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
61impl InvokeResult for UInt64Array {
62    type Item = u64;
63
64    type Builder = UInt64Builder;
65
66    // cheaper to return integers without dict-encoding them
67    const ACCEPT_DICT_RETURN: bool = false;
68
69    fn builder(capacity: usize) -> Self::Builder {
70        UInt64Builder::with_capacity(capacity)
71    }
72
73    fn append_value(builder: &mut Self::Builder, value: Option<Self::Item>) {
74        builder.append_option(value);
75    }
76
77    fn finish(mut builder: Self::Builder) -> DataFusionResult<ArrayRef> {
78        Ok(Arc::new(builder.finish()))
79    }
80
81    fn scalar(value: Option<Self::Item>) -> ScalarValue {
82        ScalarValue::UInt64(value)
83    }
84}
85
86fn jiter_json_length(opt_json: Option<&str>, path: &[JsonPath]) -> Result<u64, GetError> {
87    if let Some((mut jiter, peek)) = jiter_json_find(opt_json, path) {
88        match peek {
89            Peek::Array => {
90                let mut peek_opt = jiter.known_array()?;
91                let mut length: u64 = 0;
92                while let Some(peek) = peek_opt {
93                    jiter.known_skip(peek)?;
94                    length += 1;
95                    peek_opt = jiter.array_step()?;
96                }
97                Ok(length)
98            }
99            Peek::Object => {
100                let mut opt_key = jiter.known_object()?;
101
102                let mut length: u64 = 0;
103                while opt_key.is_some() {
104                    jiter.next_skip()?;
105                    length += 1;
106                    opt_key = jiter.next_key()?;
107                }
108                Ok(length)
109            }
110            _ => get_err!(),
111        }
112    } else {
113        get_err!()
114    }
115}
datafusion_functions_json/json_length.rs

datafusion_functions_json/
json_length.rs
