RDS Data API

Run real SQL over HTTP with the RDS Data API (ExecuteStatement) against fakecloud's real PostgreSQL and MySQL containers — typed parameters and results, including binary round-trips.

fakecloud implements the RDS Data API (rds-data) against the same real Docker-backed database containers the RDS service manages. ExecuteStatement runs your SQL on a genuine PostgreSQL or MySQL/MariaDB engine and returns AWS-shaped typed Field records — this is the wedge: other local emulators serve canned rows or return {} for binary columns, while fakecloud runs the real query.

Supported features

  • ExecuteStatement — runs the SQL against the DB the resourceArn resolves to (DB instance or Aurora cluster ARN), on the real container engine.
  • Typed parameters — :name named parameters with stringValue, longValue, doubleValue, booleanValue, blobValue, and isNull, coerced to the column's real type.
  • Typed results — records as typed Fields (integers -> longValue, text -> stringValue, bytea/BLOB -> blobValue, etc.), with optional columnMetadata (includeResultMetadata) and formattedRecords (formatRecordsAs=JSON). Writes report numberOfRecordsUpdated.
  • Binary round-trips — a bytea / BLOB column round-trips through blobValue, where rival emulators return an empty value.
  • Transactions — BeginTransaction returns a transactionId backed by a real database connection held open across requests. ExecuteStatement and BatchExecuteStatement calls that pass the transactionId run on that same connection, and CommitTransaction / RollbackTransaction commit or roll back the genuine engine transaction.
  • BatchExecuteStatement — runs one statement across many parameterSets, returning one updateResults entry per set, on a single connection (the transaction's connection when a transactionId is given).

Example

import boto3
data = boto3.client("rds-data", endpoint_url="http://localhost:4566")
arn = "arn:aws:rds:us-east-1:123456789012:db:mydb"
secret = "arn:aws:secretsmanager:us-east-1:123456789012:secret:db"

data.execute_statement(resourceArn=arn, secretArn=secret,
    sql="CREATE TABLE t (id int, name text)")
data.execute_statement(resourceArn=arn, secretArn=secret,
    sql="INSERT INTO t (id, name) VALUES (:id, :name)",
    parameters=[{"name": "id", "value": {"longValue": 1}},
                {"name": "name", "value": {"stringValue": "alice"}}])
resp = data.execute_statement(resourceArn=arn, secretArn=secret,
    sql="SELECT id, name FROM t", includeResultMetadata=True)
# resp["records"] -> [[{"longValue": 1}, {"stringValue": "alice"}]]

Transactions and batch writes share one held connection:

tx = data.begin_transaction(resourceArn=arn, secretArn=secret)["transactionId"]
data.batch_execute_statement(resourceArn=arn, secretArn=secret, transactionId=tx,
    sql="INSERT INTO t (id, name) VALUES (:id, :name)",
    parameterSets=[
        [{"name": "id", "value": {"longValue": 2}}, {"name": "name", "value": {"stringValue": "bob"}}],
        [{"name": "id", "value": {"longValue": 3}}, {"name": "name", "value": {"stringValue": "carol"}}],
    ])
data.commit_transaction(resourceArn=arn, secretArn=secret, transactionId=tx)