private void checkDefaultProperties() {
if (!Properties.containsProperty(VXLReaderPropertyKeys.CONTAINER_TYPE)) {
Map<String, String> properties = new HashMap<String, String>();
// vxl reader
properties.put(
VXLReaderPropertyKeys.CONTAINER_TYPE,
VXLReaderPropertyKeys.ContainerType.CLASS.toString());
properties.put(VXLReaderPropertyKeys.INCLUDE_CLASS, new Boolean(true).toString());
properties.put(VXLReaderPropertyKeys.INCLUDE_INTERFACE, new Boolean(true).toString());
properties.put(VXLReaderPropertyKeys.INCLUDE_SUPERCLASS, new Boolean(true).toString());
properties.put(VXLReaderPropertyKeys.INCLUDE_ENUM, new Boolean(true).toString());
properties.put(VXLReaderPropertyKeys.INCLUDE_ATTRIBUTE, new Boolean(true).toString());
properties.put(VXLReaderPropertyKeys.INCLUDE_METHOD, new Boolean(true).toString());
properties.put(VXLReaderPropertyKeys.INCLUDE_PARAMETER, new Boolean(true).toString());
properties.put(VXLReaderPropertyKeys.INCLUDE_LOCAL_VARIABLE, new Boolean(false).toString());
properties.put(VXLReaderPropertyKeys.INCLUDE_COMMENTS, new Boolean(true).toString());
properties.put(VXLReaderPropertyKeys.INCLUDE_JAVADOC, new Boolean(true).toString());
// identifier filter
final String YES = "yes";
properties.put(FilterProperties.ENGLISH_LANGUAGE, YES);
properties.put(FilterProperties.UNDERSCORE, YES);
properties.put(FilterProperties.CAMEL_CASE, YES);
properties.put(FilterProperties.STOPWORDS, YES);
properties.put(FilterProperties.STEMMING, YES);
properties.put(FilterProperties.CONVERT_TO_LOWER_CASE, YES);
properties.put(FilterProperties.LIMIT_TERM_LENGTH, new Integer(3).toString());
properties.put(FilterProperties.STOPWORDS_FILE, Properties.STOPWORDS_FILE);
// ir vector model
properties.put(
IRPropertyKeys.DISTANCE_FUNCTION, IRPropertyKeys.DistanceFunctionType.COSINE.toString());
Properties.setProperties(properties);
}
}
public JavaVocabularyExtractionController(
File[] projects,
boolean includeComment,
boolean includeJavaDoc,
boolean removeStopWords,
boolean removeSmallWords,
int minimalTermLength) {
super();
checkDefaultProperties();
this.projects = projects;
this.setNumStages(this.projects.length * 2);
Map<String, String> properties = new HashMap<String, String>();
properties.put(VXLReaderPropertyKeys.INCLUDE_COMMENTS, String.valueOf(includeComment));
properties.put(VXLReaderPropertyKeys.INCLUDE_JAVADOC, String.valueOf(includeJavaDoc));
properties.put(FilterProperties.STOPWORDS, removeStopWords ? "yes" : "no");
properties.put(
FilterProperties.LIMIT_TERM_LENGTH,
removeSmallWords ? new Integer(minimalTermLength).toString() : new Integer(0).toString());
Properties.setProperties(properties);
this.vocabularyResultFolder =
Properties.getProperty(Properties.WORKSPACE)
+ File.separator
+ Properties.VOCABULARY_OUTPUT;
this.termDocResultFolder =
Properties.getProperty(Properties.WORKSPACE)
+ File.separator
+ Properties.TERM_DOC_MATRIX_OUTPUT;
checkResultFolder(this.vocabularyResultFolder);
checkResultFolder(this.termDocResultFolder);
props = Properties.getProperties();
this.vxlReader = new VXLReader(props);
this.identifierFilter = new IdentifierFilter(props);
}