Skip to main content

datafusion_functions_json/
json_get_array.rs

1use std::sync::Arc;
2
3use datafusion::arrow::array::{ArrayRef, ListBuilder, StringBuilder};
4use datafusion::arrow::datatypes::{DataType, Field};
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;
11use crate::common_union::json_field_metadata;
12
13fn list_item_field() -> Field {
14    Field::new("item", DataType::Utf8, true).with_metadata(json_field_metadata())
15}
16
17make_udf_function!(
18    JsonGetArray,
19    json_get_array,
20    json_data path,
21    r#"Get an arrow array from a JSON string by its "path""#
22);
23
24#[derive(Debug, PartialEq, Eq, Hash)]
25pub(super) struct JsonGetArray {
26    signature: Signature,
27    aliases: [String; 1],
28}
29
30impl Default for JsonGetArray {
31    fn default() -> Self {
32        Self {
33            signature: Signature::variadic_any(Volatility::Immutable),
34            aliases: ["json_get_array".to_string()],
35        }
36    }
37}
38
39impl ScalarUDFImpl for JsonGetArray {
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::List(Arc::new(list_item_field())))
50    }
51
52    fn invoke_with_args(&self, args: ScalarFunctionArgs) -> DataFusionResult<ColumnarValue> {
53        invoke::<BuildArrayList>(&args.args, jiter_json_get_array)
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
79#[derive(Debug)]
80struct BuildArrayList;
81
82impl InvokeResult for BuildArrayList {
83    type Item = Vec<String>;
84
85    type Builder = ListBuilder<StringBuilder>;
86
87    const ACCEPT_DICT_RETURN: bool = false;
88
89    fn builder(capacity: usize) -> Self::Builder {
90        let values_builder = StringBuilder::new();
91        ListBuilder::with_capacity(values_builder, capacity).with_field(list_item_field())
92    }
93
94    fn append_value(builder: &mut Self::Builder, value: Option<Self::Item>) {
95        builder.append_option(value.map(|v| v.into_iter().map(Some)));
96    }
97
98    fn finish(mut builder: Self::Builder) -> DataFusionResult<ArrayRef> {
99        Ok(Arc::new(builder.finish()))
100    }
101
102    fn scalar(value: Option<Self::Item>) -> ScalarValue {
103        let mut builder = ListBuilder::new(StringBuilder::new()).with_field(list_item_field());
104
105        if let Some(array_items) = value {
106            for item in array_items {
107                builder.values().append_value(item);
108            }
109
110            builder.append(true);
111        } else {
112            builder.append(false);
113        }
114        let array = builder.finish();
115        ScalarValue::List(Arc::new(array))
116    }
117}
118
119fn jiter_json_get_array(opt_json: Option<&str>, path: &[JsonPath]) -> Result<Vec<String>, GetError> {
120    if let Some((mut jiter, peek)) = jiter_json_find(opt_json, path) {
121        match peek {
122            Peek::Array => {
123                let mut peek_opt = jiter.known_array()?;
124                let mut array_items: Vec<String> = Vec::new();
125
126                while let Some(element_peek) = peek_opt {
127                    // Get the raw JSON slice for each array element
128                    let start = jiter.current_index();
129                    jiter.known_skip(element_peek)?;
130                    let slice = jiter.slice_to_current(start);
131                    let element_str = std::str::from_utf8(slice)?.to_string();
132
133                    array_items.push(element_str);
134                    peek_opt = jiter.array_step()?;
135                }
136
137                Ok(array_items)
138            }
139            _ => get_err!(),
140        }
141    } else {
142        get_err!()
143    }
144}