Author: nick
Date: Mon Jun 8 12:25:15 2015
New Revision: 1684170
URL: http://svn.apache.org/r1684170
Log:
Fix indents to match http://tika.apache.org/contribute.html#Code_Formatting
TIKA-1642
Modified:
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESAnnotationProperty.java
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESConfig.java
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESContentHandler.java
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESParser.java
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESSerializer.java
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESUtils.java
Modified:
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESAnnotationProperty.java
URL:
http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESAnnotationProperty.java?rev=1684170&r1=1684169&r2=1684170&view=diff
==============================================================================
---
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESAnnotationProperty.java
(original)
+++
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESAnnotationProperty.java
Mon Jun 8 12:25:15 2015
@@ -23,24 +23,24 @@ import org.apache.ctakes.typesystem.type
*
*/
public enum CTAKESAnnotationProperty {
- BEGIN("start"),
- END("end"),
- CONDITIONAL("conditional"),
- CONFIDENCE("confidence"),
- DISCOVERY_TECNIQUE("discoveryTechnique"),
- GENERIC("generic"),
- HISTORY_OF("historyOf"),
- ID("id"),
- ONTOLOGY_CONCEPT_ARR("ontologyConceptArr"),
- POLARITY("polarity");
-
- private String name;
-
- CTAKESAnnotationProperty(String name) {
- this.name = name;
- }
-
- public String getName() {
- return name;
- }
+ BEGIN("start"),
+ END("end"),
+ CONDITIONAL("conditional"),
+ CONFIDENCE("confidence"),
+ DISCOVERY_TECNIQUE("discoveryTechnique"),
+ GENERIC("generic"),
+ HISTORY_OF("historyOf"),
+ ID("id"),
+ ONTOLOGY_CONCEPT_ARR("ontologyConceptArr"),
+ POLARITY("polarity");
+
+ private String name;
+
+ CTAKESAnnotationProperty(String name) {
+ this.name = name;
+ }
+
+ public String getName() {
+ return name;
+ }
}
\ No newline at end of file
Modified:
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESConfig.java
URL:
http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESConfig.java?rev=1684170&r1=1684169&r2=1684170&view=diff
==============================================================================
---
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESConfig.java
(original)
+++
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESConfig.java
Mon Jun 8 12:25:15 2015
@@ -24,314 +24,313 @@ import java.util.Properties;
import org.apache.tika.io.NullOutputStream;
-/*
+/**
* Configuration for {@see CTAKESContentHandler}.
*
* This class allows to enable cTAKES and set its parameters.
- *
*/
public class CTAKESConfig implements Serializable {
- /**
- * Serial version UID
- */
- private static final long serialVersionUID = -1599741171775528923L;
-
- // Path to XML descriptor for AnalysisEngine
- private String aeDescriptorPath =
"/ctakes-core/desc/analysis_engine/SentencesAndTokensAggregate.xml";
-
- // UMLS username
- private String UMLSUser = "";
-
- // UMLS password
- private String UMLSPass = "";
-
- // Enables formatted output
- private boolean prettyPrint = true;
-
- // Type of cTAKES (UIMA) serializer
- private CTAKESSerializer serializerType = CTAKESSerializer.XMI;
-
- // OutputStream object used for CAS serialization
- private OutputStream stream = NullOutputStream.NULL_OUTPUT_STREAM;
-
- // Enables CAS serialization
- private boolean serialize = false;
-
- // Enables text analysis using cTAKES
- private boolean text = true;
-
- // List of metadata to analyze using cTAKES
- private String[] metadata = null;
-
- // List of annotation properties to add to metadata in addition to text
covered by an annotation
- private CTAKESAnnotationProperty[] annotationProps = null;
-
- // Character used to separate the annotation properties into metadata
- private char separatorChar = ':';
-
- /**
- * Default constructor.
- */
- public CTAKESConfig() {
-
init(this.getClass().getResourceAsStream("CTAKESConfig.properties"));
- }
-
- /**
- * Loads properties from InputStream and then tries to close
InputStream.
- * @param stream {@see InputStream} object used to read properties.
- */
- public CTAKESConfig(InputStream stream) {
- init(stream);
- }
-
- private void init(InputStream stream) {
- if (stream == null) {
- return;
- }
- Properties props = new Properties();
-
- try {
- props.load(stream);
- } catch (IOException e) {
- // TODO warning
- } finally {
- if (stream != null) {
- try {
- stream.close();
- } catch (IOException ioe) {
- // TODO warning
- }
- }
- }
-
- setAeDescriptorPath(props.getProperty("aeDescriptorPath",
getAeDescriptorPath()));
- setUMLSUser(props.getProperty("UMLSUser", getUMLSUser()));
- setUMLSPass(props.getProperty("UMLSPass", getUMLSPass()));
- setText(Boolean.valueOf(props.getProperty("text",
Boolean.toString(isText()))));
- setMetadata(props.getProperty("metadata",
getMetadataAsString()).split(","));
- setAnnotationProps(props.getProperty("annotationProps",
getAnnotationPropsAsString()).split(","));
- setSeparatorChar(props.getProperty("separatorChar",
Character.toString(getSeparatorChar())).charAt(0));
- }
-
- /**
- * Returns the path to XML descriptor for AnalysisEngine.
- * @return the path to XML descriptor for AnalysisEngine.
- */
- public String getAeDescriptorPath() {
- return aeDescriptorPath;
- }
-
- /**
- * Returns the UMLS username.
- * @return the UMLS username.
- */
- public String getUMLSUser() {
- return UMLSUser;
- }
-
- /**
- * Returns the UMLS password.
- * @return the UMLS password.
- */
- public String getUMLSPass() {
- return UMLSPass;
- }
-
- /**
- * Returns {@code true} if formatted output is enabled, {@code false}
otherwise.
- * @return {@code true} if formatted output is enabled, {@code false}
otherwise.
- */
- public boolean isPrettyPrint() {
- return prettyPrint;
- }
-
- /**
- * Returns the type of cTAKES (UIMA) serializer used to write the CAS.
- * @return the type of cTAKES serializer.
