Delay locating .gitattributes until requested

Instead of checking every entry for .gitattributes only look for the
entry on request by TreeWalk.  This avoids impacting uses like RevWalk
filtering history.

When the attrs is requested skip to the start of the tree and look for
.gitattributes until either it is found, or it is impossible to be
present.  Due to the sorting rules of tree entries .gitattributes
should be among the first or second entries in the tree so very few
entries will need to be considered.

Waiting to find the .gitattributes file by native ordering may miss
attrs for files like .config, which sorts before .gitattributes.
Starting from the front of the tree on demand ensures the attributes
are parsed as early as necessary to process any entry in the tree.

Due to TreeWalk recursively processing up the tree of iterators we
cannot just reset the current CanonicalTreeParser to the start as
parent parsers share the same path buffer as their children.
Resetting a parent to look for .gitattributes may overwrite path
buffer data used by a child iterator.  Work around this by building a
new temporary CanonicalTreeParser instance.

Change-Id: Ife950253b687be325340d27e9915c9a40df2641c

org.eclipse.jgit.test/tst/org/eclipse/jgit/treewalk/CanonicalTreeParserTest.java

@@ -43,6 +43,8 @@
 
 package org.eclipse.jgit.treewalk;
 
+import static org.eclipse.jgit.lib.FileMode.REGULAR_FILE;
+import static org.eclipse.jgit.lib.FileMode.SYMLINK;
 import static org.junit.Assert.assertEquals;
 import static org.junit.Assert.assertFalse;
 import static org.junit.Assert.assertSame;
@@ -50,9 +52,11 @@ import static org.junit.Assert.assertTrue;
 
 import java.io.ByteArrayOutputStream;
 
+import org.eclipse.jgit.errors.CorruptObjectException;
 import org.eclipse.jgit.lib.Constants;
 import org.eclipse.jgit.lib.FileMode;
 import org.eclipse.jgit.lib.ObjectId;
+import org.eclipse.jgit.lib.TreeFormatter;
 import org.eclipse.jgit.util.RawParseUtils;
 import org.junit.Before;
 import org.junit.Test;
@@ -369,4 +373,41 @@ public class CanonicalTreeParserTest {
 		assertEquals(name, RawParseUtils.decode(Constants.CHARSET, ctp.path,
 				ctp.pathOffset, ctp.pathLen));
 	}
+
+	@Test
+	public void testFindAttributesWhenFirst() throws CorruptObjectException {
+		TreeFormatter tree = new TreeFormatter();
+		tree.append(".gitattributes", REGULAR_FILE, hash_a);
+		ctp.reset(tree.toByteArray());
+
+		assertTrue(ctp.findFile(".gitattributes"));
+		assertEquals(REGULAR_FILE.getBits(), ctp.getEntryRawMode());
+		assertEquals(".gitattributes", ctp.getEntryPathString());
+		assertEquals(hash_a, ctp.getEntryObjectId());
+	}
+
+	@Test
+	public void testFindAttributesWhenSecond() throws CorruptObjectException {
+		TreeFormatter tree = new TreeFormatter();
+		tree.append(".config", SYMLINK, hash_a);
+		tree.append(".gitattributes", REGULAR_FILE, hash_foo);
+		ctp.reset(tree.toByteArray());
+
+		assertTrue(ctp.findFile(".gitattributes"));
+		assertEquals(REGULAR_FILE.getBits(), ctp.getEntryRawMode());
+		assertEquals(".gitattributes", ctp.getEntryPathString());
+		assertEquals(hash_foo, ctp.getEntryObjectId());
+	}
+
+	@Test
+	public void testFindAttributesWhenMissing() throws CorruptObjectException {
+		TreeFormatter tree = new TreeFormatter();
+		tree.append("src", REGULAR_FILE, hash_a);
+		tree.append("zoo", REGULAR_FILE, hash_foo);
+		ctp.reset(tree.toByteArray());
+
+		assertFalse(ctp.findFile(".gitattributes"));
+		assertEquals(11, ctp.idOffset()); // Did not walk the entire tree.
+		assertEquals("src", ctp.getEntryPathString());
+	}
 }

org.eclipse.jgit/src/org/eclipse/jgit/treewalk/AbstractTreeIterator.java

@@ -327,6 +327,42 @@ public abstract class AbstractTreeIterator {
 		return pathCompare(p.path, cPos, p.pathLen, pMode, cPos);
 	}
 
