Merge branch 'datafusion-contrib:main' into main

Author: fantayeneh

Cargo.toml

@@ -1,25 +1,25 @@
 [package]
 name = "datafusion-functions-json"
-version = "0.49.0"
+version = "0.50.0"
 edition = "2021"
 description = "JSON functions for DataFusion"
 readme = "README.md"
 license = "Apache-2.0"
 keywords = ["datafusion", "JSON", "SQL"]
 categories = ["database-implementations", "parsing"]
 repository = "https://github.com/datafusion-contrib/datafusion-functions-json/"
-rust-version = "1.85.1"
+rust-version = "1.86.0"
 
 [dependencies]
-datafusion = { version = "49", default-features = false }
+datafusion = { version = "50", default-features = false }
 jiter = "0.10"
 log = "0.4"
 jsonpath-rust = "1.0.0"
 paste = "1"
 
 [dev-dependencies]
 codspeed-criterion-compat = "2.6"
-datafusion = { version = "49", default-features = false, features = [
+datafusion = { version = "50", default-features = false, features = [
     "nested_expressions",
 ] }
 tokio = { version = "1.43", features = ["full"] }

benches/main.rs

@@ -33,6 +33,7 @@ fn bench_json_contains(b: &mut Bencher) {
                 number_rows: 1,
                 arg_fields: arg_fields.clone(),
                 return_field: return_field.clone(),
+                config_options: Arc::new(datafusion::config::ConfigOptions::default()),
             })
             .unwrap()
     });
@@ -63,6 +64,7 @@ fn bench_json_get_str_scalar(b: &mut Bencher) {
                 arg_fields: arg_fields.clone(),
                 number_rows: 1,
                 return_field: return_field.clone(),
+                config_options: Arc::new(datafusion::config::ConfigOptions::default()),
             })
             .unwrap();
     });
@@ -94,6 +96,7 @@ fn bench_json_get_str_array(b: &mut Bencher) {
                 arg_fields: arg_fields.clone(),
                 number_rows: 1,
                 return_field: return_field.clone(),
+                config_options: Arc::new(datafusion::config::ConfigOptions::default()),
             })
             .unwrap();
     });
@@ -125,6 +128,7 @@ fn bench_json_get_str_view_array(b: &mut Bencher) {
                 arg_fields: arg_fields.clone(),
                 number_rows: 1,
                 return_field: return_field.clone(),
+                config_options: Arc::new(datafusion::config::ConfigOptions::default()),
             })
             .unwrap();
     });

src/common_macros.rs

@@ -30,7 +30,7 @@ macro_rules! make_udf_function {
             static [< STATIC_ $expr_fn_name:upper >]: std::sync::OnceLock<std::sync::Arc<datafusion::logical_expr::ScalarUDF>> =
                 std::sync::OnceLock::new();
 
-            /// ScalarFunction that returns a [`ScalarUDF`] for [`$udf_impl`]
+            /// `ScalarFunction` that returns a [`ScalarUDF`] for [`$udf_impl`]
             ///
             /// [`ScalarUDF`]: datafusion::logical_expr::ScalarUDF
             pub fn [< $expr_fn_name _udf >]() -> std::sync::Arc<datafusion::logical_expr::ScalarUDF> {

src/common_union.rs

@@ -1,5 +1,5 @@
 use std::collections::HashMap;
-use std::sync::{Arc, OnceLock};
+use std::sync::{Arc, LazyLock, OnceLock};
 
 use datafusion::arrow::array::{
     Array, ArrayRef, AsArray, BooleanArray, Float64Array, Int64Array, NullArray, StringArray, UnionArray,
@@ -48,6 +48,8 @@ pub(crate) fn json_from_union_scalar<'a>(
     None
 }
 
+pub static JSON_UNION_DATA_TYPE: LazyLock<DataType> = LazyLock::new(JsonUnion::data_type);
+
 #[derive(Debug)]
 pub(crate) struct JsonUnion {
     bools: Vec<Option<bool>>,
@@ -255,7 +257,7 @@ impl JsonUnionEncoder {
     ///
     /// Panics if the idx is outside the union values or an invalid type id exists in the union.
     #[must_use]
-    pub fn get_value(&self, idx: usize) -> JsonUnionValue {
+    pub fn get_value(&self, idx: usize) -> JsonUnionValue<'_> {
         let type_id = self.type_ids[idx];
         match type_id {
             TYPE_ID_NULL => JsonUnionValue::JsonNull,

src/json_as_text.rs

@@ -17,7 +17,7 @@ make_udf_function!(
     r#"Get any value from a JSON string by its "path", represented as a string"#
 );
 
-#[derive(Debug)]
+#[derive(Debug, PartialEq, Eq, Hash)]
 pub(super) struct JsonAsText {
     signature: Signature,
     aliases: [String; 1],

src/json_contains.rs

@@ -17,7 +17,7 @@ make_udf_function!(
     r#"Does the key/index exist within the JSON value as the specified "path"?"#
 );
 
-#[derive(Debug)]
+#[derive(Debug, PartialEq, Eq, Hash)]
 pub(super) struct JsonContains {
     signature: Signature,
     aliases: [String; 1],

src/json_get.rs

@@ -23,7 +23,7 @@ make_udf_function!(
 
 // build_typed_get!(JsonGet, "json_get", Union, Float64Array, jiter_json_get_float);
 
-#[derive(Debug)]
+#[derive(Debug, PartialEq, Eq, Hash)]
 pub(super) struct JsonGet {
     signature: Signature,
     aliases: [String; 1],

src/json_get_array.rs

@@ -17,7 +17,7 @@ make_udf_function!(
     r#"Get an arrow array from a JSON string by its "path""#
 );
 
-#[derive(Debug)]
+#[derive(Debug, PartialEq, Eq, Hash)]
 pub(super) struct JsonGetArray {
     signature: Signature,
     aliases: [String; 1],

src/json_get_bool.rs

@@ -16,7 +16,7 @@ make_udf_function!(
     r#"Get an boolean value from a JSON string by its "path""#
 );
 
-#[derive(Debug)]
+#[derive(Debug, PartialEq, Eq, Hash)]
 pub(super) struct JsonGetBool {
     signature: Signature,
     aliases: [String; 1],

src/json_get_float.rs

@@ -17,7 +17,7 @@ make_udf_function!(
     r#"Get a float value from a JSON string by its "path""#
 );
 
-#[derive(Debug)]
+#[derive(Debug, PartialEq, Eq, Hash)]
 pub(super) struct JsonGetFloat {
     signature: Signature,
     aliases: [String; 1],