chore: move udf_query into its own query module

Cargo.toml

@@ -6,6 +6,7 @@ members = [
   "guests/python",
   "guests/rust",
   "host",
+  "query",
 ]
 
 [workspace.package]
@@ -24,9 +25,12 @@ datafusion-sql = { version = "49.0.1", default-features = false }
 datafusion-udf-wasm-arrow2bytes = { path = "arrow2bytes", version = "0.1.0" }
 datafusion-udf-wasm-bundle = { path = "guests/bundle", version = "0.1.0" }
 datafusion-udf-wasm-guest = { path = "guests/rust", version = "0.1.0" }
+datafusion-udf-wasm-host = { path = "host", version = "0.1.0" }
 datafusion-udf-wasm-python = { path = "guests/python", version = "0.1.0" }
+datafusion-udf-wasm-query = { path = "query", version = "0.1.0" }
 http = { version = "1.3.1", default-features = false }
 hyper = { version = "1.7", default-features = false }
+insta = { version = "1.43.2", "default-features" = false }
 pyo3 = { version = "0.27.1", default-features = false, features = ["macros"] }
 sqlparser = { version = "0.55.0", default-features = false, features = [
   "std",

host/Cargo.toml

@@ -7,16 +7,13 @@ license.workspace = true
 [dependencies]
 anyhow.workspace = true
 arrow.workspace = true
-datafusion.workspace = true
 datafusion-common.workspace = true
 datafusion-expr.workspace = true
-datafusion-sql.workspace = true
 datafusion-udf-wasm-arrow2bytes.workspace = true
 http.workspace = true
 hyper.workspace = true
 rand = { version = "0.9" }
 siphasher = { version = "1", default-features = false }
-sqlparser.workspace = true
 tar.workspace = true
 tokio = { workspace = true, features = ["rt", "rt-multi-thread", "sync"] }
 wasmtime.workspace = true
@@ -29,7 +26,7 @@ datafusion-udf-wasm-bundle = { workspace = true, features = [
   "example",
   "python"
 ] }
-insta = "1.43.2"
+insta.workspace = true
 tokio = { workspace = true, features = ["fs", "macros"] }
 wiremock = "0.6.5"
 

host/src/lib.rs

@@ -47,7 +47,6 @@ mod error;
 pub mod http;
 mod linker;
 mod tokio_helpers;
-pub mod udf_query;
 pub mod vfs;
 
 /// State of the WASM payload.

host/tests/integration_tests/mod.rs

@@ -1,4 +1,3 @@
 mod python;
 mod rust;
 mod test_utils;
-mod udf_query;

host/tests/integration_tests/python/inspection/errors.rs

@@ -1,6 +1,5 @@
-use datafusion_common::DataFusionError;
-
 use crate::integration_tests::python::test_utils::python_scalar_udfs;
+use datafusion_common::DataFusionError;
 
 #[tokio::test(flavor = "multi_thread")]
 async fn test_invalid_syntax() {

host/tests/integration_tests/python/inspection/filter.rs

@@ -1,6 +1,5 @@
-use datafusion_expr::ScalarUDFImpl;
-
 use crate::integration_tests::python::test_utils::python_scalar_udfs;
+use datafusion_expr::ScalarUDFImpl;
 
 #[tokio::test(flavor = "multi_thread")]
 async fn test_underscore() {

host/tests/integration_tests/python/mod.rs

@@ -3,5 +3,5 @@ mod examples;
 mod inspection;
 mod runtime;
 mod state;
-pub(crate) mod test_utils;
+mod test_utils;
 mod types;

host/tests/integration_tests/python/state.rs

@@ -5,16 +5,15 @@
 
 use std::sync::Arc;
 
+use crate::integration_tests::{
+    python::test_utils::python_scalar_udfs, test_utils::ColumnarValueExt,
+};
 use arrow::{
     array::{Array, ArrayRef, Int64Array},
     datatypes::{DataType, Field},
 };
 use datafusion_expr::{ScalarFunctionArgs, ScalarUDFImpl};
 
-use crate::integration_tests::{
-    python::test_utils::python_scalar_udfs, test_utils::ColumnarValueExt,
-};
-
 const CODE: &str = "
 # Use system module to store our state.
 #

host/tests/integration_tests/python/test_utils.rs

@@ -2,8 +2,10 @@ use datafusion_common::DataFusionError;
 use datafusion_udf_wasm_host::{WasmComponentPrecompiled, WasmScalarUdf};
 use tokio::sync::OnceCell;
 
