Example usage for org.apache.lucene.index IndexWriter IndexWriter

List of usage examples for org.apache.lucene.index IndexWriter IndexWriter

  1. HOME
  2. Java
  3. org.apache.lucene
  4. org.apache.lucene.index.*
  5. IndexWriter
  6. IndexWriter

Introduction

In this page you can find the example usage for org.apache.lucene.index IndexWriter IndexWriter.

Prototype

public IndexWriter(Directory d, IndexWriterConfig conf) throws IOException

Source Link

Document

Constructs a new IndexWriter per the settings given in conf.

Usage

From source file:com.berico.clavin.index.IndexDirectoryBuilder.java

License:Apache License

/**
 * Turns a GeoNames gazetteer file into a Lucene index, and adds
 * some supplementary gazetteer records at the end.
 * /*w w w . j a  va  2  s  .  c  om*/
 * @param args            not used
 * @throws IOException
 */
public static void main(String[] args) throws IOException {

    logger.info("Indexing... please wait.");

    // Create a new index file on disk, allowing Lucene to choose
    // the best FSDirectory implementation given the environment.
    // TODO: delete this directory first, if it exists
    FSDirectory index = FSDirectory.open(new File("./IndexDirectory"));

    // indexing by lower-casing & tokenizing on whitespace
    Analyzer indexAnalyzer = new WhitespaceLowerCaseAnalyzer();

    // create the object that will actually build the Lucene index
    IndexWriter indexWriter = new IndexWriter(index, new IndexWriterConfig(Version.LUCENE_40, indexAnalyzer));

    // open the gazetteer files to be loaded
    BufferedReader r = new BufferedReader(
            new InputStreamReader(new FileInputStream(new File(pathToGazetteer)), "UTF-8"));
    BufferedReader r2 = new BufferedReader(new InputStreamReader(
            new FileInputStream(new File("./src/main/resources/SupplementaryGazetteer.txt")), "UTF-8"));

    String line;

    // let's see how long this takes...
    Date start = new Date();

    // load GeoNames gazetteer into Lucene index
    while ((line = r.readLine()) != null)
        addToIndex(indexWriter, line);

    // add supplementary gazetteer records to index
    while ((line = r2.readLine()) != null)
        addToIndex(indexWriter, line);

    // that wasn't so long, was it?
    Date stop = new Date();

    logger.info("[DONE]");
    logger.info(indexWriter.maxDoc() + " geonames added to index.");
    logger.info("Merging indices... please wait.");

    indexWriter.close();
    index.close();
    r.close();
    r2.close();

    logger.info("[DONE]");

    DateFormat df = new SimpleDateFormat("HH:mm:ss");
    long elapsed_MILLIS = stop.getTime() - start.getTime();
    logger.info("Process started: " + df.format(start) + ", ended: " + df.format(stop) + "; elapsed time: "
            + MILLISECONDS.toSeconds(elapsed_MILLIS) + " seconds.");
}

From source file:com.berico.clavin.resolver.impl.lucene.LuceneComponentsFactory.java

License:Apache License

/**
 * Initialize the IndexWriter (and other components).
 * @return This object (it's needlessly fluent!).
 * @throws IOException/*  ww w  . j  a va 2s.  c om*/
 */
public LuceneComponentsFactory initializeWriter() throws IOException {

    initializeCommon();

    // instantiate the index writer
    indexWriter = new IndexWriter(index, new IndexWriterConfig(Version.LUCENE_43, indexAnalyzer));

    return this;
}

From source file:com.bericotech.clavin.index.IndexDirectoryBuilder.java

License:Apache License

public void buildIndex(final File indexDir, final List<File> gazetteerFiles, final File altNamesFile)
        throws IOException {
    LOG.info("Indexing... please wait.");

    indexCount = 0;/*  w w w  .j  a v a2  s  .c o  m*/

    // Create a new index file on disk, allowing Lucene to choose
    // the best FSDirectory implementation given the environment.
    FSDirectory index = FSDirectory.open(indexDir);

    // indexing by lower-casing & tokenizing on whitespace
    Analyzer indexAnalyzer = new WhitespaceLowerCaseAnalyzer();

    // create the object that will actually build the Lucene index
    indexWriter = new IndexWriter(index, new IndexWriterConfig(Version.LUCENE_4_9, indexAnalyzer));

    // let's see how long this takes...
    Date start = new Date();

    // if we were given an alternate names file, process it
    if (altNamesFile != null) {
        loadAlternateNames(altNamesFile);
    }

