diff --git a/CHANGELOG.md b/CHANGELOG.md
index 897badb..df0bf7e 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -6,6 +6,10 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 
 ## [Unreleased]
 
+### Added
+
+- Construct `stac_api::Search` (moved from `stac_api` crate) ([#81](https://github.com/stac-utils/stacrs/pull/81))
+
 ### Fixed
 
 - Swallow broken pipe errors ([#73](https://github.com/stac-utils/stacrs/pull/73))
diff --git a/Cargo.lock b/Cargo.lock
index 3362b95..35e7864 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -2695,7 +2695,7 @@ dependencies = [
 [[package]]
 name = "pgstac"
 version = "0.3.0"
-source = "git+https://github.com/stac-utils/stac-rs?branch=main#83fd636ef3b21d65761a21228a1ff2ca094b5090"
+source = "git+https://github.com/stac-utils/stac-rs?branch=main#e59405cf4d566218cc9fa5fef52b21d96244ef5f"
 dependencies = [
  "serde",
  "serde_json",
@@ -3398,9 +3398,9 @@ dependencies = [
 
 [[package]]
 name = "rustix"
-version = "1.0.1"
+version = "1.0.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "dade4812df5c384711475be5fcd8c162555352945401aed22a35bffeab61f657"
+checksum = "f7178faa4b75a30e269c71e61c353ce2748cf3d76f0c44c393f4e60abf49b825"
 dependencies = [
  "bitflags 2.9.0",
  "errno",
@@ -3712,7 +3712,7 @@ checksum = "a8f112729512f8e442d81f95a8a7ddf2b7c6b8a1a6f509a95864142b30cab2d3"
 [[package]]
 name = "stac"
 version = "0.12.0"
-source = "git+https://github.com/stac-utils/stac-rs?branch=main#83fd636ef3b21d65761a21228a1ff2ca094b5090"
+source = "git+https://github.com/stac-utils/stac-rs?branch=main#e59405cf4d566218cc9fa5fef52b21d96244ef5f"
 dependencies = [
  "arrow-array",
  "arrow-cast",
@@ -3743,7 +3743,7 @@ dependencies = [
 [[package]]
 name = "stac-api"
 version = "0.7.1"
-source = "git+https://github.com/stac-utils/stac-rs?branch=main#83fd636ef3b21d65761a21228a1ff2ca094b5090"
+source = "git+https://github.com/stac-utils/stac-rs?branch=main#e59405cf4d566218cc9fa5fef52b21d96244ef5f"
 dependencies = [
  "async-stream",
  "chrono",
@@ -3752,8 +3752,6 @@ dependencies = [
  "geo",
  "geojson",
  "http",
- "pyo3",
- "pythonize",
  "reqwest",
  "serde",
  "serde_json",
@@ -3769,7 +3767,7 @@ dependencies = [
 [[package]]
 name = "stac-cli"
 version = "0.5.3"
-source = "git+https://github.com/stac-utils/stac-rs?branch=main#83fd636ef3b21d65761a21228a1ff2ca094b5090"
+source = "git+https://github.com/stac-utils/stac-rs?branch=main#e59405cf4d566218cc9fa5fef52b21d96244ef5f"
 dependencies = [
  "anyhow",
  "axum",
@@ -3787,7 +3785,7 @@ dependencies = [
 [[package]]
 name = "stac-derive"
 version = "0.2.0"
-source = "git+https://github.com/stac-utils/stac-rs?branch=main#83fd636ef3b21d65761a21228a1ff2ca094b5090"
+source = "git+https://github.com/stac-utils/stac-rs?branch=main#e59405cf4d566218cc9fa5fef52b21d96244ef5f"
 dependencies = [
  "quote",
  "syn 2.0.100",
@@ -3796,7 +3794,7 @@ dependencies = [
 [[package]]
 name = "stac-duckdb"
 version = "0.1.1"
-source = "git+https://github.com/stac-utils/stac-rs?branch=main#83fd636ef3b21d65761a21228a1ff2ca094b5090"
+source = "git+https://github.com/stac-utils/stac-rs?branch=main#e59405cf4d566218cc9fa5fef52b21d96244ef5f"
 dependencies = [
  "arrow",
  "chrono",
@@ -3814,7 +3812,7 @@ dependencies = [
 [[package]]
 name = "stac-server"
 version = "0.3.4"
-source = "git+https://github.com/stac-utils/stac-rs?branch=main#83fd636ef3b21d65761a21228a1ff2ca094b5090"
