Rename helper to json_field_metadata and move to common_union

`is_json_metadata()` was ambiguous about direction (check vs. mark).
The helper is a constructor of metadata that marks a field as containing
JSON-encoded data. Renaming to `json_field_metadata()` makes that
unambiguous and stays correct if more keys (e.g. canonical Arrow
extension keys) are added to the returned map in the future.

Move it to `src/common_union.rs`, which already houses JSON-typing
concerns.

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

Author: adriangb

src/common.rs

@@ -1,4 +1,3 @@
-use std::collections::HashMap;
 use std::str::Utf8Error;
 use std::sync::Arc;
 
@@ -17,13 +16,6 @@ use crate::common_union::{
     is_json_union, json_from_union_scalar, nested_json_array, nested_json_array_ref, TYPE_ID_NULL,
 };
 
-/// Field metadata that marks a `Utf8` column/field as containing raw JSON.
-/// Downstream consumers (e.g. the rewrite layer, other UDFs) use this to
-/// recognize JSON-bearing string columns.
-pub fn is_json_metadata() -> HashMap<String, String> {
-    HashMap::from([("is_json".to_string(), "true".to_string())])
-}
-
 /// General implementation of `ScalarUDFImpl::return_type`.
 ///
 /// # Arguments

src/common_union.rs

@@ -1,3 +1,4 @@
+use std::collections::HashMap;
 use std::sync::{Arc, LazyLock, OnceLock};
 
 use datafusion::arrow::array::{
@@ -8,7 +9,13 @@ use datafusion::arrow::datatypes::{DataType, Field, UnionFields, UnionMode};
 use datafusion::arrow::error::ArrowError;
 use datafusion::common::ScalarValue;
 
-use crate::common::is_json_metadata;
+/// Field metadata used to mark a `Utf8` field as containing raw JSON.
+///
+/// Attach this to any Arrow `Field` whose values are JSON-encoded strings so
+/// downstream consumers can recognize them as JSON rather than opaque text.
+pub fn json_field_metadata() -> HashMap<String, String> {
+    HashMap::from([("is_json".to_string(), "true".to_string())])
+}
 
 pub fn is_json_union(data_type: &DataType) -> bool {
     match data_type {
@@ -170,11 +177,11 @@ fn union_fields() -> UnionFields {
                 (TYPE_ID_STR, Arc::new(Field::new("str", DataType::Utf8, false))),
                 (
                     TYPE_ID_ARRAY,
-                    Arc::new(Field::new("array", DataType::Utf8, false).with_metadata(is_json_metadata())),
+                    Arc::new(Field::new("array", DataType::Utf8, false).with_metadata(json_field_metadata())),
                 ),
                 (
                     TYPE_ID_OBJECT,
-                    Arc::new(Field::new("object", DataType::Utf8, false).with_metadata(is_json_metadata())),
+                    Arc::new(Field::new("object", DataType::Utf8, false).with_metadata(json_field_metadata())),
                 ),
             ])
         })

src/json_get_array.rs

@@ -7,13 +7,12 @@ use datafusion::common::{Result as DataFusionResult, ScalarValue};
 use datafusion::logical_expr::{ColumnarValue, ScalarFunctionArgs, ScalarUDFImpl, Signature, Volatility};
 use jiter::Peek;
 
-use crate::common::{
-    get_err, invoke, is_json_metadata, jiter_json_find, return_type_check, GetError, InvokeResult, JsonPath,
-};
+use crate::common::{get_err, invoke, jiter_json_find, return_type_check, GetError, InvokeResult, JsonPath};
 use crate::common_macros::make_udf_function;
+use crate::common_union::json_field_metadata;
 
 fn list_item_field() -> Field {
-    Field::new("item", DataType::Utf8, true).with_metadata(is_json_metadata())
+    Field::new("item", DataType::Utf8, true).with_metadata(json_field_metadata())
 }
 
 make_udf_function!(

src/json_get_json.rs

@@ -8,8 +8,9 @@ use datafusion::logical_expr::{
     ColumnarValue, ReturnFieldArgs, ScalarFunctionArgs, ScalarUDFImpl, Signature, Volatility,
 };
 
-use crate::common::{get_err, invoke, is_json_metadata, jiter_json_find, return_type_check, GetError, JsonPath};
+use crate::common::{get_err, invoke, jiter_json_find, return_type_check, GetError, JsonPath};
 use crate::common_macros::make_udf_function;
+use crate::common_union::json_field_metadata;
 
 make_udf_function!(
     JsonGetJson,
@@ -54,7 +55,7 @@ impl ScalarUDFImpl for JsonGetJson {
         let arg_types: Vec<DataType> = args.arg_fields.iter().map(|f| f.data_type().clone()).collect();
         let return_type = self.return_type(&arg_types)?;
         Ok(Arc::new(
-            Field::new(self.name(), return_type, true).with_metadata(is_json_metadata()),
+            Field::new(self.name(), return_type, true).with_metadata(json_field_metadata()),
         ))
     }