    // load GeoNames gazetteer into Lucene index
    String line;
    int count = 0;
    for (File gazetteer : gazetteerFiles) {
        LOG.info("Processing Gazetteer: {}", gazetteer.getAbsolutePath());
        BufferedReader reader = new BufferedReader(
                new InputStreamReader(new FileInputStream(gazetteer), "UTF-8"));
        while ((line = reader.readLine()) != null) {
            try {
                count += 1;
                // print progress update to console
                if (count % 100000 == 0) {
                    LOG.info("rowcount: " + count);
                }
                GeoName geoName = BasicGeoName.parseFromGeoNamesRecord(line);
                resolveAncestry(geoName);
            } catch (IOException e) {
                LOG.info("Skipping... Error on line: {}", line);
            } catch (RuntimeException re) {
                LOG.info("Skipping... Error on line: {}", line);
            }
        }
        reader.close();
    }

    // that wasn't so long, was it?
    Date stop = new Date();

    LOG.info("Unresolved GeoNames (Pre-resolution)");
    logUnresolved();

    resolveUnresolved();

    LOG.info("Unresolved GeoNames (Post-resolution)");
    logUnresolved();

    LOG.info("Indexing unresolved GeoNames.");
    for (Set<GeoName> geos : unresolvedMap.values()) {
        for (GeoName nm : geos) {
            indexGeoName(nm);
        }
    }

    LOG.info("[DONE]");
    LOG.info("{} geonames added to index. ({} records)", indexWriter.maxDoc(), indexCount);
    LOG.info("Merging indices... please wait.");

    indexWriter.close();
    index.close();

    LOG.info("[DONE]");

    DateFormat df = new SimpleDateFormat("HH:mm:ss");
    long elapsed_MILLIS = stop.getTime() - start.getTime();
    LOG.info("Process started: " + df.format(start) + ", ended: " + df.format(stop) + "; elapsed time: "
            + MILLISECONDS.toSeconds(elapsed_MILLIS) + " seconds.");
}

From source file:com.bewsia.script.LuceneHandler.java

License:Open Source License

protected void createEntity(SEntity src) {
    if (src.getId().length() == 0)
        return;/*from   w ww.  j av  a  2s . c  om*/
    if (src.getKind().length() == 0)
        return;

    try {
        if (!src.getKind().equals(KIND_QUOTA)) {
            if (!quotaCreate(src))
                return;
        }
        backup(src);
        Analyzer analyzer = new StandardAnalyzer(Version.LUCENE_36);
        IndexWriterConfig iwc = new IndexWriterConfig(Version.LUCENE_36, analyzer);
        iwc.setOpenMode(OpenMode.CREATE_OR_APPEND);
        IndexWriter writer = new IndexWriter(FSDirectory.open(new File(dirIndex)), iwc);
        Document doc = new Document();
        write(src, doc);
        writer.addDocument(doc);
        writer.close();
    } catch (Exception e) {
    }
}

From source file:com.bewsia.script.LuceneHandler.java

License:Open Source License

protected void updateEntity(SEntity src) {
    if (src.getId().length() == 0)
        return;//from   w w w .j a v a 2 s .c  om
    if (src.getKind().length() == 0)
        return;

    try {
        if (!src.getKind().equals(KIND_QUOTA)) {
            if (!quotaUpdate(src))
                return;
        }
        backup(src);
        Analyzer analyzer = new StandardAnalyzer(Version.LUCENE_36);
        IndexWriterConfig iwc = new IndexWriterConfig(Version.LUCENE_36, analyzer);
        iwc.setOpenMode(OpenMode.CREATE_OR_APPEND);
        IndexWriter writer = new IndexWriter(FSDirectory.open(new File(dirIndex)), iwc);
        Document doc = new Document();
        write(src, doc);
        writer.updateDocument(new Term(SEntity.ID, src.getId()), doc);
        writer.close();
    } catch (Exception e) {
    }
}

