From 036af5facdc5b7ad3f0623975795473ce7cc0303 Mon Sep 17 00:00:00 2001
From: bbimber <bbimber@gmail.com>
Date: Thu, 30 Jan 2025 06:43:48 -0800
Subject: [PATCH] Add check against duplicated IDs in MCC ETL

---
 .../org/labkey/mcc/etl/PopulateIdsStep.java   | 25 +++++++++++++++++++
 1 file changed, 25 insertions(+)

diff --git a/mcc/src/org/labkey/mcc/etl/PopulateIdsStep.java b/mcc/src/org/labkey/mcc/etl/PopulateIdsStep.java
index 638af85df..81debad62 100644
--- a/mcc/src/org/labkey/mcc/etl/PopulateIdsStep.java
+++ b/mcc/src/org/labkey/mcc/etl/PopulateIdsStep.java
@@ -43,9 +43,34 @@ public RecordedActionSet run(@NotNull PipelineJob job) throws PipelineJobExcepti
         populateForDemographics(job);
         populateForKinship(job);
 
+        // Check for duplicate IDs and proactively error:
+        performDuplicateCheck(job);
+
         return new RecordedActionSet();
     }
 
+    private void performDuplicateCheck(PipelineJob job) throws PipelineJobException
+    {
+        // Query aggregated demographics:
+        UserSchema sourceSchema = QueryService.get().getUserSchema(_containerUser.getUser(), _containerUser.getContainer(), MccSchema.NAME);
+        if (sourceSchema == null)
+        {
+            throw new PipelineJobException("Unable to find source schema: " + MccSchema.NAME);
+        }
+
+        TableInfo sourceTi = sourceSchema.getTable("duplicatedAggregatedDemographics");
+        if (sourceTi == null)
+        {
+            throw new PipelineJobException("Unable to find table: duplicatedAggregatedDemographics");
+        }
+
+        TableSelector ts = new TableSelector(sourceTi);
+        if (ts.exists())
+        {
+            throw new PipelineJobException("There were duplicated IDs in aggregatedDemographics");
+        }
+    }
+
     private void populateForDemographics(PipelineJob job) throws PipelineJobException
     {
         // Query aggregated demographics: