EolCanonicalizingInputStream: binary detection should be optional

EolCanonicalizingInputStream may also be used in combination with
.gitattributes. If .gitattributes states that a file is of type text, line
endings have to be canonicalized even if the actual file content seems
to be binary.

Change-Id: Ie4ccdfc5cb91fbd55e06f51146cf5c7c84b8e18b

3 files changed, 30 insertions, 17 deletions

diff --git a/org.eclipse.jgit.test/tst/org/eclipse/jgit/util/io/EolCanonicalizingInputStreamTest.java b/org.eclipse.jgit.test/tst/org/eclipse/jgit/util/io/EolCanonicalizingInputStreamTest.java
index 960ca62411..52ad0139c6 100644
--- a/org.eclipse.jgit.test/tst/org/eclipse/jgit/util/io/EolCanonicalizingInputStreamTest.java
+++ b/org.eclipse.jgit.test/tst/org/eclipse/jgit/util/io/EolCanonicalizingInputStreamTest.java
@@ -57,35 +57,46 @@ public class EolCanonicalizingInputStreamTest {
 	@Test
 	public void testLF() throws IOException {
 		final byte[] bytes = asBytes("1\n2\n3");
-		test(bytes, bytes);
+		test(bytes, bytes, false);
 	}
 
 	@Test
 	public void testCR() throws IOException {
 		final byte[] bytes = asBytes("1\r2\r3");
-		test(bytes, bytes);
+		test(bytes, bytes, false);
 	}
 
 	@Test
 	public void testCRLF() throws IOException {
-		test(asBytes("1\r\n2\r\n3"), asBytes("1\n2\n3"));
+		test(asBytes("1\r\n2\r\n3"), asBytes("1\n2\n3"), false);
 	}
 
 	@Test
 	public void testLFCR() throws IOException {
 		final byte[] bytes = asBytes("1\n\r2\n\r3");
-		test(bytes, bytes);
+		test(bytes, bytes, false);
 	}
 
 	@Test
 	public void testEmpty() throws IOException {
 		final byte[] bytes = asBytes("");
-		test(bytes, bytes);
+		test(bytes, bytes, false);
 	}
 
-	private void test(byte[] input, byte[] expected) throws IOException {
+	@Test
+	public void testBinaryDetect() throws IOException {
+		final byte[] bytes = asBytes("1\r\n2\r\n3\0");
+		test(bytes, bytes, true);
+	}
+
+	@Test
+	public void testBinaryDontDetect() throws IOException {
+		test(asBytes("1\r\n2\r\n3\0"), asBytes("1\n2\n3\0"), false);
+	}
+
+	private void test(byte[] input, byte[] expected, boolean detectBinary) throws IOException {
 		final InputStream bis1 = new ByteArrayInputStream(input);
-		final InputStream cis1 = new EolCanonicalizingInputStream(bis1);
+		final InputStream cis1 = new EolCanonicalizingInputStream(bis1, detectBinary);
 		int index1 = 0;
 		for (int b = cis1.read(); b != -1; b = cis1.read()) {
 			assertEquals(expected[index1], (byte) b);
@@ -97,7 +108,7 @@ public class EolCanonicalizingInputStreamTest {
 		for (int bufferSize = 1; bufferSize < 10; bufferSize++) {
 			final byte[] buffer = new byte[bufferSize];
 			final InputStream bis2 = new ByteArrayInputStream(input);
-			final InputStream cis2 = new EolCanonicalizingInputStream(bis2);
+			final InputStream cis2 = new EolCanonicalizingInputStream(bis2, detectBinary);
 
 			int read = 0;
 			for (int readNow = cis2.read(buffer, 0, buffer.length); readNow != -1
diff --git a/org.eclipse.jgit/src/org/eclipse/jgit/treewalk/WorkingTreeIterator.java b/org.eclipse.jgit/src/org/eclipse/jgit/treewalk/WorkingTreeIterator.java
index 89403993f8..aaaf81b1d7 100644
--- a/org.eclipse.jgit/src/org/eclipse/jgit/treewalk/WorkingTreeIterator.java
+++ b/org.eclipse.jgit/src/org/eclipse/jgit/treewalk/WorkingTreeIterator.java
@@ -402,7 +402,7 @@ public abstract class WorkingTreeIterator extends AbstractTreeIterator {
 	}
 
 	private InputStream filterClean(InputStream in) throws IOException {
-		return new EolCanonicalizingInputStream(in);
+		return new EolCanonicalizingInputStream(in, true);
 	}
 
 	/**
@@ -500,7 +500,7 @@ public abstract class WorkingTreeIterator extends AbstractTreeIterator {
 		InputStream rawis = current().openInputStream();
 		InputStream is;
 		if (getOptions().getAutoCRLF() != AutoCRLF.FALSE)
-			is = new EolCanonicalizingInputStream(rawis);
+			is = new EolCanonicalizingInputStream(rawis, true);
 		else
 			is = rawis;
 		return is;
diff --git a/org.eclipse.jgit/src/org/eclipse/jgit/util/io/EolCanonicalizingInputStream.java b/org.eclipse.jgit/src/org/eclipse/jgit/util/io/EolCanonicalizingInputStream.java
index 387d9b82bf..2492393064 100644
--- a/org.eclipse.jgit/src/org/eclipse/jgit/util/io/EolCanonicalizingInputStream.java
+++ b/org.eclipse.jgit/src/org/eclipse/jgit/util/io/EolCanonicalizingInputStream.java
@@ -49,10 +49,11 @@ import java.io.InputStream;
 import org.eclipse.jgit.diff.RawText;
 
 /**
- * An input stream which canonicalizes EOLs bytes on the fly to '\n', unless the
- * first 8000 bytes indicate the stream is binary.
+ * An input stream which canonicalizes EOLs bytes on the fly to '\n'.
  *
- * Note: Make sure to apply this InputStream only to text files!
+ * Optionally, a binary check on the first 8000 bytes is performed
+ * and in case of binary files, canonicalization is turned off
+ * (for the complete file).
  */
 public class EolCanonicalizingInputStream extends InputStream {
 	private final byte[] single = new byte[1];
@@ -67,7 +68,7 @@ public class EolCanonicalizingInputStream extends InputStream {
 
 	private boolean isBinary;
 
-	private boolean modeDetected;
+	private boolean detectBinary;
 
 	/**
 	 * Creates a new InputStream, wrapping the specified stream
@@ -75,8 +76,9 @@ public class EolCanonicalizingInputStream extends InputStream {
 	 * @param in
 	 *            raw input stream
 	 */
-	public EolCanonicalizingInputStream(InputStream in) {
+	public EolCanonicalizingInputStream(InputStream in, boolean detectBinary) {
 		this.in = in;
+		this.detectBinary = detectBinary;
 	}
 
 	@Override
@@ -132,9 +134,9 @@ public class EolCanonicalizingInputStream extends InputStream {
 		cnt = in.read(buf, 0, buf.length);
 		if (cnt < 1)
 			return false;
-		if (!modeDetected) {
+		if (detectBinary) {
 			isBinary = RawText.isBinary(buf, cnt);
-			modeDetected = true;
+			detectBinary = false;
 		}
 		ptr = 0;
 		return true;