| 1 | package org.greenstone.atlas.server;
|
|---|
| 2 |
|
|---|
| 3 | import gate.Annotation;
|
|---|
| 4 | import gate.AnnotationSet;
|
|---|
| 5 | import gate.Corpus;
|
|---|
| 6 | import gate.Document;
|
|---|
| 7 | import gate.Factory;
|
|---|
| 8 | import gate.FeatureMap;
|
|---|
| 9 | import gate.Gate;
|
|---|
| 10 | import gate.GateConstants;
|
|---|
| 11 | import gate.creole.ANNIEConstants;
|
|---|
| 12 | import gate.creole.SerialAnalyserController;
|
|---|
| 13 | import gate.util.GateException;
|
|---|
| 14 | import gate.util.Out;
|
|---|
| 15 | import gate.util.persistence.PersistenceManager;
|
|---|
| 16 |
|
|---|
| 17 | import java.io.File;
|
|---|
| 18 | import java.net.URL;
|
|---|
| 19 | import java.util.ArrayList;
|
|---|
| 20 | import java.util.HashMap;
|
|---|
| 21 |
|
|---|
| 22 | import org.apache.log4j.BasicConfigurator;
|
|---|
| 23 |
|
|---|
| 24 | public class GateScanner
|
|---|
| 25 | {
|
|---|
| 26 | private SerialAnalyserController annieController;
|
|---|
| 27 |
|
|---|
| 28 | public GateScanner()
|
|---|
| 29 | {
|
|---|
| 30 | try
|
|---|
| 31 | {
|
|---|
| 32 | BasicConfigurator.configure();
|
|---|
| 33 | Gate.setGateHome(new File("C:\\Users\\sjm84\\Desktop\\Gate"));
|
|---|
| 34 | Gate.setPluginsHome(new File("C:\\Users\\sjm84\\Desktop\\Gate\\plugins"));
|
|---|
| 35 | Gate.setUserConfigFile(new File("C:\\Users\\sjm84\\Desktop\\Gate\\gate.xml"));
|
|---|
| 36 | Gate.init();
|
|---|
| 37 | Gate.getCreoleRegister().registerDirectories(new File("C:\\Users\\sjm84\\Desktop\\Gate\\plugins\\ANNIE").toURI().toURL());
|
|---|
| 38 | Gate.getUserConfig().put(GateConstants.DOCUMENT_ADD_SPACE_ON_UNPACK_FEATURE_NAME, new Boolean(false));
|
|---|
| 39 |
|
|---|
| 40 | Out.prln("Initialising ANNIE...");
|
|---|
| 41 |
|
|---|
| 42 | annieController = (SerialAnalyserController) PersistenceManager.loadObjectFromFile(new File(new File(Gate.getPluginsHome(), ANNIEConstants.PLUGIN_DIR), ANNIEConstants.DEFAULT_FILE));
|
|---|
| 43 |
|
|---|
| 44 | Out.prln("...ANNIE loaded");
|
|---|
| 45 | }
|
|---|
| 46 | catch (Exception e)
|
|---|
| 47 | {
|
|---|
| 48 | e.printStackTrace();
|
|---|
| 49 | }
|
|---|
| 50 | }
|
|---|
| 51 |
|
|---|
| 52 | /** Tell ANNIE's controller about the corpus you want to run on */
|
|---|
| 53 | public void setCorpus(Corpus corpus)
|
|---|
| 54 | {
|
|---|
| 55 | annieController.setCorpus(corpus);
|
|---|
| 56 | } // setCorpus
|
|---|
| 57 |
|
|---|
| 58 | /** Run ANNIE */
|
|---|
| 59 | public void execute() throws GateException
|
|---|
| 60 | {
|
|---|
| 61 | Out.prln("Running ANNIE...");
|
|---|
| 62 | annieController.execute();
|
|---|
| 63 | Out.prln("...ANNIE complete");
|
|---|
| 64 | } // execute()
|
|---|
| 65 |
|
|---|
| 66 | public HashMap<String, Word> classifyText(String originalText)
|
|---|
| 67 | {
|
|---|
| 68 | HashMap<String, Word> words = new HashMap<String, Word>();
|
|---|
| 69 | try
|
|---|
| 70 | {
|
|---|
| 71 | Corpus corpus = (Corpus) Factory.createResource("gate.corpora.CorpusImpl");
|
|---|
| 72 | FeatureMap params = Factory.newFeatureMap();
|
|---|
| 73 | params.put("stringContent", "<html><head></head><body>" + originalText + "</body></html>");
|
|---|
| 74 | params.put("preserveOriginalContent", new Boolean(true));
|
|---|
| 75 | params.put("collectRepositioningInfo", new Boolean(true));
|
|---|
| 76 | Document doc = (Document) Factory.createResource("gate.corpora.DocumentImpl", params);
|
|---|
| 77 | corpus.add(doc);
|
|---|
| 78 |
|
|---|
| 79 | // tell the pipeline about the corpus and run it
|
|---|
| 80 | this.setCorpus(corpus);
|
|---|
| 81 | this.execute();
|
|---|
| 82 |
|
|---|
| 83 | AnnotationSet annotations = doc.getAnnotations();
|
|---|
| 84 |
|
|---|
| 85 | for(int j = 0; j < annotations.size(); j++)
|
|---|
| 86 | {
|
|---|
| 87 | Annotation currentAnnotation = annotations.get(j);
|
|---|
| 88 |
|
|---|
| 89 | if(currentAnnotation == null)
|
|---|
| 90 | {
|
|---|
| 91 | continue;
|
|---|
| 92 | }
|
|---|
| 93 | if(currentAnnotation.getType().equals("Token"))
|
|---|
| 94 | {
|
|---|
| 95 | Word w = null;
|
|---|
| 96 | if(words.containsKey((String)currentAnnotation.getFeatures().get("string")))
|
|---|
| 97 | {
|
|---|
| 98 | w = words.get((String)currentAnnotation.getFeatures().get("string"));
|
|---|
| 99 | w.addClassification((String)currentAnnotation.getFeatures().get("category"));
|
|---|
| 100 | w.setValue((String)currentAnnotation.getFeatures().get("string"));
|
|---|
| 101 | }
|
|---|
| 102 | else
|
|---|
| 103 | {
|
|---|
| 104 | w = new Word();
|
|---|
| 105 | w.addClassification((String)currentAnnotation.getFeatures().get("category"));
|
|---|
| 106 | w.setValue((String)currentAnnotation.getFeatures().get("string"));
|
|---|
| 107 | words.put(w.getValue(), w);
|
|---|
| 108 | }
|
|---|
| 109 | }
|
|---|
| 110 | }
|
|---|
| 111 | }
|
|---|
| 112 | catch(Exception ex)
|
|---|
| 113 | {
|
|---|
| 114 | ex.printStackTrace();
|
|---|
| 115 | }
|
|---|
| 116 |
|
|---|
| 117 | return words;
|
|---|
| 118 | }
|
|---|
| 119 | } |
|---|
