Database DAO - JDBC Query Class - Code review please

Question

I use the following class (Database.java) as a DAO for handling all database calls. The class works. You can use this if you're looking for a DAO solution (i.e. a way to query databases using connection pools / JNDI). I am looking for some suggestions on improving this class. Can you help? I.e. suggestions for improving SELECT speed, logic? The class is used by a moderately heavily used Web application (ca. 2000 - 3000 queries a day - so not too much) on a Sun Web Server system.
This class had the following interfaces:
getErrors() // for retrieving any errors that occurred during an query
setSql() // one way to set the SQL that must be run
setDbName() // one way to set the JNDI name of the database resource
getInsertDelete() // run the INSERT/DELETE statement that was provided with setSql
getInsertDelete(String givenSql, String givenDb) // run with provided sql and db
getClobInsert(String givenSql, String givenDb, Hashtable clobHash, String identifierSql) // clobHash contains the column name and the value (which is a string over 4000 chars). Identifier SQL is the SQL for identifying the row, into which the Clob must be inserted. This interface is specific to Oracle.
getSelect() // Query database with set SQL. Return as a vector of hashes so I can close connection.
getSelect(String givenSql, String givenDb) // select with given sqlAnd here is the full class. I know, this is a weird post, but we don't really have a code review process here at work, and I don't have a specific problem. Just want some feedback concerning the way I query our databases. So, any tips or comments are welcome.
package melib.network;
import java.io.Writer;
import java.io.StringReader;
import java.io.IOException;
import java.util.Vector;
import java.util.Hashtable;
import java.sql.Connection;
import java.sql.Statement;
import java.sql.ResultSet;
import java.sql.ResultSetMetaData;
import java.sql.Clob;
import javax.sql.DataSource;
import javax.naming.InitialContext;
import melib.system.PropertiesReader; // This is just a simple class for pulling values from a .properties file
* Manages database connections.
* @author jetcat33
* @version 2.0, 2006-07-09
public class Database {
    protected String sql = "";
    protected String dbName = "";
    private StringBuffer errors = new StringBuffer();
    /** Creates a new instance of Database */
    public Database() {
     * Check completeness of data
     * for example to check if dbname given or
     * sql given or to make sure that yikes
     * used instead of sccweb in test conditions
    protected boolean checkData(){
        if(dbName.equals("") || sql.equals("")){
            Email.sendMail(PropertiesReader.getValue("statusEmail"),null,"MelibLibraryError","melib error: [Database]","No database selected for query (db: "+dbName+" or sql not included (sql: "+sql+")");
            setErrors("No database selected for query (db: "+dbName+" or sql not included (sql: "+sql+")");
            return false;
        }else{
            return true;
     * Sets and gets errors resulting from
     * exceptions - otherwise would have to
     * somehow include errors in the results
     * that are returned but that would include
     * other more complicated stuff.
    private void setErrors(String e){
        errors.append("\n"+e);
    public StringBuffer getErrors(){
       return errors;
     * Setter for the sql variable.
     * @param givenSql The sql required to process
    public void setSql(java.lang.String givenSql) {
        sql = givenSql;
     * Sets the dbName needed to commect
     * to the correct database.
     * @param givenDbName Name of the database - name and connections specified within this class
    public void setDbName(java.lang.String givenDbName) {
        dbName = givenDbName;
     * Processes Insert and Delete requests with given SQL and DB Name.
     * @param givenSql
     * @param givenDbName
     * @return String Number of rows affected
    public String getInsertDelete(String givenSql, String givenDbName){
        sql = givenSql;
        dbName = givenDbName;
        return getInsertDelete();
     * Takes care of insert, update and delete requests.
     * Must have set both dbName as well as the sql String.
     * Will return number of rows affected as String.
