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.util.LinkedHashMap;
24 import java.util.LinkedHashSet;
25 import java.util.List;
28 import java.util.SortedMap;
29 import java.util.TreeMap;
30 import java.util.concurrent.atomic.AtomicInteger;
31 import java.util.logging.Level;
32 import java.util.logging.Logger;
33 import java.util.regex.Pattern;
35 import com.hughes.android.dictionary.engine.EntrySource;
36 import com.hughes.android.dictionary.engine.EntryTypeName;
37 import com.hughes.android.dictionary.engine.IndexBuilder;
38 import com.hughes.android.dictionary.engine.IndexedEntry;
39 import com.hughes.android.dictionary.parser.Parser;
40 import com.hughes.android.dictionary.parser.WikiTokenizer;
41 import com.hughes.util.EnumUtil;
43 public abstract class AbstractWiktionaryParser implements Parser {
45 static final Logger LOG = Logger.getLogger("WiktionaryParser");
47 final SortedMap<String, AtomicInteger> counters = new TreeMap<String, AtomicInteger>();
48 final Set<String> pairsAdded = new LinkedHashSet<String>();
50 EntrySource entrySource;
54 abstract void parseSection(final String heading, final String text);
56 abstract void removeUselessArgs(final Map<String, String> namedArgs);
59 public void parse(final File file, final EntrySource entrySource, final int pageLimit) throws IOException {
60 this.entrySource = entrySource;
62 final DataInputStream dis = new DataInputStream(new BufferedInputStream(new FileInputStream(file)));
65 if (pageLimit >= 0 && pageCount >= pageLimit) {
70 title = dis.readUTF();
71 } catch (EOFException e) {
72 LOG.log(Level.INFO, "EOF reading split.");
76 final String heading = dis.readUTF();
77 final int bytesLength = dis.readInt();
78 final byte[] bytes = new byte[bytesLength];
80 final String text = new String(bytes, "UTF8");
82 parseSection(heading, text);
85 if (pageCount % 1000 == 0) {
86 LOG.info("pageCount=" + pageCount);
90 LOG.info("***COUNTERS***");
91 for (final Map.Entry<String, AtomicInteger> entry : counters.entrySet()) {
92 LOG.info(entry.getKey() + ": " + entry.getValue());
97 static final Pattern whitespace = Pattern.compile("\\s+");
98 static String trim(final String s) {
99 return whitespace.matcher(s).replaceAll(" ").trim();
102 public void incrementCount(final String string) {
103 AtomicInteger counter = counters.get(string);
104 if (counter == null) {
105 counter = new AtomicInteger();
106 counters.put(string, counter);
108 counter.incrementAndGet();
112 // -------------------------------------------------------------------------
114 static class AppendAndIndexWikiCallback<T extends AbstractWiktionaryParser> implements WikiTokenizer.Callback {
117 StringBuilder builder;
118 IndexedEntry indexedEntry;
119 IndexBuilder indexBuilder;
120 final Map<String,FunctionCallback<T>> functionCallbacks = new LinkedHashMap<String, FunctionCallback<T>>();
122 boolean entryTypeNameSticks = false;
123 EntryTypeName entryTypeName = null;
125 final Map<String,AtomicInteger> langCodeToTCount = new LinkedHashMap<String, AtomicInteger>();
127 final NameAndArgs<T> nameAndArgs = new NameAndArgs<T>();
129 public AppendAndIndexWikiCallback(final T parser) {
130 this.parser = parser;
133 public void reset(final StringBuilder builder, final IndexedEntry indexedEntry) {
134 this.builder = builder;
135 this.indexedEntry = indexedEntry;
136 this.indexBuilder = null;
137 entryTypeName = null;
138 entryTypeNameSticks = false;
141 public void dispatch(final String wikiText, final IndexBuilder indexBuilder, final EntryTypeName entryTypeName) {
142 final IndexBuilder oldIndexBuilder = this.indexBuilder;
143 final EntryTypeName oldEntryTypeName = this.entryTypeName;
144 this.indexBuilder = indexBuilder;
145 if (!entryTypeNameSticks) {
146 this.entryTypeName = EnumUtil.min(entryTypeName, this.entryTypeName);
148 if (entryTypeName == null) this.entryTypeName = null;
149 WikiTokenizer.dispatch(wikiText, false, this);
150 this.indexBuilder = oldIndexBuilder;
151 this.entryTypeName = oldEntryTypeName;
154 public String dispatch(final String wikiText, final EntryTypeName entryTypeName) {
155 final int start = builder.length();
156 dispatch(wikiText, this.indexBuilder, entryTypeName);
157 return builder.substring(start);
161 public void onPlainText(final String plainText) {
162 // The only non-recursive callback. Just appends to the builder, and indexes.
