org.etlunit.io.file.FileDataImpl Maven / Gradle / Ivy

Go to download
Show more of this group Show more artifacts with this name
Show all versions of etlunit-core Show documentation
There is a newer version: 1.6.9
package org.etlunit.io.file;

import java.io.BufferedReader;
import java.io.FileReader;
import java.io.IOException;
import java.util.*;
import java.util.concurrent.Semaphore;

class FileDataImpl implements DataFile.FileData
{
	private final FlatFile flatFile;
	private final List columns;

	private BufferedReader bread;
	Semaphore state = new Semaphore(1);

	public FileDataImpl(FlatFile flatFile, List columns)
	{
		this.flatFile = flatFile;

		if (columns == null)
		{
			this.columns = flatFile.getDataFileSchema().getColumnNames();
		}
		else
		{
			this.columns = columns;
		}

		// verify at least one column is selected
		if (this.columns.size() == 0)
		{
			throw new IllegalArgumentException("At least one column must be specified in the select list");
		}
	}

	@Override
	public Iterator iterator() throws IOException
	{
		if (bread != null)
		{
			throw new IOException("One per costumer, please (I.E., don't call iterator() twice)");
		}

		bread = new BufferedReader(new FileReader(flatFile.getSource()), 16384);

		Semaphore state = new Semaphore(1);

		return new DataIterator();
	}

	@Override
	public void dispose() throws IOException
	{
		if (state.availablePermits() == 1)
		{
			bread.close();
		}
	}

	private class DataIterator implements Iterator
	{
		String nullToken;
		String columnDelimiter;

		String nextLine;
		private final Map lineData;
		private final Map publicLineData;

		public DataIterator()
		{
			nullToken = flatFile.getDataFileSchema().getNullToken();
			columnDelimiter = flatFile.getDataFileSchema().getColumnDelimiter();
			nextLine = null;
			lineData = new HashMap();
			publicLineData = Collections.unmodifiableMap(lineData);
		}

		public boolean hasNext()
		{
			if (state.availablePermits() == 0)
			{
				return false;
			}

			if (nextLine != null)
			{
				return true;
			}

			while (state.availablePermits() == 1)
			{
				try
				{
					nextLine = FlatFile.readLine(bread, flatFile.getDataFileSchema().getRowDelimiter());

					if (nextLine == null)
					{
						bread.close();
						state.acquireUninterruptibly();
					}
					else
					{
						String trim = nextLine.trim();

						if (!trim.equals("") && !(trim.startsWith("/*") && trim.endsWith("*/")) && !trim.startsWith("#"))
						{
							break;
						}
					}
				}
				catch (IOException e)
				{
					throw new RuntimeException(e);
				}
			}

			return nextLine != null;
		}

		public DataFile.FileRow next()
		{
			if (nextLine == null)
			{
				throw new IllegalStateException("Next line does not exist");
			}

			// convert the line into columns
			lineData.clear();

			if (flatFile.getDataFileSchema() != null)
			{
				// defer mapping columns to the schema
				lineData.putAll(flatFile.getDataFileSchema().validateAndSplitLine(nextLine));
			}
			else
			{
				String[] columns = nextLine.split(columnDelimiter);

				for (int i = 0; i < columns.length; i++)
				{
					if (nullToken.equals(columns[i]))
					{
						lineData.put(flatFile.getDataFileSchema().getColumnNames().get(i), null);
					}
					else
					{
						lineData.put(flatFile.getDataFileSchema().getColumnNames().get(i), columns[i]);
					}
				}
			}

			if (lineData.size() != flatFile.getDataFileSchema().getColumnNames().size())
			{
				throw new IllegalStateException("Line has the wrong number of columns.  Required["
						+ flatFile.getDataFileSchema().getColumnNames().size()
						+ "], actual["
						+ lineData.size()
						+ "] - source '"
						+ nextLine
						+ "'");
			}

			// remove unwanted columns - this is ugly but better than doing it twice above in two different handlers
			Set> eSet = lineData.entrySet();

			Iterator> eSi = eSet.iterator();

			while (eSi.hasNext())
			{
				if (!columns.contains(eSi.next().getKey()))
				{
					// get rid of it
					eSi.remove();
				}
			}

			// this method also needs to know about columns
			final OrderKey orderKey = FlatFile.addOrderKey(flatFile.getDataFileSchema(), (Map) lineData, columns);

			nextLine = null;

			return new DataFile.FileRow()
			{
				@Override
				public Map getData()
				{
					return publicLineData;
				}

				@Override
				public OrderKey getOrderKey()
				{
					return orderKey;
				}
			};
		}

		public void remove()
		{
			throw new UnsupportedOperationException();
		}
	}
}