fix(cleanup): pass array values as scalar binds, incompatible with fetch_types false (#6035)

Author: TheodoreSpeaks

apps/sim/lib/execution/payloads/large-value-metadata.ts

@@ -390,13 +390,16 @@ async function pruneStaleReferences(
   batchSize: number,
   dbClient: LargeValueMetadataClient
 ): Promise<number> {
+  // `IN ()` is a syntax error; callers chunk a non-empty list, but never rely on that.
+  if (workspaceIds.length === 0) return 0
+
   const rows = await dbClient.execute<{ count: number }>(sql`
     WITH deleted AS (
       DELETE FROM ${executionLargeValueReferences} AS ref
       WHERE ref.ctid IN (
         SELECT ref.ctid
         FROM ${executionLargeValueReferences} AS ref
-        WHERE ref.workspace_id = ANY(${sql.param(workspaceIds)}::text[])
+        WHERE ref.workspace_id IN ${workspaceIds}
           AND (
             (
               ref.source = 'execution_log'
@@ -431,13 +434,16 @@ async function pruneDeletedParentDependencies(
   batchSize: number,
   dbClient: LargeValueMetadataClient
 ): Promise<number> {
+  // `IN ()` is a syntax error; callers chunk a non-empty list, but never rely on that.
+  if (workspaceIds.length === 0) return 0
+
   const rows = await dbClient.execute<{ count: number }>(sql`
     WITH deleted AS (
       DELETE FROM ${executionLargeValueDependencies} AS dependency
       WHERE dependency.ctid IN (
         SELECT dependency.ctid
         FROM ${executionLargeValueDependencies} AS dependency
-        WHERE dependency.workspace_id = ANY(${sql.param(workspaceIds)}::text[])
+        WHERE dependency.workspace_id IN ${workspaceIds}
           AND (
             EXISTS (
               SELECT 1
@@ -466,13 +472,16 @@ async function pruneDeletedLargeValueTombstones(
   batchSize: number,
   dbClient: LargeValueMetadataClient
 ): Promise<number> {
+  // `IN ()` is a syntax error; callers chunk a non-empty list, but never rely on that.
+  if (workspaceIds.length === 0) return 0
+
   const rows = await dbClient.execute<{ count: number }>(sql`
     WITH deleted AS (
       DELETE FROM ${executionLargeValues} AS value
       WHERE value.ctid IN (
         SELECT value.ctid
         FROM ${executionLargeValues} AS value
-        WHERE value.workspace_id = ANY(${sql.param(workspaceIds)}::text[])
+        WHERE value.workspace_id IN ${workspaceIds}
           AND value.deleted_at IS NOT NULL
           AND value.deleted_at < ${deletedBefore}
           AND NOT EXISTS (

apps/sim/lib/execution/payloads/prune-metadata-sql.test.ts

@@ -5,6 +5,11 @@
 // Renders the real prune statements against the real drizzle dialect and
 // schema. These are raw `sql` templates, so a rendering bug only surfaces at
 // execution time — the global drizzle/schema mocks would hide it entirely.
+//
+// The shared client sets `fetch_types: false` (packages/db/db.ts), which skips
+// loading array type OIDs. That makes an array *parameter* unserializable —
+// postgres.js sends it in a form Postgres rejects with 22P02. Only scalar bind
+// params are safe here, so these tests assert no array parameter is emitted.
 import { describe, expect, it, vi } from 'vitest'
 
 vi.unmock('drizzle-orm')
@@ -16,13 +21,18 @@ process.env.DATABASE_URL ??= 'postgresql://user:pass@localhost:5432/test'
 const { PgDialect } = await import('drizzle-orm/pg-core')
 const { pruneLargeValueMetadata } = await import('@/lib/execution/payloads/large-value-metadata')
 
