Java Lucene 4.5如何通过大小写不敏感搜索

Question

我们已经实现了Java Lucene搜索引擎4.5,我试图搜索内容,即使字段值不区分大小写(例如,如果我搜索名为"Banglore"的城市,我得到一个结果,但是当我搜索一个名字的城市时"banglore"我得到0结果).

我用来StandardAnalyzer分析数据并WildcardQuery匹配一个Like条件(我试过这里没有成功).

我不知道我哪里出错了.我很感激有关修复此区分大小写问题的任何指导.

public SearchHelper
{
    Analyzer analyzer;

    Directory index;
    public IndexSearcher searcher = null;
    public IndexWriter indexWriter = null;
    public QueryParser parser = null;
    private static int hitsPerPage = 100;

    /**
     * @param indexFileLocation
     * @throws IOException
     */
    public SearchHelper(String indexFileLocation) throws IOException
    {
//        this.analyzer =new StandardAnalyzer();
        this.analyzer = new CaseStandardAnalyzer();
//        analyzer = new ThaiAnalyzer();
        this.index = FSDirectory.open(java.nio.file.Paths.get(indexFileLocation));
    }

    /**
     * @param create
     * @return
     * @throws IOException
     */
    public IndexWriter getIndexWriter(boolean create) throws IOException
    {
        if (indexWriter == null)
        {
            IndexWriterConfig iwc = new IndexWriterConfig(this.analyzer);
            this.indexWriter = new IndexWriter(this.index, iwc);
        }
        return this.indexWriter;
    } //End of getIndexWriter

    /**
     * @throws IOException
     */
    public void closeIndexWriter() throws IOException
    {
        if (this.indexWriter != null)
        {
             this.indexWriter.commit();//optimize(); LUCENE_36
             this.indexWriter.close();
        }
    } //End closeIndexWriter

    /**
     * @param indexFileLocation
     * @throws CorruptIndexException
     * @throws IOException
     */
    public void startSearch(String indexFileLocation) throws CorruptIndexException, IOException
    {
//        searcher = new IndexSearcher(FSDirectory.open(new File(indexFileLocation)));

        IndexReader reader = DirectoryReader.open(FSDirectory.open(java.nio.file.Paths.get(indexFileLocation)));
//        IndexReader.open(this.index);
//        open(getIndexWriter(true), true);
        this.searcher = new IndexSearcher(reader);
    }

    /**
     * @param fieldNames
     * @param fieldValues
     * @return
     * @throws IOException
     * @throws ParseException
     * 
     * <p></p>
     * https://stackoverflow.com/questions/2005084/how-to-specify-two-fields-in-lucene-queryparser
     */
    public ScoreDoc[] searchSEO(String[] fieldNames, String[] fieldValues, int limitSize) throws IOException, ParseException
    {
        this.analyzer = new StandardAnalyzer();
        int searchFieldSize = (null == fieldNames) ? 0 : fieldNames.length;

        BooleanQuery booleanQuery = new BooleanQuery();

        for (int i = 0; i < searchFieldSize; i++)
        {
             Query query1 = searchIndexWithWildcardQuery(fieldNames[i], fieldValues[i]);                
             addQueries(booleanQuery, query1, 2);               
        }

        TopScoreDocCollector collector = null; // Or use by default hitsPerPage instead limitSize

        if (limitSize > 0)
        {
            collector = TopScoreDocCollector.create(limitSize);
        } else {
            collector = TopScoreDocCollector.create(hitsPerPage);
        }

        this.searcher.search(booleanQuery,collector);

        return  collector.topDocs().scoreDocs;
    }

    /**
     * @param whichField
     * @param searchString
     * @return
     * @throws IOException
     * @throws ParseException
     */
    public Query searchIndexWithWildcardQuery(String whichField, String searchString) throws IOException, ParseException
    {
        Term term = addTerm(whichField, "*" + searchString + "*");
        Query query = new WildcardQuery(term);
        return query;
    }

    /**
     * @param whichField
     * @param searchString
     * @return
     */
    public Term addTerm(String whichField, String searchString)
    {
        Term term = new Term(whichField, searchString);
        return term;
    }

    /**
     * @param searchString
     * @param operation
     * @return
     * @throws ParseException
     */
    public Query addConditionOpertaion(String searchString, String operation) throws ParseException
    {
        Query query = null;
        if ("and".equals(operation))
        {
            parser.setDefaultOperator(QueryParser.AND_OPERATOR);
        } else if("or".equals(operation)) {
            parser.setDefaultOperator(QueryParser.AND_OPERATOR);
        }

        query = parser.parse(searchString);
        return query;
    }

    /**
     * @param booleanQuery <code>BooleanQuery</code>
     * @param q <code>Query</code>
     * @param type <code>int</code> , 1--> Must, 2-->Should, 3 --> Must Not
     */
    public void addQueries(BooleanQuery booleanQuery, Query q, int type)
    {
        switch(type)
        {
            case 1: booleanQuery.add(q, Occur.MUST);
                    break;
            case 2: booleanQuery.add(q, Occur.SHOULD);
                    break;
            default:booleanQuery.add(q, Occur.MUST_NOT);
                    break;
        } //End of switch
    }

    public QueryParser getParser()
    {
        return parser;
    }

    public void setParser(String fieldName)
    {
        this.parser = new QueryParser(fieldName, this.analyzer);
    }

