Fix migration problems

Author: milanmajchrak

.github/workflows/deploy.yml

@@ -100,6 +100,88 @@ jobs:
             (psql -p 10563 -U dspace -d postgres -c \"CREATE ROLE dspace5 IN ROLE dspace;\" || true) &&
             psql -p 10563 -U dspace -d dspace -f /tmp/$FNAME &&
             psql -p 10563 -U dspace -d dspace -c 'CREATE EXTENSION IF NOT EXISTS pgcrypto;' &&
+            psql -p 10563 -U dspace -d dspace -c \"
+              -- Comprehensive cleanup of duplicate entries before migration
+
+              -- First, let's check what tables actually exist and their names
+              \\\\echo 'Checking existing tables...';
+
+              -- Clean up epersongroup2eperson duplicates (case variations)
+              DO \\\$cleanup\\\$
+              BEGIN
+                -- Try different case variations of the table name
+                IF EXISTS (SELECT 1 FROM information_schema.tables WHERE LOWER(table_name) = 'epersongroup2eperson') THEN
+                  \\\\echo 'Cleaning epersongroup2eperson duplicates...';
+
+                  -- First show how many duplicates we have
+                  PERFORM (SELECT COUNT(*) FROM (
+                    SELECT eperson_group_id, eperson_id, COUNT(*)
+                    FROM epersongroup2eperson
+                    GROUP BY eperson_group_id, eperson_id
+                    HAVING COUNT(*) > 1
+                  ) dups);
+
+                  -- Delete duplicates, keeping only the first occurrence
+                  WITH ranked AS (
+                    SELECT ctid, row_number() OVER (PARTITION BY eperson_group_id, eperson_id ORDER BY ctid) AS rn
+                    FROM epersongroup2eperson
+                  )
+                  DELETE FROM epersongroup2eperson
+                  WHERE ctid IN (
+                    SELECT ctid FROM ranked WHERE rn > 1
+                  );
+
+                  -- Verify cleanup worked
+                  IF EXISTS (
+                    SELECT 1 FROM (
+                      SELECT eperson_group_id, eperson_id, COUNT(*)
+                      FROM epersongroup2eperson
+                      GROUP BY eperson_group_id, eperson_id
+                      HAVING COUNT(*) > 1
+                    ) remaining_dups
+                  ) THEN
+                    RAISE EXCEPTION 'Failed to clean all duplicates from epersongroup2eperson';
+                  END IF;
+
+                  \\\\echo 'epersongroup2eperson cleanup completed successfully';
+                END IF;
+
+                -- Check for other possible table name variations
+                IF EXISTS (SELECT 1 FROM information_schema.tables WHERE LOWER(table_name) = 'epersongroup2workspace') THEN
+                  WITH ranked AS (
+                    SELECT ctid, row_number() OVER (PARTITION BY eperson_group_id, workspace_item_id ORDER BY ctid) AS rn
+                    FROM epersongroup2workspace
+                  )
+                  DELETE FROM epersongroup2workspace WHERE ctid IN (SELECT ctid FROM ranked WHERE rn > 1);
+                END IF;
+
+                IF EXISTS (SELECT 1 FROM information_schema.tables WHERE LOWER(table_name) = 'collection2item') THEN
+                  WITH ranked AS (
+                    SELECT ctid, row_number() OVER (PARTITION BY collection_id, item_id ORDER BY ctid) AS rn
+                    FROM collection2item
+                  )
+                  DELETE FROM collection2item WHERE ctid IN (SELECT ctid FROM ranked WHERE rn > 1);
+                END IF;
+
+                IF EXISTS (SELECT 1 FROM information_schema.tables WHERE LOWER(table_name) = 'community2collection') THEN
+                  WITH ranked AS (
+                    SELECT ctid, row_number() OVER (PARTITION BY community_id, collection_id ORDER BY ctid) AS rn
+                    FROM community2collection
+                  )
+                  DELETE FROM community2collection WHERE ctid IN (SELECT ctid FROM ranked WHERE rn > 1);
+                END IF;
+
+                IF EXISTS (SELECT 1 FROM information_schema.tables WHERE LOWER(table_name) = 'community2community') THEN
+                  WITH ranked AS (
+                    SELECT ctid, row_number() OVER (PARTITION BY parent_comm_id, child_comm_id ORDER BY ctid) AS rn
+                    FROM community2community
+                  )
+                  DELETE FROM community2community WHERE ctid IN (SELECT ctid FROM ranked WHERE rn > 1);
+                END IF;
+              END \\\$cleanup\\\$;
+
+              \\\\echo 'All duplicate cleanup completed';
+            \" &&
             rm /tmp/$FNAME
           "
           echo "====="