datafusion_functions_json/
json_length.rs

1use std::any::Any;
2use std::sync::Arc;
3
4use datafusion::arrow::array::{ArrayRef, UInt64Array, UInt64Builder};
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    JsonLength,
15    json_length,
16    json_data path,
17    r"Get the length of the array or object at the given path."
18);
19
20#[derive(Debug, PartialEq, Eq, Hash)]
21pub(super) struct JsonLength {
22    signature: Signature,
23    aliases: [String; 2],
24}
25
26impl Default for JsonLength {
27    fn default() -> Self {
28        Self {
29            signature: Signature::variadic_any(Volatility::Immutable),
30            aliases: ["json_length".to_string(), "json_len".to_string()],
31        }
32    }
33}
34
35impl ScalarUDFImpl for JsonLength {
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::UInt64)
50    }
51
52    fn invoke_with_args(&self, args: ScalarFunctionArgs) -> DataFusionResult<ColumnarValue> {
53        invoke::<UInt64Array>(&args.args, jiter_json_length)
54    }
55
56    fn aliases(&self) -> &[String] {
57        &self.aliases
58    }
59
60    fn placement(
61        &self,
62        args: &[datafusion::logical_expr::ExpressionPlacement],
63    ) -> datafusion::logical_expr::ExpressionPlacement {
64        // If the first argument is a column and the remaining arguments are literals (a path)
65        // then we can push this UDF down to the leaf nodes.
66        if args.len() >= 2
67            && matches!(args[0], datafusion::logical_expr::ExpressionPlacement::Column)
68            && args[1..]
69                .iter()
70                .all(|arg| matches!(arg, datafusion::logical_expr::ExpressionPlacement::Literal))
71        {
72            datafusion::logical_expr::ExpressionPlacement::MoveTowardsLeafNodes
73        } else {
74            datafusion::logical_expr::ExpressionPlacement::KeepInPlace
75        }
76    }
77}
78
79impl InvokeResult for UInt64Array {
80    type Item = u64;
81
82    type Builder = UInt64Builder;
83
84    // cheaper to return integers without dict-encoding them
85    const ACCEPT_DICT_RETURN: bool = false;
86
87    fn builder(capacity: usize) -> Self::Builder {
88        UInt64Builder::with_capacity(capacity)
89    }
90
91    fn append_value(builder: &mut Self::Builder, value: Option<Self::Item>) {
92        builder.append_option(value);
93    }
94
95    fn finish(mut builder: Self::Builder) -> DataFusionResult<ArrayRef> {
96        Ok(Arc::new(builder.finish()))
97    }
98
99    fn scalar(value: Option<Self::Item>) -> ScalarValue {
100        ScalarValue::UInt64(value)
101    }
102}
103
104fn jiter_json_length(opt_json: Option<&str>, path: &[JsonPath]) -> Result<u64, GetError> {
105    if let Some((mut jiter, peek)) = jiter_json_find(opt_json, path) {
106        match peek {
107            Peek::Array => {
108                let mut peek_opt = jiter.known_array()?;
109                let mut length: u64 = 0;
110                while let Some(peek) = peek_opt {
111                    jiter.known_skip(peek)?;
112                    length += 1;
113                    peek_opt = jiter.array_step()?;
114                }
115                Ok(length)
116            }
117            Peek::Object => {
118                let mut opt_key = jiter.known_object()?;
119
120                let mut length: u64 = 0;
121                while opt_key.is_some() {
122                    jiter.next_skip()?;
123                    length += 1;
124                    opt_key = jiter.next_key()?;
125                }
126                Ok(length)
127            }
128            _ => get_err!(),
129        }
130    } else {
131        get_err!()
132    }
133}
datafusion_functions_json/json_length.rs

datafusion_functions_json/
json_length.rs
