JulianJaffePinterest commented on a change in pull request #11474: URL: https://github.com/apache/druid/pull/11474#discussion_r676276706
########## File path: spark/src/main/scala/org/apache/druid/spark/clients/DruidClient.scala ########## @@ -0,0 +1,197 @@ +/* + * 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.druid.spark.clients + +import com.fasterxml.jackson.core.`type`.TypeReference +import com.google.common.net.HostAndPort +import org.apache.druid.java.util.common.{ISE, Intervals, JodaUtils, StringUtils} +import org.apache.druid.java.util.http.client.response.{StringFullResponseHandler, + StringFullResponseHolder} +import org.apache.druid.java.util.http.client.{HttpClient, Request} +import org.apache.druid.query.Druids +import org.apache.druid.query.metadata.metadata.{ColumnAnalysis, SegmentAnalysis, + SegmentMetadataQuery} +import org.apache.druid.spark.MAPPER +import org.apache.druid.spark.configuration.{Configuration, DruidConfigurationKeys} +import org.apache.druid.spark.mixins.Logging +import org.jboss.netty.handler.codec.http.{HttpMethod, HttpResponseStatus} +import org.joda.time.{Duration, Interval} + +import java.net.URL +import java.util.{List => JList} +import javax.ws.rs.core.MediaType +import scala.collection.JavaConverters.{asScalaBufferConverter, mapAsJavaMapConverter, + mapAsScalaMapConverter, seqAsJavaListConverter} + +/** + * Going with SegmentMetadataQueries despite the significant overhead because there's no other way + * to get accurate information about Druid columns. + * + * This is pulled pretty directly from the druid-hadoop-input-format project on GitHub. There is + * likely substantial room for improvement. + */ +class DruidClient( + httpClient: HttpClient, + hostAndPort: HostAndPort, + numRetries: Int, + retryWaitSeconds: Int, + timeoutWaitMilliseconds: Int + ) extends Logging { + private val druidBaseQueryURL: HostAndPort => String = + (hostAndPort: HostAndPort) => s"http://$hostAndPort/druid/v2/"; + + private val DefaultSegmentMetadataInterval = List[Interval](Intervals.utc( + JodaUtils.MIN_INSTANT, + JodaUtils.MAX_INSTANT + )) + + /** + * The SQL system catalog tables are incorrect for multivalue columns and don't have accurate + * type info, so we need to fall back to segmentMetadataQueries. Given a DATASOURCE and a range + * of INTERVALS to query over, return a map from column name to a tuple of + * (columnType, hasMultipleValues). Note that this is a very expensive operation over large + * numbers of segments. If possible, this method should only be called with the most granular + * starting and ending intervals instead of over a larger interval. + * + * @param dataSource The Druid dataSource to fetch the schema for. + * @param intervals The intervals to return the schema for, or None to query all segments. + * @return A map from column name to data type and whether or not the column is multi-value + * for the schema of DATASOURCE. + */ + def getSchema(dataSource: String, intervals: Option[List[Interval]]): Map[String, (String, Boolean)] = { + val queryInterval = intervals.getOrElse(DefaultSegmentMetadataInterval) + val body = Druids.newSegmentMetadataQueryBuilder() + .dataSource(dataSource) + .intervals(queryInterval.asJava) + .analysisTypes(SegmentMetadataQuery.AnalysisType.SIZE) + .merge(true) + .context(Map[String, AnyRef]( + "timeout" -> Int.box(timeoutWaitMilliseconds) + ).asJava) + .build() + val response = sendRequestWithRetry( + druidBaseQueryURL(hostAndPort), numRetries, Option(MAPPER.writeValueAsBytes(body)) + ) + val segments = + MAPPER.readValue[JList[SegmentAnalysis]]( + response.getContent, new TypeReference[JList[SegmentAnalysis]]() {}) + if (segments.size() == 0) { + throw new ISE( + s"No segments found for intervals [${intervals.mkString(",")}] on $dataSource" + ) + } + // Since we're setting merge to true, there should only be one item in the list + if (segments.size() > 1) { + throw new ISE("Merged segment metadata response had more than one row!") + } + log.debug(segments.asScala.map(_.toString).mkString("SegmentAnalysis: [", ", ", "]")) Review comment: The statement is evaluated lazily (the actual method signature in the `Logging` trait is `debug(msg: => String)`, i.e. a function that takes no arguments and returns a String. The argument will only be evaluated if `log.isDebugEnabled` evaluates to true) -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
