WeichenXu123 commented on a change in pull request #24851: [SPARK-27303][GRAPH] Add PropertyGraph construction API URL: https://github.com/apache/spark/pull/24851#discussion_r297716573
########## File path: graph/api/src/main/scala/org/apache/spark/graph/api/CypherSession.scala ########## @@ -0,0 +1,138 @@ +/* + * 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.spark.graph.api + +import scala.collection.JavaConverters + +import org.apache.spark.sql.{functions, DataFrame, SparkSession} + +object CypherSession { + val ID_COLUMN = "$ID" + val SOURCE_ID_COLUMN = "$SOURCE_ID" + val TARGET_ID_COLUMN = "$TARGET_ID" + val LABEL_COLUMN_PREFIX = ":" +} + +/** + * The entry point for using property graphs in Spark. + * + * Provides factory methods for creating [[PropertyGraph]] instances. + * + * Wraps a [[org.apache.spark.sql.SparkSession]]. + * + * @since 3.0.0 + */ +trait CypherSession { + + def sparkSession: SparkSession + + /** + * Creates a [[PropertyGraph]] from a sequence of [[NodeFrame]]s and [[RelationshipFrame]]s. + * At least one [[NodeFrame]] has to be provided. + * + * For each label set and relationship type there can be at most one [[NodeFrame]] and at most one + * [[RelationshipFrame]], respectively. + * + * @param nodes NodeFrames that define the nodes in the graph + * @param relationships RelationshipFrames that define the relationships in the graph + * @since 3.0.0 + */ + def createGraph(nodes: Seq[NodeFrame], relationships: Seq[RelationshipFrame]): PropertyGraph + + /** + * Creates a [[PropertyGraph]] from a sequence of [[NodeFrame]]s and [[RelationshipFrame]]s. + * At least one [[NodeFrame]] has to be provided. + * + * For each label set and relationship type there can be at most one [[NodeFrame]] and at most one + * [[RelationshipFrame]], respectively. + * + * @param nodes NodeFrames that define the nodes in the graph + * @param relationships RelationshipFrames that define the relationships in the graph + * @since 3.0.0 + */ + def createGraph( + nodes: java.util.List[NodeFrame], + relationships: java.util.List[RelationshipFrame]): PropertyGraph = { + createGraph(JavaConverters.asScalaBuffer(nodes), JavaConverters.asScalaBuffer(relationships)) + } + + /** + * Creates a [[PropertyGraph]] from nodes and relationships. + * + * The given DataFrames need to adhere to the following column naming conventions: + * + * {{{ + * Id column: `$ID` (nodes and relationships) + * SourceId column: `$SOURCE_ID` (relationships) + * TargetId column: `$TARGET_ID` (relationships) + * + * Label columns: `:{LABEL_NAME}` (nodes) + * RelType columns: `:{REL_TYPE}` (relationships) + * + * Property columns: `{Property_Key}` (nodes and relationships) + * }}} + * + * @see [[CypherSession]] + * @param nodes node DataFrame + * @param relationships relationship DataFrame + * @since 3.0.0 + */ + def createGraph(nodes: DataFrame, relationships: DataFrame): PropertyGraph = { + val idColumn = CypherSession.ID_COLUMN + val sourceIdColumn = CypherSession.SOURCE_ID_COLUMN + val targetIdColumn = CypherSession.TARGET_ID_COLUMN + + val labelColumns = nodes.columns.filter(_.startsWith(CypherSession.LABEL_COLUMN_PREFIX)).toSet + val nodeProperties = (nodes.columns.toSet - idColumn -- labelColumns) + .map(col => col -> col) + .toMap + + val trueLit = functions.lit(true) + val falseLit = functions.lit(false) + + val labelSets = labelColumns.subsets().toSet + Set.empty + val nodeFrames = labelSets.map { labelSet => + val predicate = labelColumns + .map { + case labelColumn if labelSet.contains(labelColumn) => nodes.col(labelColumn) === trueLit + case labelColumn => nodes.col(labelColumn) === falseLit + } + .reduce(_ && _) + + NodeFrame(nodes.filter(predicate), idColumn, labelSet.map(_.substring(1)), nodeProperties) + } + + val relColumns = relationships.columns.toSet + val relTypeColumns = relColumns.filter(_.startsWith(CypherSession.LABEL_COLUMN_PREFIX)) + val propertyColumns = relColumns - idColumn - sourceIdColumn - targetIdColumn -- relTypeColumns + val relProperties = propertyColumns.map(col => col -> col).toMap + val relFrames = relTypeColumns.map { relTypeColumn => + val predicate = relationships.col(relTypeColumn) === trueLit + + RelationshipFrame( + relationships.filter(predicate), + idColumn, + sourceIdColumn, + targetIdColumn, + relTypeColumn.substring(1), + relProperties) + } + + createGraph(nodeFrames.toSeq, relFrames.toSeq) Review comment: Should we cache `nodes` and `relationships` dataframe first ? The code here filter the two dataframes multiple times. ---------------------------------------------------------------- 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: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
