Remove duplicate values while keeping the smallest key.
Postgres: Identify retained keys in a grouped subquery before deleting other copies.
CREATE TABLE tags (id INTEGER PRIMARY KEY,label VARCHAR(20));
INSERT INTO tags VALUES (1,'a'),(2,'a'),(3,'b');
DELETE FROM tags WHERE id NOT IN (SELECT keep_id FROM (SELECT MIN(id) AS keep_id FROM tags GROUP BY label) kept);
SELECT id,label FROM tags ORDER BY id;
Expected output EXAMPLE
1|a 3|b
Details
Requires PostgreSQL 18. Run in an empty scratch schema using psql or a SQL client. Result rows below use | between columns, without headers; client display formatting may differ. Identify retained keys in a grouped subquery before deleting other copies. The CREATE and INSERT statements provide all input data. Run the complete example in a fresh scratch database or schema.
Tested on Ubuntu 24.04.5 LTS · · PostgreSQL 18.6 (Debian 18.6-1.pgdg13+2) · bash 5.2.21-2ubuntu4; coreutils 9.4-3ubuntu6.3; psycopg 3.2.10; pymssql 2.3.8
Test inputs, results and environment
The exact published code ran in a dedicated Linux test container with prepared inputs. Snippet process uid 65534 with a read-only root filesystem and temporary working files. Database services use a private internal Docker network without published ports. Fresh scratch schema/database for each case; test database administrator credentials are confined to these disposable services. Fixtures and output records are isolated from the application database. Testing covers this example and its stated checks, not every input, operating system, or runtime version. The Linux container shares its host's kernel; this is not a native Windows test.
- Architecture / host kernel
- x86_64 / 6.18.54.1-microsoft-standard-WSL2
- Snippet identity / locale
- uid=65534, gid=65534 (unprivileged process) / C.UTF-8
- Interpreter invocation
Execute the exact SQL in a fresh database/schema; serialize result rows with | separators- Container image ID
sha256:ec91410a8f6ec3e985830b5e3a0560306d36f8c1c5ccd91a6699abe707123fa3- Code SHA-256
289a2e04797b541c6b98e0c0b393cfb4fef9d9e5e4057a7b7f8d72b58dc42eab- Dependencies
- bash 5.2.21-2ubuntu4; coreutils 9.4-3ubuntu6.3; psycopg 3.2.10; pymssql 2.3.8
Test input setup
These are the prepared inputs used by the test harness. See Details above for the prerequisites and how to run this example.
# No input files required.
Observed standard output
1|a 3|b
Exit code: 0. Standard error: empty. Validation: Exact stdout comparison. System-dependent values above are one recorded run.