datafusion_functions_json/
json_as_text.rs

1use 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
61impl InvokeResult for StringArray {
62    type Item = String;
63
64    type Builder = StringBuilder;
65
66    const ACCEPT_DICT_RETURN: bool = true;
67
68    fn builder(capacity: usize) -> Self::Builder {
69        StringBuilder::with_capacity(capacity, 0)
70    }
71
72    fn append_value(builder: &mut Self::Builder, value: Option<Self::Item>) {
73        builder.append_option(value);
74    }
75
76    fn finish(mut builder: Self::Builder) -> DataFusionResult<ArrayRef> {
77        Ok(Arc::new(builder.finish()))
78    }
79
80    fn scalar(value: Option<Self::Item>) -> ScalarValue {
81        ScalarValue::Utf8(value)
82    }
83}
84
85fn jiter_json_as_text(opt_json: Option<&str>, path: &[JsonPath]) -> Result<String, GetError> {
86    if let Some((mut jiter, peek)) = jiter_json_find(opt_json, path) {
87        match peek {
88            Peek::Null => {
89                jiter.known_null()?;
90                get_err!()
91            }
92            Peek::String => Ok(jiter.known_str()?.to_owned()),
93            _ => {
94                let start = jiter.current_index();
95                jiter.known_skip(peek)?;
96                let object_slice = jiter.slice_to_current(start);
97                let object_string = std::str::from_utf8(object_slice)?;
98                Ok(object_string.to_owned())
99            }
100        }
101    } else {
102        get_err!()
103    }
104}