- */
- public CTAKESSerializer getSerializerType() {
- return serializerType;
- }
-
- /**
- * Returns an {@see OutputStream} object used write the CAS.
- * @return {@see OutputStream} object used write the CAS.
- */
- public OutputStream getOutputStream() {
- return stream;
- }
-
- /**
- * Returns {@code true} if CAS serialization is enabled, {@code false}
otherwise.
- * @return {@code true} if CAS serialization output is enabled, {@code
false} otherwise.
- */
- public boolean isSerialize() {
- return serialize;
- }
-
- /**
- * Returns {@code true} if content text analysis is enabled {@code
false} otherwise.
- * @return {@code true} if content text analysis is enabled {@code
false} otherwise.
- */
- public boolean isText() {
- return text;
- }
-
- /**
- * Returns an array of metadata whose values will be analyzed using
cTAKES.
- * @return an array of metadata whose values will be analyzed using
cTAKES.
- */
- public String[] getMetadata() {
- return metadata;
- }
-
- /**
- * Returns a string containing a comma-separated list of metadata whose
values will be analyzed using cTAKES.
- * @return a string containing a comma-separated list of metadata whose
values will be analyzed using cTAKES.
- */
- public String getMetadataAsString() {
- if (metadata == null) {
- return "";
- }
- StringBuilder sb = new StringBuilder();
- for (int i = 0; i < metadata.length; i++) {
- sb.append(metadata[i]);
- if (i < metadata.length-1) {
- sb.append(",");
- }
- }
- return sb.toString();
- }
-
- /**
- * Returns an array of {@see CTAKESAnnotationProperty}'s that will be
included into cTAKES metadata.
- * @return an array of {@see CTAKESAnnotationProperty}'s that will be
included into cTAKES metadata.
- */
- public CTAKESAnnotationProperty[] getAnnotationProps() {
- return annotationProps;
- }
-
- /**
- * Returns a string containing a comma-separated list of {@see
CTAKESAnnotationProperty} names that will be included into cTAKES metadata.
- * @return
- */
- public String getAnnotationPropsAsString() {
- StringBuilder sb = new StringBuilder();
- sb.append("coveredText");
- if (annotationProps != null) {
- for (CTAKESAnnotationProperty property :
annotationProps) {
- sb.append(separatorChar);
- sb.append(property.getName());
- }
- }
- return sb.toString();
- }
-
- /**
- * Returns the separator character used for annotation properties.
- * @return the separator character used for annotation properties.
- */
- public char getSeparatorChar() {
- return separatorChar;
- }
-
- /**
- * Sets the path to XML descriptor for AnalysisEngine.
- * @param aeDescriptorPath the path to XML descriptor for
AnalysisEngine.
- */
- public void setAeDescriptorPath(String aeDescriptorPath) {
- this.aeDescriptorPath = aeDescriptorPath;
- }
-
- /**
- * Sets the UMLS username.
- * @param uMLSUser the UMLS username.
- */
- public void setUMLSUser(String uMLSUser) {
- this.UMLSUser = uMLSUser;
- }
-
- /**
- * Sets the UMLS password.
- * @param uMLSPass the UMLS password.
- */
- public void setUMLSPass(String uMLSPass) {
- this.UMLSPass = uMLSPass;
- }
-
- /**
- * Enables the formatted output for serializer.
- * @param prettyPrint {@true} to enable formatted output, {@code false}
otherwise.
- */
- public void setPrettyPrint(boolean prettyPrint) {
- this.prettyPrint = prettyPrint;
- }
-
- /**
- * Sets the type of cTAKES (UIMA) serializer used to write CAS.
- * @param serializerType the type of cTAKES serializer.
- */
- public void setSerializerType(CTAKESSerializer serializerType) {
- this.serializerType = serializerType;
- }
-
- /**
- * Sets the {@see OutputStream} object used to write the CAS.
- * @param stream the {@see OutputStream} object used to write the CAS.
- */
- public void setOutputStream(OutputStream stream) {
- this.stream = stream;
- }
-
- /**
- * Enables CAS serialization.
- * @param serialize {@true} to enable CAS serialization, {@code false}
otherwise.
- */
- public void setSerialize(boolean serialize) {
- this.serialize = serialize;
- }
-
- /**
- * Enables content text analysis using cTAKES.
- * @param text {@true} to enable content text analysis, {@code false}
otherwise.
- */
- public void setText(boolean text) {
- this.text = text;
- }
-
- /**
- * Sets the metadata whose values will be analyzed using cTAKES.
- * @param metadata the metadata whose values will be analyzed using
cTAKES.
- */
- public void setMetadata(String[] metadata) {
- this.metadata = metadata;
- }
-
- /**
- * Sets the {@see CTAKESAnnotationProperty}'s that will be included
into cTAKES metadata.
- * @param annotationProps the {@see CTAKESAnnotationProperty}'s that
will be included into cTAKES metadata.
- */
- public void setAnnotationProps(CTAKESAnnotationProperty[]
annotationProps) {
- this.annotationProps = annotationProps;
- }
-
- /**
- * ets the {@see CTAKESAnnotationProperty}'s that will be included into
cTAKES metadata.
- * @param annotationProps the {@see CTAKESAnnotationProperty}'s that
will be included into cTAKES metadata.
- */
- public void setAnnotationProps(String[] annotationProps) {
- CTAKESAnnotationProperty[] properties = new
CTAKESAnnotationProperty[annotationProps.length];
- for (int i = 0; i < annotationProps.length; i++) {
- properties[i] =
CTAKESAnnotationProperty.valueOf(annotationProps[i]);
- }
- setAnnotationProps(properties);
- }
-
- /**
- * Sets the separator character used for annotation properties.
- * @param separatorChar the separator character used for annotation
properties.