163 builder.append(plainText);
164 if (indexBuilder != null && entryTypeName != null && indexedEntry != null) {
165 indexBuilder.addEntryWithString(indexedEntry, plainText, entryTypeName);
170 public void onWikiLink(WikiTokenizer wikiTokenizer) {
171 final String text = wikiTokenizer.wikiLinkText();
172 @SuppressWarnings("unused")
173 final String link = wikiTokenizer.wikiLinkDest();
174 dispatch(text, entryTypeName);
178 public void onFunction(
179 final WikiTokenizer wikiTokenizer,
181 final List<String> args,
182 final Map<String, String> namedArgs) {
184 FunctionCallback<T> functionCallback = functionCallbacks.get(name);
185 if (functionCallback == null || !functionCallback.onWikiFunction(wikiTokenizer, name, args, namedArgs, parser, this)) {
186 // Default function handling:
187 parser.removeUselessArgs(namedArgs);
188 final boolean single = args.isEmpty() && namedArgs.isEmpty();
189 builder.append(single ? "{" : "{{");
191 final IndexBuilder oldIndexBuilder = indexBuilder;
193 nameAndArgs.onWikiFunction(wikiTokenizer, name, args, namedArgs, parser, this);
194 indexBuilder = oldIndexBuilder;
196 builder.append(single ? "}" : "}}");
201 public void onHtml(WikiTokenizer wikiTokenizer) {
202 // Unindexed for now.
203 builder.append(wikiTokenizer.token());
207 public void onMarkup(WikiTokenizer wikiTokenizer) {
212 public final void onComment(WikiTokenizer wikiTokenizer) {
217 public final void onNewline(WikiTokenizer wikiTokenizer) {
222 public final void onHeading(WikiTokenizer wikiTokenizer) {
227 public final void onListItem(WikiTokenizer wikiTokenizer) {
232 // --------------------------------------------------------------------
234 static final class NameAndArgs<T extends AbstractWiktionaryParser> implements FunctionCallback<T> {
236 public boolean onWikiFunction(final WikiTokenizer wikiTokenizer, final String name, final List<String> args,
237 final Map<String, String> namedArgs, final T parser,
238 final AppendAndIndexWikiCallback<T> appendAndIndexWikiCallback) {
241 appendAndIndexWikiCallback.builder.append(name);
243 for (int i = 0; i < args.size(); ++i) {
244 if (args.get(i).length() > 0) {
245 appendAndIndexWikiCallback.builder.append("|");
246 appendAndIndexWikiCallback.dispatch(args.get(i), null, null);
249 appendNamedArgs(namedArgs, appendAndIndexWikiCallback);
253 static NameAndArgs<AbstractWiktionaryParser> NAME_AND_ARGS = new NameAndArgs<AbstractWiktionaryParser>();
255 static void appendNamedArgs(final Map<String, String> namedArgs,
256 final AppendAndIndexWikiCallback<?> appendAndIndexWikiCallback) {
257 for (final Map.Entry<String, String> entry : namedArgs.entrySet()) {
258 appendAndIndexWikiCallback.builder.append("|");
259 appendAndIndexWikiCallback.dispatch(entry.getKey(), null, null);
260 appendAndIndexWikiCallback.builder.append("=");
261 EntryTypeName entryTypeName = null;
262 IndexBuilder indexBuilder = null;
263 // This doesn't work: we'd need to add to word-forms.
264 // System.out.println(entry.getKey());
265 // if (entry.getKey().equals("tr")) {
266 // entryTypeName = EntryTypeName.WIKTIONARY_TRANSLITERATION;
267 // indexBuilder = appendAndIndexWikiCallback.parser.foreignIndexBuilder;
269 appendAndIndexWikiCallback.dispatch(entry.getValue(), indexBuilder, entryTypeName);