feat: add config args to duckdb client

.github/workflows/ci.yaml

@@ -27,7 +27,9 @@ jobs:
       - uses: astral-sh/setup-uv@v3
       - uses: Swatinem/rust-cache@v2
       - name: Sync
-        run: uv sync
+        run: uv sync --no-install-project
+      - name: Build directly with maturin
+        run: uv run maturin dev --uv
       - name: Lint
         run: scripts/lint
       - name: Test

Cargo.toml

@@ -13,7 +13,8 @@ default = ["duckdb-bundled"]
 duckdb-bundled = ["duckdb/bundled"]
 
 [dependencies]
-duckdb = { version = "1.1.1" }
+duckdb = { version = "=1.1.1" }
+libduckdb-sys = { version = "=1.1.1" }
 geojson = "0.24.1"
 pyo3 = { version = "0.23.4", features = ["abi3-py310"] }
 pyo3-async-runtimes = { version = "0.23.0", features = [

src/duckdb.rs

@@ -5,7 +5,7 @@ use pyo3::{
     types::{PyDict, PyList},
 };
 use stac_api::python::{StringOrDict, StringOrList};
-use stac_duckdb::Client;
+use stac_duckdb::{Client, Config};
 use std::sync::Mutex;
 
 #[pyclass(frozen)]
@@ -14,8 +14,13 @@ pub struct DuckdbClient(Mutex<Client>);
 #[pymethods]
 impl DuckdbClient {
     #[new]
-    fn new() -> Result<DuckdbClient> {
-        let client = Client::new()?;
+    #[pyo3(signature = (use_s3_credential_chain=true, use_hive_partitioning=false))]
+    fn new(use_s3_credential_chain: bool, use_hive_partitioning: bool) -> Result<DuckdbClient> {
+        let config = Config {
+            use_s3_credential_chain,
+            use_hive_partitioning,
+        };
+        let client = Client::with_config(config)?;
         Ok(DuckdbClient(Mutex::new(client)))
     }
 

src/lib.rs

@@ -10,6 +10,7 @@ mod write;
 
 use ::duckdb as _;
 use error::Error;
+use libduckdb_sys as _;
 use pyo3::prelude::*;
 
 type Result<T> = std::result::Result<T, Error>;

stacrs.pyi

@@ -3,6 +3,18 @@ from typing import Any, Optional, Tuple
 class DuckdbClient:
     """A client for querying stac-geoparquet with DuckDB."""
 
+    def __init__(
+        self, use_s3_credential_chain: bool = True, use_hive_partitioning: bool = False
+    ) -> None:
+        """Creates a new duckdb client.
+
+        Args:
+            use_s3_credential_chain: If true, configures DuckDB to correctly
+                handle s3:// urls.
+            use_hive_partitioning: If true, enables queries on hive partitioned
+                geoparquet files.
+        """
+
     def search(
         self,
         href: str,

tests/test_duckdb.py

@@ -33,3 +33,7 @@ def test_search_offset(client: DuckdbClient) -> None:
 def test_get_collections(client: DuckdbClient) -> None:
     collections = client.get_collections("data/100-sentinel-2-items.parquet")
     assert len(collections) == 1
+
+
+def test_init_with_config() -> None:
+    DuckdbClient(use_s3_credential_chain=True, use_hive_partitioning=True)