X-Git-Url: https://git.mdrn.pl/pylucene.git/blobdiff_plain/a2e61f0c04805cfcb8706176758d1283c7e3a55c..aaeed5504b982cf3545252ab528713250aa33eed:/lucene-java-3.4.0/lucene/contrib/analyzers/common/src/java/org/apache/lucene/analysis/ru/RussianLetterTokenizer.java
diff --git a/lucene-java-3.4.0/lucene/contrib/analyzers/common/src/java/org/apache/lucene/analysis/ru/RussianLetterTokenizer.java b/lucene-java-3.4.0/lucene/contrib/analyzers/common/src/java/org/apache/lucene/analysis/ru/RussianLetterTokenizer.java
deleted file mode 100644
index 0082831..0000000
--- a/lucene-java-3.4.0/lucene/contrib/analyzers/common/src/java/org/apache/lucene/analysis/ru/RussianLetterTokenizer.java
+++ /dev/null
@@ -1,132 +0,0 @@
-package org.apache.lucene.analysis.ru;
-
-/**
- * Licensed to the Apache Software Foundation (ASF) under one or more
- * contributor license agreements. See the NOTICE file distributed with
- * this work for additional information regarding copyright ownership.
- * The ASF licenses this file to You under the Apache License, Version 2.0
- * (the "License"); you may not use this file except in compliance with
- * the License. You may obtain a copy of the License at
- *
- * http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-
-import java.io.Reader;
-import org.apache.lucene.analysis.CharTokenizer;
-import org.apache.lucene.analysis.Tokenizer; // for javadocs
-import org.apache.lucene.analysis.LetterTokenizer; // for javadocs
-import org.apache.lucene.analysis.standard.StandardTokenizer; // for javadocs
-import org.apache.lucene.util.AttributeSource;
-import org.apache.lucene.util.Version;
-
-/**
- * A RussianLetterTokenizer is a {@link Tokenizer} that extends {@link LetterTokenizer}
- * by also allowing the basic Latin digits 0-9.
- *
- *
- * You must specify the required {@link Version} compatibility when creating
- * {@link RussianLetterTokenizer}:
- *
- * - As of 3.1, {@link CharTokenizer} uses an int based API to normalize and
- * detect token characters. See {@link CharTokenizer#isTokenChar(int)} and
- * {@link CharTokenizer#normalize(int)} for details.
- *
- * @deprecated Use {@link StandardTokenizer} instead, which has the same functionality.
- * This filter will be removed in Lucene 5.0
- */
-@Deprecated
-public class RussianLetterTokenizer extends CharTokenizer
-{
- private static final int DIGIT_0 = '0';
- private static final int DIGIT_9 = '9';
-
- /**
- * Construct a new RussianLetterTokenizer. * @param matchVersion Lucene version
- * to match See {@link above}
- *
- * @param in
- * the input to split up into tokens
- */
- public RussianLetterTokenizer(Version matchVersion, Reader in) {
- super(matchVersion, in);
- }
-
- /**
- * Construct a new RussianLetterTokenizer using a given {@link AttributeSource}.
- *
- * @param matchVersion
- * Lucene version to match See {@link above}
- * @param source
- * the attribute source to use for this {@link Tokenizer}
- * @param in
- * the input to split up into tokens
- */
- public RussianLetterTokenizer(Version matchVersion, AttributeSource source, Reader in) {
- super(matchVersion, source, in);
- }
-
- /**
- * Construct a new RussianLetterTokenizer using a given
- * {@link org.apache.lucene.util.AttributeSource.AttributeFactory}. * @param
- * matchVersion Lucene version to match See
- * {@link above}
- *
- * @param factory
- * the attribute factory to use for this {@link Tokenizer}
- * @param in
- * the input to split up into tokens
- */
- public RussianLetterTokenizer(Version matchVersion, AttributeFactory factory, Reader in) {
- super(matchVersion, factory, in);
- }
-
- /**
- * Construct a new RussianLetterTokenizer.
- *
- * @deprecated use {@link #RussianLetterTokenizer(Version, Reader)} instead. This will
- * be removed in Lucene 4.0.
- */
- @Deprecated
- public RussianLetterTokenizer(Reader in) {
- super(in);
- }
-
- /**
- * Construct a new RussianLetterTokenizer using a given {@link AttributeSource}.
- *
- * @deprecated use {@link #RussianLetterTokenizer(Version, AttributeSource, Reader)}
- * instead. This will be removed in Lucene 4.0.
- */
- @Deprecated
- public RussianLetterTokenizer(AttributeSource source, Reader in) {
- super(source, in);
- }
-
- /**
- * Construct a new RussianLetterTokenizer using a given
- * {@link org.apache.lucene.util.AttributeSource.AttributeFactory}.
- *
- * @deprecated use {@link #RussianLetterTokenizer(Version, AttributeSource.AttributeFactory, Reader)}
- * instead. This will be removed in Lucene 4.0.
- */
- @Deprecated
- public RussianLetterTokenizer(AttributeFactory factory, Reader in) {
- super(factory, in);
- }
-
-
- /**
- * Collects only characters which satisfy
- * {@link Character#isLetter(int)}.
- */
- @Override
- protected boolean isTokenChar(int c) {
- return Character.isLetter(c) || (c >= DIGIT_0 && c <= DIGIT_9);
- }
-}