Author: rfrovarp
Date: Mon Dec 5 17:51:07 2011
New Revision: 1210543
URL: http://svn.apache.org/viewvc?rev=1210543&view=rev
Log:
Document the interface.
Modified:
incubator/droids/trunk/droids-tika/src/main/java/org/apache/droids/tika/api/TikaParse.java
Modified:
incubator/droids/trunk/droids-tika/src/main/java/org/apache/droids/tika/api/TikaParse.java
URL:
http://svn.apache.org/viewvc/incubator/droids/trunk/droids-tika/src/main/java/org/apache/droids/tika/api/TikaParse.java?rev=1210543&r1=1210542&r2=1210543&view=diff
==============================================================================
---
incubator/droids/trunk/droids-tika/src/main/java/org/apache/droids/tika/api/TikaParse.java
(original)
+++
incubator/droids/trunk/droids-tika/src/main/java/org/apache/droids/tika/api/TikaParse.java
Mon Dec 5 17:51:07 2011
@@ -5,13 +5,36 @@ import org.apache.tika.metadata.Metadata
public interface TikaParse extends Parse {
+ /**
+ * Retrieves the main content of the parsed document.
+ * Uses Tika's plugin in for Boilerpipe.
+ * @return plain text result with boilerplate removed
+ */
public String getMainContent();
+ /**
+ * Extracted meta data from the document. This can include
+ * meta tags from within an HTML document
+ * @return metadata object from the parse
+ */
public Metadata getMetadata();
+ /**
+ * The HTML representation of the document.
+ * @return The HTML representation of the document.
+ */
public String getXml();
+ /**
+ * Plain text representation of the document.
+ * @return plain text version without formatting
+ */
public String getPlainText();
+ /**
+ * If the document should be indexed or not.
+ * This can be determined from metadata or other methods
+ * @return false if the document shouldn't be indexed, true otherwise
+ */
public boolean isIndexed();
}
