[GitHub] [arrow] fsaintjacques commented on a change in pull request #7030: ARROW-7808: [Java][Dataset] Implement Datasets Java API by JNI to C++

[GitBox <gi...@apache.org>]

fsaintjacques commented on a change in pull request #7030:
URL: https://github.com/apache/arrow/pull/7030#discussion_r439413765



##########
File path: cpp/src/arrow/dataset/discovery.h
##########
@@ -216,6 +216,16 @@ class ARROW_DS_EXPORT FileSystemDatasetFactory : public DatasetFactory {
       std::shared_ptr<fs::FileSystem> filesystem, fs::FileSelector selector,
       std::shared_ptr<FileFormat> format, FileSystemFactoryOptions options);
 
+  /// \brief Build a FileSystemDatasetFactory from an uri including filesystem
+  /// information.
+  ///
+  /// \param[in] uri passed to FileSystemDataset
+  /// \param[in] format passed to FileSystemDataset
+  /// \param[in] options see FileSystemFactoryOptions for more information.
+  static Result<std::shared_ptr<DatasetFactory>> Make(std::string uri,

Review comment:
       :+1: 

##########
File path: java/dataset/src/main/java/org/apache/arrow/dataset/jni/NativeScanner.java
##########
@@ -0,0 +1,143 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *    http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.arrow.dataset.jni;
+
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.NoSuchElementException;
+import java.util.concurrent.atomic.AtomicBoolean;
+import java.util.stream.Collectors;
+
+import org.apache.arrow.dataset.scanner.ScanTask;
+import org.apache.arrow.dataset.scanner.Scanner;
+import org.apache.arrow.memory.ArrowBuf;
+import org.apache.arrow.memory.BufferAllocator;
+import org.apache.arrow.memory.BufferLedger;
+import org.apache.arrow.memory.NativeUnderlingMemory;
+import org.apache.arrow.memory.Ownerships;
+import org.apache.arrow.vector.ipc.message.ArrowFieldNode;
+import org.apache.arrow.vector.ipc.message.ArrowRecordBatch;
+import org.apache.arrow.vector.types.pojo.Schema;
+import org.apache.arrow.vector.util.SchemaUtility;
+
+/**
+ * Native implementation of {@link Scanner}. Note that it currently emits only a single scan task of type
+ * {@link NativeScanTask}, which is internally a combination of all scan task instances returned by the
+ * native scanner.
+ */
+public class NativeScanner implements Scanner {
+
+  private final AtomicBoolean closed = new AtomicBoolean(false);
+  private final AtomicBoolean executed = new AtomicBoolean(false);
+  private final NativeContext context;
+  private final long scannerId;
+
+  public NativeScanner(NativeContext context, long scannerId) {
+    this.context = context;
+    this.scannerId = scannerId;
+  }
+
+  ScanTask.BatchIterator execute() {
+    if (closed.get()) {
+      throw new NativeInstanceClosedException();
+    }
+    if (!executed.compareAndSet(false, true)) {
+      throw new UnsupportedOperationException("NativeScanner cannot be executed more than once. Consider creating " +
+          "new scanner instead");
+    }
+    return new ScanTask.BatchIterator() {
+      private ArrowRecordBatch peek = null;
+
+      @Override
+      public void close() {
+        NativeScanner.this.close();
+      }
+
+      @Override
+      public boolean hasNext() {
+        if (closed.get()) {

Review comment:
       What's the point of checking close before the method if any concurrent thread can come and close said object while this method is running? In other words, this doesn't protect against "use-after-close". Since `close()` will delete the `shared_ptr` underneath, you're very open to user-after-free in C++.
   
   I think you need to protect with close state with a read-write-mutex, or document that the Scanner is single threaded only, which it is in C++. The important part is that ScanTasks can be dispatched concurrently.

##########
File path: java/dataset/src/main/java/org/apache/arrow/dataset/jni/NativeScanner.java
##########
@@ -0,0 +1,143 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *    http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.arrow.dataset.jni;
+
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.NoSuchElementException;
+import java.util.concurrent.atomic.AtomicBoolean;
+import java.util.stream.Collectors;
+
+import org.apache.arrow.dataset.scanner.ScanTask;
+import org.apache.arrow.dataset.scanner.Scanner;
+import org.apache.arrow.memory.ArrowBuf;
+import org.apache.arrow.memory.BufferAllocator;
+import org.apache.arrow.memory.BufferLedger;
+import org.apache.arrow.memory.NativeUnderlingMemory;
+import org.apache.arrow.memory.Ownerships;
+import org.apache.arrow.vector.ipc.message.ArrowFieldNode;
+import org.apache.arrow.vector.ipc.message.ArrowRecordBatch;
+import org.apache.arrow.vector.types.pojo.Schema;
+import org.apache.arrow.vector.util.SchemaUtility;
+
+/**
+ * Native implementation of {@link Scanner}. Note that it currently emits only a single scan task of type
+ * {@link NativeScanTask}, which is internally a combination of all scan task instances returned by the
+ * native scanner.
+ */
+public class NativeScanner implements Scanner {
+
+  private final AtomicBoolean closed = new AtomicBoolean(false);
+  private final AtomicBoolean executed = new AtomicBoolean(false);
+  private final NativeContext context;
+  private final long scannerId;
+
+  public NativeScanner(NativeContext context, long scannerId) {
+    this.context = context;
+    this.scannerId = scannerId;
+  }
+
+  ScanTask.BatchIterator execute() {
+    if (closed.get()) {
+      throw new NativeInstanceClosedException();
+    }
+    if (!executed.compareAndSet(false, true)) {
+      throw new UnsupportedOperationException("NativeScanner cannot be executed more than once. Consider creating " +
+          "new scanner instead");
+    }
+    return new ScanTask.BatchIterator() {
+      private ArrowRecordBatch peek = null;
+
+      @Override
+      public void close() {
+        NativeScanner.this.close();
+      }
+
+      @Override
+      public boolean hasNext() {
+        if (closed.get()) {
+          throw new NativeInstanceClosedException();
+        }
+        if (peek != null) {
+          return true;
+        }
+        NativeRecordBatchHandle handle = JniWrapper.get().nextRecordBatch(scannerId);
+        if (handle == null) {
+          return false;
+        }
+        final ArrayList<ArrowBuf> buffers = new ArrayList<>();
+        for (NativeRecordBatchHandle.Buffer buffer : handle.getBuffers()) {
+          final BufferAllocator allocator = context.getAllocator();
+          final NativeUnderlingMemory am = NativeUnderlingMemory.create(allocator,
+              (int) buffer.size, buffer.nativeInstanceId, buffer.memoryAddress);

Review comment:
       You should minimally check if any of the size properties are bigger than INT32_MAX and raise accordingly. This will avoid nasty segfault/corrupted data down the road.




----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org