Example usage for org.apache.lucene.document Document getField

List of usage examples for org.apache.lucene.document Document getField

  1. HOME
  2. Java
  3. org.apache.lucene
  4. org.apache.lucene.document.*
  5. Document
  6. getField

Introduction

In this page you can find the example usage for org.apache.lucene.document Document getField.

Prototype

public final IndexableField getField(String name)

Source Link

Document

Returns a field with the given name if any exist in this document, or null.

Usage

From source file:application.SentimentLexiconGenerator.java

License:Open Source License

public RatingHistogram makeRatingHistogramForSynset(Synset synset, ReviewStats stats) throws IOException {
    RatingHistogram histogram = new RatingHistogram(stats);
    TermSpans ts = getIndexTermInstances(synset.toString());

    while (ts.next()) {
        histogram.incrementUniqueDocuments();

        Document doc = searcher.doc(ts.doc());
        int doc_rating = Integer.parseInt(doc.getField(Globals.IndexFieldNames.rating).stringValue());

        Iterator<byte[]> p_i = ts.getPayload().iterator();
        while (p_i.hasNext()) {
            ReviewTermPayload p = new ReviewTermPayload();
            p.decode(p_i.next());/*www .j a  v a  2s  .co  m*/
            if (filter.filterPayload(p)) {
                histogram.addObeservation(doc_rating, 1);
            }
        }
    }

    return histogram;
}

From source file:application.SentimentLexiconGenerator.java

License:Open Source License

public HashMap<SynsetCategory, RatingHistogram> makeRatingHistogramsForTerm(String term, ReviewStats stats)
        throws IOException {
    HashMap<SynsetCategory, RatingHistogram> histograms = new HashMap<SynsetCategory, RatingHistogram>();
    for (SynsetCategory synsetcat : Synset.getSynsetCategories()) {
        histograms.put(synsetcat, new RatingHistogram(stats));
    }/*  w ww .j  av  a  2 s .  co m*/

    TermSpans ts = getIndexTermInstances(term);

    while (ts.next()) {
        Document doc = searcher.doc(ts.doc());
        int doc_rating = Integer.parseInt(doc.getField(Globals.IndexFieldNames.rating).stringValue());

        Iterator<byte[]> p_i = ts.getPayload().iterator();
        while (p_i.hasNext()) {
            ReviewTermPayload p = new ReviewTermPayload();
            p.decode(p_i.next());
            if (filter.filterPayload(p)) {
                RatingHistogram histogram = histograms.get(Synset.convertPosCategory(p.getPosCat()));
                if (histogram != null) {
                    histogram.incrementUniqueDocuments();
                    histogram.addObeservation(doc_rating, 1);
                } else {
                    // TODO Remove message??
                    AppLogger.error.log(Level.SEVERE, "Histogram not found");
                }
            }
        }
    }

    return histograms;
}

From source file:au.org.ala.names.search.ALANameSearcher.java

License:Open Source License

/**
 * Dumps a list of the species LSID's that are contained in the index.
 *///w  ww  .  ja  va  2  s  .  c om
public void dumpSpecies() {
    try {
        OutputStreamWriter fileOut = new OutputStreamWriter(new FileOutputStream("/data/species.txt"), "UTF-8");
        Term term = new Term("rank", "species");
        TopDocs hits = cbSearcher.search(new TermQuery(term), 2000000);

        for (ScoreDoc sdoc : hits.scoreDocs) {
            Document doc = cbReader.document(sdoc.doc);

            if (doc.getField("synonym") == null) {
                String lsid = StringUtils.trimToNull(doc.getField("lsid").stringValue());
                if (lsid == null)
                    lsid = doc.getField("id").stringValue();
                fileOut.write(lsid + "\n");
            }

        }
        fileOut.flush();

