/**
* Creates a new child Metadata object with this object as its parent, including provided parent
* ID.
*
* @return a new child Metadata object
*/
public DocumentMetadata asParent(String id, ParentType parentType) {
setProperty("id", id);
FlexibleParameters newParameters = new FlexibleParameters();
for (String key : parameters.getKeys()) {
newParameters.setParameter("parent_" + key, parameters.getParameterValues(key));
}
newParameters.setParameter("parentType", parentType.name().toLowerCase());
return new DocumentMetadata(newParameters);
}
public void setQueryParameters(FlexibleParameters storedparams)
throws UnsupportedEncodingException {
setProperty("queryParameters", storedparams.getAsQueryString());
}
private void setProperty(String key, String[] values) {
parameters.setParameter(key, values);
}
private String getProperty(String key, String defaultValue) {
return parameters.getParameterValue(key, defaultValue);
}
private String getProperty(String key) {
return parameters.getParameterValue(key);
// return properties.getProperty(key);
}
public boolean equals(DocumentMetadata metadata) {
return parameters.equals(metadata.parameters);
}
public boolean containsKey(String string) {
return parameters.containsKey(string);
}
@Test
public void test() throws IOException {
Storage storage = new MemoryStorage();
Document document;
LuceneManager luceneManager = storage.getLuceneManager();
document = new Document();
document.add(
new TextField("lexical", "dark and stormy night in document one", Field.Store.YES));
luceneManager.addDocument(document);
DocumentTerm documentTerm;
FlexibleParameters parameters;
parameters = new FlexibleParameters();
parameters.addParameter("string", "It was a dark and stormy night.");
parameters.addParameter("string", "It was the best of times it was the worst of times.");
parameters.addParameter("tool", "StepEnabledIndexedCorpusCreator");
CorpusCreator creator = new CorpusCreator(storage, parameters);
creator.run();
parameters.setParameter("corpus", creator.getStoredId());
parameters.setParameter("tool", "DocumentTermFrequencies");
DocumentTerms documentTermFrequencies;
List<DocumentTerm> documentTerms;
parameters.setParameter("query", "dar*");
documentTermFrequencies = new DocumentTerms(storage, parameters);
documentTermFrequencies.run();
documentTerms = documentTermFrequencies.getDocumentTerms();
assertEquals(1, documentTerms.size());
documentTerm = documentTerms.get(0);
assertEquals("dark", documentTerm.getTerm());
assertEquals(1, documentTerm.getRawFrequency());
assertEquals(0, documentTerm.getDocumentIndex());
parameters.setParameter("query", "it was");
documentTermFrequencies = new DocumentTerms(storage, parameters);
documentTermFrequencies.run();
// we sort by reverse frequency by default
documentTerms = documentTermFrequencies.getDocumentTerms();
assertEquals(2, documentTerms.size());
documentTerm = documentTerms.get(0);
assertEquals(1, documentTerm.getDocumentIndex());
assertEquals("it was", documentTerm.getTerm());
assertEquals(2, documentTerm.getRawFrequency());
documentTerm = documentTerms.get(1);
assertEquals(0, documentTerm.getDocumentIndex());
assertEquals("it was", documentTerm.getTerm());
assertEquals(1, documentTerm.getRawFrequency());
parameters.removeParameter("query");
documentTermFrequencies = new DocumentTerms(storage, parameters);
documentTermFrequencies.run();
documentTerms = documentTermFrequencies.getDocumentTerms();
assertEquals(14, documentTerms.size());
documentTerm = documentTerms.get(0);
assertEquals("it", documentTerm.getTerm());
assertEquals(2, documentTerm.getRawFrequency());
parameters.setParameter("limit", 1);
documentTermFrequencies = new DocumentTerms(storage, parameters);
documentTermFrequencies.run();
documentTerms = documentTermFrequencies.getDocumentTerms();
assertEquals(1, documentTerms.size());
documentTerm = documentTerms.get(0);
assertEquals("it", documentTerm.getTerm());
assertEquals(2, documentTerm.getRawFrequency());
parameters.setParameter("start", 1);
documentTermFrequencies = new DocumentTerms(storage, parameters);
documentTermFrequencies.run();
documentTerms = documentTermFrequencies.getDocumentTerms();
assertEquals(1, documentTerms.size());
documentTerm = documentTerms.get(0);
assertEquals("of", documentTerm.getTerm());
assertEquals(2, documentTerm.getRawFrequency());
parameters.setParameter("start", 50);
documentTermFrequencies = new DocumentTerms(storage, parameters);
documentTermFrequencies.run();
documentTerms = documentTermFrequencies.getDocumentTerms();
assertEquals(0, documentTerms.size());
// with stopwords
parameters.setParameter("stopList", "stop.en.taporware.txt");
parameters.removeParameter("start");
parameters.removeParameter("limit");
documentTermFrequencies = new DocumentTerms(storage, parameters);
documentTermFrequencies.run();
documentTerms = documentTermFrequencies.getDocumentTerms();
assertEquals(6, documentTerms.size());
documentTerm = documentTerms.get(0);
assertEquals("times", documentTerm.getTerm());
documentTerm = documentTerms.get(documentTerms.size() - 1);
assertEquals("worst", documentTerm.getTerm());
storage.destroy();
}