datafusion_functions_json/
json_get_json.rs

1use std::any::Any;
2
3use datafusion::arrow::array::StringArray;
4use datafusion::arrow::datatypes::DataType;
5use datafusion::common::Result as DataFusionResult;
6use datafusion::logical_expr::{ColumnarValue, ScalarFunctionArgs, ScalarUDFImpl, Signature, Volatility};
7
8use crate::common::{get_err, invoke, jiter_json_find, return_type_check, GetError, JsonPath};
9use crate::common_macros::make_udf_function;
10
11make_udf_function!(
12    JsonGetJson,
13    json_get_json,
14    json_data path,
15    r#"Get a nested raw JSON string from a JSON string by its "path""#
16);
17
18#[derive(Debug)]
19pub(super) struct JsonGetJson {
20    signature: Signature,
21    aliases: [String; 1],
22}
23
24impl Default for JsonGetJson {
25    fn default() -> Self {
26        Self {
27            signature: Signature::variadic_any(Volatility::Immutable),
28            aliases: ["json_get_json".to_string()],
29        }
30    }
31}
32
33impl ScalarUDFImpl for JsonGetJson {
34    fn as_any(&self) -> &dyn Any {
35        self
36    }
37
38    fn name(&self) -> &str {
39        self.aliases[0].as_str()
40    }
41
42    fn signature(&self) -> &Signature {
43        &self.signature
44    }
45
46    fn return_type(&self, arg_types: &[DataType]) -> DataFusionResult<DataType> {
47        return_type_check(arg_types, self.name(), DataType::Utf8)
48    }
49
50    fn invoke_with_args(&self, args: ScalarFunctionArgs) -> DataFusionResult<ColumnarValue> {
51        invoke::<StringArray>(&args.args, jiter_json_get_json)
52    }
53
54    fn aliases(&self) -> &[String] {
55        &self.aliases
56    }
57}
58
59fn jiter_json_get_json(opt_json: Option<&str>, path: &[JsonPath]) -> Result<String, GetError> {
60    if let Some((mut jiter, peek)) = jiter_json_find(opt_json, path) {
61        let start = jiter.current_index();
62        jiter.known_skip(peek)?;
63        let object_slice = jiter.slice_to_current(start);
64        let object_string = std::str::from_utf8(object_slice)?;
65        Ok(object_string.to_owned())
66    } else {
67        get_err!()
68    }
69}