datafusion_functions_json/
json_get_array.rs

1use std::any::Any;
2use std::sync::Arc;
3
4use datafusion::arrow::array::{ArrayRef, ListBuilder, 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    JsonGetArray,
15    json_get_array,
16    json_data path,
17    r#"Get an arrow array from a JSON string by its "path""#
18);
19
20#[derive(Debug, PartialEq, Eq, Hash)]
21pub(super) struct JsonGetArray {
22    signature: Signature,
23    aliases: [String; 1],
24}
25
26impl Default for JsonGetArray {
27    fn default() -> Self {
28        Self {
29            signature: Signature::variadic_any(Volatility::Immutable),
30            aliases: ["json_get_array".to_string()],
31        }
32    }
33}
34
35impl ScalarUDFImpl for JsonGetArray {
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(
50            arg_types,
51            self.name(),
52            DataType::List(Arc::new(datafusion::arrow::datatypes::Field::new(
53                "item",
54                DataType::Utf8,
55                true,
56            ))),
57        )
58    }
59
60    fn invoke_with_args(&self, args: ScalarFunctionArgs) -> DataFusionResult<ColumnarValue> {
61        invoke::<BuildArrayList>(&args.args, jiter_json_get_array)
62    }
63
64    fn aliases(&self) -> &[String] {
65        &self.aliases
66    }
67}
68
69#[derive(Debug)]
70struct BuildArrayList;
71
72impl InvokeResult for BuildArrayList {
73    type Item = Vec<String>;
74
75    type Builder = ListBuilder<StringBuilder>;
76
77    const ACCEPT_DICT_RETURN: bool = false;
78
79    fn builder(capacity: usize) -> Self::Builder {
80        let values_builder = StringBuilder::new();
81        ListBuilder::with_capacity(values_builder, capacity)
82    }
83
84    fn append_value(builder: &mut Self::Builder, value: Option<Self::Item>) {
85        builder.append_option(value.map(|v| v.into_iter().map(Some)));
86    }
87
88    fn finish(mut builder: Self::Builder) -> DataFusionResult<ArrayRef> {
89        Ok(Arc::new(builder.finish()))
90    }
91
92    fn scalar(value: Option<Self::Item>) -> ScalarValue {
93        let mut builder = ListBuilder::new(StringBuilder::new());
94
95        if let Some(array_items) = value {
96            for item in array_items {
97                builder.values().append_value(item);
98            }
99
100            builder.append(true);
101        } else {
102            builder.append(false);
103        }
104        let array = builder.finish();
105        ScalarValue::List(Arc::new(array))
106    }
107}
108
109fn jiter_json_get_array(opt_json: Option<&str>, path: &[JsonPath]) -> Result<Vec<String>, GetError> {
110    if let Some((mut jiter, peek)) = jiter_json_find(opt_json, path) {
111        match peek {
112            Peek::Array => {
113                let mut peek_opt = jiter.known_array()?;
114                let mut array_items: Vec<String> = Vec::new();
115
116                while let Some(element_peek) = peek_opt {
117                    // Get the raw JSON slice for each array element
118                    let start = jiter.current_index();
119                    jiter.known_skip(element_peek)?;
120                    let slice = jiter.slice_to_current(start);
121                    let element_str = std::str::from_utf8(slice)?.to_string();
122
123                    array_items.push(element_str);
124                    peek_opt = jiter.array_step()?;
125                }
126
127                Ok(array_items)
128            }
129            _ => get_err!(),
130        }
131    } else {
132        get_err!()
133    }
134}