Ignore Header and Trailer Lines During Java InputStream Read

Ignore first + last lines in a file

After beating my head against the wall for some time, I was finally able to come up with a solution for this.

Say you want to hand off an InputStream or InputStreamReader to an underlying API. The problem is that your input stream may have a header or trailer. Those header/trailers differ from the underlying format. This is commonly true for CSV, EDI, and fixed length file types. So if you want a CSV file parsed by a Java API like the Smooks api, you will need to hand the framework an InputStream.

To resolve this issue, you need to create your own class that extends the FilterInputStream in Java. This will allow you to get between the framework API and the core InputStream. The code below is an example Input stream that buffers each line in a file. It then consumes the line if it is determined the header or trailer line in the file.

import java.io.FilterInputStream;
import java.io.IOException;
import java.io.InputStream;
import java.util.LinkedList;
import java.util.Queue;

 * Used for filtering (consuming) header and trailer lines from Input stream.
public class FilterHeaderTrailerInputStream extends FilterInputStream
    private Queue<Integer> readBuffer = new LinkedList <Integer>();
    int lastByte;
    boolean hasReadHeader;

    public FilterHeaderTrailerInputStream(InputStream inputStream)
        this.lastByte = -1;
        this.hasReadHeader = false;

    public int read(byte[] bytes) throws IOException {
        return read(bytes, 0, bytes.length) ;

    public int read(byte[] bytes, int offset, int length) throws IOException {
        int bytesRead = 0;
        final byte[] readBytes = new byte[length - offset];
        while (bytesRead < length - offset) {
            final int readByte = read();
            if (readByte != -1) {
                readBytes[bytesRead++] = (byte) readByte;
            else {
        if (bytesRead == 0) {
            return -1;
        System.arraycopy(readBytes, 0, bytes, offset, bytesRead);
        return bytesRead;

    public int read() throws IOException {

        // The buffer has stuff in it, so serve it off.
        if (!this.readBuffer.isEmpty()) {
            return readBuffer.remove();

        //A previous buffered read contained an extra byte. Place it on the queue.
        if(lastByte != -1){

        //Read bytes until reaching a line feed or end of file
        int currentRead;
            currentRead = super.read();
        }while (currentRead != '\n' && currentRead != '\r' && currentRead != -1);

        //The byte after a newline can either be another '\n', the first byte on the next line or 
        //there are no more bytes left.
         int afterLineByte = currentRead == -1 ? currentRead : super.read();

        //The byte signifies end of file. Guaranteed to have the trailer line in the buffer.
        if(afterLineByte == -1){
            readBuffer.clear(); // Clear out the buffer as to consume the line
            return -1; //return end of file indicator
        //A double line break is found. Add it to the read buffer. 
        //If the next bite is not available, end of file is reached.
        else if(afterLineByte == '\n'){
            if(super.available() == 0){//end of file.  Guaranteed to have the trailer line in the buffer.
                readBuffer.clear();// Clear out the buffer as to consume the line 
            //Hold the last byte. Will be added to the queue on the next recursive call.
            lastByte = afterLineByte;

        //When header hasn't been read, take the current line and consume it. Set indicator that header's been read.
            hasReadHeader = true;

        return this.read();

And this would be an example use case
FilterHeaderTrailerInputStream inputStream = 
        new FilterHeaderTrailerInputStream(

Dealing with the case of '\r\n' was the most difficult piece. Overall, this makes a difficult problem trivial. Hope it helps.


Popular posts from this blog

Atmosphere Websockets & Comet with Spring MVC