datafusion_functions/string/
bit_length.rs1use arrow::compute::kernels::length::bit_length;
19use arrow::datatypes::DataType;
20
21use crate::utils::{transform_leaf_type_preserving_encoding, utf8_to_int_type};
22use datafusion_common::types::logical_string;
23use datafusion_common::utils::take_function_args;
24use datafusion_common::{Result, ScalarValue};
25use datafusion_expr::{
26 Coercion, ColumnarValue, Documentation, EncodingPreservation, ScalarFunctionArgs,
27 ScalarUDFImpl, Signature, TypeSignatureClass, Volatility,
28};
29use datafusion_macros::user_doc;
30
31#[user_doc(
32 doc_section(label = "String Functions"),
33 description = "Returns the bit length of a string.",
34 syntax_example = "bit_length(str)",
35 sql_example = r#"```sql
36> select bit_length('datafusion');
37+--------------------------------+
38| bit_length(Utf8("datafusion")) |
39+--------------------------------+
40| 80 |
41+--------------------------------+
42```"#,
43 standard_argument(name = "str", prefix = "String"),
44 related_udf(name = "length"),
45 related_udf(name = "octet_length")
46)]
47#[derive(Debug, PartialEq, Eq, Hash)]
48pub struct BitLengthFunc {
49 signature: Signature,
50}
51
52impl Default for BitLengthFunc {
53 fn default() -> Self {
54 Self::new()
55 }
56}
57
58impl BitLengthFunc {
59 pub fn new() -> Self {
60 Self {
61 signature: Signature::coercible(
62 vec![
63 Coercion::new_exact(TypeSignatureClass::Native(logical_string()))
64 .with_encoding_preservation(EncodingPreservation::dictionary()),
65 ],
66 Volatility::Immutable,
67 ),
68 }
69 }
70}
71
72impl ScalarUDFImpl for BitLengthFunc {
73 fn name(&self) -> &str {
74 "bit_length"
75 }
76
77 fn signature(&self) -> &Signature {
78 &self.signature
79 }
80
81 fn return_type(&self, arg_types: &[DataType]) -> Result<DataType> {
82 transform_leaf_type_preserving_encoding(&arg_types[0], &|data_type| {
83 utf8_to_int_type(data_type, "bit_length")
84 })
85 }
86
87 fn invoke_with_args(&self, args: ScalarFunctionArgs) -> Result<ColumnarValue> {
88 let [array] = take_function_args(self.name(), &args.args)?;
89
90 match array {
91 ColumnarValue::Array(v) => Ok(ColumnarValue::Array(bit_length(v.as_ref())?)),
92 ColumnarValue::Scalar(v) => Ok(ColumnarValue::Scalar(bit_length_scalar(v))),
93 }
94 }
95
96 fn documentation(&self) -> Option<&Documentation> {
97 self.doc()
98 }
99}
100
101fn bit_length_scalar(value: &ScalarValue) -> ScalarValue {
102 match value {
103 ScalarValue::Utf8(v) => {
104 ScalarValue::Int32(v.as_ref().map(|x| (x.len() * 8) as i32))
105 }
106 ScalarValue::LargeUtf8(v) => {
107 ScalarValue::Int64(v.as_ref().map(|x| (x.len() * 8) as i64))
108 }
109 ScalarValue::Utf8View(v) => {
110 ScalarValue::Int32(v.as_ref().map(|x| (x.len() * 8) as i32))
111 }
112 ScalarValue::Dictionary(key_type, value) => {
113 ScalarValue::Dictionary(key_type.clone(), Box::new(bit_length_scalar(value)))
114 }
115 _ => unreachable!("bit length"),
116 }
117}