+source = "git+https://github.com/stac-utils/stac-rs?branch=main#e59405cf4d566218cc9fa5fef52b21d96244ef5f"
 dependencies = [
  "axum",
  "bb8",
diff --git a/Cargo.toml b/Cargo.toml
index 3674b0b..9e4562d 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -30,7 +30,6 @@ stac = { features = [
 ], git = "https://github.com/stac-utils/stac-rs", branch = "main" }
 stac-api = { features = [
     "client",
-    "python",
 ], git = "https://github.com/stac-utils/stac-rs", branch = "main" }
 stac-cli = { git = "https://github.com/stac-utils/stac-rs", features = [
     "pgstac",
diff --git a/src/duckdb.rs b/src/duckdb.rs
index 71382a6..e42ec86 100644
--- a/src/duckdb.rs
+++ b/src/duckdb.rs
@@ -1,4 +1,7 @@
-use crate::Result;
+use crate::{
+    search::{PySortby, StringOrDict, StringOrList},
+    Result,
+};
 use pyo3::{
     exceptions::PyException,
     prelude::*,
@@ -6,7 +9,6 @@ use pyo3::{
     IntoPyObjectExt,
 };
 use pyo3_arrow::PyTable;
-use stac_api::python::{StringOrDict, StringOrList};
 use stac_duckdb::{Client, Config};
 use std::sync::Mutex;
 
@@ -16,11 +18,24 @@ pub struct DuckdbClient(Mutex<Client>);
 #[pymethods]
 impl DuckdbClient {
     #[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> {
+    #[pyo3(signature = (*, use_s3_credential_chain=true, use_azure_credential_chain=true, use_httpfs=true, use_hive_partitioning=false, install_extensions=true, custom_extension_repository=None, extension_directory=None))]
+    fn new(
+        use_s3_credential_chain: bool,
+        use_azure_credential_chain: bool,
+        use_httpfs: bool,
+        use_hive_partitioning: bool,
+        install_extensions: bool,
+        custom_extension_repository: Option<String>,
+        extension_directory: Option<String>,
+    ) -> Result<DuckdbClient> {
         let config = Config {
             use_s3_credential_chain,
+            use_azure_credential_chain,
+            use_httpfs,
             use_hive_partitioning,
+            install_extensions,
+            custom_extension_repository,
+            extension_directory,
             convert_wkb: true,
         };
         let client = Client::with_config(config)?;
@@ -40,12 +55,12 @@ impl DuckdbClient {
         datetime: Option<String>,
         include: Option<StringOrList>,
         exclude: Option<StringOrList>,
-        sortby: Option<StringOrList>,
+        sortby: Option<PySortby<'py>>,
         filter: Option<StringOrDict>,
         query: Option<Bound<'py, PyDict>>,
         kwargs: Option<Bound<'py, PyDict>>,
     ) -> Result<Bound<'py, PyDict>> {
-        let search = stac_api::python::search(
+        let search = crate::search::build(
             intersects,
             ids,
             collections,
@@ -84,12 +99,12 @@ impl DuckdbClient {
         datetime: Option<String>,
         include: Option<StringOrList>,
         exclude: Option<StringOrList>,
-        sortby: Option<StringOrList>,
+        sortby: Option<PySortby<'py>>,
         filter: Option<StringOrDict>,
         query: Option<Bound<'py, PyDict>>,
         kwargs: Option<Bound<'py, PyDict>>,
     ) -> Result<PyObject> {
-        let search = stac_api::python::search(
+        let search = crate::search::build(
             intersects,
             ids,
             collections,
diff --git a/src/search.rs b/src/search.rs
index 339e2f9..af8a3d9 100644
--- a/src/search.rs
+++ b/src/search.rs
@@ -1,10 +1,10 @@
 use crate::{Error, Json, Result};
-use pyo3::{prelude::*, types::PyDict};
+use geojson::Geometry;