- */
- public void setSeparatorChar(char separatorChar) {
- this.separatorChar = separatorChar;
- }
+ /**
+ * Serial version UID
+ */
+ private static final long serialVersionUID = -1599741171775528923L;
+
+ // Path to XML descriptor for AnalysisEngine
+ private String aeDescriptorPath =
"/ctakes-core/desc/analysis_engine/SentencesAndTokensAggregate.xml";
+
+ // UMLS username
+ private String UMLSUser = "";
+
+ // UMLS password
+ private String UMLSPass = "";
+
+ // Enables formatted output
+ private boolean prettyPrint = true;
+
+ // Type of cTAKES (UIMA) serializer
+ private CTAKESSerializer serializerType = CTAKESSerializer.XMI;
+
+ // OutputStream object used for CAS serialization
+ private OutputStream stream = NullOutputStream.NULL_OUTPUT_STREAM;
+
+ // Enables CAS serialization
+ private boolean serialize = false;
+
+ // Enables text analysis using cTAKES
+ private boolean text = true;
+
+ // List of metadata to analyze using cTAKES
+ private String[] metadata = null;
+
+ // List of annotation properties to add to metadata in addition to text
covered by an annotation
+ private CTAKESAnnotationProperty[] annotationProps = null;
+
+ // Character used to separate the annotation properties into metadata
+ private char separatorChar = ':';
+
+ /**
+ * Default constructor.
+ */
+ public CTAKESConfig() {
+ init(this.getClass().getResourceAsStream("CTAKESConfig.properties"));
+ }
+
+ /**
+ * Loads properties from InputStream and then tries to close InputStream.
+ * @param stream {@see InputStream} object used to read properties.
+ */
+ public CTAKESConfig(InputStream stream) {
+ init(stream);
+ }
+
+ private void init(InputStream stream) {
+ if (stream == null) {
+ return;
+ }
+ Properties props = new Properties();
+
+ try {
+ props.load(stream);
+ } catch (IOException e) {
+ // TODO warning
+ } finally {
+ if (stream != null) {
+ try {
+ stream.close();
+ } catch (IOException ioe) {
+ // TODO warning
+ }
+ }
+ }
+
+ setAeDescriptorPath(props.getProperty("aeDescriptorPath",
getAeDescriptorPath()));
+ setUMLSUser(props.getProperty("UMLSUser", getUMLSUser()));
+ setUMLSPass(props.getProperty("UMLSPass", getUMLSPass()));
+ setText(Boolean.valueOf(props.getProperty("text",
Boolean.toString(isText()))));
+ setMetadata(props.getProperty("metadata",
getMetadataAsString()).split(","));
+ setAnnotationProps(props.getProperty("annotationProps",
getAnnotationPropsAsString()).split(","));
+ setSeparatorChar(props.getProperty("separatorChar",
Character.toString(getSeparatorChar())).charAt(0));
+ }
+
+ /**
+ * Returns the path to XML descriptor for AnalysisEngine.
+ * @return the path to XML descriptor for AnalysisEngine.
+ */
+ public String getAeDescriptorPath() {
+ return aeDescriptorPath;
+ }
+
+ /**
+ * Returns the UMLS username.
+ * @return the UMLS username.
+ */
+ public String getUMLSUser() {
+ return UMLSUser;
+ }
+
+ /**
+ * Returns the UMLS password.
+ * @return the UMLS password.
+ */
+ public String getUMLSPass() {
+ return UMLSPass;
+ }
+
+ /**
+ * Returns {@code true} if formatted output is enabled, {@code false}
otherwise.
+ * @return {@code true} if formatted output is enabled, {@code false}
otherwise.
+ */
+ public boolean isPrettyPrint() {
+ return prettyPrint;
+ }
+
+ /**
+ * Returns the type of cTAKES (UIMA) serializer used to write the CAS.
+ * @return the type of cTAKES serializer.
+ */
+ public CTAKESSerializer getSerializerType() {
+ return serializerType;
+ }
+
+ /**
+ * Returns an {@see OutputStream} object used write the CAS.
+ * @return {@see OutputStream} object used write the CAS.
+ */
+ public OutputStream getOutputStream() {
+ return stream;
+ }
+
+ /**
+ * Returns {@code true} if CAS serialization is enabled, {@code false}
otherwise.
+ * @return {@code true} if CAS serialization output is enabled, {@code
false} otherwise.
+ */
+ public boolean isSerialize() {
+ return serialize;
+ }
+
+ /**
+ * Returns {@code true} if content text analysis is enabled {@code false}
otherwise.
+ * @return {@code true} if content text analysis is enabled {@code false}
otherwise.
+ */
+ public boolean isText() {
+ return text;
+ }
+
+ /**
+ * Returns an array of metadata whose values will be analyzed using cTAKES.
+ * @return an array of metadata whose values will be analyzed using cTAKES.
+ */
+ public String[] getMetadata() {
+ return metadata;
+ }
+
+ /**
+ * Returns a string containing a comma-separated list of metadata whose
values will be analyzed using cTAKES.
+ * @return a string containing a comma-separated list of metadata whose
values will be analyzed using cTAKES.
+ */
+ public String getMetadataAsString() {
+ if (metadata == null) {
+ return "";
+ }
+ StringBuilder sb = new StringBuilder();
+ for (int i = 0; i < metadata.length; i++) {
+ sb.append(metadata[i]);
+ if (i < metadata.length-1) {
+ sb.append(",");
+ }
+ }
+ return sb.toString();
+ }
+
+ /**
+ * Returns an array of {@see CTAKESAnnotationProperty}'s that will be
included into cTAKES metadata.
+ * @return an array of {@see CTAKESAnnotationProperty}'s that will be
included into cTAKES metadata.
+ */
+ public CTAKESAnnotationProperty[] getAnnotationProps() {
+ return annotationProps;
+ }
+
+ /**
+ * Returns a string containing a comma-separated list of {@see
CTAKESAnnotationProperty} names that will be included into cTAKES metadata.
+ * @return
+ */
+ public String getAnnotationPropsAsString() {
+ StringBuilder sb = new StringBuilder();
+ sb.append("coveredText");
+ if (annotationProps != null) {
+ for (CTAKESAnnotationProperty property : annotationProps) {
+ sb.append(separatorChar);
+ sb.append(property.getName());
+ }
+ }
+ return sb.toString();
+ }
+
+ /**
+ * Returns the separator character used for annotation properties.
+ * @return the separator character used for annotation properties.
+ */
+ public char getSeparatorChar() {
+ return separatorChar;
+ }
+
+ /**
+ * Sets the path to XML descriptor for AnalysisEngine.