     * @return String Number of rows affected
     * @exception Exception
    public String getInsertDelete() {
        int returnValue = 0;
        if(checkData()){
            Connection conn = null;
            InitialContext initContext = null;
            DataSource source = null;
            Statement stmt = null;
            try{
                // Get connection from configured pool
                initContext = new InitialContext();
                source = (DataSource) initContext.lookup("java:comp/env/jdbc/" + dbName);
                conn = source.getConnection();
                if(conn != null){
                    stmt = conn.createStatement();
                    returnValue = stmt.executeUpdate(sql);
            }catch (Exception e){
                Email.sendMail(PropertiesReader.getValue("statusEmail"),null,"MelibLibraryError","melib error: [Database.getInsertDelete]","getInsertDelete Exception: "+e.toString()+"\nWith: "+sql);
                setErrors(e.toString());
            }finally{
                try{
                    stmt.close();
                }catch(Exception e){
                    e.printStackTrace();
                try{
                    conn.close();
                }catch(Exception e){
                   e.printStackTrace();
                try{
                    initContext.close();
                }catch(Exception e){
                    e.printStackTrace();
        return returnValue+"";
     * Processes Insert requests for SQL containing CLOBS
     * @param givenSql
     * @param givenDbName
     * @param clobHash Contains column name of clob and clob text
     * @param identifierSql Contains SQL to identify the just entered SQL so the clobs can be filled in.
     * @return String Number of rows affected
    public String getClobInsert(String givenSql, String givenDbName, Hashtable clobHash, String identifierSql){
        sql = givenSql;
        dbName = givenDbName;
        String returnValue="";
        int rv = 0;
        if(checkData()){
            Connection conn = null;
            InitialContext initContext = null;
            DataSource source = null;
            Statement stmt = null;
            try{
                // Get connection from configured pool
                initContext = new InitialContext();
                source = (DataSource) initContext.lookup("java:comp/env/jdbc/" + dbName);
                conn = source.getConnection();
                if(conn != null){
                    conn.setAutoCommit(false);
                    stmt = conn.createStatement();
                    rv = stmt.executeUpdate(sql); // write first time
                    // Now get and overwrite "EMPTY_CLOB()"
                    ResultSet lobDetails = stmt.executeQuery(identifierSql);
                    ResultSetMetaData rsmd = lobDetails.getMetaData();
                    if(lobDetails.next()){
                        for(int i = 1; i <= rsmd.getColumnCount(); i++){
                            if(clobHash.get(rsmd.getColumnName(i))!=null && !clobHash.get(rsmd.getColumnName(i)).equals("")){
                                Clob theClob = lobDetails.getClob(i);
                                Writer clobWriter = ((oracle.sql.CLOB)theClob).getCharacterOutputStream();
                                StringReader clobReader = new StringReader((String) clobHash.get(rsmd.getColumnName(i)));
                                char[] cbuffer = new char[30* 1024]; // Buffer to hold chunks of data to be written to Clob, the slob
                                int nread = 0;
                                try{
                                    while((nread=clobReader.read(cbuffer)) != -1){
                                        clobWriter.write(cbuffer,0,nread);
                                }catch(IOException ioe){
                                   //System.out.println("E: clobWriter exception - " + ioe.toString());
                                }finally{
                                    try{
                                        returnValue+=" Writing: "+rsmd.getColumnName(i);
                                        clobReader.close();
                                        clobWriter.close();
                                    }catch(IOException ioe2){
                                        //System.out.println("E: clobWriter close exception - " + ioe2.toString());
                    conn.commit();
            }catch (Exception e){
                Email.sendMail(PropertiesReader.getValue("statusEmail"),null,"MelibLibraryError","melib error: [Database.getClobInsert]","getClobInsert Exception: "+e.toString()+"\nWith: "+sql+"\nAND\n"+identifierSql);
                setErrors(e.toString());
            }finally{
                try{
                    stmt.close();
                }catch(Exception e){
                    e.printStackTrace();
                try{
                    conn.close();
                }catch(Exception e){
                   e.printStackTrace();
                try{
                    initContext.close();
                }catch(Exception e){
                    e.printStackTrace();
            returnValue=rv+" "+returnValue;
        return returnValue;
     * Takes care of Select statements.
     * Must have set both dbName as well as the sql String.
