Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Flink supports udf function #5067

Merged
merged 3 commits into from
Jan 12, 2024
Merged
Show file tree
Hide file tree
Changes from 1 commit
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Original file line number Diff line number Diff line change
Expand Up @@ -51,7 +51,7 @@ public abstract class ClusterDescriptorAdapter implements Closeable {
public static final long CLIENT_REQUEST_TIMEOUT =
FlinkEnvConfiguration.FLINK_CLIENT_REQUEST_TIMEOUT().getValue().toLong();

protected final ExecutionContext executionContext;
public final ExecutionContext executionContext;
// jobId is not null only after job is submitted
private JobID jobId;
protected ApplicationId clusterID;
Expand Down
Original file line number Diff line number Diff line change
@@ -0,0 +1,116 @@
/*
* 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.linkis.engineconnplugin.flink.client.utils;

import org.apache.flink.configuration.Configuration;
import org.apache.flink.configuration.PipelineOptions;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.table.functions.UserDefinedFunction;

import java.lang.reflect.Field;
import java.lang.reflect.Method;
import java.net.URL;
import java.net.URLClassLoader;
import java.util.ArrayList;
import java.util.List;
import java.util.Map;
import java.util.regex.Matcher;
import java.util.regex.Pattern;

import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

public class FlinkUdfUtils {

private static final Logger logger = LoggerFactory.getLogger(FlinkUdfUtils.class);

private static final String CREATE_TEMP_FUNCTION_PATTERN =
"create\\s+temporary\\s+function\\s+(\\w+)\\s+as\\s+\"(.*?)\"";

private static final String CREATE_TEMP_FUNCTION_SQL =
"CREATE TEMPORARY FUNCTION IF NOT EXISTS %s AS '%s' ";

public static void addFlinkPipelineClasspaths(StreamExecutionEnvironment env, String path) {
logger.info("Flink udf start add pipeline classpaths, jar path: {}", path);

try {
Field configuration = StreamExecutionEnvironment.class.getDeclaredField("configuration");
configuration.setAccessible(true);
Configuration conf = (Configuration) configuration.get(env);

Field confData = Configuration.class.getDeclaredField("confData");
confData.setAccessible(true);
Map<String, Object> map = (Map<String, Object>) confData.get(conf);
List<String> jarList = new ArrayList<>();
jarList.add(path);
map.put(PipelineOptions.CLASSPATHS.key(), jarList);
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is it necessary to consider the situation where the value was originally changed to append list?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ok, thanks for reviewing the code

} catch (Exception e) {
logger.warn("Flink udf add pipeline classpaths failed", e);
}
}

public static void loadJar(String jarPath) {
logger.info("Flink udf URLClassLoader start loadJar: {}", jarPath);

Method method = null;
Boolean accessible = null;
try {
method = URLClassLoader.class.getDeclaredMethod("addURL", URL.class);
accessible = method.isAccessible();

if (accessible == false) {
method.setAccessible(true);
}
URLClassLoader classLoader = (URLClassLoader) ClassLoader.getSystemClassLoader();
method.invoke(classLoader, new URL(jarPath));

} catch (Exception e) {
logger.warn("Flink udf URLClassLoader loadJar failed", e);
} finally {
if (accessible != null) {
method.setAccessible(accessible);
}
}
}

public static String extractUdfClass(String statement) {
Pattern pattern = Pattern.compile(CREATE_TEMP_FUNCTION_PATTERN);
Matcher matcher = pattern.matcher(statement);
if (matcher.find() && matcher.groupCount() >= 2) {
return matcher.group(2);
}
return "";
}

public static boolean isFlinkUdf(ClassLoader classLoader, String className) {
try {
Class<?> udfClass = classLoader.loadClass(className);
if (UserDefinedFunction.class.isAssignableFrom(udfClass)) {
return true;
}

