| 1 | |
|
| 2 | |
|
| 3 | |
package org.galagosearch.core.parse; |
| 4 | |
|
| 5 | |
import java.io.IOException; |
| 6 | |
import org.galagosearch.core.types.AdditionalDocumentText; |
| 7 | |
import org.galagosearch.core.types.ExtractedLink; |
| 8 | |
import org.galagosearch.tupleflow.InputClass; |
| 9 | |
import org.galagosearch.tupleflow.OutputClass; |
| 10 | |
import org.galagosearch.tupleflow.StandardStep; |
| 11 | |
import org.galagosearch.tupleflow.execution.Verified; |
| 12 | |
|
| 13 | |
|
| 14 | |
|
| 15 | |
|
| 16 | |
|
| 17 | |
@Verified |
| 18 | |
@InputClass(className = "org.galagosearch.core.parse.DocumentLinkData") |
| 19 | |
@OutputClass(className = "org.galagosearch.core.types.AdditionalDocumentText") |
| 20 | 0 | public class AnchorTextCreator extends StandardStep<DocumentLinkData, AdditionalDocumentText> { |
| 21 | |
@Override |
| 22 | |
public void process(DocumentLinkData object) throws IOException { |
| 23 | 0 | AdditionalDocumentText additional = new AdditionalDocumentText(); |
| 24 | 0 | StringBuilder extraText = new StringBuilder(); |
| 25 | |
|
| 26 | 0 | additional.identifier = object.identifier; |
| 27 | 0 | for (ExtractedLink link : object.links) { |
| 28 | 0 | extraText.append("<anchor>"); |
| 29 | 0 | extraText.append(link.anchorText); |
| 30 | 0 | extraText.append("</anchor>"); |
| 31 | |
} |
| 32 | 0 | additional.text = extraText.toString(); |
| 33 | |
|
| 34 | 0 | processor.process(additional); |
| 35 | 0 | } |
| 36 | |
} |