+ * @param aeDescriptorPath the path to XML descriptor for AnalysisEngine.
+ */
+ public void setAeDescriptorPath(String aeDescriptorPath) {
+ this.aeDescriptorPath = aeDescriptorPath;
+ }
+
+ /**
+ * Sets the UMLS username.
+ * @param uMLSUser the UMLS username.
+ */
+ public void setUMLSUser(String uMLSUser) {
+ this.UMLSUser = uMLSUser;
+ }
+
+ /**
+ * Sets the UMLS password.
+ * @param uMLSPass the UMLS password.
+ */
+ public void setUMLSPass(String uMLSPass) {
+ this.UMLSPass = uMLSPass;
+ }
+
+ /**
+ * Enables the formatted output for serializer.
+ * @param prettyPrint {@true} to enable formatted output, {@code false}
otherwise.
+ */
+ public void setPrettyPrint(boolean prettyPrint) {
+ this.prettyPrint = prettyPrint;
+ }
+
+ /**
+ * Sets the type of cTAKES (UIMA) serializer used to write CAS.
+ * @param serializerType the type of cTAKES serializer.
+ */
+ public void setSerializerType(CTAKESSerializer serializerType) {
+ this.serializerType = serializerType;
+ }
+
+ /**
+ * Sets the {@see OutputStream} object used to write the CAS.
+ * @param stream the {@see OutputStream} object used to write the CAS.
+ */
+ public void setOutputStream(OutputStream stream) {
+ this.stream = stream;
+ }
+
+ /**
+ * Enables CAS serialization.
+ * @param serialize {@true} to enable CAS serialization, {@code false}
otherwise.
+ */
+ public void setSerialize(boolean serialize) {
+ this.serialize = serialize;
+ }
+
+ /**
+ * Enables content text analysis using cTAKES.
+ * @param text {@true} to enable content text analysis, {@code false}
otherwise.
+ */
+ public void setText(boolean text) {
+ this.text = text;
+ }
+
+ /**
+ * Sets the metadata whose values will be analyzed using cTAKES.
+ * @param metadata the metadata whose values will be analyzed using cTAKES.
+ */
+ public void setMetadata(String[] metadata) {
+ this.metadata = metadata;
+ }
+
+ /**
+ * Sets the {@see CTAKESAnnotationProperty}'s that will be included into
cTAKES metadata.
+ * @param annotationProps the {@see CTAKESAnnotationProperty}'s that will
be included into cTAKES metadata.
+ */
+ public void setAnnotationProps(CTAKESAnnotationProperty[] annotationProps)
{
+ this.annotationProps = annotationProps;
+ }
+
+ /**
+ * ets the {@see CTAKESAnnotationProperty}'s that will be included into
cTAKES metadata.
+ * @param annotationProps the {@see CTAKESAnnotationProperty}'s that will
be included into cTAKES metadata.
+ */
+ public void setAnnotationProps(String[] annotationProps) {
+ CTAKESAnnotationProperty[] properties = new
CTAKESAnnotationProperty[annotationProps.length];
+ for (int i = 0; i < annotationProps.length; i++) {
+ properties[i] =
CTAKESAnnotationProperty.valueOf(annotationProps[i]);
+ }
+ setAnnotationProps(properties);
+ }
+
+ /**
+ * Sets the separator character used for annotation properties.
+ * @param separatorChar the separator character used for annotation
properties.
+ */
+ public void setSeparatorChar(char separatorChar) {
+ this.separatorChar = separatorChar;
+ }
}
\ No newline at end of file
Modified:
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESContentHandler.java
URL:
http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESContentHandler.java?rev=1684170&r1=1684169&r2=1684170&view=diff
==============================================================================
---
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESContentHandler.java
(original)
+++
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESContentHandler.java
Mon Jun 8 12:25:15 2015
@@ -37,112 +37,111 @@ import org.xml.sax.helpers.DefaultHandle
* that is a natural language processing system for extraction of information
* from electronic medical record clinical free-text.
* </p>
- *
*/
public class CTAKESContentHandler extends ContentHandlerDecorator {
- // Prefix used for metadata including cTAKES annotations
- public static String CTAKES_META_PREFIX = "ctakes:";
-
- // Configuration object for CTAKESContentHandler
- private CTAKESConfig config = null;
-
- // StringBuilder object used to build the clinical free-text for cTAKES
- private StringBuilder sb = null;
-
- // Metadata object used for cTAKES annotations
- private Metadata metadata = null;
-
- /**
- * Creates a new {@see CTAKESContentHandler} for the given {@see
ContentHandler} and Metadata objects.
- * @param handler the {@see ContentHandler} object to be decorated.
- * @param metadata the {@see Metadata} object that will be populated
using biomedical information extracted by cTAKES.
- * @param config the {@see CTAKESConfig} object used to configure the
handler.
- */
- public CTAKESContentHandler(ContentHandler handler, Metadata metadata,
CTAKESConfig config) {
- super(handler);
- this.metadata = metadata;
- this.config = config;
- this.sb = new StringBuilder();
- }
-
- /**
- * Creates a new {@see CTAKESContentHandler} for the given {@see
ContentHandler} and Metadata objects.
- * @param handler the {@see ContentHandler} object to be decorated.
- * @param metadata the {@see Metadata} object that will be populated
using biomedical information extracted by cTAKES.
- */
- public CTAKESContentHandler(ContentHandler handler, Metadata metadata) {
- this(handler, metadata, new CTAKESConfig());
- }
-
- /**
- * Default constructor.