+/// Static precompiled Python WASM component for tests
 static COMPONENT: OnceCell<WasmComponentPrecompiled> = OnceCell::const_new();
 
+/// Returns a static reference to the precompiled Python WASM component.
 pub(crate) async fn python_component() -> &'static WasmComponentPrecompiled {
     COMPONENT
         .get_or_init(async || {
@@ -14,12 +16,14 @@ pub(crate) async fn python_component() -> &'static WasmComponentPrecompiled {
         .await
 }
 
+/// Compiles the provided Python UDF code into a list of WasmScalarUdf instances.
 pub(crate) async fn python_scalar_udfs(code: &str) -> Result<Vec<WasmScalarUdf>, DataFusionError> {
     let component = python_component().await;
 
     WasmScalarUdf::new(component, &Default::default(), code.to_owned()).await
 }
 
+/// Compiles the provided Python UDF code into a single WasmScalarUdf instance.
 pub(crate) async fn python_scalar_udf(code: &str) -> Result<WasmScalarUdf, DataFusionError> {
     let udfs = python_scalar_udfs(code).await?;
     assert_eq!(udfs.len(), 1);

query/Cargo.toml

@@ -0,0 +1,22 @@
+[package]
+name = "datafusion-udf-wasm-query"
+version.workspace = true
+edition.workspace = true
+license.workspace = true
+
+[dependencies]
+datafusion.workspace = true
+datafusion-common.workspace = true
+datafusion-expr.workspace = true
+datafusion-sql.workspace = true
+datafusion-udf-wasm-host.workspace = true
+sqlparser.workspace = true
+tokio.workspace = true
+
+[dev-dependencies]
+datafusion-udf-wasm-bundle = { workspace = true, features = ["python"] }
+datafusion-udf-wasm-host = { workspace = true }
+insta.workspace = true
+
+[lints]
+workspace = true

host/src/udf_query.rs → query/src/lib.rs

@@ -1,4 +1,5 @@
 //! Embedded SQL approach for executing UDFs within SQL queries.
+#![allow(unused_crate_dependencies)]
 
 use std::collections::HashMap;
 
@@ -8,7 +9,7 @@ use datafusion_sql::parser::{DFParserBuilder, Statement};
 use sqlparser::ast::{CreateFunctionBody, Expr, Statement as SqlStatement, Value};
 use sqlparser::dialect::dialect_from_str;
 
-use crate::{WasmComponentPrecompiled, WasmPermissions, WasmScalarUdf};
+use datafusion_udf_wasm_host::{WasmComponentPrecompiled, WasmPermissions, WasmScalarUdf};
 
 /// A [ParsedQuery] contains the extracted UDFs and SQL query string
 #[derive(Debug)]

host/tests/integration_tests/udf_query.rs → query/tests/integration.rs

@@ -1,14 +1,21 @@
+#![expect(
+    // Docs are not strictly required for tests.
+    missing_docs,
+    // unused-crate-dependencies false positives
+    unused_crate_dependencies,
+)]
+
 use std::collections::HashMap;
 
 use datafusion::{
     assert_batches_eq,
     prelude::{DataFrame, SessionContext},
 };
 use datafusion_common::{Result as DataFusionResult, test_util::batches_to_string};
-use datafusion_udf_wasm_host::{
-    WasmPermissions,
-    udf_query::{ParsedQuery, UdfQueryParser},
-};
+use datafusion_udf_wasm_host::WasmPermissions;
+use datafusion_udf_wasm_query::{ParsedQuery, UdfQueryParser};
+
+mod integration_tests;
 
 use crate::integration_tests::python::test_utils::python_component;
 

query/tests/integration_tests/mod.rs

@@ -0,0 +1 @@
+pub(crate) mod python;

query/tests/integration_tests/python/mod.rs

@@ -0,0 +1 @@
+pub(crate) mod test_utils;

query/tests/integration_tests/python/test_utils.rs

@@ -0,0 +1,16 @@
+use datafusion_udf_wasm_host::WasmComponentPrecompiled;
+use tokio::sync::OnceCell;
+
+/// Static precompiled Python WASM component for tests
+static COMPONENT: OnceCell<WasmComponentPrecompiled> = OnceCell::const_new();
+
+/// Returns a static reference to the precompiled Python WASM component.
+pub(crate) async fn python_component() -> &'static WasmComponentPrecompiled {
+    COMPONENT
+        .get_or_init(async || {
+            WasmComponentPrecompiled::new(datafusion_udf_wasm_bundle::BIN_PYTHON.into())
+                .await
+                .unwrap()
+        })
+        .await
+}