D-Net

package eu.dnetlib.data.mapreduce.hbase.statsExport.drivers;

import java.io.File;

import java.util.Map.Entry;

import org.apache.hadoop.fs.FileSystem;

import org.apache.hadoop.fs.LocatedFileStatus;

import org.apache.hadoop.fs.Path;

import org.apache.hadoop.fs.RemoteIterator;

import org.apache.sqoop.Sqoop;

public class SqoopDriver {

	private Logger log = Logger.getLogger(this.getClass());

	private Properties tableMappings = new Properties();

	private String dbPass;

	private String connectionUrl;

	// Number of statements in each batch insert op

	private String RecsPerStatement;

	// Number of statements for each commit ( commit every 1000 inserts, for

	// instance). Set to high value to reduce I/O costs.

	private String StatementPerTrans;

	// Number of reducer classes that Sqoop can use.

	private String sqoopReducersCount;

	 * Driver for the Sqoop tool. Calls the Sqoop Client for each <input file,

	 * destination table> pair given in the @tables argument.

	 *

	 * Needed parameters ( connection, database name, etc are set in class

	 * parameters when an new Sqoop Driver instance is created.

	 *

	 * @param tables

	 */

		for (Entry<String, String> table : tables.entrySet()) {

			"--connect", connectionUrl, "--table", table.getKey(),

					// outputs writer appends a \t char after writing the KEY

					// field in HDFS.

					// use this so sqoop can ignore it

					"--optionally-enclosed-by", "	", "--input-fields-terminated-by", delim,

					"--verbose", "--username", dbUser, "--password", dbPass, "--batch", "--mapreduce-job-name", "Sqoop Stats Import", "-m", sqoopReducersCount };

				log.error("Could not run Sqoop Tool " + ex.getMessage());

				throw new Exception("Could not run Sqoop Tool ", ex);

			}

		}

	}

	{

		file.close();

			throw new Exception("Could not load Table Mappings in sqoop init job");

		}

	public void initSqoopJob() throws Exception {

		loadTableMappings();

		if (useHdfsStore) {

			fileNames = listHdfsDir();

		} else {

			fileNames = listFilesystemDir();

		}

		// Table mappings containt the mapping between HDFS files and the Stats

		// DB table that each should be imported to

		for (Entry<Object, Object> e : tableMappings.entrySet()) {

			String name = (String) e.getKey();

			for (String filename : fileNames) {

				String split[] = filename.split("-");

				split[0] = split[0].replaceAll(".*/", "");

				if (split[0].equals(name)) {

					tables.put((String) e.getValue(), filename);

				}

			}

		}

		long startTime = System.currentTimeMillis();

		try {

			this.run(tables);

		} catch (Exception e) {

			log.error("Error while importing tables in Sqoop: ", e);

		long endtime = System.currentTimeMillis();

		log.info("Time taken for Sqoop Import : " + (endtime - startTime) / 60000 + " minutes");

	}

				throw new Exception("No files generated in " + outputPath + " at " + exportPath.getName() + " in " + hdfs.resolvePath(exportPath));

		}

		ArrayList<String> fileNames = new ArrayList<String>();

			log.info(fileName);

			fileNames.add(fileName);

		}

		return fileNames;

	}

	private ArrayList<String> listFilesystemDir() throws Exception {

		ArrayList<String> fileNames = new ArrayList<String>();

		String files;

		File folder = new File(outputPath);

		File[] listOfFiles = folder.listFiles();

			log.error("No files generated in " + outputPath);

			throw new Exception("No files generated in " + outputPath);

		}

			if (listOfFiles[i].isFile()) {

				fileNames.add(files);

				System.out.println(files);

			}

		}

		return fileNames;

	}

	 * Cleans up auto-generated Sqoop class files

	 *

	 * @param table

	 */

	private void cleanUp(String table) {

		try {

			File file = new File(table + ".java");

			if (file.delete()) {

				log.info(file.getName() + " is deleted!");

			} else {

				log.info("Delete operation   failed.");

			}

		} catch (Exception e) {

			log.error("Delete operation   failed" + table);

		}

	}

	public String getConnectionUrl() {

		return connectionUrl;

	}

	public void setConnectionUrl(String connectionUrl) {

		this.connectionUrl = connectionUrl;

	}

	public String getDbUser() {

		return dbUser;

	}

	public void setDbUser(String dbUser) {

		this.dbUser = dbUser;

	}

	public String getDbPass() {

		return dbPass;

	}

	public void setDbPass(String dbPass) {

		this.dbPass = dbPass;

	}

	public String getDelim() {

		return delim;

	}

	public void setDelim(String delim) {

		this.delim = delim;

	}

	public String getReducersCount() {

		return sqoopReducersCount;

	}

	public void setReducersCount(String reducersCount) {

		this.sqoopReducersCount = reducersCount;

	}

	public String getRecsPerStatement() {

		return RecsPerStatement;

	}

	public void setRecsPerStatement(String recsPerStatement) {

		RecsPerStatement = recsPerStatement;

	}

	public String getStatementPerTrans() {

		return StatementPerTrans;

	}

	public void setStatementPerTrans(String statementPerTrans) {

		StatementPerTrans = statementPerTrans;

	}

		return log;

	}

	public void setLog(Logger log) {

		this.log = log;

	}

	public String getTABLE_MAP_PATH() {

		return TABLE_MAP_PATH;

	}

	public void setTABLE_MAP_PATH(String tABLE_MAP_PATH) {

		TABLE_MAP_PATH = tABLE_MAP_PATH;

	}

	public Properties getTableMappings() {

		return tableMappings;

	}

	public void setTableMappings(Properties tableMappings) {

		this.tableMappings = tableMappings;

	}

	public String getOutputPath() {

		return outputPath;

	}

	public void setOutputPath(String outputPath) {

		this.outputPath = outputPath;

	}

	public String getSqoopReducersCount() {

		return sqoopReducersCount;

	}

	public void setSqoopReducersCount(String sqoopReducersCount) {

		this.sqoopReducersCount = sqoopReducersCount;

	}

	public FileSystem getHdfs() {

		return hdfs;

	}

	public void setHdfs(FileSystem hdfs) {

		this.hdfs = hdfs;

	}

		return useHdfsStore;

	}

	public void setUseHdfsStore(boolean useHdfsStore) {

		this.useHdfsStore = useHdfsStore;

	}

		return conf;

	}

	public void setConf(Configuration conf) {

		this.conf = conf;

	}