datafusion_functions_json/
json_as_text.rs1use std::any::Any;
2use std::sync::Arc;
3
4use datafusion::arrow::array::{ArrayRef, StringArray, StringBuilder};
5use datafusion::arrow::datatypes::DataType;
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 JsonAsText,
15 json_as_text,
16 json_data path,
17 r#"Get any value from a JSON string by its "path", represented as a string"#
18);
19
20#[derive(Debug, PartialEq, Eq, Hash)]
21pub(super) struct JsonAsText {
22 signature: Signature,
23 aliases: [String; 1],
24}
25
26impl Default for JsonAsText {
27 fn default() -> Self {
28 Self {
29 signature: Signature::variadic_any(Volatility::Immutable),
30 aliases: ["json_as_text".to_string()],
31 }
32 }
33}
34
35impl ScalarUDFImpl for JsonAsText {
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(arg_types, self.name(), DataType::Utf8)
50 }
51
52 fn invoke_with_args(&self, args: ScalarFunctionArgs) -> DataFusionResult<ColumnarValue> {
53 invoke::<StringArray>(&args.args, jiter_json_as_text)
54 }
55
56 fn aliases(&self) -> &[String] {
57 &self.aliases
58 }
59
60 fn placement(
61 &self,
62 args: &[datafusion::logical_expr::ExpressionPlacement],
63 ) -> datafusion::logical_expr::ExpressionPlacement {
64 if args.len() >= 2
67 && matches!(args[0], datafusion::logical_expr::ExpressionPlacement::Column)
68 && args[1..]
69 .iter()
70 .all(|arg| matches!(arg, datafusion::logical_expr::ExpressionPlacement::Literal))
71 {
72 datafusion::logical_expr::ExpressionPlacement::MoveTowardsLeafNodes
73 } else {
74 datafusion::logical_expr::ExpressionPlacement::KeepInPlace
75 }
76 }
77}
78
79impl InvokeResult for StringArray {
80 type Item = String;
81
82 type Builder = StringBuilder;
83
84 const ACCEPT_DICT_RETURN: bool = true;
85
86 fn builder(capacity: usize) -> Self::Builder {
87 StringBuilder::with_capacity(capacity, 0)
88 }
89
90 fn append_value(builder: &mut Self::Builder, value: Option<Self::Item>) {
91 builder.append_option(value);
92 }
93
94 fn finish(mut builder: Self::Builder) -> DataFusionResult<ArrayRef> {
95 Ok(Arc::new(builder.finish()))
96 }
97
98 fn scalar(value: Option<Self::Item>) -> ScalarValue {
99 ScalarValue::Utf8(value)
100 }
101}
102
103fn jiter_json_as_text(opt_json: Option<&str>, path: &[JsonPath]) -> Result<String, GetError> {
104 if let Some((mut jiter, peek)) = jiter_json_find(opt_json, path) {
105 match peek {
106 Peek::Null => {
107 jiter.known_null()?;
108 get_err!()
109 }
110 Peek::String => Ok(jiter.known_str()?.to_owned()),
111 _ => {
112 let start = jiter.current_index();
113 jiter.known_skip(peek)?;
114 let object_slice = jiter.slice_to_current(start);
115 let object_string = std::str::from_utf8(object_slice)?;
116 Ok(object_string.to_owned())
117 }
118 }
119 } else {
120 get_err!()
121 }
122}