    } catch (Exception e) {
        e.printStackTrace();
    }

}

From source file:axiom.objectmodel.dom.convert.LuceneVersion9Convertor.java

License:Open Source License

public Document convertDocument(Document doc) {
    Field protoField = doc.getField(LuceneManager.PROTOTYPE);
    if (protoField != null && ("CMSTask".equalsIgnoreCase(protoField.stringValue())
            || "CMSTaskContainer".equalsIgnoreCase(protoField.stringValue()))) {
        return null;
    }//from w  w w .j a  va  2  s  .c o m

    Document ndoc = new Document();
    Enumeration e = doc.fields();

    String id = null, layer = null;
    Field idField = doc.getField(LuceneManager.ID);
    Field layerField = doc.getField(LuceneManager.LAYER_OF_SAVE);
    if (idField != null && layerField != null) {
        id = idField.stringValue();
        layer = layerField.stringValue();
    }

    while (e.hasMoreElements()) {
        Field f = (Field) e.nextElement();
        Field.Store currstore = Field.Store.YES;
        if (!f.isStored()) {
            currstore = Field.Store.NO;
        } else if (f.isCompressed()) {
            currstore = Field.Store.COMPRESS;
        }
        Field.Index curridx = Field.Index.UN_TOKENIZED;
        if (!f.isIndexed()) {
            curridx = Field.Index.NO;
        } else if (f.isTokenized()) {
            curridx = Field.Index.TOKENIZED;
        }

        String name = f.name();
        String value = f.stringValue();

        if (!("84".equals(id) && "1".equals(layer) && "_task".equals(name))
                && !("71".equals(id) && "1".equals(layer) && "_task".equals(name))) {

            ndoc.add(new Field(name, value, currstore, curridx));
        }
    }

    return ndoc;
}

From source file:axiom.objectmodel.dom.LuceneManager.java

License:Open Source License

private static String getFieldValue(String fieldname, Document doc, HashSet unchecked) {
    String value = null;//  ww  w.  j a va2 s .  c o m
    Field field = doc.getField(fieldname);
    if (field != null) {
        value = field.stringValue();
        unchecked.remove(field);
    }
    return value;
}

From source file:axiom.objectmodel.dom.LuceneManager.java

License:Open Source License

private void getTheChildren(Node parent, String relChildren, NodeManager nmgr, int mode, int layerInStorage)
        throws Exception {

    IndexSearcher searcher = null;/*w  w w. j ava 2s .  co  m*/
    BooleanQuery bq = null;

    try {
        searcher = this.getIndexSearcher();
        String pid = parent.getID();

        HashMap ids = new HashMap();
        int length;
        Query query1 = new TermQuery(new Term(ISCHILD, "true"));
        Query query2 = new TermQuery(new Term(PARENTID, pid));
        for (int i = LIVE_MODE; i <= mode; i++) {
            bq = new BooleanQuery();
            bq.add(query1, BooleanClause.Occur.MUST);
            bq.add(query2, BooleanClause.Occur.MUST);
            bq.add(new TermQuery(new Term(LAYER_OF_SAVE, i + "")), BooleanClause.Occur.MUST);
            Hits hits = searcher.search(bq);
            length = hits.length();

            /*if (app.debug()) 
               app.logEvent("LuceneManager.getTheChildren(), parent = " + parent.getKey() 
             + ", layer = " + mode + ", layerInStorage = " + layerInStorage 
             + ", executed query " + bq + " which produced " + length 
             + " results");*/

            for (int j = 0; j < length; j++) {
                Document doc = hits.doc(j);
                ids.put(doc.getField(ID).stringValue(), doc.getField(PROTOTYPE).stringValue());
            }
        }

