datafusion_functions/string/
ascii.rs1use crate::utils::make_scalar_function;
19use arrow::array::{ArrayRef, AsArray, Int32Array, StringArrayType};
20use arrow::datatypes::DataType;
21use arrow::error::ArrowError;
22use datafusion_common::types::logical_string;
23use datafusion_common::{internal_err, Result};
24use datafusion_expr::{ColumnarValue, Documentation, TypeSignatureClass};
25use datafusion_expr::{ScalarFunctionArgs, ScalarUDFImpl, Signature, Volatility};
26use datafusion_expr_common::signature::Coercion;
27use datafusion_macros::user_doc;
28use std::any::Any;
29use std::sync::Arc;
30
31#[user_doc(
32 doc_section(label = "String Functions"),
33 description = "Returns the Unicode character code of the first character in a string.",
34 syntax_example = "ascii(str)",
35 sql_example = r#"```sql
36> select ascii('abc');
37+--------------------+
38| ascii(Utf8("abc")) |
39+--------------------+
40| 97 |
41+--------------------+
42> select ascii('🚀');
43+-------------------+
44| ascii(Utf8("🚀")) |
45+-------------------+
46| 128640 |
47+-------------------+
48```"#,
49 standard_argument(name = "str", prefix = "String"),
50 related_udf(name = "chr")
51)]
52#[derive(Debug)]
53pub struct AsciiFunc {
54 signature: Signature,
55}
56
57impl Default for AsciiFunc {
58 fn default() -> Self {
59 Self::new()
60 }
61}
62
63impl AsciiFunc {
64 pub fn new() -> Self {
65 Self {
66 signature: Signature::coercible(
67 vec![Coercion::new_exact(TypeSignatureClass::Native(
68 logical_string(),
69 ))],
70 Volatility::Immutable,
71 ),
72 }
73 }
74}
75
76impl ScalarUDFImpl for AsciiFunc {
77 fn as_any(&self) -> &dyn Any {
78 self
79 }
80
81 fn name(&self) -> &str {
82 "ascii"
83 }
84
85 fn signature(&self) -> &Signature {
86 &self.signature
87 }
88
89 fn return_type(&self, _arg_types: &[DataType]) -> Result<DataType> {
90 use DataType::*;
91
92 Ok(Int32)
93 }
94
95 fn invoke_with_args(&self, args: ScalarFunctionArgs) -> Result<ColumnarValue> {
96 make_scalar_function(ascii, vec![])(&args.args)
97 }
98
99 fn documentation(&self) -> Option<&Documentation> {
100 self.doc()
101 }
102}
103
104fn calculate_ascii<'a, V>(array: V) -> Result<ArrayRef, ArrowError>
105where
106 V: StringArrayType<'a, Item = &'a str>,
107{
108 let values: Vec<_> = (0..array.len())
109 .map(|i| {
110 if array.is_null(i) {
111 0
112 } else {
113 let s = array.value(i);
114 s.chars().next().map_or(0, |c| c as i32)
115 }
116 })
117 .collect();
118
119 let array = Int32Array::new(values.into(), array.nulls().cloned());
120
121 Ok(Arc::new(array))
122}
123
124pub fn ascii(args: &[ArrayRef]) -> Result<ArrayRef> {
126 match args[0].data_type() {
127 DataType::Utf8 => {
128 let string_array = args[0].as_string::<i32>();
129 Ok(calculate_ascii(string_array)?)
130 }
131 DataType::LargeUtf8 => {
132 let string_array = args[0].as_string::<i64>();
133 Ok(calculate_ascii(string_array)?)
134 }
135 DataType::Utf8View => {
136 let string_array = args[0].as_string_view();
137 Ok(calculate_ascii(string_array)?)
138 }
139 _ => internal_err!("Unsupported data type"),
140 }
141}
142
143#[cfg(test)]
144mod tests {
145 use crate::string::ascii::AsciiFunc;
146 use crate::utils::test::test_function;
147 use arrow::array::{Array, Int32Array};
148 use arrow::datatypes::DataType::Int32;
149 use datafusion_common::{Result, ScalarValue};
150 use datafusion_expr::{ColumnarValue, ScalarUDFImpl};
151
152 macro_rules! test_ascii {
153 ($INPUT:expr, $EXPECTED:expr) => {
154 test_function!(
155 AsciiFunc::new(),
156 vec![ColumnarValue::Scalar(ScalarValue::Utf8($INPUT))],
157 $EXPECTED,
158 i32,
159 Int32,
160 Int32Array
161 );
162
163 test_function!(
164 AsciiFunc::new(),
165 vec![ColumnarValue::Scalar(ScalarValue::LargeUtf8($INPUT))],
166 $EXPECTED,
167 i32,
168 Int32,
169 Int32Array
170 );
171
172 test_function!(
173 AsciiFunc::new(),
174 vec![ColumnarValue::Scalar(ScalarValue::Utf8View($INPUT))],
175 $EXPECTED,
176 i32,
177 Int32,
178 Int32Array
179 );
180 };
181 }
182
183 #[test]
184 fn test_functions() -> Result<()> {
185 test_ascii!(Some(String::from("x")), Ok(Some(120)));
186 test_ascii!(Some(String::from("a")), Ok(Some(97)));
187 test_ascii!(Some(String::from("")), Ok(Some(0)));
188 test_ascii!(Some(String::from("🚀")), Ok(Some(128640)));
189 test_ascii!(None, Ok(None));
190 Ok(())
191 }
192}