- */
- public CTAKESContentHandler() {
- this(new DefaultHandler(), new Metadata());
- }
-
- @Override
- public void characters(char[] ch, int start, int length) throws
SAXException {
- if (config.isText()) {
- sb.append(ch, start, length);
- }
- super.characters(ch, start, length);
- }
-
- @Override
- public void endDocument() throws SAXException {
- try {
- // create an Analysis Engine
- AnalysisEngine ae =
CTAKESUtils.getAnalysisEngine(config.getAeDescriptorPath(),
config.getUMLSUser(), config.getUMLSPass());
-
- // create a JCas, given an AE
- JCas jcas = CTAKESUtils.getJCas(ae);
-
- StringBuilder metaText = new StringBuilder();
- for (String name : config.getMetadata()) {
- for (String value : metadata.getValues(name)) {
- metaText.append(value);
- metaText.append(System.lineSeparator());
- }
- }
-
- // analyze text
- jcas.setDocumentText(metaText.toString() +
sb.toString());
- ae.process(jcas);
-
- // add annotations to metadata
- metadata.add(CTAKES_META_PREFIX + "schema",
config.getAnnotationPropsAsString());
- CTAKESAnnotationProperty[] annotationPros =
config.getAnnotationProps();
- Collection<IdentifiedAnnotation> collection =
JCasUtil.select(jcas, IdentifiedAnnotation.class);
- Iterator<IdentifiedAnnotation> iterator =
collection.iterator();
- while (iterator.hasNext()) {
- IdentifiedAnnotation annotation =
iterator.next();
- StringBuilder annotationBuilder = new
StringBuilder();
-
annotationBuilder.append(annotation.getCoveredText());
- if (annotationPros != null) {
- for (CTAKESAnnotationProperty property
: annotationPros) {
-
annotationBuilder.append(config.getSeparatorChar());
-
annotationBuilder.append(CTAKESUtils.getAnnotationProperty(annotation,
property));
- }
- }
- metadata.add(CTAKES_META_PREFIX +
annotation.getType().getShortName(), annotationBuilder.toString());
- }
-
- if (config.isSerialize()) {
- // serialize data
-
CTAKESUtils.serialize(config.getSerializerType(), config.isPrettyPrint(),
config.getOutputStream());
- }
- } catch (Exception e) {
- e.printStackTrace();
- throw new SAXException(e.getMessage());
- }
- }
-
- /**
- * Returns metadata that includes cTAKES annotations.
- * @return {@Metadata} object that includes cTAKES annotations.
- */
- public Metadata getMetadata() {
- return metadata;
- }
+ // Prefix used for metadata including cTAKES annotations
+ public static String CTAKES_META_PREFIX = "ctakes:";
+
+ // Configuration object for CTAKESContentHandler
+ private CTAKESConfig config = null;
+
+ // StringBuilder object used to build the clinical free-text for cTAKES
+ private StringBuilder sb = null;
+
+ // Metadata object used for cTAKES annotations
+ private Metadata metadata = null;
+
+ /**
+ * Creates a new {@see CTAKESContentHandler} for the given {@see
ContentHandler} and Metadata objects.
+ * @param handler the {@see ContentHandler} object to be decorated.
+ * @param metadata the {@see Metadata} object that will be populated using
biomedical information extracted by cTAKES.
+ * @param config the {@see CTAKESConfig} object used to configure the
handler.
+ */
+ public CTAKESContentHandler(ContentHandler handler, Metadata metadata,
CTAKESConfig config) {
+ super(handler);
+ this.metadata = metadata;
+ this.config = config;
+ this.sb = new StringBuilder();
+ }
+
+ /**
+ * Creates a new {@see CTAKESContentHandler} for the given {@see
ContentHandler} and Metadata objects.
+ * @param handler the {@see ContentHandler} object to be decorated.
+ * @param metadata the {@see Metadata} object that will be populated using
biomedical information extracted by cTAKES.
+ */
+ public CTAKESContentHandler(ContentHandler handler, Metadata metadata) {
+ this(handler, metadata, new CTAKESConfig());
+ }
+
+ /**
+ * Default constructor.
+ */
+ public CTAKESContentHandler() {
+ this(new DefaultHandler(), new Metadata());
+ }
+
+ @Override
+ public void characters(char[] ch, int start, int length) throws
SAXException {
+ if (config.isText()) {
+ sb.append(ch, start, length);
+ }
+ super.characters(ch, start, length);
+ }
+
+ @Override
+ public void endDocument() throws SAXException {
+ try {
+ // create an Analysis Engine
+ AnalysisEngine ae =
CTAKESUtils.getAnalysisEngine(config.getAeDescriptorPath(),
config.getUMLSUser(), config.getUMLSPass());
+
+ // create a JCas, given an AE
+ JCas jcas = CTAKESUtils.getJCas(ae);
+
+ StringBuilder metaText = new StringBuilder();
+ for (String name : config.getMetadata()) {
+ for (String value : metadata.getValues(name)) {
+ metaText.append(value);
+ metaText.append(System.lineSeparator());
+ }
+ }
+
+ // analyze text
+ jcas.setDocumentText(metaText.toString() + sb.toString());
+ ae.process(jcas);
+
+ // add annotations to metadata
+ metadata.add(CTAKES_META_PREFIX + "schema",
config.getAnnotationPropsAsString());
+ CTAKESAnnotationProperty[] annotationPros =
config.getAnnotationProps();
+ Collection<IdentifiedAnnotation> collection =
JCasUtil.select(jcas, IdentifiedAnnotation.class);
+ Iterator<IdentifiedAnnotation> iterator = collection.iterator();
+ while (iterator.hasNext()) {
+ IdentifiedAnnotation annotation = iterator.next();
+ StringBuilder annotationBuilder = new StringBuilder();
+ annotationBuilder.append(annotation.getCoveredText());
+ if (annotationPros != null) {
+ for (CTAKESAnnotationProperty property : annotationPros) {
+ annotationBuilder.append(config.getSeparatorChar());
+
annotationBuilder.append(CTAKESUtils.getAnnotationProperty(annotation,
property));
+ }
+ }
+ metadata.add(CTAKES_META_PREFIX +
annotation.getType().getShortName(), annotationBuilder.toString());
+ }
+
+ if (config.isSerialize()) {
+ // serialize data
+ CTAKESUtils.serialize(config.getSerializerType(),
config.isPrettyPrint(), config.getOutputStream());
+ }
+ } catch (Exception e) {
+ e.printStackTrace();
+ throw new SAXException(e.getMessage());
+ }
+ }
+
+ /**
+ * Returns metadata that includes cTAKES annotations.
+ * @return {@Metadata} object that includes cTAKES annotations.