        Collection<NodeHandle> subnodes = null;
        length = ids.size();
        if (length > 0) {
            subnodes = parent.createSubnodeList();
            Iterator iter = ids.keySet().iterator();
            while (iter.hasNext()) {
                String id = (String) iter.next();
                NodeHandle handle = makeNodeHandle(nmgr, id, (String) ids.get(id), mode);
                if (subnodes instanceof SubnodeList) {
                    ((SubnodeList) subnodes).addSorted(handle);
                } else {
                    subnodes.add(handle);
                }
            }
        }
        ids.clear();
        ids = null;

        if (relChildren != null) {
            String[] charr = relChildren.split(NULL_DELIM);
            if (charr.length > 0 && charr.length % 2 == 0) {
                if (subnodes == null) {
                    subnodes = parent.createSubnodeList();
                }
                for (int i = 0; i < charr.length; i += 2) {
                    if (subnodes instanceof SubnodeList) {
                        ((SubnodeList) subnodes).addSorted(makeNodeHandle(nmgr, charr[i], charr[i + 1], mode));
                    } else {
                        subnodes.add(makeNodeHandle(nmgr, charr[i], charr[i + 1], mode));
                    }
                }
            }
        }
    } catch (IOException ioe) {
        throw new Exception("Searcher failed when attempting to retrieve children of " + "id '" + parent.getID()
                + "', query = " + bq);
    } finally {
        this.releaseIndexSearcher(searcher);
    }
}

From source file:axiom.objectmodel.dom.LuceneManager.java

License:Open Source License

private void createDocumentFromNode(ArrayList ids, ArrayList docs, INode node, HashMap analyzerMap)
        throws Exception {
    Document doc = this.createDocument(node, analyzerMap);
    if (doc != null) {
        String key = doc.getField(ID).stringValue() + DeletedInfos.KEY_SEPERATOR
                + doc.getField(LAYER_OF_SAVE).stringValue();
        ids.add(key);/*  w w  w .  ja  v  a2s  .  c om*/
        docs.add(doc);
    }
}

From source file:axiom.objectmodel.dom.LuceneManager.java

License:Open Source License

private void addToDoc(Document doc, String key, IProperty prop, ResourceProperties rprops, ArrayList ref_list,
        HashMap analyzerMap, String accessprop) throws Exception {
    if (prop.getValue() == null) {
        return;/*from   ww w. jav  a 2 s  .  com*/
    }

    final Field.Store store;
    final Field.Index index;
    if (accessprop != null && (key.equals(accessprop)
            || (this.app.isPropertyFilesIgnoreCase() && key.equalsIgnoreCase(accessprop)))) {
        store = Field.Store.YES;
        index = Field.Index.UN_TOKENIZED;
    } else if (key.startsWith("_") && !key.equals(FileObject.CONTENT)) {
        store = Field.Store.YES;
        index = Field.Index.UN_TOKENIZED;
    } else {
        store = getStore(rprops, key);
        index = getIndex(rprops, key);
    }

    int type = getType(rprops, key);
    if (type < 0) {
        type = prop.getType();
    }

    final Analyzer analyzer = (type == IProperty.STRING) ? getAnalyzer(rprops, key) : null;
    final float boost = getBoost(rprops, key);
    Field f;

    switch (type) {

    case IProperty.BOOLEAN:
        f = new Field(key, serializeBoolean(prop.getBooleanValue()), store, index);
        if (boost > -1f) {
            f.setBoost(boost);
        }
        doc.add(f);
        break;

    case IProperty.DATE:
        f = new Field(key, serializeDate(prop.getDateValue()), store, index);
        if (boost > -1f) {
            f.setBoost(boost);
        }
        doc.add(f);
        break;

    case IProperty.TIME:
        f = new Field(key, serializeTime(prop.getDateValue()), store, index);
        if (boost > -1f) {
            f.setBoost(boost);
        }
        doc.add(f);
        break;

    case IProperty.TIMESTAMP:
        f = new Field(key, serializeTimestamp(prop.getDateValue()), store, index);
        if (boost > -1f) {
            f.setBoost(boost);
        }
        doc.add(f);
        break;

    case IProperty.FLOAT:
        f = new Field(key, serializeFloat(prop.getFloatValue()), store, index);
        if (boost > -1f) {
            f.setBoost(boost);
        }
        doc.add(f);
        break;

