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

Introduction

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

Prototype

IndexOptions

Source Link

Usage

From source file:edu.stanford.muse.index.Indexer.java

License:Apache License

private static void testQueries()
        throws IOException, ParseException, GeneralSecurityException, ClassNotFoundException {
    Indexer li = new Indexer("/tmp", new IndexOptions());

    // public EmailDocument(String id, String folderName, Address[] to, Address[] cc, Address[] bcc, Address[] from, String subject, String messageID, Date date)
    EmailDocument ed = new EmailDocument("1", "dummy", "dummy", new Address[0], new Address[0], new Address[0],
            new Address[0], "", "", new Date());
    li.indexSubdoc(" ssn 123-45 6789 ", "name 1 is John Smith.  credit card # 1234 5678 9012 3456 ", ed, null);
    ed = new EmailDocument("2", "dummy", "dummy", new Address[0], new Address[0], new Address[0],
            new Address[0], "", "", new Date());
    li.indexSubdoc(" ssn 123 45 6789",
            "name 1 is John Smith.  credit card not ending with a non-digit # 1234 5678 9012 345612 ", ed,
            null);//from w w w . j  a va2  s  .co m
    ed = new EmailDocument("3", "dummy", "dummy", new Address[0], new Address[0], new Address[0],
            new Address[0], "", "", new Date());
    li.indexSubdoc(" ssn 123 45 6789", "name 1 is John Smith.  credit card # 111234 5678 9012 3456 ", ed, null);
    ed = new EmailDocument("4", "dummy", "dummy", new Address[0], new Address[0], new Address[0],
            new Address[0], "", "", new Date());
    li.indexSubdoc(" ssn 123 45 6789", "\nmy \nfirst \n book is \n something ", ed, null);
    ed = new EmailDocument("5", "dummy", "dummy", new Address[0], new Address[0], new Address[0],
            new Address[0], "", "", new Date());
    li.indexSubdoc("passport number k4190893", "\nmy \nfirst \n book is \n something ", ed, null);

    li.close();

    li.setupForRead();
    String q = "john";
    Collection<EmailDocument> docs = li.lookupDocs(q, QueryType.FULL);
    System.out.println("hits for: " + q + " = " + docs.size());

    q = "/j..n/\\\\*";
    docs = li.lookupDocs(q, QueryType.FULL);
    System.out.println("hits for: " + q + " = " + docs.size());

    q = "\"john\"";
    docs = li.lookupDocs(q, QueryType.FULL);
    System.out.println("hits for: " + q + " = " + docs.size());

    q = "\"john smith\"";
    docs = li.lookupDocs(q, QueryType.FULL);
    System.out.println("hits for: " + q + " = " + docs.size());

    q = "john*smith";
    docs = li.lookupDocs(q, QueryType.FULL);
    System.out.println("hits for: " + q + " = " + docs.size());

    q = "title:john";
    docs = li.lookupDocs(q, QueryType.FULL);
    System.out.println("hits for: " + q + " = " + docs.size());

    q = "title:subject";
    docs = li.lookupDocs(q, QueryType.FULL);
    System.out.println("hits for: " + q + " = " + docs.size());

    q = "body:johns";
    docs = li.lookupDocs(q, QueryType.FULL);
    System.out.println("hits for: " + q + " = " + docs.size());

    q = "title:johns";
    docs = li.lookupDocs(q, QueryType.FULL);
    System.out.println("hits for: " + q + " = " + docs.size());

    q = "joh*";
    docs = li.lookupDocs(q, QueryType.FULL);
    System.out.println("hits for: " + q + " = " + docs.size());

    q = "/j..n/";
    //      q = "/\\b(\\d{9}|\\d{3}-\\d{2}-\\d{4})\\b/";
    docs = li.lookupDocs(q, QueryType.FULL);
    System.out.println("hits for: " + q + " = " + docs.size());

    // look for sequence of 4-4-4-4 . the .* at the beginning and end is needed.
    q = "[0-9]{3}[\\- ]*[0-9]{2}[ \\-]*[0-9]{4}";
    docs = li.lookupDocs(q, QueryType.REGEX);
    System.out.println("hits for: " + q + " = " + docs.size());

    // look for sequence of 3-2-4
    //q = "[0-9]{3}[ \\-]*[0-9]{2}[ \\-]*[0-9]{4}";
    q = "123-45[ \\-]*[0-9]{4}";
    System.out.println("hits for: " + q + " = " + li.lookupDocs(q, QueryType.REGEX).size());

    // look for sequence of 3-2-4
    q = "first\\sbook";
    docs = li.lookupDocs(q, QueryType.REGEX);
    System.out.println("hits for: " + q + " = " + docs.size());

    q = "ssn";
    int numHits = li.getNumHits(q, false, QueryType.FULL);
    System.err.println("Number of hits for: " + q + " is " + numHits);

    q = "[A-Za-z][0-9]{7}";
    docs = li.lookupDocs(q, QueryType.REGEX);
    System.out.println("hits for: " + q + " = " + docs.size());

    li.analyzer = null;
    li.isearcher = null;
    li.parser = null;
    li.parserOriginal = null;
    li.parserSubject = null;
    li.parserCorrespondents = null;
    Util.writeObjectToFile("/tmp/1", li);
    li = (Indexer) Util.readObjectFromFile("/tmp/1");
}

---