@Override
public void init(Map<String, String> args) {
super.init(args);
withOriginal = getBoolean("withOriginal", true);
maxPosAsterisk = getInt("maxPosAsterisk", 2);
maxPosQuestion = getInt("maxPosQuestion", 1);
minTrailing = getInt("minTrailing", 2);
maxFractionAsterisk = getFloat("maxFractionAsterisk", 0.0f);
}
@Override
public void init(Map<String, String> args) {
super.init(args);
assureMatchVersion();
}
/**
* Returns a description of the given analyzer, by either reporting the Analyzer class name (and
* optionally luceneMatchVersion) if it's not a TokenizerChain, or if it is, querying each
* analysis factory for its name and args.
*/
protected static SimpleOrderedMap<Object> getAnalyzerProperties(Analyzer analyzer) {
SimpleOrderedMap<Object> analyzerProps = new SimpleOrderedMap<>();
if (analyzer instanceof TokenizerChain) {
Map<String, String> factoryArgs;
TokenizerChain tokenizerChain = (TokenizerChain) analyzer;
CharFilterFactory[] charFilterFactories = tokenizerChain.getCharFilterFactories();
if (0 < charFilterFactories.length) {
List<SimpleOrderedMap<Object>> charFilterProps = new ArrayList<>();
for (CharFilterFactory charFilterFactory : charFilterFactories) {
SimpleOrderedMap<Object> props = new SimpleOrderedMap<>();
props.add(CLASS_NAME, charFilterFactory.getClassArg());
factoryArgs = charFilterFactory.getOriginalArgs();
if (null != factoryArgs) {
for (String key : factoryArgs.keySet()) {
if (!CLASS_NAME.equals(key)) {
if (LUCENE_MATCH_VERSION_PARAM.equals(key)) {
if (charFilterFactory.isExplicitLuceneMatchVersion()) {
props.add(key, factoryArgs.get(key));
}
} else {
props.add(key, factoryArgs.get(key));
}
}
}
}
charFilterProps.add(props);
}
analyzerProps.add(CHAR_FILTERS, charFilterProps);
}
SimpleOrderedMap<Object> tokenizerProps = new SimpleOrderedMap<>();
TokenizerFactory tokenizerFactory = tokenizerChain.getTokenizerFactory();
tokenizerProps.add(CLASS_NAME, tokenizerFactory.getClassArg());
factoryArgs = tokenizerFactory.getOriginalArgs();
if (null != factoryArgs) {
for (String key : factoryArgs.keySet()) {
if (!CLASS_NAME.equals(key)) {
if (LUCENE_MATCH_VERSION_PARAM.equals(key)) {
if (tokenizerFactory.isExplicitLuceneMatchVersion()) {
tokenizerProps.add(key, factoryArgs.get(key));
}
} else {
tokenizerProps.add(key, factoryArgs.get(key));
}
}
}
}
analyzerProps.add(TOKENIZER, tokenizerProps);
TokenFilterFactory[] filterFactories = tokenizerChain.getTokenFilterFactories();
if (0 < filterFactories.length) {
List<SimpleOrderedMap<Object>> filterProps = new ArrayList<>();
for (TokenFilterFactory filterFactory : filterFactories) {
SimpleOrderedMap<Object> props = new SimpleOrderedMap<>();
props.add(CLASS_NAME, filterFactory.getClassArg());
factoryArgs = filterFactory.getOriginalArgs();
if (null != factoryArgs) {
for (String key : factoryArgs.keySet()) {
if (!CLASS_NAME.equals(key)) {
if (LUCENE_MATCH_VERSION_PARAM.equals(key)) {
if (filterFactory.isExplicitLuceneMatchVersion()) {
props.add(key, factoryArgs.get(key));
}
} else {
props.add(key, factoryArgs.get(key));
}
}
}
}
filterProps.add(props);
}
analyzerProps.add(FILTERS, filterProps);
}
} else { // analyzer is not instanceof TokenizerChain
analyzerProps.add(CLASS_NAME, analyzer.getClass().getName());
if (analyzer.getVersion() != Version.LATEST) {
analyzerProps.add(LUCENE_MATCH_VERSION_PARAM, analyzer.getVersion().toString());
}
}
return analyzerProps;
}
