public class Preprocessor
extends java.lang.Object
Constructor and Description |
---|
Preprocessor() |
Modifier and Type | Method and Description |
---|---|
Preprocessor |
addCustomAE(org.apache.uima.analysis_engine.AnalysisEngineDescription customAE) |
PreprocessorService |
asService(Lang lang) |
PreprocessorService |
asService(Lang lang,
CorpusMetadata corpusMetadata) |
PreprocessorService |
asService(TXTCorpus textCorpus) |
java.util.stream.Stream<org.apache.uima.jcas.JCas> |
asStream(TXTCorpus textCorpus) |
Preprocessor |
setDocumentLoggingEnabled(boolean documentLoggingEnabled) |
Preprocessor |
setFixedExpressionEnabled(boolean fixedExpressionEnabled) |
Preprocessor |
setHistory(TermHistory history) |
Preprocessor |
setListener(PipelineListener listener) |
Preprocessor |
setPreprocessedCorpusCache(java.nio.file.Path cachedPath) |
Preprocessor |
setResourceOptions(ResourceConfig resourceOptions) |
Preprocessor |
setTagger(Tagger tagger) |
Preprocessor |
setTaggerPath(java.nio.file.Path taggerPath) |
static org.apache.uima.jcas.JCas |
toCas(Document doc,
java.lang.String documentText) |
IndexedCorpus |
toIndexedCorpus(Lang lang,
java.util.stream.Stream<org.apache.uima.jcas.JCas> blankCasStream,
int maxSize) |
IndexedCorpus |
toIndexedCorpus(Lang lang,
java.util.stream.Stream<org.apache.uima.jcas.JCas> blankCasStream,
int maxSize,
IndexedCorpus indexedCorpus) |
IndexedCorpus |
toIndexedCorpus(TextualCorpus textCorpus,
int maxSize) |
IndexedCorpus |
toIndexedCorpus(TextualCorpus textCorpus,
int maxSize,
IndexedCorpus indexedCorpus) |
IndexedCorpus |
toIndexedCorpus(TXTCorpus textCorpus,
int maxSize) |
Preprocessor |
toJSON(java.nio.file.Path jsonPath) |
IndexedCorpus |
toPersistentIndexedCorpus(TXTCorpus textCorpus,
java.lang.String storeUrl,
int maxSize) |
XMICorpus |
toPreparedCorpusJSON(TXTCorpus textCorpus,
java.nio.file.Path jsonDir) |
XMICorpus |
toPreparedCorpusXMI(TXTCorpus textCorpus,
java.nio.file.Path xmiDir) |
Preprocessor |
toTSV(java.nio.file.Path tsvPath) |
Preprocessor |
toXMI(java.nio.file.Path xmiPath) |
public Preprocessor setTaggerPath(java.nio.file.Path taggerPath)
public Preprocessor setTagger(Tagger tagger)
public Preprocessor setDocumentLoggingEnabled(boolean documentLoggingEnabled)
public Preprocessor setFixedExpressionEnabled(boolean fixedExpressionEnabled)
public Preprocessor setListener(PipelineListener listener)
public Preprocessor setHistory(TermHistory history)
public Preprocessor setResourceOptions(ResourceConfig resourceOptions)
public Preprocessor addCustomAE(org.apache.uima.analysis_engine.AnalysisEngineDescription customAE)
public IndexedCorpus toPersistentIndexedCorpus(TXTCorpus textCorpus, java.lang.String storeUrl, int maxSize)
public IndexedCorpus toIndexedCorpus(TXTCorpus textCorpus, int maxSize)
public IndexedCorpus toIndexedCorpus(TextualCorpus textCorpus, int maxSize, IndexedCorpus indexedCorpus)
public IndexedCorpus toIndexedCorpus(TextualCorpus textCorpus, int maxSize)
public static org.apache.uima.jcas.JCas toCas(Document doc, java.lang.String documentText)
public IndexedCorpus toIndexedCorpus(Lang lang, java.util.stream.Stream<org.apache.uima.jcas.JCas> blankCasStream, int maxSize)
public IndexedCorpus toIndexedCorpus(Lang lang, java.util.stream.Stream<org.apache.uima.jcas.JCas> blankCasStream, int maxSize, IndexedCorpus indexedCorpus)
public XMICorpus toPreparedCorpusJSON(TXTCorpus textCorpus, java.nio.file.Path jsonDir)
public XMICorpus toPreparedCorpusXMI(TXTCorpus textCorpus, java.nio.file.Path xmiDir)
public java.util.stream.Stream<org.apache.uima.jcas.JCas> asStream(TXTCorpus textCorpus)
public Preprocessor setPreprocessedCorpusCache(java.nio.file.Path cachedPath)
public PreprocessorService asService(Lang lang)
public PreprocessorService asService(TXTCorpus textCorpus)
public PreprocessorService asService(Lang lang, CorpusMetadata corpusMetadata)
public Preprocessor toXMI(java.nio.file.Path xmiPath)
public Preprocessor toTSV(java.nio.file.Path tsvPath)
public Preprocessor toJSON(java.nio.file.Path jsonPath)