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)]
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
67impl InvokeResult for JsonUnion {
68 type Item = JsonUnionField;
69
70 type Builder = JsonUnion;
71
72 const ACCEPT_DICT_RETURN: bool = true;
73
74 fn builder(capacity: usize) -> Self::Builder {
75 JsonUnion::new(capacity)
76 }
77
78 fn append_value(builder: &mut Self::Builder, value: Option<Self::Item>) {
79 if let Some(value) = value {
80 builder.push(value);
81 } else {
82 builder.push_none();
83 }
84 }
85
86 fn finish(builder: Self::Builder) -> DataFusionResult<ArrayRef> {
87 let array: UnionArray = builder.try_into()?;
88 Ok(Arc::new(array) as ArrayRef)
89 }
90
91 fn scalar(value: Option<Self::Item>) -> ScalarValue {
92 JsonUnionField::scalar_value(value)
93 }
94}
95
96fn jiter_json_get_union(opt_json: Option<&str>, path: &[JsonPath]) -> Result<JsonUnionField, GetError> {
97 if let Some((mut jiter, peek)) = jiter_json_find(opt_json, path) {
98 build_union(&mut jiter, peek)
99 } else {
100 get_err!()
101 }
102}
103
104fn build_union(jiter: &mut Jiter, peek: Peek) -> Result<JsonUnionField, GetError> {
105 match peek {
106 Peek::Null => {
107 jiter.known_null()?;
108 Ok(JsonUnionField::JsonNull)
109 }
110 Peek::True | Peek::False => {
111 let value = jiter.known_bool(peek)?;
112 Ok(JsonUnionField::Bool(value))
113 }
114 Peek::String => {
115 let value = jiter.known_str()?;
116 Ok(JsonUnionField::Str(value.to_owned()))
117 }
118 Peek::Array => {
119 let start = jiter.current_index();
120 jiter.known_skip(peek)?;
121 let array_slice = jiter.slice_to_current(start);
122 let array_string = std::str::from_utf8(array_slice)?;
123 Ok(JsonUnionField::Array(array_string.to_owned()))
124 }
125 Peek::Object => {
126 let start = jiter.current_index();
127 jiter.known_skip(peek)?;
128 let object_slice = jiter.slice_to_current(start);
129 let object_string = std::str::from_utf8(object_slice)?;
130 Ok(JsonUnionField::Object(object_string.to_owned()))
131 }
132 _ => match jiter.known_number(peek)? {
133 NumberAny::Int(NumberInt::Int(value)) => Ok(JsonUnionField::Int(value)),
134 NumberAny::Int(NumberInt::BigInt(_)) => todo!("BigInt not supported yet"),
135 NumberAny::Float(value) => Ok(JsonUnionField::Float(value)),
136 },
137 }
138}