+interface RenderedQuery {
+  sql: string
+  params: unknown[]
+}
+
 /** Captures the raw statements `pruneLargeValueMetadata` issues, without a database. */
-async function renderPruneStatements(workspaceIds: string[]): Promise<string[]> {
+async function renderPruneStatements(workspaceIds: string[]): Promise<RenderedQuery[]> {
   const dialect = new PgDialect()
-  const rendered: string[] = []
+  const rendered: RenderedQuery[] = []
   const capturingClient = {
     execute: async (query: Parameters<PgDialect['sqlToQuery']>[0]) => {
-      rendered.push(dialect.sqlToQuery(query).sql)
+      rendered.push(dialect.sqlToQuery(query))
       return [{ count: 0 }]
     },
   }
@@ -39,25 +49,33 @@ async function renderPruneStatements(workspaceIds: string[]): Promise<string[]>
 }
 
 describe('pruneLargeValueMetadata SQL', () => {
-  it('binds workspace ids as one array parameter, not a value list', async () => {
-    const statements = await renderPruneStatements(['ws-1', 'ws-2'])
-
-    expect(statements.length).toBeGreaterThan(0)
-    for (const statement of statements) {
-      // `ANY(($1, $2)::text[])` is what interpolating the raw JS array yields —
-      // Postgres rejects it ("cannot cast type record to text[]"), and with a
-      // single id `ANY(($1)::text[])` fails as 22P02 instead.
-      expect(statement).not.toMatch(/ANY\(\(\$\d+/)
-      expect(statement).toMatch(/ANY\(\$\d+::text\[\]\)/)
-    }
-  })
+  for (const [label, ids] of [
+    ['multiple workspace ids', ['ws-1', 'ws-2']],
+    ['a single workspace id', ['ws-only']],
+  ] as const) {
+    describe(label, () => {
+      it('matches workspaces with an IN list of scalar params', async () => {
+        const statements = await renderPruneStatements([...ids])
 
-  it('renders identically for a single workspace id', async () => {
-    const statements = await renderPruneStatements(['ws-only'])
+        expect(statements.length).toBeGreaterThan(0)
+        for (const { sql: text } of statements) {
+          expect(text).toMatch(/workspace_id IN \(\$\d+/)
+          // `= IN (...)` is a syntax error Postgres only reports at execution.
+          expect(text).not.toMatch(/=\s*IN\s*\(/)
+          // `ANY(($1)::text[])` / `ANY(($1, $2)::text[])` — the row-constructor cast.
+          expect(text).not.toMatch(/ANY\(\(\$\d+/)
+        }
+      })
 
-    expect(statements.length).toBeGreaterThan(0)
-    for (const statement of statements) {
-      expect(statement).toMatch(/ANY\(\$\d+::text\[\]\)/)
-    }
-  })
+      it('never binds an array as a parameter', async () => {
+        const statements = await renderPruneStatements([...ids])
+
+        for (const { params } of statements) {
+          for (const param of params) {
+            expect(Array.isArray(param)).toBe(false)
+          }
+        }
+      })
+    })
+  }
 })

apps/sim/lib/webhooks/polling/utils.test.ts

@@ -15,8 +15,9 @@ vi.mock('drizzle-orm', () => {
     sqlCalls.push(node)
     return node
   }
-  // Identity, so a `sql.param(arr)` value still shows up verbatim in `values`.
+  // Identity, so an interpolated value still shows up verbatim in `values`.
   sql.param = (value: unknown) => value
+  sql.join = (fragments: unknown[], separator: unknown) => ({ fragments, separator })
   return {
     sql,
     and: vi.fn(),
@@ -69,7 +70,7 @@ describe('updateWebhookProviderConfig (atomic jsonb merge)', () => {
 
     expect(dbChainMockFns.update).toHaveBeenCalledTimes(1)
     expect(allInterpolatedValues()).toContain(JSON.stringify({ historyId: 'h1', nulled: null }))
-    expect(allInterpolatedValues()).toContainEqual(['cleared'])
+    expect(allInterpolatedValues()).toContain('cleared')
   })
 
   it('uses merge only (no key-removal expression) when nothing is undefined', async () => {

apps/sim/lib/webhooks/polling/utils.ts

@@ -165,7 +165,12 @@ export async function updateWebhookProviderConfig(
 
     const merged = sql`COALESCE(${webhook.providerConfig}::jsonb, '{}'::jsonb) || ${JSON.stringify(defined)}::jsonb`
     const nextConfig =
-      removedKeys.length > 0 ? sql`(${merged}) - ${sql.param(removedKeys)}::text[]` : merged
+      removedKeys.length > 0
+        ? sql`(${merged}) - ARRAY[${sql.join(
+            removedKeys.map((key) => sql`${key}`),
+            sql`, `
+          )}]::text[]`
+        : merged
 
     await db
       .update(webhook)