Fix Files.walk behavior

jean-philippe-martin · jean-philippe-martin · commit 9c8e6a201e8a · 2017-06-07T15:18:56.000-07:00
Java's Files.walk() ends up in an infinite loop when iterating
over files on GCS because the iterator returns the directory
itself as the first result, so Files.walk() tries to cd there
and we're stuck.

The fix is to check for that path and not return it.
diff --git a/google-cloud-contrib/google-cloud-nio/src/main/java/com/google/cloud/storage/contrib/nio/CloudStorageFileSystemProvider.java b/google-cloud-contrib/google-cloud-nio/src/main/java/com/google/cloud/storage/contrib/nio/CloudStorageFileSystemProvider.java
@@ -97,9 +97,12 @@ private static class LazyPathIterator extends AbstractIterator<Path> {
     private final Iterator<Blob> blobIterator;
     private final Filter<? super Path> filter;
     private final CloudStorageFileSystem fileSystem;
+    private final String prefix;
 
-    LazyPathIterator(CloudStorageFileSystem fileSystem, Iterator<Blob> blobIterator,
+    LazyPathIterator(CloudStorageFileSystem fileSystem, String prefix,
+                     Iterator<Blob> blobIterator,
                      Filter<? super Path> filter) {
+      this.prefix = prefix;
       this.blobIterator = blobIterator;
       this.filter = filter;
       this.fileSystem = fileSystem;
@@ -110,6 +113,10 @@ protected Path computeNext() {
       while (blobIterator.hasNext()) {
         Path path = fileSystem.getPath(blobIterator.next().getName());
         try {
+          if (path.toString().equals(prefix)) {
+            // do not return ourselves, because that confuses recursive descents.
+            continue;
+          }
           if (filter.accept(path)) {
             return path;
           }
@@ -617,14 +624,14 @@ public DirectoryStream<Path> newDirectoryStream(Path dir, final Filter<? super P
     final CloudStoragePath cloudPath = CloudStorageUtil.checkPath(dir);
     checkNotNull(filter);
     initStorage();
-    String prefix = cloudPath.toRealPath().toString();
+    final String prefix = cloudPath.toRealPath().toString();
     final Iterator<Blob> blobIterator = storage.list(cloudPath.bucket(),
         Storage.BlobListOption.prefix(prefix), Storage.BlobListOption.currentDirectory(),
         Storage.BlobListOption.fields()).iterateAll().iterator();
     return new DirectoryStream<Path>() {
       @Override
       public Iterator<Path> iterator() {
-        return new LazyPathIterator(cloudPath.getFileSystem(), blobIterator, filter);
+        return new LazyPathIterator(cloudPath.getFileSystem(), prefix, blobIterator, filter);
       }
 
       @Override