+ */
+ public Metadata getMetadata() {
+ return metadata;
+ }
}
Modified:
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESParser.java
URL:
http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESParser.java?rev=1684170&r1=1684169&r2=1684170&view=diff
==============================================================================
---
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESParser.java
(original)
+++
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESParser.java
Mon Jun 8 12:25:15 2015
@@ -33,26 +33,26 @@ import org.xml.sax.SAXException;
*
*/
public class CTAKESParser extends ParserDecorator {
- /**
- * Serial version UID
- */
- private static final long serialVersionUID = -2313482748027097961L;
+ /**
+ * Serial version UID
+ */
+ private static final long serialVersionUID = -2313482748027097961L;
- /**
- * Default constructor.
- */
- public CTAKESParser() {
- super(new AutoDetectParser());
- }
+ /**
+ * Default constructor.
+ */
+ public CTAKESParser() {
+ super(new AutoDetectParser());
+ }
- @Override
- public void parse(InputStream stream, ContentHandler handler,
- Metadata metadata, ParseContext context) throws
IOException,
- SAXException, TikaException {
- CTAKESConfig config = context.get(CTAKESConfig.class,
- new CTAKESConfig());
- CTAKESContentHandler ctakesHandler = new
CTAKESContentHandler(handler,
- metadata, config);
- super.parse(stream, ctakesHandler, metadata, context);
- }
+ @Override
+ public void parse(InputStream stream, ContentHandler handler,
+ Metadata metadata, ParseContext context) throws IOException,
+ SAXException, TikaException {
+ CTAKESConfig config = context.get(CTAKESConfig.class,
+ new CTAKESConfig());
+ CTAKESContentHandler ctakesHandler = new CTAKESContentHandler(handler,
+ metadata, config);
+ super.parse(stream, ctakesHandler, metadata, context);
+ }
}
Modified:
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESSerializer.java
URL:
http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESSerializer.java?rev=1684170&r1=1684169&r2=1684170&view=diff
==============================================================================
---
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESSerializer.java
(original)
+++
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESSerializer.java
Mon Jun 8 12:25:15 2015
@@ -24,20 +24,19 @@ import org.apache.uima.util.XmlCasSerial
* Enumeration for types of cTAKES (UIMA) CAS serializer supported by cTAKES.
*
* A CAS serializer writes a CAS in the given format.
- *
*/
public enum CTAKESSerializer {
- XCAS(XCASSerializer.class.getName()),
- XMI(XmiCasSerializer.class.getName()),
- XML(XmlCasSerializer.class.getName());
-
- private final String className;
-
- private CTAKESSerializer(String className) {
- this.className = className;
- }
-
- public String getClassName() {
- return className;
- }
+ XCAS(XCASSerializer.class.getName()),
+ XMI(XmiCasSerializer.class.getName()),
+ XML(XmlCasSerializer.class.getName());
+
+ private final String className;
+
+ private CTAKESSerializer(String className) {
+ this.className = className;
+ }
+
+ public String getClassName() {
+ return className;
+ }
}
Modified:
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESUtils.java
URL:
http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESUtils.java?rev=1684170&r1=1684169&r2=1684170&view=diff
==============================================================================
---
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESUtils.java
(original)
+++
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/ctakes/CTAKESUtils.java
Mon Jun 8 12:25:15 2015
@@ -45,198 +45,197 @@ import org.xml.sax.SAXException;
* UIMA</a> framework and <a href="https://opennlp.apache.org/";>OpenNLP</a>
* toolkit.
* </p>
- *
*/
public class CTAKESUtils {
- // UIMA Analysis Engine
- private static AnalysisEngine ae = null;
+ // UIMA Analysis Engine
+ private static AnalysisEngine ae = null;
- // JCas object for working with the CAS (Common Analysis System)
- private static JCas jcas = null;
+ // JCas object for working with the CAS (Common Analysis System)
+ private static JCas jcas = null;
- // UMLS username property
- private final static String CTAKES_UMLS_USER = "ctakes.umlsuser";
+ // UMLS username property
+ private final static String CTAKES_UMLS_USER = "ctakes.umlsuser";
- // UMLS password property
- private final static String CTAKES_UMLS_PASS = "ctakes.umlspw";
-
- /**
- * Returns a new UIMA Analysis Engine (AE). This method ensures that
only
- * one instance of an AE is created.
- *
- * <p>
- * An Analysis Engine is a component responsible for analyzing
unstructured
- * information, discovering and representing semantic content.
Unstructured
- * information includes, but is not restricted to, text documents.
- * </p>
- *
- * @param aeDescriptor
- * pathname for XML file including an
AnalysisEngineDescription
- * that contains all of the information needed to
instantiate and
- * use an AnalysisEngine.
- * @param umlsUser
- * UMLS username for NLM database
- * @param umlsPass
- * UMLS password for NLM database
- * @return an Analysis Engine for analyzing unstructured information.
- * @throws IOException
- * if any I/O error occurs.
- * @throws InvalidXMLException
- * if the input XML is not valid or does not specify a valid
- * ResourceSpecifier.
- * @throws ResourceInitializationException
- * if a failure occurred during production of the resource.
- * @throws URISyntaxException
- * if URL of the resource is not formatted strictly
according to
- * to RFC2396 and cannot be converted to a URI.
- */
- public static AnalysisEngine getAnalysisEngine(String aeDescriptor,
- String umlsUser, String umlsPass) throws IOException,
- InvalidXMLException, ResourceInitializationException,
- URISyntaxException {
- if (ae == null) {
- // UMLS user ID and password.
- String aeDescriptorPath = CTAKESUtils.class
-
.getResource(aeDescriptor).toURI().getPath();
-
- // get Resource Specifier from XML
- XMLInputSource aeIputSource = new
XMLInputSource(aeDescriptorPath);
- ResourceSpecifier aeSpecifier =
UIMAFramework.getXMLParser()
- .parseResourceSpecifier(aeIputSource);
-
- // UMLS user ID and password
- if ((umlsUser != null) && (!umlsUser.isEmpty())
- && (umlsPass != null) &&
(!umlsPass.isEmpty())) {
- /*
- * It is highly recommended that you change
UMLS credentials in
- * the XML configuration file instead of giving
user and
- * password using CTAKESConfig.
