Merge pull request #99 from sfu-cl-lab/new-database-framework

Initial changes required to change our code to use the new database framework and some cleanup.

code/factorbase/src/main/java/ca/sfu/cs/factorbase/app/RunBB.java

@@ -3,67 +3,64 @@
 import java.util.logging.Logger;
 
 import ca.sfu.cs.common.Configuration.Config;
+import ca.sfu.cs.factorbase.database.FactorBaseDataBase;
+import ca.sfu.cs.factorbase.database.MySQLFactorBaseDataBase;
 import ca.sfu.cs.factorbase.exporter.csvexporter.CSVPrecomputor;
 import ca.sfu.cs.factorbase.tables.BayesBaseCT_SortMerge;
 import ca.sfu.cs.factorbase.tables.BayesBaseH;
 import ca.sfu.cs.factorbase.tables.KeepTablesOnly;
-import ca.sfu.cs.factorbase.tables.MakeSetup;
 import ca.sfu.cs.factorbase.util.LoggerConfig;
 
-/* July 3rd, 2014, zqian
- * input: 
+/**
+ * July 3rd, 2014, zqian
+ * input:
  * @database@ (original data based on ER diagram)
- * output: 
+ * output:
  * @database@_BN (e.g. _CP, Path_BayesNets, Score)
- * 			@database@_CT (e.g. BiggestRchain_CT)
- * 			@database@_setup (preconditions for learning)
- * 
- * */
+ * @database@_CT (e.g. BiggestRchain_CT)
+ * @database@_setup (preconditions for learning)
+ */
 public class RunBB {
-	static String isAutomaticSetup;
-	private static Logger logger = Logger.getLogger(RunBB.class.getName());
-
-	public static void main(String[] args) throws Exception {
-		LoggerConfig.setGlobalLevel();
-		long t1 = System.currentTimeMillis(); 
-		logger.info("Start Program...");
-		setVarsFromConfig();
-		if (isAutomaticSetup.equals("1")) {
-			MakeSetup.runMS();
-			logger.info("Setup database is ready.");
-		} else {
-			logger.info("Setup database exists.");
-		}
-		runBBLearner();
-
-		long t2 = System.currentTimeMillis(); 
-		logger.info("Total Running time is " + (t2-t1) + "ms.");
-	}
-
-
-	public static void setVarsFromConfig(){
-		Config conf = new Config();
-		//1: run Setup; 0: not run
-		isAutomaticSetup = conf.getProperty("AutomaticSetup");
-	}
-
-	public static void runBBLearner() throws Exception {
-
-		//assumes that dbname is in config file and that dbname_setup exists.
+    private static Logger logger = Logger.getLogger(RunBB.class.getName());
 
-		BayesBaseCT_SortMerge.buildCT();
-		logger.info("The CT database is ready for use.");
-		logger.info("*********************************************************");
-		CSVPrecomputor.runCSV();
-		logger.info("CSV files are generated.");
-		logger.info("*********************************************************");
-		BayesBaseH.runBBH();
-		logger.info("\nFinish running BayesBaseH.");
-		logger.info("*********************************************************");
-		logger.info("Cleaning CT database");
-		//Now eliminate temporary tables. Keep only the tables for the longest Rchain. Turn this off for debugging.//
-		KeepTablesOnly.Drop_tmpTables();
 
-	}
+    public static void main(String[] args) throws Exception {
+        long t1 = System.currentTimeMillis();
+        LoggerConfig.setGlobalLevel();
+        Config config = new Config();
+        logger.info("Start Program...");
 
-}
+        FactorBaseDataBase factorBaseDatabase = new MySQLFactorBaseDataBase(
+            config.getProperty("dbaddress"),
+            config.getProperty("dbname"),
+            config.getProperty("dbusername"),
+            config.getProperty("dbpassword")
+        );
+
+        // Generate the setup database if specified to.
+        if (config.getProperty("AutomaticSetup").equals("1")) {
+            factorBaseDatabase.setupDatabase();
+            logger.info("Setup database is ready.");
+        } else {
+            logger.info("Setup database exists.");
+        }
+
+        // Learn a Bayesian Network.
+        BayesBaseCT_SortMerge.buildCT();
+        logger.info("The CT database is ready for use.");
+        logger.info("*********************************************************");
+        CSVPrecomputor.runCSV();
+        logger.info("CSV files are generated.");
+        logger.info("*********************************************************");
+        BayesBaseH.runBBH();
+        logger.info("\nFinish running BayesBaseH.");
+        logger.info("*********************************************************");
+
+        // Now eliminate temporary tables. Keep only the tables for the longest Rchain. Turn this off for debugging.
+        logger.info("Cleaning CT database");
+        KeepTablesOnly.Drop_tmpTables();
+
+        long t2 = System.currentTimeMillis();
+        logger.info("Total Running time is " + (t2 - t1) + "ms.");
+    }
+}

code/factorbase/src/main/java/ca/sfu/cs/factorbase/database/FactorBaseDataBase.java