+	/**
+	 * Seek the iterator on a file, if present.
+	 *
+	 * @param name
+	 *            file name to find (will not find a directory).
+	 * @return true if the file exists in this tree; false otherwise.
+	 * @throws CorruptObjectException
+	 *             tree is invalid.
+	 * @since 4.2
+	 */
+	public boolean findFile(String name) throws CorruptObjectException {
+		return findFile(Constants.encode(name));
+	}
+
+	/**
+	 * Seek the iterator on a file, if present.
+	 *
+	 * @param name
+	 *            file name to find (will not find a directory).
+	 * @return true if the file exists in this tree; false otherwise.
+	 * @throws CorruptObjectException
+	 *             tree is invalid.
+	 * @since 4.2
+	 */
+	public boolean findFile(byte[] name) throws CorruptObjectException {
+		for (; !eof(); next(1)) {
+			int cmp = pathCompare(name, 0, name.length, 0, pathOffset);
+			if (cmp == 0) {
+				return true;
+			} else if (cmp > 0) {
+				return false;
+			}
+		}
+		return false;
+	}
+
 	/**
 	 * Compare the path of this current entry to a raw buffer.
 	 *

org.eclipse.jgit/src/org/eclipse/jgit/treewalk/CanonicalTreeParser.java

@@ -44,6 +44,13 @@
 
 package org.eclipse.jgit.treewalk;
 
+import static org.eclipse.jgit.lib.Constants.DOT_GIT_ATTRIBUTES;
+import static org.eclipse.jgit.lib.Constants.OBJECT_ID_LENGTH;
+import static org.eclipse.jgit.lib.Constants.OBJ_BLOB;
+import static org.eclipse.jgit.lib.Constants.OBJ_TREE;
+import static org.eclipse.jgit.lib.Constants.TYPE_TREE;
+import static org.eclipse.jgit.lib.Constants.encode;
+
 import java.io.IOException;
 import java.io.InputStream;
 import java.util.Arrays;
@@ -54,20 +61,15 @@ import org.eclipse.jgit.attributes.AttributesRule;
 import org.eclipse.jgit.errors.IncorrectObjectTypeException;
 import org.eclipse.jgit.errors.MissingObjectException;
 import org.eclipse.jgit.lib.AnyObjectId;
-import org.eclipse.jgit.lib.Constants;
 import org.eclipse.jgit.lib.FileMode;
 import org.eclipse.jgit.lib.MutableObjectId;
 import org.eclipse.jgit.lib.ObjectId;
-import org.eclipse.jgit.lib.ObjectLoader;
 import org.eclipse.jgit.lib.ObjectReader;
-import org.eclipse.jgit.util.RawParseUtils;
 
 /** Parses raw Git trees from the canonical semi-text/semi-binary format. */
 public class CanonicalTreeParser extends AbstractTreeIterator {
 	private static final byte[] EMPTY = {};
-
-	private static final byte[] ATTRS = Constants
-			.encode(Constants.DOT_GIT_ATTRIBUTES);
+	private static final byte[] ATTRS = encode(DOT_GIT_ATTRIBUTES);
 
 	private byte[] raw;
 
@@ -133,6 +135,7 @@ public class CanonicalTreeParser extends AbstractTreeIterator {
 	 *            the raw tree content.
 	 */
 	public void reset(final byte[] treeData) {
+		attributesNode = null;
 		raw = treeData;
 		prevPtr = -1;
 		currPtr = 0;
@@ -208,7 +211,7 @@ public class CanonicalTreeParser extends AbstractTreeIterator {
 	 */
 	public void reset(final ObjectReader reader, final AnyObjectId id)
 			throws IncorrectObjectTypeException, IOException {
-		reset(reader.open(id, Constants.OBJ_TREE).getCachedBytes());
+		reset(reader.open(id, OBJ_TREE).getCachedBytes());
 	}
 
