datafusion_functions_json/
json_get.rs1use std::any::Any;
2use std::sync::Arc;
3
4use datafusion::arrow::array::ArrayRef;
5use datafusion::arrow::array::UnionArray;
6use datafusion::arrow::datatypes::DataType;
7use datafusion::common::Result as DataFusionResult;
8use datafusion::logical_expr::{ColumnarValue, ScalarFunctionArgs, ScalarUDFImpl, Signature, Volatility};
9use datafusion::scalar::ScalarValue;
10use jiter::{Jiter, NumberAny, NumberInt, Peek};
11
12use crate::common::InvokeResult;
13use crate::common::{get_err, invoke, jiter_json_find, return_type_check, GetError, JsonPath};
14use crate::common_macros::make_udf_function;
15use crate::common_union::{JsonUnion, JsonUnionField};
16
17make_udf_function!(
18 JsonGet,
19 json_get,
20 json_data path,
21 r#"Get a value from a JSON string by its "path""#
22);
23
24#[derive(Debug, PartialEq, Eq, Hash)]
27pub(super) struct JsonGet {
28 signature: Signature,
29 aliases: [String; 1],
30}
31
32impl Default for JsonGet {
33 fn default() -> Self {
34 Self {
35 signature: Signature::variadic_any(Volatility::Immutable),
36 aliases: ["json_get".to_string()],
37 }
38 }
39}
40
41impl ScalarUDFImpl for JsonGet {
42 fn as_any(&self) -> &dyn Any {
43 self
44 }
45
46 fn name(&self) -> &str {
47 self.aliases[0].as_str()
48 }
49
50 fn signature(&self) -> &Signature {
51 &self.signature
52 }
53
54 fn return_type(&self, arg_types: &[DataType]) -> DataFusionResult<DataType> {
55 return_type_check(arg_types, self.name(), JsonUnion::data_type())
56 }
57
58 fn invoke_with_args(&self, args: ScalarFunctionArgs) -> DataFusionResult<ColumnarValue> {
59 invoke::<JsonUnion>(&args.args, jiter_json_get_union)
60 }
61
62 fn aliases(&self) -> &[String] {
63 &self.aliases
64 }
65
66 fn placement(
67 &self,
68 args: &[datafusion::logical_expr::ExpressionPlacement],
69 ) -> datafusion::logical_expr::ExpressionPlacement {
70 if args.len() >= 2
73 && matches!(args[0], datafusion::logical_expr::ExpressionPlacement::Column)
74 && args[1..]
75 .iter()
76 .all(|arg| matches!(arg, datafusion::logical_expr::ExpressionPlacement::Literal))
77 {
78 datafusion::logical_expr::ExpressionPlacement::MoveTowardsLeafNodes
79 } else {
80 datafusion::logical_expr::ExpressionPlacement::KeepInPlace
81 }
82 }
83}
84
85impl InvokeResult for JsonUnion {
86 type Item = JsonUnionField;
87
88 type Builder = JsonUnion;
89
90 const ACCEPT_DICT_RETURN: bool = true;
91
92 fn builder(capacity: usize) -> Self::Builder {
93 JsonUnion::new(capacity)
94 }
95
96 fn append_value(builder: &mut Self::Builder, value: Option<Self::Item>) {
97 if let Some(value) = value {
98 builder.push(value);
99 } else {
100 builder.push_none();
101 }
102 }
103
104 fn finish(builder: Self::Builder) -> DataFusionResult<ArrayRef> {
105 let array: UnionArray = builder.try_into()?;
106 Ok(Arc::new(array) as ArrayRef)
107 }
108
109 fn scalar(value: Option<Self::Item>) -> ScalarValue {
110 JsonUnionField::scalar_value(value)
111 }
112}
113
114fn jiter_json_get_union(opt_json: Option<&str>, path: &[JsonPath]) -> Result<JsonUnionField, GetError> {
115 if let Some((mut jiter, peek)) = jiter_json_find(opt_json, path) {
116 build_union(&mut jiter, peek)
117 } else {
118 get_err!()
119 }
120}
121
122fn build_union(jiter: &mut Jiter, peek: Peek) -> Result<JsonUnionField, GetError> {
123 match peek {
124 Peek::Null => {
125 jiter.known_null()?;
126 Ok(JsonUnionField::JsonNull)
127 }
128 Peek::True | Peek::False => {
129 let value = jiter.known_bool(peek)?;
130 Ok(JsonUnionField::Bool(value))
131 }
132 Peek::String => {
133 let value = jiter.known_str()?;
134 Ok(JsonUnionField::Str(value.to_owned()))
135 }
136 Peek::Array => {
137 let start = jiter.current_index();
138 jiter.known_skip(peek)?;
139 let array_slice = jiter.slice_to_current(start);
140 let array_string = std::str::from_utf8(array_slice)?;
141 Ok(JsonUnionField::Array(array_string.to_owned()))
142 }
143 Peek::Object => {
144 let start = jiter.current_index();
145 jiter.known_skip(peek)?;
146 let object_slice = jiter.slice_to_current(start);
147 let object_string = std::str::from_utf8(object_slice)?;
148 Ok(JsonUnionField::Object(object_string.to_owned()))
149 }
150 _ => match jiter.known_number(peek)? {
151 NumberAny::Int(NumberInt::Int(value)) => Ok(JsonUnionField::Int(value)),
152 NumberAny::Int(NumberInt::BigInt(_)) => todo!("BigInt not supported yet"),
153 NumberAny::Float(value) => Ok(JsonUnionField::Float(value)),
154 },
155 }
156}