[jira] [Work logged] (BEAM-10925) ZetaSQL: Support Java UDF

["ASF GitHub Bot (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/BEAM-10925?focusedWorklogId=531387&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-531387 ]

ASF GitHub Bot logged work on BEAM-10925:
-----------------------------------------

                Author: ASF GitHub Bot
            Created on: 05/Jan/21 18:07
            Start Date: 05/Jan/21 18:07
    Worklog Time Spent: 10m 
      Work Description: apilloud commented on a change in pull request #13629:
URL: https://github.com/apache/beam/pull/13629#discussion_r552102591



##########
File path: sdks/java/extensions/sql/src/main/java/org/apache/beam/sdk/extensions/sql/impl/JavaUdfLoader.java
##########
@@ -0,0 +1,206 @@
+/*
+ * 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.beam.sdk.extensions.sql.impl;
+
+import com.google.auto.value.AutoValue;
+import java.io.File;
+import java.io.FileInputStream;
+import java.io.IOException;
+import java.io.InputStream;
+import java.net.URL;
+import java.net.URLClassLoader;
+import java.nio.channels.ReadableByteChannel;
+import java.nio.channels.WritableByteChannel;
+import java.nio.file.ProviderNotFoundException;
+import java.util.HashMap;
+import java.util.Iterator;
+import java.util.List;
+import java.util.Map;
+import java.util.ServiceLoader;
+import org.apache.beam.sdk.extensions.sql.udf.ScalarFn;
+import org.apache.beam.sdk.extensions.sql.udf.UdfProvider;
+import org.apache.beam.sdk.io.FileSystems;
+import org.apache.beam.sdk.io.fs.ResourceId;
+import org.apache.beam.sdk.util.common.ReflectHelpers;
+import org.apache.beam.vendor.calcite.v1_20_0.org.apache.commons.codec.digest.DigestUtils;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.collect.ImmutableList;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.collect.ImmutableMap;
+import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.io.ByteStreams;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Loads {@link UdfProvider} implementations from user-provided jars.
+ *
+ * <p>All UDFs are loaded and cached for each jar to mitigate IO costs.
+ */
+public class JavaUdfLoader {
+  private static final Logger LOG = LoggerFactory.getLogger(JavaUdfLoader.class);
+
+  /**
+   * Maps the external jar location to the functions the jar defines. Static so it can persist
+   * across multiple SQL transforms.
+   */
+  private static final Map<String, FunctionDefinitions> cache = new HashMap<>();
+
+  private static final ClassLoader originalClassLoader = ReflectHelpers.findClassLoader();
+
+  /**
+   * Load a user-defined scalar function from the specified jar.
+   *
+   * <p><strong>WARNING</strong>: The first time a jar is loaded, it is added to the thread's
+   * context {@link ClassLoader} so that the jar can be staged by the runner.
+   */
+  public ScalarFn loadScalarFunction(List<String> functionPath, String jarPath) {
+    String functionFullName = String.join(".", functionPath);
+    try {
+      FunctionDefinitions functionDefinitions = loadJar(jarPath);
+      if (!functionDefinitions.scalarFunctions().containsKey(functionPath)) {
+        throw new IllegalArgumentException(
+            String.format(
+                "No implementation of scalar function %s found in %s.%n"
+                    + " 1. Create a class implementing %s and annotate it with @AutoService(%s.class).%n"
+                    + " 2. Add function %s to the class's userDefinedScalarFunctions implementation.",
+                functionFullName,
+                jarPath,
+                UdfProvider.class.getSimpleName(),
+                UdfProvider.class.getSimpleName(),
+                functionFullName));
+      }
+      return functionDefinitions.scalarFunctions().get(functionPath);
+    } catch (IOException e) {
+      throw new RuntimeException(
+          String.format(
+              "Failed to load user-defined scalar function %s from %s", functionFullName, jarPath),
+          e);
+    }
+  }
+
+  /**
+   * Creates a temporary local copy of the file at {@code inputPath}, and returns a handle to the
+   * local copy.
+   */
+  private File downloadFile(String inputPath, String mimeType) throws IOException {
+    Preconditions.checkArgument(!inputPath.isEmpty(), "Path cannot be empty.");
+    ResourceId inputResource = FileSystems.matchNewResource(inputPath, false /* is directory */);
+    try (ReadableByteChannel inputChannel = FileSystems.open(inputResource)) {
+      File outputFile = File.createTempFile("sql-udf-", inputResource.getFilename());
+      ResourceId outputResource =
+          FileSystems.matchNewResource(outputFile.getAbsolutePath(), false /* is directory */);
+      try (WritableByteChannel outputChannel = FileSystems.create(outputResource, mimeType)) {
+        ByteStreams.copy(inputChannel, outputChannel);
+      }
+      // Compute and log checksum.
+      try (InputStream inputStream = new FileInputStream(outputFile)) {
+        LOG.info(
+            "Copied {} to {} with md5 hash {}.",
+            inputPath,
+            outputFile.getAbsolutePath(),
+            DigestUtils.md5Hex(inputStream));
+      }
+      return outputFile;
+    }
+  }
+
+  private ClassLoader createAndSetClassLoader(String inputJarPath) throws IOException {
+    File tmpJar = downloadFile(inputJarPath, "application/java-archive");
+    // Set the thread's context class loader so that the jar can be staged by the runner.
+    Thread.currentThread()
+        .setContextClassLoader(

Review comment:
       Not resetting the ContextClassLoader after your method returns is unsafe. You need to find another way to communicate the classes upstream. Our current UDF loading happens in SqlTransform.expand(), seems like this should be called from there too. So option 1 should be trivial.




----------------------------------------------------------------
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


Issue Time Tracking
-------------------

    Worklog Id:     (was: 531387)
    Time Spent: 26h 50m  (was: 26h 40m)

> ZetaSQL: Support Java UDF
> -------------------------
>
>                 Key: BEAM-10925
>                 URL: https://issues.apache.org/jira/browse/BEAM-10925
>             Project: Beam
>          Issue Type: New Feature
>          Components: dsl-sql-zetasql
>            Reporter: Kyle Weaver
>            Assignee: Kyle Weaver
>            Priority: P2
>          Time Spent: 26h 50m
>  Remaining Estimate: 0h
>




--
This message was sent by Atlassian Jira
(v8.3.4#803005)