Skip to main content

datafusion_functions_json/
json_as_text.rs

1use std::sync::Arc;
2
3use datafusion::arrow::array::{ArrayRef, StringArray, StringBuilder};
4use datafusion::arrow::datatypes::DataType;
5use datafusion::common::{Result as DataFusionResult, ScalarValue};
6use datafusion::logical_expr::{ColumnarValue, ScalarFunctionArgs, ScalarUDFImpl, Signature, Volatility};
7use jiter::Peek;
8
9use crate::common::{get_err, invoke, jiter_json_find, return_type_check, GetError, InvokeResult, JsonPath};
10use crate::common_macros::make_udf_function;
11
12make_udf_function!(
13    JsonAsText,
14    json_as_text,
15    json_data path,
16    r#"Get any value from a JSON string by its "path", represented as a string"#
17);
18
19#[derive(Debug, PartialEq, Eq, Hash)]
20pub(super) struct JsonAsText {
21    signature: Signature,
22    aliases: [String; 1],
23}
24
25impl Default for JsonAsText {
26    fn default() -> Self {
27        Self {
28            signature: Signature::variadic_any(Volatility::Immutable),
29            aliases: ["json_as_text".to_string()],
30        }
31    }
32}
33
34impl ScalarUDFImpl for JsonAsText {
35    fn name(&self) -> &str {
36        self.aliases[0].as_str()
37    }
38
39    fn signature(&self) -> &Signature {
40        &self.signature
41    }
42
43    fn return_type(&self, arg_types: &[DataType]) -> DataFusionResult<DataType> {
44        return_type_check(arg_types, self.name(), DataType::Utf8)
45    }
46
47    fn invoke_with_args(&self, args: ScalarFunctionArgs) -> DataFusionResult<ColumnarValue> {
48        invoke::<StringArray>(&args.args, jiter_json_as_text)
49    }
50
51    fn aliases(&self) -> &[String] {
52        &self.aliases
53    }
54
55    fn placement(
56        &self,
57        args: &[datafusion::logical_expr::ExpressionPlacement],
58    ) -> datafusion::logical_expr::ExpressionPlacement {
59        // If the first argument is a column and the remaining arguments are literals (a path)
60        // then we can push this UDF down to the leaf nodes.
61        if args.len() >= 2
62            && matches!(args[0], datafusion::logical_expr::ExpressionPlacement::Column)
63            && args[1..]
64                .iter()
65                .all(|arg| matches!(arg, datafusion::logical_expr::ExpressionPlacement::Literal))
66        {
67            datafusion::logical_expr::ExpressionPlacement::MoveTowardsLeafNodes
68        } else {
69            datafusion::logical_expr::ExpressionPlacement::KeepInPlace
70        }
71    }
72}
73
74impl InvokeResult for StringArray {
75    type Item = String;
76
77    type Builder = StringBuilder;
78
79    const ACCEPT_DICT_RETURN: bool = true;
80
81    fn builder(capacity: usize) -> Self::Builder {
82        StringBuilder::with_capacity(capacity, 0)
83    }
84
85    fn append_value(builder: &mut Self::Builder, value: Option<Self::Item>) {
86        builder.append_option(value);
87    }
88
89    fn finish(mut builder: Self::Builder) -> DataFusionResult<ArrayRef> {
90        Ok(Arc::new(builder.finish()))
91    }
92
93    fn scalar(value: Option<Self::Item>) -> ScalarValue {
94        ScalarValue::Utf8(value)
95    }
96}
97
98fn jiter_json_as_text(opt_json: Option<&str>, path: &[JsonPath]) -> Result<String, GetError> {
99    if let Some((mut jiter, peek)) = jiter_json_find(opt_json, path) {
100        match peek {
101            Peek::Null => {
102                jiter.known_null()?;
103                get_err!()
104            }
105            Peek::String => Ok(jiter.known_str()?.to_owned()),
106            _ => {
107                let start = jiter.current_index();
108                jiter.known_skip(peek)?;
109                let object_slice = jiter.slice_to_current(start);
110                let object_string = std::str::from_utf8(object_slice)?;
111                Ok(object_string.to_owned())
112            }
113        }
114    } else {
115        get_err!()
116    }
117}