+use pyo3::prelude::*;
+use pyo3::{exceptions::PyValueError, types::PyDict, Bound, FromPyObject, PyErr, PyResult};
+use stac::Bbox;
 use stac::Format;
-use stac_api::{
-    python::{StringOrDict, StringOrList},
-    Search,
-};
+use stac_api::{Fields, Filter, Items, Search, Sortby};
 
 #[pyfunction]
 #[pyo3(signature = (href, *, intersects=None, ids=None, collections=None, max_items=None, limit=None, bbox=None, datetime=None, include=None, exclude=None, sortby=None, filter=None, query=None, use_duckdb=None, **kwargs))]
@@ -21,13 +21,13 @@ pub fn search<'py>(
     datetime: Option<String>,
     include: Option<StringOrList>,
     exclude: Option<StringOrList>,
-    sortby: Option<StringOrList>,
+    sortby: Option<PySortby<'py>>,
     filter: Option<StringOrDict>,
     query: Option<Bound<'py, PyDict>>,
     use_duckdb: Option<bool>,
     kwargs: Option<Bound<'_, PyDict>>,
 ) -> PyResult<Bound<'py, PyAny>> {
-    let search = stac_api::python::search(
+    let search = build(
         intersects,
         ids,
         collections,
@@ -72,7 +72,7 @@ pub fn search_to<'py>(
     datetime: Option<String>,
     include: Option<StringOrList>,
     exclude: Option<StringOrList>,
-    sortby: Option<StringOrList>,
+    sortby: Option<PySortby<'py>>,
     filter: Option<StringOrDict>,
     query: Option<Bound<'py, PyDict>>,
     format: Option<String>,
@@ -80,7 +80,7 @@ pub fn search_to<'py>(
     use_duckdb: Option<bool>,
     kwargs: Option<Bound<'_, PyDict>>,
 ) -> PyResult<Bound<'py, PyAny>> {
-    let search = stac_api::python::search(
+    let search = build(
         intersects,
         ids,
         collections,
@@ -150,3 +150,144 @@ async fn search_api(
     let value = stac_api::client::search(&href, search, max_items).await?;
     Ok(value)
 }
+
+/// Creates a [Search] from Python arguments.
+#[allow(clippy::too_many_arguments)]
+pub fn build<'py>(
+    intersects: Option<StringOrDict<'py>>,
+    ids: Option<StringOrList>,
+    collections: Option<StringOrList>,
+    limit: Option<u64>,
+    bbox: Option<Vec<f64>>,
+    datetime: Option<String>,
+    include: Option<StringOrList>,
+    exclude: Option<StringOrList>,
+    sortby: Option<PySortby<'py>>,
+    filter: Option<StringOrDict<'py>>,
+    query: Option<Bound<'py, PyDict>>,
+    kwargs: Option<Bound<'py, PyDict>>,
+) -> PyResult<Search> {
+    let mut fields = Fields::default();
+    if let Some(include) = include {
+        fields.include = include.into();
+    }
+    if let Some(exclude) = exclude {
+        fields.exclude = exclude.into();
+    }
+    let fields = if fields.include.is_empty() && fields.exclude.is_empty() {
+        None
+    } else {
+        Some(fields)
+    };
+    let query = query
+        .map(|query| pythonize::depythonize(&query))
+        .transpose()?;
+    let bbox = bbox.map(Bbox::try_from).transpose().map_err(Error::from)?;
+    let sortby: Vec<Sortby> = sortby
+        .map(|sortby| match sortby {
+            PySortby::ListOfDicts(list) => list
+                .into_iter()
+                .map(|d| pythonize::depythonize(&d).map_err(Error::from))
+                .collect::<Result<Vec<_>>>(),
+            PySortby::ListOfStrings(list) => list
+                .into_iter()
+                .map(|s| Ok(s.parse().unwrap())) // infallible
+                .collect::<Result<Vec<_>>>(),
+            PySortby::String(s) => Ok(vec![s.parse().unwrap()]),
+        })
+        .transpose()?
