implement a mechanism for validating graph stores by injecting hooks

neo4j · Jul 1, 2024 · 68ee9a6 · 68ee9a6
1 parent ec3cce3
commit 68ee9a6
Show file tree

Hide file tree

Showing 29 changed files with 214 additions and 22 deletions.
diff --git a/algo/src/main/java/org/neo4j/gds/algorithms/estimation/AlgorithmEstimator.java b/algo/src/main/java/org/neo4j/gds/algorithms/estimation/AlgorithmEstimator.java
@@ -84,7 +84,8 @@ public <G, A extends Algorithm<?>, C extends AlgoBaseConfig> MemoryEstimateResul
                 config,
                 maybeRelationshipProperty,
                 requestScopedDependencies.getUser(),
-                requestScopedDependencies.getDatabaseId()
+                requestScopedDependencies.getDatabaseId(),
+                Optional.empty()
             ).graphStore();
             dimensions = GraphDimensionsComputer.of(graphStore, config);
         } else {

diff --git a/algo/src/main/java/org/neo4j/gds/algorithms/runner/AlgorithmRunner.java b/algo/src/main/java/org/neo4j/gds/algorithms/runner/AlgorithmRunner.java
@@ -86,7 +86,8 @@ public <A extends Algorithm<R>, R, C extends AlgoBaseConfig> AlgorithmComputatio
             config,
             relationshipProperty,
             requestScopedDependencies.getUser(),
-            requestScopedDependencies.getDatabaseId()
+            requestScopedDependencies.getDatabaseId(),
+            Optional.empty()
         );
 
         var graph = graphResources.graph();

diff --git a/algo/src/test/java/org/neo4j/gds/algorithms/runner/AlgorithmRunnerTest.java b/algo/src/test/java/org/neo4j/gds/algorithms/runner/AlgorithmRunnerTest.java
@@ -33,6 +33,7 @@
 import org.neo4j.gds.config.AlgoBaseConfig;
 import org.neo4j.gds.core.loading.GraphResources;
 import org.neo4j.gds.core.loading.GraphStoreCatalogService;
+import org.neo4j.gds.core.loading.PostGraphStoreLoadValidationHook;
 import org.neo4j.gds.logging.Log;
 import org.neo4j.gds.metrics.ExecutionMetric;
 import org.neo4j.gds.metrics.algorithms.AlgorithmMetricsService;
@@ -157,7 +158,8 @@ public GraphResources getGraphResources(
                 AlgoBaseConfig config,
                 Optional<String> relationshipProperty,
                 User user,
-                DatabaseId databaseId
+                DatabaseId databaseId,
+                Optional<Iterable<PostGraphStoreLoadValidationHook>> postGraphStoreLoadValidationHooks
             ) {
                 // this gets called right after the preconditions check
                 // so, we save ourselves trouble and shunt the rest of the method

diff --git a/.../gds/applications/algorithms/centrality/CentralityAlgorithmsMutateModeBusinessFacade.java b/.../gds/applications/algorithms/centrality/CentralityAlgorithmsMutateModeBusinessFacade.java
@@ -76,6 +76,7 @@ public <RESULT> RESULT articleRank(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             ArticleRank,
             estimation::pageRank,
             graph -> algorithms.articleRank(graph, configuration),
@@ -94,6 +95,7 @@ public <RESULT> RESULT betweennessCentrality(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             BetweennessCentrality,
             () -> estimation.betweennessCentrality(configuration),
             graph -> algorithms.betweennessCentrality(graph, configuration),
@@ -112,6 +114,7 @@ public <RESULT> RESULT celf(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             CELF,
             () -> estimation.celf(configuration),
             graph -> algorithms.celf(graph, configuration),
@@ -130,6 +133,7 @@ public <RESULT> RESULT closenessCentrality(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             ClosenessCentrality,
             () -> estimation.closenessCentrality(configuration),
             graph -> algorithms.closenessCentrality(graph, configuration),
@@ -148,6 +152,7 @@ public <RESULT> RESULT degreeCentrality(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             DegreeCentrality,
             () -> estimation.degreeCentrality(configuration),
             graph -> algorithms.degreeCentrality(graph, configuration),
@@ -166,6 +171,7 @@ public <RESULT> RESULT eigenVector(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             EigenVector,
             estimation::pageRank,
             graph -> algorithms.eigenVector(graph, configuration),
@@ -184,6 +190,7 @@ public <RESULT> RESULT harmonicCentrality(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             HarmonicCentrality,
             estimation::harmonicCentrality,
             graph -> algorithms.harmonicCentrality(graph, configuration),
@@ -202,6 +209,7 @@ public <RESULT> RESULT pageRank(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             PageRank,
             estimation::pageRank,
             graph -> algorithms.pageRank(graph, configuration),