- */
- System.setProperty(CTAKES_UMLS_USER, umlsUser);
- System.setProperty(CTAKES_UMLS_PASS, umlsPass);
- }
-
- // create AE
- ae = UIMAFramework.produceAnalysisEngine(aeSpecifier);
- }
- return ae;
- }
-
- /**
- * Returns a new JCas () appropriate for the given Analysis Engine. This
- * method ensures that only one instance of a JCas is created. A Jcas
is a
- * Java Cover Classes based Object-oriented CAS (Common Analysis System)
- * API.
- *
- * <p>
- * Important: It is highly recommended that you reuse CAS objects rather
- * than creating new CAS objects prior to each analysis. This is
because CAS
- * objects may be expensive to create and may consume a significant
amount
- * of memory.
- * </p>
- *
- * @param ae
- * AnalysisEngine used to create an appropriate JCas object.
- * @return a JCas object appropriate for the given AnalysisEngine.
- * @throws ResourceInitializationException
- * if a CAS could not be created because this
AnalysisEngine's
- * CAS metadata (type system, type priorities, or FS
indexes)
- * are invalid.
- */
- public static JCas getJCas(AnalysisEngine ae)
- throws ResourceInitializationException {
- if (jcas == null) {
- jcas = ae.newJCas();
- }
- return jcas;
- }
-
- /**
- * Serializes a CAS in the given format.
- *
- * @param type
- * type of cTAKES (UIMA) serializer used to write CAS.
- * @param prettyPrint
- * {@code true} to do pretty printing of output.
- * @param stream
- * {@see OutputStream} object used to print out information
- * extracted by using cTAKES.
- * @throws SAXException
- * if there was a SAX exception.
- * @throws IOException
- * if any I/O error occurs.
- */
- public static void serialize(CTAKESSerializer type, boolean prettyPrint,
- OutputStream stream) throws SAXException, IOException {
- if (type == CTAKESSerializer.XCAS) {
- XCASSerializer.serialize(jcas.getCas(), stream,
prettyPrint);
- } else if (type == CTAKESSerializer.XMI) {
- XmiCasSerializer.serialize(jcas.getCas(),
jcas.getTypeSystem(),
- stream, prettyPrint, new
XmiSerializationSharedData());
- } else {
- XmlCasSerializer.serialize(jcas.getCas(),
jcas.getTypeSystem(),
- stream);
- }
- }
-
- /**
- * Returns the annotation value based on the given annotation type.
- * @param annotation {@see IdentifiedAnnotation} object.
- * @param property {@see CTAKESAnnotationProperty} enum used to
identify the annotation type.
- * @return the annotation value.
- */
- public static String getAnnotationProperty(IdentifiedAnnotation
annotation,
- CTAKESAnnotationProperty property) {
- String value = null;
- if (property == CTAKESAnnotationProperty.BEGIN) {
- value = Integer.toString(annotation.getBegin());
- } else if (property == CTAKESAnnotationProperty.END) {
- value = Integer.toString(annotation.getEnd());
- } else if (property == CTAKESAnnotationProperty.CONDITIONAL) {
- value = Boolean.toString(annotation.getConditional());
- } else if (property == CTAKESAnnotationProperty.CONFIDENCE) {
- value = Float.toString(annotation.getConfidence());
- } else if (property ==
CTAKESAnnotationProperty.DISCOVERY_TECNIQUE) {
- value =
Integer.toString(annotation.getDiscoveryTechnique());
- } else if (property == CTAKESAnnotationProperty.GENERIC) {
- value = Boolean.toString(annotation.getGeneric());
- } else if (property == CTAKESAnnotationProperty.HISTORY_OF) {
- value = Integer.toString(annotation.getHistoryOf());
- } else if (property == CTAKESAnnotationProperty.ID) {
- value = Integer.toString(annotation.getId());
- } else if (property ==
CTAKESAnnotationProperty.ONTOLOGY_CONCEPT_ARR) {
- FSArray mentions = annotation.getOntologyConceptArr();
- StringBuilder sb = new StringBuilder();
- if (mentions != null) {
- for (int i = 0; i < mentions.size(); i++) {
- if (mentions.get(i) instanceof
UmlsConcept) {
- UmlsConcept concept =
(UmlsConcept) mentions.get(i);
- sb.append(concept.getCui());
- if (i < mentions.size()-1) {
- sb.append(",");
- }
- }
- }
- }
- value = sb.toString();
- } else if (property == CTAKESAnnotationProperty.POLARITY) {
- value = Integer.toString(annotation.getPolarity());
- }
- return value;
- }
-
- /**
- * Resets cTAKES objects, if created. This method ensures that new
cTAKES
- * objects (a.k.a., Analysis Engine and JCas) will be created if
getters of
- * this class are called.
- */
- public static void reset() {
- // Analysis Engine
- ae.destroy();
- ae = null;
-
- // JCas
- jcas.reset();
- jcas = null;
- }
+ // UMLS password property
+ private final static String CTAKES_UMLS_PASS = "ctakes.umlspw";
+
+ /**
+ * Returns a new UIMA Analysis Engine (AE). This method ensures that only
+ * one instance of an AE is created.
+ *
+ * <p>
+ * An Analysis Engine is a component responsible for analyzing unstructured
+ * information, discovering and representing semantic content. Unstructured
+ * information includes, but is not restricted to, text documents.
+ * </p>
+ *
+ * @param aeDescriptor
+ * pathname for XML file including an AnalysisEngineDescription
+ * that contains all of the information needed to instantiate
and
+ * use an AnalysisEngine.
+ * @param umlsUser
+ * UMLS username for NLM database
+ * @param umlsPass
+ * UMLS password for NLM database
+ * @return an Analysis Engine for analyzing unstructured information.
+ * @throws IOException
+ * if any I/O error occurs.
+ * @throws InvalidXMLException
+ * if the input XML is not valid or does not specify a valid
+ * ResourceSpecifier.
+ * @throws ResourceInitializationException
+ * if a failure occurred during production of the resource.
+ * @throws URISyntaxException
+ * if URL of the resource is not formatted strictly according
to
+ * to RFC2396 and cannot be converted to a URI.
