# docker-images

**Repository Path**: ymjake/docker-images

## Basic Information

- **Project Name**: docker-images
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2026-04-13
- **Last Updated**: 2026-04-13

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# PostgreSQL AI Extensions Image

This directory contains a focused PostgreSQL 18 image for AI and RAG scenarios.

## Included extensions

- `vector`
- `age`
- `pg_textsearch`
- `pg_trgm`
- `unaccent`

## Why this image exists

- Keep the database image focused on the extensions used by the knowledge-base POC.
- Avoid carrying unrelated spatial, routing, analytics, and TIGER geocoder dependencies.
- Make hybrid retrieval and graph experimentation reproducible across projects.

## Files

- `Dockerfile`: builds PostgreSQL 18 with the five required extensions
- `docker-compose.yml`: local run configuration
- `init-scripts/01-init-extensions-ai.sql`: initializes the extensions on first startup

## Build

```powershell
docker build -t local/postgres-ai:18 .
```

## Run

```powershell
docker compose up -d --build
```

## Verify

```sql
SELECT * FROM pg_available_extensions
WHERE name IN ('vector', 'age', 'pg_textsearch', 'pg_trgm', 'unaccent');

SELECT extname, extversion
FROM pg_extension
ORDER BY extname;
```

## Smoke test

```sql
SELECT '[1,2,3]'::vector;

LOAD 'age';
SET search_path = ag_catalog, "$user", public;
SELECT create_graph('smoke_graph');

CREATE TABLE docs (
    id bigserial PRIMARY KEY,
    content text NOT NULL
);

INSERT INTO docs (content)
VALUES ('warehouse allocation rule'), ('delayed shipment support');

CREATE INDEX docs_bm25_idx
ON docs USING bm25(content)
WITH (text_config = 'english');

SELECT *
FROM docs
ORDER BY content <@> 'allocation rule'
LIMIT 5;
```

## Notes

- `pgvector` is installed from the PostgreSQL 18 package.
- `Apache AGE` is compiled from source for PostgreSQL 18.
- `pg_textsearch` is compiled from source and preloaded at startup.
- If you upgrade PostgreSQL major versions, re-check the Apache AGE tag and `pg_textsearch` compatibility first.