feat: add pgvector support with typed numpy arrays

  Add support for PostgreSQL vector type with proper numpy typing:
  - Map 'vector' type to NDArray[numpy.float32]
  - Auto-import numpy and numpy.typing.NDArray
  - Add test case for vector type generation
  - Add .DS_Store to .gitignore

  This enables type-safe pgvector usage in generated Python code.

.gitignore

@@ -1,5 +1,8 @@
 bin
 
+# macOS
+.DS_Store
+
 # Devenv
 .envrc
 .direnv

internal/endtoend/testdata/emit_numpy_array/python/models.py

@@ -0,0 +1,13 @@
+# Code generated by sqlc. DO NOT EDIT.
+# versions:
+#   sqlc v1.30.0
+import dataclasses
+import numpy
+from numpy.typing import NDArray
+from typing import Optional
+
+
+@dataclasses.dataclass()
+class Item:
+    id: int
+    embedding: Optional[NDArray[numpy.float32]]

internal/endtoend/testdata/emit_numpy_array/python/query.py

@@ -0,0 +1,68 @@
+# Code generated by sqlc. DO NOT EDIT.
+# versions:
+#   sqlc v1.30.0
+# source: query.sql
+import numpy
+from numpy.typing import NDArray
+from typing import Optional
+
+import sqlalchemy
+import sqlalchemy.ext.asyncio
+
+from python import models
+
+
+CREATE_ITEM = """-- name: create_item \\:one
+INSERT INTO items (embedding) VALUES (:p1) RETURNING id, embedding
+"""
+
+
+GET_ITEM = """-- name: get_item \\:one
+SELECT id, embedding FROM items WHERE id = :p1
+"""
+
+
+class Querier:
+    def __init__(self, conn: sqlalchemy.engine.Connection):
+        self._conn = conn
+
+    def create_item(self, *, embedding: Optional[NDArray[numpy.float32]]) -> Optional[models.Item]:
+        row = self._conn.execute(sqlalchemy.text(CREATE_ITEM), {"p1": embedding}).first()
+        if row is None:
+            return None
+        return models.Item(
+            id=row[0],
+            embedding=row[1],
+        )
+
+    def get_item(self, *, id: int) -> Optional[models.Item]:
+        row = self._conn.execute(sqlalchemy.text(GET_ITEM), {"p1": id}).first()
+        if row is None:
+            return None
+        return models.Item(
+            id=row[0],
+            embedding=row[1],
+        )
+
+
+class AsyncQuerier:
+    def __init__(self, conn: sqlalchemy.ext.asyncio.AsyncConnection):
+        self._conn = conn
+
+    async def create_item(self, *, embedding: Optional[NDArray[numpy.float32]]) -> Optional[models.Item]:
+        row = (await self._conn.execute(sqlalchemy.text(CREATE_ITEM), {"p1": embedding})).first()
+        if row is None:
+            return None
+        return models.Item(
+            id=row[0],
+            embedding=row[1],
+        )
+
+    async def get_item(self, *, id: int) -> Optional[models.Item]:
+        row = (await self._conn.execute(sqlalchemy.text(GET_ITEM), {"p1": id})).first()
+        if row is None:
+            return None
+        return models.Item(
+            id=row[0],
+            embedding=row[1],
+        )

internal/endtoend/testdata/emit_numpy_array/query.sql

@@ -0,0 +1,6 @@
+-- name: GetItem :one
+SELECT * FROM items WHERE id = $1;
+
+
+-- name: CreateItem :one
+INSERT INTO items (embedding) VALUES ($1) RETURNING *;

internal/endtoend/testdata/emit_numpy_array/schema.sql

@@ -0,0 +1,4 @@
+CREATE TABLE items (
+    id SERIAL PRIMARY KEY,
+    embedding vector(3)
+);

internal/endtoend/testdata/emit_numpy_array/sqlc.yaml

@@ -0,0 +1,17 @@
+version: '2'
+plugins:
+- name: py
+  wasm:
+    url: file://../../../../bin/sqlc-gen-python.wasm
+    sha256: "839af1f07c31644548192fc095569e62f1511d72c1c30c1a958ddc9c9429edbc"
+sql:
+- schema: schema.sql
+  queries: query.sql
+  engine: postgresql
+  codegen:
+  - plugin: py
+    out: python
+    options:
+      package: python
+      emit_sync_querier: true
+      emit_async_querier: true

internal/imports.go

@@ -272,5 +272,9 @@ func stdImports(uses func(name string) bool) map[string]importSpec {
 	if uses("Any") {
 		std["typing.Any"] = importSpec{Module: "typing", Name: "Any"}
 	}
+	if uses("NDArray[numpy.float32]") {
+		std["numpy"] = importSpec{Module: "numpy"}
+		std["numpy.typing.NDArray"] = importSpec{Module: "numpy.typing", Name: "NDArray"}
+	}
 	return std
 }

internal/postgresql_type.go

@@ -42,6 +42,8 @@ func postgresType(req *plugin.GenerateRequest, col *plugin.Column) string {
 		return "str"
 	case "ltree", "lquery", "ltxtquery":
 		return "str"
+	case "vector":
+		return "NDArray[numpy.float32]"
 	default:
 		for _, schema := range req.Catalog.Schemas {
 			if schema.Name == "pg_catalog" || schema.Name == "information_schema" {