java - 如何从 CSV 读取 H2 结果集，并重置光标以进行第二遍

Question

我正在编写一个 Java 库，用于从 CSV 文件导入数据，并根据一些正则表达式测试的结果以 INTEGER、DECIMAL、VARCHAR 和其他列类型返回结果。

因此，我需要读取 Csv.read() 返回的 ResultSet 两次；一次确定列类型，然后再次填充新的 SimpleResultSet (或者最好将原始 ResultSet 包装在 SimpleRowSource 中，所以我不'不必存储表数据)。我尝试调用 ResultSet.beforeFirst()，但第二遍没有可用行。

// Create an input stream with test data.
String data =
    "name,value\n" +
    "one,1\n" +
    "two,2\n" +
    "three,3\n";
InputStream stream = new ByteArrayInputStream(data.getBytes());
Reader reader = new InputStreamReader(stream);

// Read the test data into a new ResultSet.
Csv csv = new Csv();
ResultSet resultSet = csv.read(reader, null);

// Iterate over the ResultSet on first pass.
int firstPass = 0;
while (resultSet.next())
{
    // Test all values for integer, decimal etc.
    ++firstPass;
}
System.out.println("Read " + firstPass + " row(s) on first pass");

// Move the cursor to before the first row; this doesn't work!
resultSet.beforeFirst();

// Check if the cursor is before the first row; throws exception!
/* resultSet.isBeforeFirst(); */

// Iterate over the ResultSet on second pass.
int secondPass = 0;
while (resultSet.next())
{
    // Add all values to new SimpleResultSet, or wrap this code
    // in an H2 SimpleRowSource, and pass to the constructor of
    // a new SimpleResultSet.
    ++secondPass;
}
System.out.println("Read " + secondPass + " row(s) on second pass");

这会产生以下输出:

Read 3 row(s) on first pass
Read 0 row(s) on second pass

我注意到 beforeFirst() 不会抛出异常，但 isBeforeFirst() 会抛出异常:

org.h2.jdbc.JdbcSQLException: Feature not supported: null [50100-172]

如果有任何关于为什么这行不通的建议，或者其他方法的建议，我将不胜感激!

更新

由于 H2 Csv.read() 函数生成一个无法回滚的 ResultSet，因此我最终复制了源 InputStream，因为它是在第一遍读取，然后在第二遍重播生成的 byte[]。这样，我就可以确保导入相同的数据。这是我为此编写的辅助类:

public class CopyStream extends InputStream
{
    private final InputStream mInputStream;
    private final ByteArrayOutputStream mOutputStream;

    public CopyStream(InputStream inputStream)
    {
        // Initialise the class.
        mInputStream = inputStream;
        mOutputStream = new ByteArrayOutputStream();
    }

    public InputStream copy()
    {
        // Return a new input stream based on the copied data.
        byte[] buffer = mOutputStream.toByteArray();
        return new ByteArrayInputStream(buffer);
    }

    @Override
    public int read() throws IOException
    {
        // Read a byte from the input stream.
        int b = mInputStream.read();
        if (b >= 0)
        {
            // Copy the byte to the output stream.
            mOutputStream.write(b);
        }
        return b;
    }

    @Override
    public int read(byte[] buffer) throws IOException
    {
        // Read some bytes from the input stream.
        int length = mInputStream.read(buffer);
        if (length > 0)
        {
            // Copy the bytes to the output stream.
            mOutputStream.write(buffer, 0, length);
        }
        return length;
    }

    @Override
    public int read(byte[] buffer, int offset, int length) throws IOException
    {
        // Read some bytes from the input stream.
        length = mInputStream.read(buffer, offset, length);
        if (length > 0)
        {
            // Copy the bytes to the output stream.
            mOutputStream.write(buffer, offset, length);
        }
        return length;
    }
}

其用法如下:

Csv csv = new Csv();
CopyStream copyStream = new CopyStream(inputStream);
ResultSet resultSet1 = csv.read(new InputStreamReader(copyStream), null);
ResultSet resultSet2 = csv.read(new InputStreamReader(copyStream.copy()), null);

Answer 1

您不需要读取 ResultSet 两次。 您可以使用非标准 H2 命令 CSVREAD 从文件填充表格。

或者， 您可以再次创建 Csv 对象。