datafusion_functions_json/
json_object_keys.rs

1use std::any::Any;
2use std::sync::Arc;
3
4use datafusion::arrow::array::{ArrayRef, ListBuilder, StringBuilder};
5use datafusion::arrow::datatypes::{DataType, Field};
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    JsonObjectKeys,
15    json_object_keys,
16    json_data path,
17    r"Get the keys of a JSON object as an array."
18);
19
20#[derive(Debug, PartialEq, Eq, Hash)]
21pub(super) struct JsonObjectKeys {
22    signature: Signature,
23    aliases: [String; 2],
24}
25
26impl Default for JsonObjectKeys {
27    fn default() -> Self {
28        Self {
29            signature: Signature::variadic_any(Volatility::Immutable),
30            aliases: ["json_object_keys".to_string(), "json_keys".to_string()],
31        }
32    }
33}
34
35impl ScalarUDFImpl for JsonObjectKeys {
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(Field::new("item", DataType::Utf8, true))),
53        )
54    }
55
56    fn invoke_with_args(&self, args: ScalarFunctionArgs) -> DataFusionResult<ColumnarValue> {
57        invoke::<BuildListArray>(&args.args, jiter_json_object_keys)
58    }
59
60    fn aliases(&self) -> &[String] {
61        &self.aliases
62    }
63}
64
65/// Struct used to build a `ListArray` from the result of `jiter_json_object_keys`.
66#[derive(Debug)]
67struct BuildListArray;
68
69impl InvokeResult for BuildListArray {
70    type Item = Vec<String>;
71
72    type Builder = ListBuilder<StringBuilder>;
73
74    const ACCEPT_DICT_RETURN: bool = true;
75
76    fn builder(capacity: usize) -> Self::Builder {
77        let values_builder = StringBuilder::new();
78        ListBuilder::with_capacity(values_builder, capacity)
79    }
80
81    fn append_value(builder: &mut Self::Builder, value: Option<Self::Item>) {
82        builder.append_option(value.map(|v| v.into_iter().map(Some)));
83    }
84
85    fn finish(mut builder: Self::Builder) -> DataFusionResult<ArrayRef> {
86        Ok(Arc::new(builder.finish()))
87    }
88
89    fn scalar(value: Option<Self::Item>) -> ScalarValue {
90        keys_to_scalar(value)
91    }
92}
93
94fn keys_to_scalar(opt_keys: Option<Vec<String>>) -> ScalarValue {
95    let values_builder = StringBuilder::new();
96    let mut builder = ListBuilder::new(values_builder);
97    if let Some(keys) = opt_keys {
98        for value in keys {
99            builder.values().append_value(value);
100        }
101        builder.append(true);
102    } else {
103        builder.append(false);
104    }
105    let array = builder.finish();
106    ScalarValue::List(Arc::new(array))
107}
108
109fn jiter_json_object_keys(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::Object => {
113                let mut opt_key = jiter.known_object()?;
114
115                let mut keys = Vec::new();
116                while let Some(key) = opt_key {
117                    keys.push(key.to_string());
118                    jiter.next_skip()?;
119                    opt_key = jiter.next_key()?;
120                }
121                Ok(keys)
122            }
123            _ => get_err!(),
124        }
125    } else {
126        get_err!()
127    }
128}