    case IProperty.SMALLFLOAT:
        f = new Field(key, serializeSmallFloat(prop.getFloatValue()), store, index);
        if (boost > -1f) {
            f.setBoost(boost);
        }
        doc.add(f);
        break;

    case IProperty.INTEGER:
        f = new Field(key, serializeInt(prop.getIntegerValue()), store, index);
        if (boost > -1f) {
            f.setBoost(boost);
        }
        doc.add(f);
        break;

    case IProperty.SMALLINT:
        f = new Field(key, serializeSmallInt(prop.getIntegerValue()), store, index);
        if (boost > -1f) {
            f.setBoost(boost);
        }
        doc.add(f);
        break;

    case IProperty.STRING:
        f = new Field(key, prop.getStringValue(), store, index);
        if (boost > -1f) {
            f.setBoost(boost);
        }
        doc.add(f);
        if (analyzer != null) {
            String docid = doc.getField(ID).stringValue() + DeletedInfos.KEY_SEPERATOR
                    + doc.getField(LAYER_OF_SAVE).stringValue();
            PerFieldAnalyzerWrapper ret = (PerFieldAnalyzerWrapper) analyzerMap.get(docid);
            if (ret == null) {
                ret = buildAnalyzer();
                analyzerMap.put(docid, ret);
            }
            ret.addAnalyzer(key, analyzer);
        }
        break;

    case IProperty.NODE:
        INode propNode = prop.getNodeValue();
        if (propNode != null) {
            String value = serializeNodeProp(propNode.getID(), propNode.getPrototype());
            f = new Field(key, value, store, index);
            if (boost > -1f) {
                f.setBoost(boost);
            }
            doc.add(f);
        }
        break;

    case IProperty.REFERENCE:
        if (prop instanceof axiom.objectmodel.db.Property) {
            axiom.objectmodel.db.Property p = (axiom.objectmodel.db.Property) prop;
            Reference relobj = p.getReferenceValue();
            if (relobj != null) {
                String serialized_ref = serializeReference(relobj);
                f = new Field(key, serialized_ref, store, index);
                if (boost > -1f) {
                    f.setBoost(boost);
                }
                doc.add(f);

                StringBuffer sb = new StringBuffer(serialized_ref);
                sb.append(NULL_DELIM).append(key);
                ref_list.add(sb.toString());
            }
        }
        break;

    case IProperty.MULTI_VALUE:
        if (prop instanceof axiom.objectmodel.db.Property) {
            axiom.objectmodel.db.Property p = (axiom.objectmodel.db.Property) prop;
            MultiValue mvobj = p.getMultiValue();
            if (mvobj != null) {
                addMultiValueToDoc(doc, key, mvobj, store, index, ref_list);
            }
        }
        break;

    case IProperty.XML:
        if (prop instanceof axiom.objectmodel.db.Property) {
            axiom.objectmodel.db.Property p = (axiom.objectmodel.db.Property) prop;
            Object xml = p.getXMLValue();
            if (xml != null) {
                f = new Field(key, XmlUtils.objectToXMLString(xml), store, index);
                if (boost > -1f) {
                    f.setBoost(boost);
                }
                doc.add(f);
            }
        }
    case IProperty.XHTML:
        if (prop instanceof axiom.objectmodel.db.Property) {
            axiom.objectmodel.db.Property p = (axiom.objectmodel.db.Property) prop;
            Object xml = p.getXHTMLValue();
            if (xml != null) {
                f = new Field(key, XmlUtils.objectToXMLString(xml), store, index);
                if (boost > -1f) {
                    f.setBoost(boost);
                }
                doc.add(f);
                addXhtmlRefs(key, xml, ref_list);
            }
        }
    default:
        break;
    }
}