+        .unwrap_or_default();
+    let filter = filter
+        .map(|filter| match filter {
+            StringOrDict::Dict(cql_json) => pythonize::depythonize(&cql_json).map(Filter::Cql2Json),
+            StringOrDict::String(cql2_text) => Ok(Filter::Cql2Text(cql2_text)),
+        })
+        .transpose()?;
+    let filter = filter
+        .map(|filter| filter.into_cql2_json())
+        .transpose()
+        .map_err(Error::from)?;
+    let mut items = Items {
+        limit,
+        bbox,
+        datetime,
+        query,
+        fields,
+        sortby,
+        filter,
+        ..Default::default()
+    };
+    if let Some(kwargs) = kwargs {
+        items.additional_fields = pythonize::depythonize(&kwargs)?;
+    }
+
+    let intersects = intersects
+        .map(|intersects| match intersects {
+            StringOrDict::Dict(json) => pythonize::depythonize(&json)
+                .map_err(PyErr::from)
+                .and_then(|json| {
+                    Geometry::from_json_object(json)
+                        .map_err(|err| PyValueError::new_err(err.to_string()))
+                }),
+            StringOrDict::String(s) => s
+                .parse::<Geometry>()
+                .map_err(|err| PyValueError::new_err(err.to_string())),
+        })
+        .transpose()?;
+    let ids = ids.map(|ids| ids.into()).unwrap_or_default();
+    let collections = collections.map(|ids| ids.into()).unwrap_or_default();
+    Ok(Search {
+        items,
+        intersects,
+        ids,
+        collections,
+    })
+}
+
+/// A string or dictionary.
+///
+/// Used for the CQL2 filter argument and for intersects.
+#[derive(Debug, FromPyObject)]
+pub enum StringOrDict<'py> {
+    /// Text
+    String(String),
+
+    /// Json
+    Dict(Bound<'py, PyDict>),
+}
+
+/// A string or a list.
+///
+/// Used for collections, ids, etc.
+#[derive(Debug, FromPyObject)]
+pub enum StringOrList {
+    /// A string.
+    String(String),
+
+    /// A list.
+    List(Vec<String>),
+}
+
+/// A sortby structure.
+///
+/// This can be a string, a list of strings, or a list of dictionaries.
+#[derive(Debug, FromPyObject)]
+pub enum PySortby<'py> {
+    /// A string.
+    String(String),
+
+    /// A list.
+    ListOfStrings(Vec<String>),
+
+    /// A list.
+    ListOfDicts(Vec<Bound<'py, PyDict>>),
+}
+
+impl From<StringOrList> for Vec<String> {
+    fn from(value: StringOrList) -> Vec<String> {
+        match value {
+            StringOrList::List(list) => list,
+            StringOrList::String(s) => vec![s],
+        }
+    }
+}
diff --git a/stacrs.pyi b/stacrs.pyi
index c04f81b..b3425c4 100644
--- a/stacrs.pyi
+++ b/stacrs.pyi
@@ -6,15 +6,30 @@ class DuckdbClient:
     """A client for querying stac-geoparquet with DuckDB."""
 
     def __init__(
-        self, use_s3_credential_chain: bool = True, use_hive_partitioning: bool = False
+        self,
+        *,
+        use_s3_credential_chain: bool = True,
+        use_azure_credential_chain: bool = True,
+        use_httpfs: bool = True,
+        use_hive_partitioning: bool = False,
+        install_extensions: bool = True,
+        custom_extension_repository: str | None = None,
+        extension_directory: str | None = None,
     ) -> None:
         """Creates a new duckdb client.
 
         Args:
             use_s3_credential_chain: If true, configures DuckDB to correctly
                 handle s3:// urls.
+            use_azure_credential_chain: If true, configures DuckDB to correctly
+                handle azure urls.
+            use_https_credential_chain: If true, configures DuckDB to correctly
+                handle https urls.
             use_hive_partitioning: If true, enables queries on hive partitioned
                 geoparquet files.
