get back the key collision check

Signed-off-by: Fredrik Adelöw <freben@gmail.com>

plugins/catalog-backend/src/database/operations/stitcher/buildEntitySearch.test.ts

@@ -236,5 +236,57 @@ describe('buildEntitySearch', () => {
         },
       ]);
     });
+
+    it('rejects duplicate keys', () => {
+      expect(() =>
+        buildEntitySearch('eid', {
+          relations: [],
+          apiVersion: 'a',
+          kind: 'b',
+          metadata: {
+            name: 'n',
+            Namespace: 'ns',
+          },
+        }),
+      ).toThrow(
+        `Entity has duplicate keys that vary only in casing, 'metadata.namespace'`,
+      );
+
+      expect(() =>
+        buildEntitySearch('eid', {
+          relations: [
+            {
+              type: 'dup',
+              targetRef: 'k:ns/a',
+            },
+            {
+              type: 'DUP',
+              targetRef: 'k:ns/b',
+            },
+          ],
+          apiVersion: 'a',
+          kind: 'b',
+          metadata: { name: 'n' },
+        }),
+      ).toThrow(
+        `Entity has duplicate keys that vary only in casing, 'relations.dup'`,
+      );
+
+      expect(() =>
+        buildEntitySearch('eid', {
+          apiVersion: 'a',
+          kind: 'b',
+          metadata: { name: 'n' },
+          spec: {
+            owner: 'o',
+            OWNER: 'o',
+            lifecycle: 'production',
+            lifeCycle: 'production',
+          },
+        }),
+      ).toThrow(
+        `Entity has duplicate keys that vary only in casing, 'spec.owner', 'spec.lifecycle'`,
+      );
+    });
   });
 });

plugins/catalog-backend/src/database/operations/stitcher/buildEntitySearch.ts

@@ -15,6 +15,7 @@
  */
 
 import { DEFAULT_NAMESPACE, Entity } from '@backstage/catalog-model';
+import { InputError } from '@backstage/errors';
 import { DbSearchRow } from '../../tables';
 
 // These are excluded in the generic loop, either because they do not make sense
@@ -199,5 +200,23 @@ export function buildEntitySearch(
     });
   }
 
+  // This validates that there are no keys that vary only in casing, such
+  // as `spec.foo` and `spec.Foo`.
+  const keys = new Set(raw.map(r => r.key));
+  const lowerKeys = new Set(raw.map(r => r.key.toLocaleLowerCase('en-US')));
+  if (keys.size !== lowerKeys.size) {
+    const difference = [];
+    for (const key of keys) {
+      const lower = key.toLocaleLowerCase('en-US');
+      if (!lowerKeys.delete(lower)) {
+        difference.push(lower);
+      }
+    }
+    const badKeys = `'${difference.join("', '")}'`;
+    throw new InputError(
+      `Entity has duplicate keys that vary only in casing, ${badKeys}`,
+    );
+  }
+
   return mapToRows(raw, entityId);
 }

plugins/catalog-backend/src/database/operations/stitcher/performStitching.ts

@@ -209,6 +209,11 @@ export async function performStitching(options: {
       entity.metadata.etag = hash;
     }
 
+    // This may throw if the entity is invalid, so we call it before
+    // the final_entities write, even though we may end up not needing
+    // to write the search index.
+    const searchEntries = buildEntitySearch(entityId, entity);
+
     const amountOfRowsChanged = await knex<DbFinalEntitiesRow>('final_entities')
       .update({
         final_entity: JSON.stringify(entity),
@@ -223,7 +228,6 @@ export async function performStitching(options: {
       return 'abandoned';
     }
 
-    const searchEntries = buildEntitySearch(entityId, entity);
     await knex.transaction(async trx => {
       await trx<DbSearchRow>('search').where({ entity_id: entityId }).delete();
       await trx.batchInsert('search', searchEntries, BATCH_SIZE);