diff --git a/...j/gds/applications/algorithms/centrality/CentralityAlgorithmsStatsModeBusinessFacade.java b/...j/gds/applications/algorithms/centrality/CentralityAlgorithmsStatsModeBusinessFacade.java
@@ -66,6 +66,7 @@ public <RESULT> RESULT articleRank(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             ArticleRank,
             estimationFacade::pageRank,
             graph -> centralityAlgorithms.articleRank(graph, configuration),
@@ -82,6 +83,7 @@ public <RESULT> RESULT betweennessCentrality(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             BetweennessCentrality,
             () -> estimationFacade.betweennessCentrality(configuration),
             graph -> centralityAlgorithms.betweennessCentrality(graph, configuration),
@@ -98,6 +100,7 @@ public <RESULT> RESULT celf(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             CELF,
             () -> estimationFacade.celf(configuration),
             graph -> centralityAlgorithms.celf(graph, configuration),
@@ -114,6 +117,7 @@ public <RESULT> RESULT closenessCentrality(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             ClosenessCentrality,
             () -> estimationFacade.closenessCentrality(configuration),
             graph -> centralityAlgorithms.closenessCentrality(graph, configuration),
@@ -130,6 +134,7 @@ public <RESULT> RESULT degreeCentrality(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             DegreeCentrality,
             () -> estimationFacade.degreeCentrality(configuration),
             graph -> centralityAlgorithms.degreeCentrality(graph, configuration),
@@ -146,6 +151,7 @@ public <RESULT> RESULT eigenVector(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             EigenVector,
             estimationFacade::pageRank,
             graph -> centralityAlgorithms.eigenVector(graph, configuration),
@@ -162,6 +168,7 @@ public <RESULT> RESULT harmonicCentrality(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             HarmonicCentrality,
             estimationFacade::harmonicCentrality,
             graph -> centralityAlgorithms.harmonicCentrality(graph, configuration),
@@ -178,6 +185,7 @@ public <RESULT> RESULT pageRank(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             PageRank,
             estimationFacade::pageRank,
             graph -> centralityAlgorithms.pageRank(graph, configuration),

diff --git a/.../gds/applications/algorithms/centrality/CentralityAlgorithmsStreamModeBusinessFacade.java b/.../gds/applications/algorithms/centrality/CentralityAlgorithmsStreamModeBusinessFacade.java
@@ -67,6 +67,7 @@ public <RESULT> RESULT articleRank(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             ArticleRank,
             estimationFacade::pageRank,
             graph -> centralityAlgorithms.articleRank(graph, configuration),
@@ -83,6 +84,7 @@ public <RESULT> RESULT betweennessCentrality(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             BetweennessCentrality,
             () -> estimationFacade.betweennessCentrality(configuration),
             graph -> centralityAlgorithms.betweennessCentrality(graph, configuration),
@@ -99,6 +101,7 @@ public <RESULT> RESULT celf(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             CELF,
             () -> estimationFacade.celf(configuration),
             graph -> centralityAlgorithms.celf(graph, configuration),
@@ -115,6 +118,7 @@ public <RESULT> RESULT closenessCentrality(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             ClosenessCentrality,
             () -> estimationFacade.closenessCentrality(configuration),
             graph -> centralityAlgorithms.closenessCentrality(graph, configuration),
@@ -131,6 +135,7 @@ public <RESULT> RESULT degreeCentrality(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             DegreeCentrality,
             () -> estimationFacade.degreeCentrality(configuration),
             graph -> centralityAlgorithms.degreeCentrality(graph, configuration),
@@ -147,6 +152,7 @@ public <RESULT> RESULT eigenvector(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             EigenVector,
             estimationFacade::pageRank,
             graph -> centralityAlgorithms.eigenVector(graph, configuration),
@@ -163,6 +169,7 @@ public <RESULT> RESULT harmonicCentrality(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             HarmonicCentrality,
             estimationFacade::harmonicCentrality,
             graph -> centralityAlgorithms.harmonicCentrality(graph, configuration),
@@ -179,6 +186,7 @@ public <RESULT> RESULT pageRank(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             PageRank,
             estimationFacade::pageRank,
             graph -> centralityAlgorithms.pageRank(graph, configuration),