+ */
+ public static AnalysisEngine getAnalysisEngine(String aeDescriptor,
+ String umlsUser, String umlsPass) throws IOException,
+ InvalidXMLException, ResourceInitializationException,
+ URISyntaxException {
+ if (ae == null) {
+ // UMLS user ID and password.
+ String aeDescriptorPath = CTAKESUtils.class
+ .getResource(aeDescriptor).toURI().getPath();
+
+ // get Resource Specifier from XML
+ XMLInputSource aeIputSource = new XMLInputSource(aeDescriptorPath);
+ ResourceSpecifier aeSpecifier = UIMAFramework.getXMLParser()
+ .parseResourceSpecifier(aeIputSource);
+
+ // UMLS user ID and password
+ if ((umlsUser != null) && (!umlsUser.isEmpty())
+ && (umlsPass != null) && (!umlsPass.isEmpty())) {
+ /*
+ * It is highly recommended that you change UMLS credentials in
+ * the XML configuration file instead of giving user and
+ * password using CTAKESConfig.
+ */
+ System.setProperty(CTAKES_UMLS_USER, umlsUser);
+ System.setProperty(CTAKES_UMLS_PASS, umlsPass);
+ }
+
+ // create AE
+ ae = UIMAFramework.produceAnalysisEngine(aeSpecifier);
+ }
+ return ae;
+ }
+
+ /**
+ * Returns a new JCas () appropriate for the given Analysis Engine. This
+ * method ensures that only one instance of a JCas is created. A Jcas is a
+ * Java Cover Classes based Object-oriented CAS (Common Analysis System)
+ * API.
+ *
+ * <p>
+ * Important: It is highly recommended that you reuse CAS objects rather
+ * than creating new CAS objects prior to each analysis. This is because
CAS
+ * objects may be expensive to create and may consume a significant amount
+ * of memory.
+ * </p>
+ *
+ * @param ae
+ * AnalysisEngine used to create an appropriate JCas object.
+ * @return a JCas object appropriate for the given AnalysisEngine.
+ * @throws ResourceInitializationException
+ * if a CAS could not be created because this AnalysisEngine's
+ * CAS metadata (type system, type priorities, or FS indexes)
+ * are invalid.
+ */
+ public static JCas getJCas(AnalysisEngine ae)
+ throws ResourceInitializationException {
+ if (jcas == null) {
+ jcas = ae.newJCas();
+ }
+ return jcas;
+ }
+
+ /**
+ * Serializes a CAS in the given format.
+ *
+ * @param type
+ * type of cTAKES (UIMA) serializer used to write CAS.
+ * @param prettyPrint
+ * {@code true} to do pretty printing of output.
+ * @param stream
+ * {@see OutputStream} object used to print out information
+ * extracted by using cTAKES.
+ * @throws SAXException
+ * if there was a SAX exception.
+ * @throws IOException
+ * if any I/O error occurs.
+ */
+ public static void serialize(CTAKESSerializer type, boolean prettyPrint,
+ OutputStream stream) throws SAXException, IOException {
+ if (type == CTAKESSerializer.XCAS) {
+ XCASSerializer.serialize(jcas.getCas(), stream, prettyPrint);
+ } else if (type == CTAKESSerializer.XMI) {
+ XmiCasSerializer.serialize(jcas.getCas(), jcas.getTypeSystem(),
+ stream, prettyPrint, new XmiSerializationSharedData());
+ } else {
+ XmlCasSerializer.serialize(jcas.getCas(), jcas.getTypeSystem(),
+ stream);
+ }
+ }
+
+ /**
+ * Returns the annotation value based on the given annotation type.
+ * @param annotation {@see IdentifiedAnnotation} object.
+ * @param property {@see CTAKESAnnotationProperty} enum used to identify
the annotation type.
+ * @return the annotation value.
+ */
+ public static String getAnnotationProperty(IdentifiedAnnotation annotation,
+ CTAKESAnnotationProperty property) {
+ String value = null;
+ if (property == CTAKESAnnotationProperty.BEGIN) {
+ value = Integer.toString(annotation.getBegin());
+ } else if (property == CTAKESAnnotationProperty.END) {
+ value = Integer.toString(annotation.getEnd());
+ } else if (property == CTAKESAnnotationProperty.CONDITIONAL) {
+ value = Boolean.toString(annotation.getConditional());
+ } else if (property == CTAKESAnnotationProperty.CONFIDENCE) {
+ value = Float.toString(annotation.getConfidence());
+ } else if (property == CTAKESAnnotationProperty.DISCOVERY_TECNIQUE) {
+ value = Integer.toString(annotation.getDiscoveryTechnique());
+ } else if (property == CTAKESAnnotationProperty.GENERIC) {
+ value = Boolean.toString(annotation.getGeneric());
+ } else if (property == CTAKESAnnotationProperty.HISTORY_OF) {
+ value = Integer.toString(annotation.getHistoryOf());
+ } else if (property == CTAKESAnnotationProperty.ID) {
+ value = Integer.toString(annotation.getId());
+ } else if (property == CTAKESAnnotationProperty.ONTOLOGY_CONCEPT_ARR) {
+ FSArray mentions = annotation.getOntologyConceptArr();
+ StringBuilder sb = new StringBuilder();
+ if (mentions != null) {
+ for (int i = 0; i < mentions.size(); i++) {
+ if (mentions.get(i) instanceof UmlsConcept) {
+ UmlsConcept concept = (UmlsConcept) mentions.get(i);
+ sb.append(concept.getCui());
+ if (i < mentions.size()-1) {
+ sb.append(",");
+ }
+ }
+ }
+ }
+ value = sb.toString();
+ } else if (property == CTAKESAnnotationProperty.POLARITY) {
+ value = Integer.toString(annotation.getPolarity());
+ }
+ return value;
+ }
+
+ /**
+ * Resets cTAKES objects, if created. This method ensures that new cTAKES
+ * objects (a.k.a., Analysis Engine and JCas) will be created if getters of
+ * this class are called.
+ */
+ public static void reset() {
+ // Analysis Engine
+ ae.destroy();
+ ae = null;
+
+ // JCas
+ jcas.reset();
+ jcas = null;
+ }
}