} catch (ClassNotFoundException e) {
logger.warn("flink udf load isFlinkUdf failed, ClassNotFoundException: {}", className);
}
return false;
}

public static String generateFlinkUdfSql(String name, String className) {
return String.format(CREATE_TEMP_FUNCTION_SQL, name, className);
}
}
Original file line number Diff line number Diff line change
Expand Up @@ -39,6 +39,7 @@ import org.apache.linkis.engineconnplugin.flink.client.sql.operation.{
import org.apache.linkis.engineconnplugin.flink.client.sql.operation.impl.InsertOperation
import org.apache.linkis.engineconnplugin.flink.client.sql.operation.result.ResultKind
import org.apache.linkis.engineconnplugin.flink.client.sql.parser.SqlCommandParser
import org.apache.linkis.engineconnplugin.flink.client.utils.FlinkUdfUtils
import org.apache.linkis.engineconnplugin.flink.config.{
FlinkEnvConfiguration,
FlinkExecutionTargetType
Expand All @@ -50,18 +51,22 @@ import org.apache.linkis.engineconnplugin.flink.listener.{
}
import org.apache.linkis.engineconnplugin.flink.listener.RowsType.RowsType
import org.apache.linkis.governance.common.paser.SQLCodeParser
import org.apache.linkis.manager.label.entity.engine.UserCreatorLabel
import org.apache.linkis.protocol.engine.JobProgressInfo
import org.apache.linkis.scheduler.executer.{
ErrorExecuteResponse,
ExecuteResponse,
SuccessExecuteResponse
}
import org.apache.linkis.storage.resultset.ResultSetFactory
import org.apache.linkis.udf.UDFClient

import org.apache.calcite.rel.metadata.{JaninoRelMetadataProvider, RelMetadataQueryBase}
import org.apache.commons.lang3.StringUtils
import org.apache.flink.api.common.JobStatus._
import org.apache.flink.configuration.DeploymentOptions
import org.apache.flink.kubernetes.configuration.KubernetesConfigOptions
import org.apache.flink.table.api.TableResult
import org.apache.flink.table.planner.plan.metadata.FlinkDefaultRelMetadataProvider
import org.apache.flink.yarn.configuration.YarnConfigOptions
import org.apache.hadoop.yarn.util.ConverterUtils
Expand Down Expand Up @@ -125,6 +130,9 @@ class FlinkSQLComputationExecutor(
engineExecutionContext: EngineExecutionContext,
code: String
): ExecuteResponse = {
// The load flink udf failure does not affect task execution
Utils.tryAndWarn(loadFlinkUdf(engineExecutionContext))

val callOpt = SqlCommandParser.getSqlCommandParser.parse(code.trim, true)
val callSQL =
if (!callOpt.isPresent)
Expand Down Expand Up @@ -194,6 +202,91 @@ class FlinkSQLComputationExecutor(
}
}

private def getExecSqlUser(engineExecutionContext: EngineExecutionContext): String = {
val userCreatorLabel = engineExecutionContext.getLabels
.find(_.isInstanceOf[UserCreatorLabel])
.get
.asInstanceOf[UserCreatorLabel]
userCreatorLabel.getUser
}

private def loadFlinkUdf(engineExecutionContext: EngineExecutionContext) = {
logger.info("Flink start load udf")

val execSqlUser = getExecSqlUser(engineExecutionContext)
val udfAllLoad: String =
engineExecutionContext.getProperties.getOrDefault("linkis.user.udf.all.load", "true").toString
val udfIdStr: String =
engineExecutionContext.getProperties.getOrDefault("linkis.user.udf.custom.ids", "").toString
val udfIds = udfIdStr.split(",").filter(StringUtils.isNotBlank).map(s => s.toLong)

logger.info(s"start loading UDFs, user: $execSqlUser, load all: $udfAllLoad, udfIds: ${udfIds
.mkString("Array(", ", ", ")")}")

