DfsFsck: reduce memory usage during verifyIndex

Don't convert a lot of ObjectId to String stored in generic
java.util.HashSet.  This is a very expensive way to store objects.

Instead rely on "this" from the FsckPackParser to lookup information
about the objects in this pack file, which lets the verify code avoid
sorting the object list.

Use ObjectIdOwnerMap, which is the most efficient format JGit has
for storing lots of objects.

Change-Id: Ib68f93acb4d91b96d0a44c0612f704500d332ac1

2 files changed, 15 insertions, 9 deletions

diff --git a/org.eclipse.jgit/src/org/eclipse/jgit/internal/fsck/FsckPackParser.java b/org.eclipse.jgit/src/org/eclipse/jgit/internal/fsck/FsckPackParser.java
index e6ec6814b3..184bf416e0 100644
--- a/org.eclipse.jgit/src/org/eclipse/jgit/internal/fsck/FsckPackParser.java
+++ b/org.eclipse.jgit/src/org/eclipse/jgit/internal/fsck/FsckPackParser.java
@@ -49,7 +49,6 @@ import java.nio.channels.Channels;
 import java.text.MessageFormat;
 import java.util.Arrays;
 import java.util.HashSet;
-import java.util.List;
 import java.util.Set;
 import java.util.zip.CRC32;
 
@@ -65,6 +64,7 @@ import org.eclipse.jgit.internal.storage.file.PackIndex.MutableEntry;
 import org.eclipse.jgit.lib.AnyObjectId;
 import org.eclipse.jgit.lib.ObjectChecker;
 import org.eclipse.jgit.lib.ObjectDatabase;
+import org.eclipse.jgit.lib.ObjectIdOwnerMap;
 import org.eclipse.jgit.transport.PackParser;
 import org.eclipse.jgit.transport.PackedObjectInfo;
 
@@ -265,18 +265,18 @@ public class FsckPackParser extends PackParser {
 	/**
 	 * Verify the existing index file with all objects from the pack.
 	 *
-	 * @param entries
-	 *            all the entries that are expected in the index file
 	 * @param idx
 	 *            index file associate with the pack
 	 * @throws CorruptPackIndexException
 	 *             when the index file is corrupt.
 	 */
-	public void verifyIndex(List<PackedObjectInfo> entries, PackIndex idx)
+	public void verifyIndex(PackIndex idx)
 			throws CorruptPackIndexException {
-		Set<String> all = new HashSet<>();
-		for (PackedObjectInfo entry : entries) {
-			all.add(entry.getName());
+		ObjectIdOwnerMap<ObjFromPack> inPack = new ObjectIdOwnerMap<>();
+		for (int i = 0; i < getObjectCount(); i++) {
+			PackedObjectInfo entry = getObject(i);
+			inPack.add(new ObjFromPack(entry));
+
 			long offset = idx.findOffset(entry);
 			if (offset == -1) {
 				throw new CorruptPackIndexException(
@@ -305,7 +305,7 @@ public class FsckPackParser extends PackParser {
 		}
 
 		for (MutableEntry entry : idx) {
-			if (!all.contains(entry.name())) {
+			if (!inPack.contains(entry.toObjectId())) {
 				throw new CorruptPackIndexException(MessageFormat.format(
 						JGitText.get().unknownObjectInIndex, entry.name()),
 						ErrorType.UNKNOWN_OBJ);
@@ -323,4 +323,10 @@ public class FsckPackParser extends PackParser {
 	public void overwriteObjectCount(long expectedObjectCount) {
 		this.expectedObjectCount = expectedObjectCount;
 	}
+
+	static class ObjFromPack extends ObjectIdOwnerMap.Entry {
+		ObjFromPack(AnyObjectId id) {
+			super(id);
+		}
+	}
 }
diff --git a/org.eclipse.jgit/src/org/eclipse/jgit/internal/storage/dfs/DfsFsck.java b/org.eclipse.jgit/src/org/eclipse/jgit/internal/storage/dfs/DfsFsck.java
index 2580047b15..75eade2273 100644
--- a/org.eclipse.jgit/src/org/eclipse/jgit/internal/storage/dfs/DfsFsck.java
+++ b/org.eclipse.jgit/src/org/eclipse/jgit/internal/storage/dfs/DfsFsck.java
@@ -128,7 +128,7 @@ public class DfsFsck {
 		fpp.parse(pm);
 		errors.getCorruptObjects().addAll(fpp.getCorruptObjects());
 
-		fpp.verifyIndex(fpp.getSortedObjectList(null), pack.getPackIndex(ctx));
+		fpp.verifyIndex(pack.getPackIndex(ctx));
 	}
 
 	private void checkConnectivity(ProgressMonitor pm, FsckError errors)