@@ -0,0 +1,16 @@
+package ca.sfu.cs.factorbase.database;
+
+import ca.sfu.cs.factorbase.exception.DataBaseException;
+
+/**
+ * Methods expected to be implemented to enable the extraction of data from a database for FactorBase.
+ */
+public interface FactorBaseDataBase {
+    /**
+     * This method should setup all the extra tables required for FactorBase to learn a Bayesian
+     * network for the provided database.
+     *
+     * @throws DataBaseException if an error occurs when attempting to access the database.
+     */
+    void setupDatabase() throws DataBaseException;
+}

code/factorbase/src/main/java/ca/sfu/cs/factorbase/database/MySQLFactorBaseDataBase.java

@@ -0,0 +1,53 @@
+package ca.sfu.cs.factorbase.database;
+
+import java.io.IOException;
+import java.sql.DriverManager;
+import java.sql.SQLException;
+import java.text.MessageFormat;
+
+import ca.sfu.cs.common.Configuration.Config;
+import ca.sfu.cs.factorbase.exception.DataBaseException;
+import ca.sfu.cs.factorbase.util.BZScriptRunner;
+
+import com.mysql.jdbc.Connection;
+
+public class MySQLFactorBaseDataBase implements FactorBaseDataBase {
+
+    private static final String CONNECTION_STRING = "jdbc:{0}/{1}";
+    private String baseDatabaseName;
+    private Connection baseConnection;
+
+
+    /**
+     * Create connections to the databases required by FactorBase to learn a Bayesian Network.
+     *
+     * @param dbaddress - the address of the MySQL database to connect to. e.g. mysql://127.0.0.1
+     * @param dbname - the name of the database with the original data. e.g. unielwin
+     * @param username - the username to use when accessing the database.
+     * @param password - the password to use when accessing the database.
+     * @throws SQLException if there is a problem connecting to the required databases.
+     */
+    public MySQLFactorBaseDataBase(String dbaddress, String dbname, String username, String password) throws DataBaseException {
+        this.baseDatabaseName = dbname;
+        String baseConnectionString = MessageFormat.format(CONNECTION_STRING, dbaddress, dbname);
+
+        try {
+            this.baseConnection = (Connection) DriverManager.getConnection(baseConnectionString, username, password);
+        } catch (SQLException e) {
+            throw new DataBaseException("Unable to connect to the provided database.", e);
+        }
+    }
+
+
+    @Override
+    public void setupDatabase() throws DataBaseException {
+        BZScriptRunner bzsr = new BZScriptRunner(this.baseDatabaseName, this.baseConnection);
+        try {
+            bzsr.runScript(Config.SCRIPTS_DIRECTORY + "setup.sql");
+            bzsr.createSP(Config.SCRIPTS_DIRECTORY + "storedprocs.sql");
+            bzsr.callSP("find_values");
+        } catch (SQLException | IOException e) {
+            throw new DataBaseException("An error occurred when attempting to setup the database for FactorBase.", e);
+        }
+    }
+}

code/factorbase/src/main/java/ca/sfu/cs/factorbase/exception/DataBaseException.java

@@ -0,0 +1,26 @@
+package ca.sfu.cs.factorbase.exception;
+
+
+/**
+ * Custom exception to wrap and throw the specific cause of an error.  Should be thrown when an
+ * error occurs when attempting to access a database.
+ */
+public class DataBaseException extends Exception {
+
+
+    /**
+     * Serial version - auto-generated by Eclipse.
+     */
+    private static final long serialVersionUID = -6623897407612813075L;
+
+
+    /**
+     * Used to wrap the root cause of an exception when attempting to access a database.
+     *
+     * @param message - useful message telling the user what has potentially gone wrong.
+     * @param cause - the exception that caused the error to occur when accessing a database.
+     */
+    public DataBaseException(String message, Throwable cause) {
+        super(message, cause);
+    }
+}

travis-resources/config.cfg

@@ -2,27 +2,15 @@ dbaddress = mysql://127.0.0.1
 dbusername = root
 dbpassword = 123456
 
-dbBNschema = unielwin_BN
-dbDataSchema = unielwin
-dbOutputSchema = unielwin_db
-pathBayesNet = Path_BayesNets
-
 LinkAnalysis = 1
 
 Continuous = 0
-Grounded = 0
 
 dbname = unielwin
 AutomaticSetup = 1
 LinkCorrelations = 1
 ComputeKLD = 0
 
-dbschema = unielwin_BN
-dbcounts = unielwin_db
-dbcondprob = unielwin_db
-
 UseLocal_CT = 0
 
-CrossValidation = 0
-
 LoggingLevel = info