From source file:com.bewsia.script.LuceneHandler.java

License:Open Source License

protected void deleteEntity(String id) {
    if (id.length() == 0)
        return;/*from  w w  w. ja  v a 2s. c  om*/
    String kind = "";

    try {
        IndexReader reader = IndexReader.open(FSDirectory.open(new File(dirIndex)));
        IndexSearcher searcher = new IndexSearcher(reader);
        TopDocs td = searcher.search(new TermQuery(new Term(SEntity.ID, id)), 1);
        if (td.totalHits > 0) {
            Document doc = searcher.doc(td.scoreDocs[0].doc);
            kind = doc.get(SEntity.KIND);
        }
        searcher.close();
        reader.close();
    } catch (Exception e) {
    }
    if (kind.length() == 0)
        return;
    if (!allowDelete(id, kind))
        return;

    try {
        if (!kind.equals(KIND_QUOTA)) {
            if (!quotaDelete(id, kind))
                return;
        }
        removeBackup(id, kind);
        Analyzer analyzer = new StandardAnalyzer(Version.LUCENE_36);
        IndexWriterConfig iwc = new IndexWriterConfig(Version.LUCENE_36, analyzer);
        iwc.setOpenMode(OpenMode.CREATE_OR_APPEND);
        IndexWriter writer = new IndexWriter(FSDirectory.open(new File(dirIndex)), iwc);
        writer.deleteDocuments(new Term(SEntity.ID, id));
        writer.close();
    } catch (Exception e) {
    }
}

From source file:com.bitranger.parknshop.common.fulltext.BuildIndexForCustomer.java

License:Open Source License

@SuppressWarnings("deprecation")
public void createIndex() throws IOException {
    // store in the directory
    Directory directory = new SimpleFSDirectory(indexFile);
    IndexWriterConfig indexWriterConfig = new IndexWriterConfig(Version.LUCENE_45, analyzer);
    indexWriterConfig.setOpenMode(IndexWriterConfig.OpenMode.CREATE);
    IndexWriter writer = new IndexWriter(directory, indexWriterConfig);// 
    FetchOption option = new FetchOption();
    customerlist = customerDao.findAll(option);
    if (customerlist.size() > 0) {
        long time1 = System.currentTimeMillis();
        for (int i = 0; i < customerlist.size(); i++) {
            Document document = new Document();
            document.add(new Field("nickname", customerlist.get(i).getNickname(), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(/*www . j  a v  a 2  s.com*/
                    new Field("email", customerlist.get(i).getEmail(), Field.Store.YES, Field.Index.ANALYZED));
            document.add(new Field("password", customerlist.get(i).getPassword(), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(new Field("gender", String.valueOf(customerlist.get(i).getGender()), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(
                    new Field("name", customerlist.get(i).getName(), Field.Store.YES, Field.Index.ANALYZED));

            writer.addDocument(document);
        }
        long time2 = System.currentTimeMillis();
        System.out.println("" + writer.numDocs() + "");
        System.out.println("" + (time2 - time1) + "");
    }
    writer.close();
}

From source file:com.bitranger.parknshop.common.fulltext.BuildIndexForItem.java

License:Open Source License

@SuppressWarnings("deprecation")
public void createIndex() throws IOException {
    // store in the directory
    Directory directory = new SimpleFSDirectory(indexFile);
    IndexWriterConfig indexWriterConfig = new IndexWriterConfig(Version.LUCENE_45, analyzer);
    indexWriterConfig.setOpenMode(IndexWriterConfig.OpenMode.CREATE);
    IndexWriter writer = new IndexWriter(directory, indexWriterConfig);// 
    itemlist = itemDao.findAll();//w  ww  .  java2  s . c o m
    if (itemlist.size() > 0) {
        long time1 = System.currentTimeMillis();
        for (int i = 0; i < itemlist.size(); i++) {
            Document document = new Document();
            document.add(new Field("name", itemlist.get(i).getName(), Field.Store.YES, Field.Index.ANALYZED));
            document.add(new Field("introduction", itemlist.get(i).getIntroduction(), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(new Field("price", String.valueOf(itemlist.get(i).getPrice()), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(new Field("urlPicture", itemlist.get(i).getUrlPicture(), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(new Field("extra1", String.valueOf(itemlist.get(i).getExtra1()), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(new Field("extra2", String.valueOf(itemlist.get(i).getExtra2()), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(new Field("countPurchase", String.valueOf(itemlist.get(i).getCountPurchase()),
                    Field.Store.YES, Field.Index.ANALYZED));
            document.add(new Field("countFavourite", String.valueOf(itemlist.get(i).getCountFavourite()),
                    Field.Store.YES, Field.Index.ANALYZED));
            document.add(new Field("countClick", String.valueOf(itemlist.get(i).getCountClick()),
                    Field.Store.YES, Field.Index.ANALYZED));
            document.add(new Field("vote", String.valueOf(itemlist.get(i).getVote()), Field.Store.YES,
                    Field.Index.ANALYZED));

            writer.addDocument(document);
        }
        long time2 = System.currentTimeMillis();
        System.out.println("" + writer.numDocs() + "");
        System.out.println("" + (time2 - time1) + "");
    }
    writer.close();
}

