datafusion_functions_json/
json_object_keys.rs1use std::any::Any;
2use std::sync::Arc;
3
4use datafusion::arrow::array::{ArrayRef, ListBuilder, StringBuilder};
5use datafusion::arrow::datatypes::{DataType, Field};
6use datafusion::common::{Result as DataFusionResult, ScalarValue};
7use datafusion::logical_expr::{ColumnarValue, ScalarFunctionArgs, ScalarUDFImpl, Signature, Volatility};
8use jiter::Peek;
9
10use crate::common::{get_err, invoke, jiter_json_find, return_type_check, GetError, InvokeResult, JsonPath};
11use crate::common_macros::make_udf_function;
12
13make_udf_function!(
14 JsonObjectKeys,
15 json_object_keys,
16 json_data path,
17 r"Get the keys of a JSON object as an array."
18);
19
20#[derive(Debug, PartialEq, Eq, Hash)]
21pub(super) struct JsonObjectKeys {
22 signature: Signature,
23 aliases: [String; 2],
24}
25
26impl Default for JsonObjectKeys {
27 fn default() -> Self {
28 Self {
29 signature: Signature::variadic_any(Volatility::Immutable),
30 aliases: ["json_object_keys".to_string(), "json_keys".to_string()],
31 }
32 }
33}
34
35impl ScalarUDFImpl for JsonObjectKeys {
36 fn as_any(&self) -> &dyn Any {
37 self
38 }
39
40 fn name(&self) -> &str {
41 self.aliases[0].as_str()
42 }
43
44 fn signature(&self) -> &Signature {
45 &self.signature
46 }
47
48 fn return_type(&self, arg_types: &[DataType]) -> DataFusionResult<DataType> {
49 return_type_check(
50 arg_types,
51 self.name(),
52 DataType::List(Arc::new(Field::new("item", DataType::Utf8, true))),
53 )
54 }
55
56 fn invoke_with_args(&self, args: ScalarFunctionArgs) -> DataFusionResult<ColumnarValue> {
57 invoke::<BuildListArray>(&args.args, jiter_json_object_keys)
58 }
59
60 fn aliases(&self) -> &[String] {
61 &self.aliases
62 }
63
64 fn placement(
65 &self,
66 args: &[datafusion::logical_expr::ExpressionPlacement],
67 ) -> datafusion::logical_expr::ExpressionPlacement {
68 if args.len() >= 2
71 && matches!(args[0], datafusion::logical_expr::ExpressionPlacement::Column)
72 && args[1..]
73 .iter()
74 .all(|arg| matches!(arg, datafusion::logical_expr::ExpressionPlacement::Literal))
75 {
76 datafusion::logical_expr::ExpressionPlacement::MoveTowardsLeafNodes
77 } else {
78 datafusion::logical_expr::ExpressionPlacement::KeepInPlace
79 }
80 }
81}
82
83#[derive(Debug)]
85struct BuildListArray;
86
87impl InvokeResult for BuildListArray {
88 type Item = Vec<String>;
89
90 type Builder = ListBuilder<StringBuilder>;
91
92 const ACCEPT_DICT_RETURN: bool = true;
93
94 fn builder(capacity: usize) -> Self::Builder {
95 let values_builder = StringBuilder::new();
96 ListBuilder::with_capacity(values_builder, capacity)
97 }
98
99 fn append_value(builder: &mut Self::Builder, value: Option<Self::Item>) {
100 builder.append_option(value.map(|v| v.into_iter().map(Some)));
101 }
102
103 fn finish(mut builder: Self::Builder) -> DataFusionResult<ArrayRef> {
104 Ok(Arc::new(builder.finish()))
105 }
106
107 fn scalar(value: Option<Self::Item>) -> ScalarValue {
108 keys_to_scalar(value)
109 }
110}
111
112fn keys_to_scalar(opt_keys: Option<Vec<String>>) -> ScalarValue {
113 let values_builder = StringBuilder::new();
114 let mut builder = ListBuilder::new(values_builder);
115 if let Some(keys) = opt_keys {
116 for value in keys {
117 builder.values().append_value(value);
118 }
119 builder.append(true);
120 } else {
121 builder.append(false);
122 }
123 let array = builder.finish();
124 ScalarValue::List(Arc::new(array))
125}
126
127fn jiter_json_object_keys(opt_json: Option<&str>, path: &[JsonPath]) -> Result<Vec<String>, GetError> {
128 if let Some((mut jiter, peek)) = jiter_json_find(opt_json, path) {
129 match peek {
130 Peek::Object => {
131 let mut opt_key = jiter.known_object()?;
132
133 let mut keys = Vec::new();
134 while let Some(key) = opt_key {
135 keys.push(key.to_string());
136 jiter.next_skip()?;
137 opt_key = jiter.next_key()?;
138 }
139 Ok(keys)
140 }
141 _ => get_err!(),
142 }
143 } else {
144 get_err!()
145 }
146}