datafusion_functions_json/
json_length.rs

1use std::sync::Arc;
2
3use datafusion::arrow::array::{ArrayRef, UInt64Array, UInt64Builder};
4use datafusion::arrow::datatypes::DataType;
5use datafusion::common::{Result as DataFusionResult, ScalarValue};
6use datafusion::logical_expr::{ColumnarValue, ScalarFunctionArgs, ScalarUDFImpl, Signature, Volatility};
7use jiter::Peek;
8
9use crate::common::{get_err, invoke, jiter_json_find, return_type_check, GetError, InvokeResult, JsonPath};
10use crate::common_macros::make_udf_function;
11
12make_udf_function!(
13    JsonLength,
14    json_length,
15    json_data path,
16    r"Get the length of the array or object at the given path."
17);
18
19#[derive(Debug, PartialEq, Eq, Hash)]
20pub(super) struct JsonLength {
21    signature: Signature,
22    aliases: [String; 2],
23}
24
25impl Default for JsonLength {
26    fn default() -> Self {
27        Self {
28            signature: Signature::variadic_any(Volatility::Immutable),
29            aliases: ["json_length".to_string(), "json_len".to_string()],
30        }
31    }
32}
33
34impl ScalarUDFImpl for JsonLength {
35    fn name(&self) -> &str {
36        self.aliases[0].as_str()
37    }
38
39    fn signature(&self) -> &Signature {
40        &self.signature
41    }
42
43    fn return_type(&self, arg_types: &[DataType]) -> DataFusionResult<DataType> {
44        return_type_check(arg_types, self.name(), DataType::UInt64)
45    }
46
47    fn invoke_with_args(&self, args: ScalarFunctionArgs) -> DataFusionResult<ColumnarValue> {
48        invoke::<UInt64Array>(&args.args, jiter_json_length)
49    }
50
51    fn aliases(&self) -> &[String] {
52        &self.aliases
53    }
54
55    fn placement(
56        &self,
57        args: &[datafusion::logical_expr::ExpressionPlacement],
58    ) -> datafusion::logical_expr::ExpressionPlacement {
59        // If the first argument is a column and the remaining arguments are literals (a path)
60        // then we can push this UDF down to the leaf nodes.
61        if args.len() >= 2
62            && matches!(args[0], datafusion::logical_expr::ExpressionPlacement::Column)
63            && args[1..]
64                .iter()
65                .all(|arg| matches!(arg, datafusion::logical_expr::ExpressionPlacement::Literal))
66        {
67            datafusion::logical_expr::ExpressionPlacement::MoveTowardsLeafNodes
68        } else {
69            datafusion::logical_expr::ExpressionPlacement::KeepInPlace
70        }
71    }
72}
73
74impl InvokeResult for UInt64Array {
75    type Item = u64;
76
77    type Builder = UInt64Builder;
78
79    // cheaper to return integers without dict-encoding them
80    const ACCEPT_DICT_RETURN: bool = false;
81
82    fn builder(capacity: usize) -> Self::Builder {
83        UInt64Builder::with_capacity(capacity)
84    }
85
86    fn append_value(builder: &mut Self::Builder, value: Option<Self::Item>) {
87        builder.append_option(value);
88    }
89
90    fn finish(mut builder: Self::Builder) -> DataFusionResult<ArrayRef> {
91        Ok(Arc::new(builder.finish()))
92    }
93
94    fn scalar(value: Option<Self::Item>) -> ScalarValue {
95        ScalarValue::UInt64(value)
96    }
97}
98
99fn jiter_json_length(opt_json: Option<&str>, path: &[JsonPath]) -> Result<u64, GetError> {
100    if let Some((mut jiter, peek)) = jiter_json_find(opt_json, path) {
101        match peek {
102            Peek::Array => {
103                let mut peek_opt = jiter.known_array()?;
104                let mut length: u64 = 0;
105                while let Some(peek) = peek_opt {
106                    jiter.known_skip(peek)?;
107                    length += 1;
108                    peek_opt = jiter.array_step()?;
109                }
110                Ok(length)
111            }
112            Peek::Object => {
113                let mut opt_key = jiter.known_object()?;
114
115                let mut length: u64 = 0;
116                while opt_key.is_some() {
117                    jiter.next_skip()?;
118                    length += 1;
119                    opt_key = jiter.next_key()?;
120                }
121                Ok(length)
122            }
123            _ => get_err!(),
124        }
125    } else {
126        get_err!()
127    }
128}
datafusion_functions_json/json_length.rs

datafusion_functions_json/
json_length.rs