 	@Override
@@ -218,7 +221,7 @@ public class CanonicalTreeParser extends AbstractTreeIterator {
 		idBuffer.fromRaw(idBuffer(), idOffset());
 		if (!FileMode.TREE.equals(mode)) {
 			final ObjectId me = idBuffer.toObjectId();
-			throw new IncorrectObjectTypeException(me, Constants.TYPE_TREE);
+			throw new IncorrectObjectTypeException(me, TYPE_TREE);
 		}
 		return createSubtreeIterator0(reader, idBuffer);
 	}
@@ -263,7 +266,7 @@ public class CanonicalTreeParser extends AbstractTreeIterator {
 
 	@Override
 	public int idOffset() {
-		return nextPtr - Constants.OBJECT_ID_LENGTH;
+		return nextPtr - OBJECT_ID_LENGTH;
 	}
 
 	@Override
@@ -301,7 +304,7 @@ public class CanonicalTreeParser extends AbstractTreeIterator {
 			prevPtr = ptr;
 			while (raw[ptr] != 0)
 				ptr++;
-			ptr += Constants.OBJECT_ID_LENGTH + 1;
+			ptr += OBJECT_ID_LENGTH + 1;
 		}
 		if (delta != 0)
 			throw new ArrayIndexOutOfBoundsException(delta);
@@ -337,7 +340,7 @@ public class CanonicalTreeParser extends AbstractTreeIterator {
 			trace[delta] = ptr;
 			while (raw[ptr] != 0)
 				ptr++;
-			ptr += Constants.OBJECT_ID_LENGTH + 1;
+			ptr += OBJECT_ID_LENGTH + 1;
 		}
 		if (trace[1] == -1)
 			throw new ArrayIndexOutOfBoundsException(delta);
@@ -372,12 +375,7 @@ public class CanonicalTreeParser extends AbstractTreeIterator {
 			}
 		}
 		pathLen = tmp;
-		nextPtr = ptr + Constants.OBJECT_ID_LENGTH;
-
-		// Check if this entry is a .gitattributes file
-		if (path[pathOffset] == '.'
-				&& RawParseUtils.match(path, pathOffset, ATTRS) > 0)
-			attributesNode = new LazyLoadingAttributesNode(idOffset());
+		nextPtr = ptr + OBJECT_ID_LENGTH;
 	}
 
 	/**
@@ -391,36 +389,32 @@ public class CanonicalTreeParser extends AbstractTreeIterator {
 	 */
 	public AttributesNode getEntryAttributesNode(ObjectReader reader)
 			throws IOException {
-		if (attributesNode instanceof LazyLoadingAttributesNode)
-			attributesNode = ((LazyLoadingAttributesNode) attributesNode)
-					.load(reader);
-		return attributesNode;
+		if (attributesNode == null) {
+			attributesNode = findAttributes(reader);
+		}
+		return attributesNode.getRules().isEmpty() ? null : attributesNode;
 	}
 
-	/**
-	 * {@link AttributesNode} implementation that provides lazy loading
-	 */
-	private class LazyLoadingAttributesNode extends AttributesNode {
-		private final int idOffset;
-
-		LazyLoadingAttributesNode(int idOffset) {
-			super(Collections.<AttributesRule> emptyList());
-			this.idOffset = idOffset;
+	private AttributesNode findAttributes(ObjectReader reader)
+			throws IOException {
+		CanonicalTreeParser itr = new CanonicalTreeParser();
+		itr.reset(raw);
+		if (itr.findFile(ATTRS)) {
+			return loadAttributes(reader, itr.getEntryObjectId());
 		}
+		return noAttributes();
+	}
 
-		AttributesNode load(ObjectReader reader) throws IOException {
-			AttributesNode r = new AttributesNode();
-			ObjectId id = ObjectId.fromRaw(raw, idOffset);
-			ObjectLoader loader = reader.open(id);
-			if (loader != null) {
-				InputStream in = loader.openStream();
-				try {
-					r.parse(in);
-				} finally {
-					in.close();
-				}
-			}
-			return r.getRules().isEmpty() ? null : r;
+	private static AttributesNode loadAttributes(ObjectReader reader,
+			AnyObjectId id) throws IOException {
+		AttributesNode r = new AttributesNode();
+		try (InputStream in = reader.open(id, OBJ_BLOB).openStream()) {
+			r.parse(in);
 		}
+		return r.getRules().isEmpty() ? noAttributes() : r;
+	}
+
+	private static AttributesNode noAttributes() {
+		return new AttributesNode(Collections.<AttributesRule> emptyList());
 	}
 }