You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@commons.apache.org by "X Liu (Jira)" <ji...@apache.org> on 2023/10/27 17:13:00 UTC
[jira] [Updated] (IO-818) NullInputStream breaks InputStream's read method contract
[ https://issues.apache.org/jira/browse/IO-818?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
X Liu updated IO-818:
---------------------
Description:
A java.io.InputStream instance's read(...) method will always return -1 after it reaches its end.
However, after a NullInputStream instance reaches its end, if you call its read(...) method again, you'll get IOException with message {{{}Read after end of file{}}}. It's document says the read(...) method should {{{}return -1 if the end of file has been reached and throwEofException is set to false{}}}. However, its code doesn't check throwEofException flag before it throws the IOException:
{noformat}
public int read() throws IOException {
if (eof) {
throw new IOException("Read after end of file");
}
....
}{noformat}
I'm not sure whether its read method should strictly follow what InputStream does, i.e., always return -1 after it reaches end, or should it check throwEofException flag before it throws an IOException.
Following is a snippet that can reproduce this problem:
{noformat}
import org.apache.commons.io.input.NullInputStream;
import java.io.ByteArrayInputStream;
import java.io.InputStream;
public class InputStreamTest {
private static void testOneStream(InputStream inputStream) throws Exception {
final byte[] buffer = new byte[4096];
int readLen;
while ((readLen = inputStream.read(buffer)) != -1) {
// do nothing
}
readLen = inputStream.read(buffer);
if (readLen != -1) {
throw new Exception("Failed to read after reaching the end of the stream.");
}
}
public static void main(String[] args) {
try { // Test ByteArrayInputStream
final InputStream byteArrayInputStream = new ByteArrayInputStream(new byte[1024]);
testOneStream(byteArrayInputStream);
System.out.println("ByteArrayInputStream succeeded.");
} catch (Exception e) {
System.out.println("ByteArrayInputStream failed");
}
try { //Test NullInputStream
final InputStream nullInputStream = new NullInputStream(1024, true, false);
testOneStream(nullInputStream);
System.out.println("NullInputStream succeeded.");
} catch (Exception e) {
System.out.println("NullInputStream failed");
}
}
}{noformat}
was:
A java.io.InputStream instance's read(...) method will always return -1 after it reaches its end.
However, after a NullInputStream instance reaches its end, if you call its read(...) method again, you'll get IOException with message {{{}Read after end of file{}}}. It's document says the read(...) method should {{{}return -1 if the end of file has been reached and throwEofException is set to false{}}}. However, its code doesn't check throwEofException flag before it throws the IOException:
public int read() throws IOException \{
if (eof) {
throw new IOException("Read after end of file");
}
I'm not sure whether its read method should strictly follow what InputStream does, i.e., always return -1 after it reaches end, or should it check throwEofException flag before it throws an IOException.
Following is a snippet that can reproduce this problem:
import org.apache.commons.io.input.NullInputStream;
import java.io.ByteArrayInputStream;
import java.io.InputStream;
public class InputStreamTest \{
private static void testOneStream(InputStream inputStream) throws Exception {
final byte[] buffer = new byte[4096];
int readLen;
while ((readLen = inputStream.read(buffer)) != -1) {
// do nothing
}
readLen = inputStream.read(buffer);
if (readLen != -1) \{
throw new Exception("Failed to read after reaching the end of the stream.");
}
}
public static void main(String[] args) \{
try { // Test ByteArrayInputStream
final InputStream byteArrayInputStream = new ByteArrayInputStream(new byte[1024]);
testOneStream(byteArrayInputStream);
System.out.println("ByteArrayInputStream succeeded.");
} catch (Exception e) \{
System.out.println("ByteArrayInputStream failed");
}
try \{ //Test NullInputStream
final InputStream nullInputStream = new NullInputStream(1024, true, false);
testOneStream(nullInputStream);
System.out.println("NullInputStream succeeded.");
} catch (Exception e) \{
System.out.println("NullInputStream failed");
}
}
}
> NullInputStream breaks InputStream's read method contract
> ---------------------------------------------------------
>
> Key: IO-818
> URL: https://issues.apache.org/jira/browse/IO-818
> Project: Commons IO
> Issue Type: Bug
> Components: Streams/Writers
> Affects Versions: 2.13.0
> Environment: JDK 17.0
> Apache commons-io-2.13.0, but it looks like this apply to 2.15.0 too.
> Reporter: X Liu
> Priority: Major
>
> A java.io.InputStream instance's read(...) method will always return -1 after it reaches its end.
>
> However, after a NullInputStream instance reaches its end, if you call its read(...) method again, you'll get IOException with message {{{}Read after end of file{}}}. It's document says the read(...) method should {{{}return -1 if the end of file has been reached and throwEofException is set to false{}}}. However, its code doesn't check throwEofException flag before it throws the IOException:
> {noformat}
> public int read() throws IOException {
> if (eof) {
> throw new IOException("Read after end of file");
> }
> ....
> }{noformat}
> I'm not sure whether its read method should strictly follow what InputStream does, i.e., always return -1 after it reaches end, or should it check throwEofException flag before it throws an IOException.
> Following is a snippet that can reproduce this problem:
> {noformat}
> import org.apache.commons.io.input.NullInputStream;
> import java.io.ByteArrayInputStream;
> import java.io.InputStream;
> public class InputStreamTest {
> private static void testOneStream(InputStream inputStream) throws Exception {
> final byte[] buffer = new byte[4096];
> int readLen;
> while ((readLen = inputStream.read(buffer)) != -1) {
> // do nothing
> }
> readLen = inputStream.read(buffer);
> if (readLen != -1) {
> throw new Exception("Failed to read after reaching the end of the stream.");
> }
> }
> public static void main(String[] args) {
> try { // Test ByteArrayInputStream
> final InputStream byteArrayInputStream = new ByteArrayInputStream(new byte[1024]);
> testOneStream(byteArrayInputStream);
> System.out.println("ByteArrayInputStream succeeded.");
> } catch (Exception e) {
> System.out.println("ByteArrayInputStream failed");
> }
> try { //Test NullInputStream
> final InputStream nullInputStream = new NullInputStream(1024, true, false);
> testOneStream(nullInputStream);
> System.out.println("NullInputStream succeeded.");
> } catch (Exception e) {
> System.out.println("NullInputStream failed");
> }
> }
> }{noformat}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)