    public void getDefaultByStatus(int status)
    {
        this.analyzer = new StandardAnalyzer();
        this.parser = new QueryParser("status", this.analyzer);
    }

    protected void doClear(File dir,boolean deleteSubDir)
    {
        for (File file: dir.listFiles())
        {
            if (file.isDirectory() && deleteSubDir)
            {
                doClear(file,deleteSubDir);
            }
            file.delete();
        }
    } //End of doClear();

    protected void doClose() throws IOException
    {
        this.searcher.getIndexReader().close();
    }

    public boolean add(Object Obj) throws Exception
    {
        User currentUser = (User)Obj;
        boolean isAdded = false;

        org.apache.lucene.document.Document luceneDoc = new org.apache.lucene.document.Document();
        luceneDoc.add(new IntField("oid", currentUser.getOid(), Field.Store.YES));
        luceneDoc.add(new IntField("status", currentUser.getStatus(), Field.Store.YES));
        luceneDoc.add(new StringField("login", currentUser.getLogin(), Field.Store.YES));
        luceneDoc.add(new StringField("fName", currentUser.getFirstName(), Field.Store.YES));
        luceneDoc.add(new StringField("lName", currentUser.getLastName(), Field.Store.NO));
        luceneDoc.add(new StringField("email", currentUser.getEmailId(), Field.Store.YES));
        luceneDoc.add(new StringField("city", currentUser.getCity(), Field.Store.YES));

//        addRelatedFields(luceneDoc,city.getStateCode());

        IndexWriter writer = getIndexWriter(false);
        writer.addDocument(luceneDoc);

        closeIndexWriter();

        isAdded = true;
        System.out.println(isAdded);
        return isAdded;
    } // End of add

    public boolean update(Object Obj) throws Exception
    {
        boolean isUpdated = false;
        User currentUser = (User) Obj;

        org.apache.lucene.document.Document luceneDoc = new org.apache.lucene.document.Document();
//        luceneDoc.add(new IntField("oid", currentUser.getOid(), Field.Store.YES));
        luceneDoc.add(new IntField("oid", currentUser.getOid(), Field.Store.YES));
        luceneDoc.add(new StringField("login", currentUser.getLogin(), Field.Store.YES));
        luceneDoc.add(new IntField("status", currentUser.getStatus(), Field.Store.YES));
        luceneDoc.add(new StringField("fName", currentUser.getFirstName(), Field.Store.YES));
        luceneDoc.add(new StringField("lName", currentUser.getLastName(), Field.Store.NO));
        luceneDoc.add(new StringField("email", currentUser.getEmailId(), Field.Store.YES));
        luceneDoc.add(new StringField("city", currentUser.getCity(), Field.Store.YES));

//        addRelatedFields(luceneDoc,city.getStateCode());

        IndexWriter writer = getIndexWriter(false);
        writer.updateDocument(new Term("login", currentUser.getLogin()),luceneDoc); 
        closeIndexWriter();

        isUpdated = true;
        return isUpdated;
    } // End of update

    public boolean delete(Object Obj) throws Exception
    {
        boolean isDeleted = false;
        User currentUser = (User) Obj;      

        Term deleteTerm = new Term("login", currentUser.getLogin());

        IndexWriter writer = getIndexWriter(false);
        writer.deleteDocuments(deleteTerm); // Or use Query
        writer.forceMergeDeletes();
        closeIndexWriter();

        isDeleted = true;

        return isDeleted;
    } // End of delete

    @Override
    public Object search(String[] fieldNames, String[] fieldValues, int returnType, int limit) throws Exception
    {
        Object obj = null;
        org.apache.lucene.search.ScoreDoc[] hits =  searchSEO(fieldNames,fieldValues,  limit);
        int hitSize = (null == hits) ? 0 : hits.length;

        System.out.println("total:" + hitSize);

        doClose();
        return obj;
    } // End of search

    public void addThreadUser()
    {
        User user = new User();
        addUserPojo(user);    
        add(user);
    }

    public void updateThreadUser()
    {
        User user = new User();
        addUserPojo(user);
        update(user);
    }

    public void deleteThreadUser()
    {
        User user = new User();
        addUserPojo(user);   
        delete(user);
    }

    private void addUserPojo(User user)
    {
        user.setOid(3);
        user.setLogin("senthil");
        user.setFirstName("Semthil");
        user.setLastName("Semthil");
        user.setStatus(1);
        user.setCity("Combiatore");
        user.setEmailId("semthil@xyz.com");
    }

    public void searchUser()
    {
        searchUser(new String[] {"login"}, new String[] {"Se"}, null);
    }

    public static void main(String[] args)
    {
        SearchHelper test = new SearchHelper();
        test.searchUser();
    }
}

Answer 1

您正在使用StringField索引数据,但此字段将绕过分析器链并始终将您的术语逐字索引为一个标记,无论您的分析器如何.TextField如果您想分析数据并且StandardAnalyzer已经进行了较低的套管,则应该使用.除此之外,WildcardQuery它不会分析它的术语,所以如果你搜索Banglore,它将与索引中现在较小的banglore不匹配.您必须自己小写搜索项(或使用分析器).

Answer 2

按照您引用的帖子的建议使用LowerCaseFilter ：

    TokenStream stream = new StandardFilter(Version.LUCENE_CURRENT, tokenizer);
    stream = new LowerCaseFilter(Version.LUCENE_CURRENT, stream);

这篇文章中有一个更完整的例子。