikiTokenizer wikiTokenizer) {
288 public void onListItem(WikiTokenizer wikiTokenizer) {
294 // --------------------------------------------------------------------
296 static final class NameAndArgs<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
298 public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
299 final Map<String, String> namedArgs, final T parser,
300 final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
303 appendAndIndexWikiCallback.dispatch(name, null);
305 for (int i = 0; i < args.size(); ++i) {
306 if (args.get(i).length() > 0) {
307 appendAndIndexWikiCallback.builder.append("|");
308 appendAndIndexWikiCallback.dispatch(args.get(i), null, null);
311 appendNamedArgs(namedArgs, appendAndIndexWikiCallback);
315 static NameAndArgs<AbstractWiktionaryParser> NAME_AND_ARGS = new NameAndArgs<AbstractWiktionaryParser>();
317 static void appendNamedArgs(final Map<String, String> namedArgs,
318 final AppendAndIndexWikiCallback<?> appendAndIndexWikiCallback) {
319 for (final Map.Entry<String, String> entry : namedArgs.entrySet()) {
320 appendAndIndexWikiCallback.builder.append("|");
321 appendAndIndexWikiCallback.dispatch(entry.getKey(), null, null);
322 appendAndIndexWikiCallback.builder.append("=");
323 EntryTypeName entryTypeName = null;
324 IndexBuilder indexBuilder = null;
325 // This doesn't work: we'd need to add to word-forms.
326 // System.out.println(entry.getKey());
327 // if (entry.getKey().equals("tr")) {
328 // entryTypeName = EntryTypeName.WIKTIONARY_TRANSLITERATION;
329 // indexBuilder = appendAndIndexWikiCallback.parser.foreignIndexBuilder;
331 appendAndIndexWikiCallback.dispatch(entry.getValue(), indexBuilder, entryTypeName);