From source file:com.bitranger.parknshop.common.fulltext.BuildIndexForOrder.java

License:Open Source License

@SuppressWarnings("deprecation")
public void createIndex() throws IOException {
    // store in the directory
    Directory directory = new SimpleFSDirectory(indexFile);
    IndexWriterConfig indexWriterConfig = new IndexWriterConfig(Version.LUCENE_45, analyzer);
    indexWriterConfig.setOpenMode(IndexWriterConfig.OpenMode.CREATE);
    IndexWriter writer = new IndexWriter(directory, indexWriterConfig);// 
    Date from = new Date(), to = new Date();
    orderlist = orderDao.findAll(from, to);
    if (orderlist.size() > 0) {
        long time1 = System.currentTimeMillis();
        for (int i = 0; i < orderlist.size(); i++) {
            Document document = new Document();
            document.add(new Field("id", String.valueOf(orderlist.get(i).getId()), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(new Field("psCustomer", String.valueOf(orderlist.get(i).getPsCustomer()),
                    Field.Store.YES, Field.Index.ANALYZED));
            document.add(new Field("idShop", String.valueOf(orderlist.get(i).getId()), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(new Field("status", String.valueOf(orderlist.get(i).getStatus()), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(new Field("trackingNumber", orderlist.get(i).getTrackingNumber(), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(new Field("address", orderlist.get(i).getPsRecipient().getAddresss(), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(new Field("postalCode", orderlist.get(i).getTrackingNumber(), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(new Field("nameRecipient", orderlist.get(i).getPsRecipient().getPsCustomer().getName(),
                    Field.Store.YES, Field.Index.ANALYZED));
            document.add(new Field("phoneRecipient", orderlist.get(i).getPsRecipient().getPhoneNumber(),
                    Field.Store.YES, Field.Index.ANALYZED));
            document.add(new Field("priceTotal", String.valueOf(orderlist.get(i).getPriceTotal()),
                    Field.Store.YES, Field.Index.ANALYZED));
            document.add(new Field("timeCreated", String.valueOf(orderlist.get(i).getTimeCreated()),
                    Field.Store.YES, Field.Index.ANALYZED));

            writer.addDocument(document);
        }/*from w w w . jav a  2 s . c o  m*/
        long time2 = System.currentTimeMillis();
        System.out.println("" + writer.numDocs() + "");
        System.out.println("" + (time2 - time1) + "");
    }
    writer.close();
}

From source file:com.bitranger.parknshop.common.fulltext.BuildIndexForSeller.java

License:Open Source License

@SuppressWarnings("deprecation")
public void createIndex() throws IOException {
    // store in the directory
    Directory directory = new SimpleFSDirectory(indexFile);
    IndexWriterConfig indexWriterConfig = new IndexWriterConfig(Version.LUCENE_45, analyzer);
    indexWriterConfig.setOpenMode(IndexWriterConfig.OpenMode.CREATE);
    IndexWriter writer = new IndexWriter(directory, indexWriterConfig);// 
    FetchOption option = new FetchOption();
    sellerlist = sellerDao.findAll(new FetchOption().limit(100));
    if (sellerlist.size() > 0) {
        long time1 = System.currentTimeMillis();
        for (int i = 0; i < sellerlist.size(); i++) {
            Document document = new Document();
            document.add(new Field("id", String.valueOf(sellerlist.get(i).getId()), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(new Field("nickname", sellerlist.get(i).getNickname(), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(new Field("personIdNum", String.valueOf(sellerlist.get(i).getPersonIdNum()),
                    Field.Store.YES, Field.Index.ANALYZED));
            document.add(/*from   w w w.j  a va  2s. c  o m*/
                    new Field("email", sellerlist.get(i).getEmail(), Field.Store.YES, Field.Index.ANALYZED));
            document.add(new Field("password", sellerlist.get(i).getPassword(), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(new Field("status", String.valueOf(sellerlist.get(i).getStatus()), Field.Store.YES,
                    Field.Index.ANALYZED));
            document.add(new Field("timeCreated", String.valueOf(sellerlist.get(i).getTimeCreated()),
                    Field.Store.YES, Field.Index.ANALYZED));

            writer.addDocument(document);
        }
        long time2 = System.currentTimeMillis();
        System.out.println("" + writer.numDocs() + "");
        System.out.println("" + (time2 - time1) + "");
    }
    writer.close();
}