+            install_extensions: If true, installs extensions before loading them.
+            custom_extension_repository: A custom extension repository to use.
+            extension_directory: A non-standard extension directory to use.
         """
 
     def search(
@@ -30,7 +45,7 @@ class DuckdbClient:
         datetime: Optional[str] = None,
         include: Optional[str | list[str]] = None,
         exclude: Optional[str | list[str]] = None,
-        sortby: Optional[str | list[str]] = None,
+        sortby: Optional[str | list[str | dict[str, str]]] = None,
         filter: Optional[str | dict[str, Any]] = None,
         query: Optional[dict[str, Any]] = None,
         **kwargs: str,
@@ -79,7 +94,7 @@ class DuckdbClient:
         datetime: Optional[str] = None,
         include: Optional[str | list[str]] = None,
         exclude: Optional[str | list[str]] = None,
-        sortby: Optional[str | list[str]] = None,
+        sortby: Optional[str | list[str | dict[str, str]]] = None,
         filter: Optional[str | dict[str, Any]] = None,
         query: Optional[dict[str, Any]] = None,
         **kwargs: str,
@@ -260,7 +275,7 @@ async def search(
     datetime: Optional[str] = None,
     include: Optional[str | list[str]] = None,
     exclude: Optional[str | list[str]] = None,
-    sortby: Optional[str | list[str]] = None,
+    sortby: Optional[str | list[str | dict[str, str]]] = None,
     filter: Optional[str | dict[str, Any]] = None,
     query: Optional[dict[str, Any]] = None,
     use_duckdb: Optional[bool] = None,
@@ -329,7 +344,7 @@ async def search_to(
     datetime: Optional[str] = None,
     include: Optional[str | list[str]] = None,
     exclude: Optional[str | list[str]] = None,
-    sortby: Optional[str | list[str]] = None,
+    sortby: Optional[str | list[str | dict[str, str]]] = None,
     filter: Optional[str | dict[str, Any]] = None,
     query: Optional[dict[str, Any]] = None,
     format: Optional[str] = None,
diff --git a/tests/test_search.py b/tests/test_search.py
index 472aae7..f31c8f9 100644
--- a/tests/test_search.py
+++ b/tests/test_search.py
@@ -11,7 +11,6 @@ async def test_search() -> None:
         "https://landsatlook.usgs.gov/stac-server",
         collections="landsat-c2l2-sr",
         intersects={"type": "Point", "coordinates": [-105.119, 40.173]},
-        sortby="-properties.datetime",
         max_items=1,
     )
     assert len(items) == 1
@@ -23,7 +22,6 @@ async def test_search_to(tmp_path: Path) -> None:
         "https://landsatlook.usgs.gov/stac-server",
         collections="landsat-c2l2-sr",
         intersects={"type": "Point", "coordinates": [-105.119, 40.173]},
-        sortby="-properties.datetime",
         max_items=1,
     )
     with open(tmp_path / "out.json") as f:
@@ -37,7 +35,6 @@ async def test_search_to_geoparquet(tmp_path: Path) -> None:
         "https://landsatlook.usgs.gov/stac-server",
         collections="landsat-c2l2-sr",
         intersects={"type": "Point", "coordinates": [-105.119, 40.173]},
-        sortby="-properties.datetime",
         max_items=1,
     )
     assert count == 1
@@ -49,3 +46,16 @@ async def test_search_to_geoparquet(tmp_path: Path) -> None:
 async def test_search_geoparquet(data: Path) -> None:
     items = await stacrs.search(str(data / "extended-item.parquet"))
     assert len(items) == 1
+
+
+async def test_sortby_list_of_dict() -> None:
+    items = await stacrs.search(
+        "https://landsatlook.usgs.gov/stac-server",
+        collections="landsat-c2l2-sr",
+        intersects={"type": "Point", "coordinates": [-105.119, 40.173]},
+        sortby=[
+            {"field": "properties.datetime", "direction": "asc"},
+        ],
+        max_items=1,
+    )
+    assert len(items) == 1