datafusion_functions/string/
ascii.rs1use crate::utils::make_scalar_function;
19use arrow::array::{ArrayAccessor, ArrayIter, ArrayRef, AsArray, Int32Array};
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: ArrayAccessor<Item = &'a str>,
107{
108 let iter = ArrayIter::new(array);
109 let result = iter
110 .map(|string| {
111 string.map(|s| {
112 let mut chars = s.chars();
113 chars.next().map_or(0, |v| v as i32)
114 })
115 })
116 .collect::<Int32Array>();
117
118 Ok(Arc::new(result) as ArrayRef)
119}
120
121pub fn ascii(args: &[ArrayRef]) -> Result<ArrayRef> {
123 match args[0].data_type() {
124 DataType::Utf8 => {
125 let string_array = args[0].as_string::<i32>();
126 Ok(calculate_ascii(string_array)?)
127 }
128 DataType::LargeUtf8 => {
129 let string_array = args[0].as_string::<i64>();
130 Ok(calculate_ascii(string_array)?)
131 }
132 DataType::Utf8View => {
133 let string_array = args[0].as_string_view();
134 Ok(calculate_ascii(string_array)?)
135 }
136 _ => internal_err!("Unsupported data type"),
137 }
138}
139
140#[cfg(test)]
141mod tests {
142 use crate::string::ascii::AsciiFunc;
143 use crate::utils::test::test_function;
144 use arrow::array::{Array, Int32Array};
145 use arrow::datatypes::DataType::Int32;
146 use datafusion_common::{Result, ScalarValue};
147 use datafusion_expr::{ColumnarValue, ScalarUDFImpl};
148
149 macro_rules! test_ascii {
150 ($INPUT:expr, $EXPECTED:expr) => {
151 test_function!(
152 AsciiFunc::new(),
153 vec![ColumnarValue::Scalar(ScalarValue::Utf8($INPUT))],
154 $EXPECTED,
155 i32,
156 Int32,
157 Int32Array
158 );
159
160 test_function!(
161 AsciiFunc::new(),
162 vec![ColumnarValue::Scalar(ScalarValue::LargeUtf8($INPUT))],
163 $EXPECTED,
164 i32,
165 Int32,
166 Int32Array
167 );
168
169 test_function!(
170 AsciiFunc::new(),
171 vec![ColumnarValue::Scalar(ScalarValue::Utf8View($INPUT))],
172 $EXPECTED,
173 i32,
174 Int32,
175 Int32Array
176 );
177 };
178 }
179
180 #[test]
181 fn test_functions() -> Result<()> {
182 test_ascii!(Some(String::from("x")), Ok(Some(120)));
183 test_ascii!(Some(String::from("a")), Ok(Some(97)));
184 test_ascii!(Some(String::from("")), Ok(Some(0)));
185 test_ascii!(None, Ok(None));
186 Ok(())
187 }
188}