From source file:axiom.objectmodel.dom.LuceneManager.java

License:Open Source License

public Key[] getSourceNodeIds(final String id, final int mode, ArrayList protos, BooleanQuery append, Sort sort)
        throws Exception {
    IndexSearcher searcher = null;/*from   ww  w.  j a v a2s . c  o  m*/
    Hits hits = null;
    Key[] keys = null;
    BooleanQuery query = null;

    try {
        searcher = this.getIndexSearcher();
        query = new BooleanQuery();
        final int sizeOfProtos;
        if ((sizeOfProtos = protos.size()) > 0) {
            BooleanQuery proto_query = new BooleanQuery();
            for (int i = 0; i < sizeOfProtos; i++) {
                proto_query.add(new TermQuery(new Term(PROTOTYPE, (String) protos.get(i))),
                        BooleanClause.Occur.SHOULD);
            }
            query.add(proto_query, BooleanClause.Occur.MUST);
        }

        query.add(new TermQuery(new Term(REF_LIST_FIELD, id)), BooleanClause.Occur.MUST);

        if (append != null && append.getClauses().length > 0) {
            query.add(append, BooleanClause.Occur.MUST);
        }

        hits = searcher.search(query, sort);

        /*if (app.debug())
           app.logEvent("LuceneManager.getSourceNodeIds(): id=" + id + ",layer=" + mode
          + " executed query [" + query + " which resulted in " 
          + hits.length() + " hits");*/

        int size = hits.length();
        ArrayList<Key> list = new ArrayList<Key>();
        for (int i = 0; i < size; i++) {
            Document doc = hits.doc(i);

            if (!isIdInDocumentRefs(doc, id)) {
                continue;
            }

            Field id_field = doc.getField(ID);
            Field proto_field = doc.getField(PROTOTYPE);
            Field layer_field = doc.getField(LAYER_OF_SAVE);
            if (layer_field != null) {
                try {
                    if (mode < Integer.parseInt(layer_field.stringValue())) {
                        continue;
                    }
                } catch (Exception nfe) {
                }
            }
            if (id_field != null && proto_field != null) {
                list.add(new DbKey(this.app.getDbMapping(proto_field.stringValue()), id_field.stringValue(),
                        mode));
            }
        }

        keys = new Key[list.size()];
        list.toArray(keys);
    } catch (Exception ex) {
        app.logError(
                ErrorReporter.errorMsg(this.getClass(), "getSourceNodeIds") + "Could not retrieve document "
                        + id + " from Lucene index with query = " + (query != null ? query : "null"),
                ex);
        throw ex;
    } finally {
        this.releaseIndexSearcher(searcher);
    }

    return keys;
}

From source file:axiom.objectmodel.dom.LuceneManager.java

License:Open Source License

public ArrayList getChildrenIds(INode node) throws Exception {
    ArrayList childrenIds = new ArrayList();
    IndexSearcher searcher = null;//from w ww.  j  av  a2s.  co  m
    BooleanQuery bq = new BooleanQuery();

    try {
        searcher = this.getIndexSearcher();
        String id = node.getID();
        final Query query1 = new TermQuery(new Term(PARENTID, isSpecialNode(id) ? id : node.getID()));
        final Query query2 = new TermQuery(new Term(ISCHILD, "true"));
        bq.add(query1, BooleanClause.Occur.MUST);
        bq.add(query2, BooleanClause.Occur.MUST);

        final Hits hits = searcher.search(bq);

        /*if (app.debug())
           app.logEvent("LuceneManager.getChildrenIds() executed query [" + bq 
             + " which resulted in " + hits.length() + " hits");*/

        final int length = hits.length();
        for (int i = 0; i < length; i++) {
            Document doc = hits.doc(i);
            childrenIds.add(doc.getField(ID).stringValue());
        }
    } catch (IOException ioe) {
        throw new Exception("Searcher failed when attempting to retrieve children of " + "id '" + node.getID()
                + "', query = " + bq);
    } finally {
        this.releaseIndexSearcher(searcher);
    }

    return childrenIds;
}