val udfInfos =
if (udfAllLoad.toBoolean) UDFClient.getJarUdf(execSqlUser)
else UDFClient.getJarUdfByIds(execSqlUser, udfIds)

if (udfInfos.nonEmpty) {
import scala.util.control.Breaks._
udfInfos.foreach { udfInfo =>
val path: String = udfInfo.getPath
val registerFormat: String = udfInfo.getRegisterFormat

breakable {
if (StringUtils.isBlank(path) && StringUtils.isBlank(registerFormat)) {
logger.warn("flink udf udfInfo path or RegisterFormat cannot is empty")
break()
}
logger.info(
s"udfName:${udfInfo.getUdfName}, bml_resource_id:${udfInfo.getBmlResourceId}, bml_id:${udfInfo.getId}\n"
)

val udfClassName: String = FlinkUdfUtils.extractUdfClass(registerFormat)
if (StringUtils.isBlank(udfClassName)) {
logger.warn("flink udf extract Udf Class cannot is empty")
break()
}

FlinkUdfUtils.loadJar(path)

if (!FlinkUdfUtils.isFlinkUdf(ClassLoader.getSystemClassLoader(), udfClassName)) {
logger.warn(
"There is no extends UserDefinedFunction, skip loading flink udf: {} ",
path
)
break()
}

val context = clusterDescriptor.executionContext
val flinkUdfSql: String =
FlinkUdfUtils.generateFlinkUdfSql(udfInfo.getUdfName, udfClassName)

FlinkUdfUtils.addFlinkPipelineClasspaths(context.getStreamExecutionEnvironment, path)
val tableEnv = context.getTableEnvironment
logger.info("Flink execute udf sql:{}", flinkUdfSql)
val tableResult: TableResult = tableEnv.executeSql(flinkUdfSql)

var loadUdfLog =
s"udfName:${udfInfo.getUdfName}, udfJar:${path}, udfClass:${udfClassName}, Flink load udf %s ."
if (
tableResult.getResultKind != null && tableResult.getResultKind
.name()
.equalsIgnoreCase("SUCCESS")
) {
loadUdfLog = String.format(loadUdfLog, "success")
logger.info(loadUdfLog)
} else {
loadUdfLog = String.format(loadUdfLog, "failed")
logger.error(loadUdfLog)
}
engineExecutionContext.appendStdout(loadUdfLog)
}
}

}
}

override def executeCompletely(
engineExecutorContext: EngineExecutionContext,
code: String,
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -37,12 +37,15 @@ import org.apache.linkis.manager.engineplugin.common.launch.process.LaunchConsta
addPathToClassPath,
CLASS_PATH_SEPARATOR
}
import org.apache.linkis.manager.label.entity.engine.UserCreatorLabel
import org.apache.linkis.manager.label.entity.engine.{EngineConnMode, UserCreatorLabel}
import org.apache.linkis.manager.label.utils.LabelUtil

import java.util

import scala.collection.JavaConverters._

import com.google.common.collect.Lists

class FlinkEngineConnLaunchBuilder extends JavaProcessEngineConnLaunchBuilder {

override protected def getCommands(implicit
Expand Down Expand Up @@ -136,4 +139,17 @@ class FlinkEngineConnLaunchBuilder extends JavaProcessEngineConnLaunchBuilder {

override protected def ifAddHiveConfigPath: Boolean = true

override protected def getEngineConnManagerHooks(implicit
engineConnBuildRequest: EngineConnBuildRequest
): java.util.List[String] = if (isOnceMode) {
super.getEngineConnManagerHooks(engineConnBuildRequest)
} else {
Lists.newArrayList("JarUDFLoadECMHook")
}

def isOnceMode: Boolean = {
val engineConnMode = LabelUtil.getEngineConnMode(engineConnBuildRequest.labels)
EngineConnMode.toEngineConnMode(engineConnMode) == EngineConnMode.Once
}

}
Loading