diff --git a/...j/gds/applications/algorithms/centrality/CentralityAlgorithmsWriteModeBusinessFacade.java b/...j/gds/applications/algorithms/centrality/CentralityAlgorithmsWriteModeBusinessFacade.java
@@ -97,6 +97,7 @@ public <RESULT> RESULT articleRank(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             ArticleRank,
             estimationFacade::pageRank,
             graph -> centralityAlgorithms.articleRank(graph, configuration),
@@ -115,6 +116,7 @@ public <RESULT> RESULT betweennessCentrality(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             BetweennessCentrality,
             () -> estimationFacade.betweennessCentrality(configuration),
             graph -> centralityAlgorithms.betweennessCentrality(graph, configuration),
@@ -133,6 +135,7 @@ public <CONFIGURATION extends InfluenceMaximizationWriteConfig, RESULT> RESULT c
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             CELF,
             () -> estimationFacade.celf(configuration),
             graph -> centralityAlgorithms.celf(graph, configuration),
@@ -151,6 +154,7 @@ public <RESULT> RESULT closenessCentrality(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             ClosenessCentrality,
             () -> estimationFacade.closenessCentrality(configuration),
             graph -> centralityAlgorithms.closenessCentrality(graph, configuration),
@@ -169,6 +173,7 @@ public <RESULT> RESULT degreeCentrality(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             DegreeCentrality,
             () -> estimationFacade.degreeCentrality(configuration),
             graph -> centralityAlgorithms.degreeCentrality(graph, configuration),
@@ -187,6 +192,7 @@ public <RESULT> RESULT eigenvector(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             EigenVector,
             estimationFacade::pageRank,
             graph -> centralityAlgorithms.eigenVector(graph, configuration),
@@ -205,6 +211,7 @@ public <CONFIGURATION extends HarmonicCentralityWriteConfig, RESULT> RESULT harm
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             HarmonicCentrality,
             estimationFacade::harmonicCentrality,
             graph -> centralityAlgorithms.harmonicCentrality(graph, configuration),
@@ -223,6 +230,7 @@ public <RESULT> RESULT pageRank(
         return algorithmProcessingTemplate.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             PageRank,
             estimationFacade::pageRank,
             graph -> centralityAlgorithms.pageRank(graph, configuration),

diff --git a/...4j/gds/applications/algorithms/community/CommunityAlgorithmsMutateModeBusinessFacade.java b/...4j/gds/applications/algorithms/community/CommunityAlgorithmsMutateModeBusinessFacade.java
@@ -94,6 +94,7 @@ public <RESULT> RESULT approximateMaximumKCut(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             ApproximateMaximumKCut,
             () -> estimation.approximateMaximumKCut(configuration),
             graph -> algorithms.approximateMaximumKCut(graph, configuration),
@@ -112,6 +113,7 @@ public <RESULT> RESULT k1Coloring(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             K1Coloring,
             estimation::k1Coloring,
             graph -> algorithms.k1Coloring(graph, configuration),
@@ -130,6 +132,7 @@ public <RESULT> RESULT kCore(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             KCore,
             estimation::kCore,
             graph -> algorithms.kCore(graph, configuration),
@@ -148,6 +151,7 @@ public <RESULT> RESULT kMeans(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             KMeans,
             () -> estimation.kMeans(configuration),
             graph -> algorithms.kMeans(graph, configuration),
@@ -166,6 +170,7 @@ public <RESULT> RESULT labelPropagation(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             LabelPropagation,
             estimation::labelPropagation,
             graph -> algorithms.labelPropagation(graph, configuration),
@@ -184,6 +189,7 @@ public <RESULT> RESULT lcc(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             LCC,
             () -> estimation.lcc(configuration),
             graph -> algorithms.lcc(graph, configuration),
@@ -202,6 +208,7 @@ public <RESULT> RESULT leiden(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             Leiden,
             () -> estimation.leiden(configuration),
             graph -> algorithms.leiden(graph, configuration),
@@ -220,6 +227,7 @@ public <RESULT> RESULT louvain(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             Louvain,
             () -> estimation.louvain(configuration),
             graph -> algorithms.louvain(graph, configuration),
@@ -238,6 +246,7 @@ public <RESULT> RESULT modularityOptimization(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             ModularityOptimization,
             estimation::modularityOptimization,
             graph -> algorithms.modularityOptimization(graph, configuration),
@@ -256,6 +265,7 @@ public <RESULT> RESULT scc(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             SCC,
             estimation::scc,
             graph -> algorithms.scc(graph, configuration),
@@ -274,6 +284,7 @@ public <RESULT> RESULT triangleCount(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             TriangleCount,
             estimation::triangleCount,
             graph -> algorithms.triangleCount(graph, configuration),
@@ -292,6 +303,7 @@ public <RESULT> RESULT wcc(
         return template.processAlgorithm(
             graphName,
             configuration,
+            Optional.empty(),
             WCC,
             () -> estimation.wcc(configuration),
             graph -> algorithms.wcc(graph, configuration),