     * Will return a vector.
     * @return Vector of Hashes containing the Results of the query
     * @exception SQLException
     * @exception Exception
    public Vector getSelect(){
        Vector returnValue = new Vector();
        if(checkData()){
            Connection conn = null;
            InitialContext initContext = null;
            DataSource source = null;
            ResultSet result = null;
            ResultSetMetaData rsmd = null;
            Statement stmt = null;
            try{
                // Get connection from configured pool
                initContext = new InitialContext();
                source = (DataSource) initContext.lookup("java:comp/env/jdbc/" + dbName);
                conn = source.getConnection();
                if(conn != null){
                    stmt = conn.createStatement();
                    result = stmt.executeQuery(sql);
                    rsmd = result.getMetaData();
                    while(result.next()){
                        Hashtable hash = new Hashtable();
                        for(int i = 1; i <= rsmd.getColumnCount(); i++){
                            if(result.getString(i) != null){
                                hash.put(rsmd.getColumnName(i),result.getString(i));
                            }else{
                                hash.put(rsmd.getColumnName(i),"");
                        returnValue.addElement(hash);
            }catch (Exception e){
                Email.sendMail(PropertiesReader.getValue("statusEmail"),null,"MelibLibraryError","melib error: [Database.getSelect]","getSelect Exception: "+e.toString()+"\nWith: "+sql);
                setErrors(e.toString());
            }finally{
                try{
                    result.close();
                }catch(Exception e){
                    e.printStackTrace();
                    setErrors(e.toString());
                try{
                    stmt.close();
                }catch(Exception e){
                    e.printStackTrace();
                    setErrors(e.toString());
                try{
                    conn.close();
                }catch(Exception e){
                   e.printStackTrace();
                   setErrors(e.toString());
                try{
                    initContext.close();
                }catch(Exception e){
                    e.printStackTrace();
                    setErrors(e.toString());
        return returnValue;
     * Takes care of Select statements with given SQL.
     * Must have set both dbName as well as the sql String.
     * Will return a vector.
     * @return Vector with Results of the query
     * @exception SQLException
     * @exception Exception
    public Vector getSelect(String givenSql, String givenDbName){
        sql = givenSql;
        dbName = givenDbName;
        return getSelect();
}Thank you,
dailysun

Answer 1

too much code that's repeated. refactor all the code you have for closing resources into a utility class and simply call its methods.
your dao creates the connection, so it can't participate in a transaction. if there are several daos that should be one unit of work, you can't manage it with this framework.
transactions are under the control of a service layer. I think it's better to have the service object get the connection, pass it to all the DAOs needed for that unit of work, and then have the service close it.
e-mailing errors? I don't like it. if you have 2-3K queries a day failing because the database you'll have 2-3K e-mails to delete. better to log messages. if you really want e-mail, you can have Log4J add an e-mail appender. at least that way it's configurable outside the code.
CRUD operations without an UPDATE? You're missing something important.
What if I don't want to get the database from a JNDI datasource? Now you can't use this DAO without an app server.
Your error messages are less informative than a stack trace. Why not throw an exception with detailed info?
have a look at Spring and how it handles persistence. if you insist on writing your own, might want to look at Spring. Rod Johnson has developed a much better, more general way to handle persistence.
http://www.springframework.org
You return a Vector of Hashtables for queries? Two old-school data structures. You understand the implication of choosing those, right? I'd recommend that you change that to a List of Maps. Let the concrete types be ArrayList and HashMap. Interfaces are preferred, because they let you change the implementation without affecting clients. Vector and Hashtable are synchronized by default; ArrayList and HashMap are not. Synchronization is thread safe but slower.
I understand why you chose to return Vector, because you wanted something general that wouldn't force you to deal with object-relational mapping. This way you can just worry about ResultSetMetaData.
A Vector of Hashtables is a row-based view of the ResultSet (one Hashtable per row). Sometimes a column based view can be more convenient. Then it's a Map of Lists.
You gave it a good try, but I think there's a lot of room for improvement.
%