datafusion_functions/unicode/
translate.rs1use std::any::Any;
19use std::sync::Arc;
20
21use arrow::array::{
22 ArrayAccessor, ArrayIter, ArrayRef, AsArray, GenericStringArray, OffsetSizeTrait,
23};
24use arrow::datatypes::DataType;
25use datafusion_common::HashMap;
26use unicode_segmentation::UnicodeSegmentation;
27
28use crate::utils::{make_scalar_function, utf8_to_str_type};
29use datafusion_common::{exec_err, Result};
30use datafusion_expr::TypeSignature::Exact;
31use datafusion_expr::{
32 ColumnarValue, Documentation, ScalarUDFImpl, Signature, Volatility,
33};
34use datafusion_macros::user_doc;
35
36#[user_doc(
37 doc_section(label = "String Functions"),
38 description = "Translates characters in a string to specified translation characters.",
39 syntax_example = "translate(str, chars, translation)",
40 sql_example = r#"```sql
41> select translate('twice', 'wic', 'her');
42+--------------------------------------------------+
43| translate(Utf8("twice"),Utf8("wic"),Utf8("her")) |
44+--------------------------------------------------+
45| there |
46+--------------------------------------------------+
47```"#,
48 standard_argument(name = "str", prefix = "String"),
49 argument(name = "chars", description = "Characters to translate."),
50 argument(
51 name = "translation",
52 description = "Translation characters. Translation characters replace only characters at the same position in the **chars** string."
53 )
54)]
55#[derive(Debug)]
56pub struct TranslateFunc {
57 signature: Signature,
58}
59
60impl Default for TranslateFunc {
61 fn default() -> Self {
62 Self::new()
63 }
64}
65
66impl TranslateFunc {
67 pub fn new() -> Self {
68 use DataType::*;
69 Self {
70 signature: Signature::one_of(
71 vec![
72 Exact(vec![Utf8View, Utf8, Utf8]),
73 Exact(vec![Utf8, Utf8, Utf8]),
74 ],
75 Volatility::Immutable,
76 ),
77 }
78 }
79}
80
81impl ScalarUDFImpl for TranslateFunc {
82 fn as_any(&self) -> &dyn Any {
83 self
84 }
85
86 fn name(&self) -> &str {
87 "translate"
88 }
89
90 fn signature(&self) -> &Signature {
91 &self.signature
92 }
93
94 fn return_type(&self, arg_types: &[DataType]) -> Result<DataType> {
95 utf8_to_str_type(&arg_types[0], "translate")
96 }
97
98 fn invoke_with_args(
99 &self,
100 args: datafusion_expr::ScalarFunctionArgs,
101 ) -> Result<ColumnarValue> {
102 make_scalar_function(invoke_translate, vec![])(&args.args)
103 }
104
105 fn documentation(&self) -> Option<&Documentation> {
106 self.doc()
107 }
108}
109
110fn invoke_translate(args: &[ArrayRef]) -> Result<ArrayRef> {
111 match args[0].data_type() {
112 DataType::Utf8View => {
113 let string_array = args[0].as_string_view();
114 let from_array = args[1].as_string::<i32>();
115 let to_array = args[2].as_string::<i32>();
116 translate::<i32, _, _>(string_array, from_array, to_array)
117 }
118 DataType::Utf8 => {
119 let string_array = args[0].as_string::<i32>();
120 let from_array = args[1].as_string::<i32>();
121 let to_array = args[2].as_string::<i32>();
122 translate::<i32, _, _>(string_array, from_array, to_array)
123 }
124 DataType::LargeUtf8 => {
125 let string_array = args[0].as_string::<i64>();
126 let from_array = args[1].as_string::<i64>();
127 let to_array = args[2].as_string::<i64>();
128 translate::<i64, _, _>(string_array, from_array, to_array)
129 }
130 other => {
131 exec_err!("Unsupported data type {other:?} for function translate")
132 }
133 }
134}
135
136fn translate<'a, T: OffsetSizeTrait, V, B>(
139 string_array: V,
140 from_array: B,
141 to_array: B,
142) -> Result<ArrayRef>
143where
144 V: ArrayAccessor<Item = &'a str>,
145 B: ArrayAccessor<Item = &'a str>,
146{
147 let string_array_iter = ArrayIter::new(string_array);
148 let from_array_iter = ArrayIter::new(from_array);
149 let to_array_iter = ArrayIter::new(to_array);
150
151 let result = string_array_iter
152 .zip(from_array_iter)
153 .zip(to_array_iter)
154 .map(|((string, from), to)| match (string, from, to) {
155 (Some(string), Some(from), Some(to)) => {
156 let from_map: HashMap<&str, usize> = from
158 .graphemes(true)
159 .collect::<Vec<&str>>()
160 .iter()
161 .enumerate()
162 .map(|(index, c)| (c.to_owned(), index))
163 .collect();
164
165 let to = to.graphemes(true).collect::<Vec<&str>>();
166
167 Some(
168 string
169 .graphemes(true)
170 .collect::<Vec<&str>>()
171 .iter()
172 .flat_map(|c| match from_map.get(*c) {
173 Some(n) => to.get(*n).copied(),
174 None => Some(*c),
175 })
176 .collect::<Vec<&str>>()
177 .concat(),
178 )
179 }
180 _ => None,
181 })
182 .collect::<GenericStringArray<T>>();
183
184 Ok(Arc::new(result) as ArrayRef)
185}
186
187#[cfg(test)]
188mod tests {
189 use arrow::array::{Array, StringArray};
190 use arrow::datatypes::DataType::Utf8;
191
192 use datafusion_common::{Result, ScalarValue};
193 use datafusion_expr::{ColumnarValue, ScalarUDFImpl};
194
195 use crate::unicode::translate::TranslateFunc;
196 use crate::utils::test::test_function;
197
198 #[test]
199 fn test_functions() -> Result<()> {
200 test_function!(
201 TranslateFunc::new(),
202 vec![
203 ColumnarValue::Scalar(ScalarValue::from("12345")),
204 ColumnarValue::Scalar(ScalarValue::from("143")),
205 ColumnarValue::Scalar(ScalarValue::from("ax"))
206 ],
207 Ok(Some("a2x5")),
208 &str,
209 Utf8,
210 StringArray
211 );
212 test_function!(
213 TranslateFunc::new(),
214 vec![
215 ColumnarValue::Scalar(ScalarValue::Utf8(None)),
216 ColumnarValue::Scalar(ScalarValue::from("143")),
217 ColumnarValue::Scalar(ScalarValue::from("ax"))
218 ],
219 Ok(None),
220 &str,
221 Utf8,
222 StringArray
223 );
224 test_function!(
225 TranslateFunc::new(),
226 vec![
227 ColumnarValue::Scalar(ScalarValue::from("12345")),
228 ColumnarValue::Scalar(ScalarValue::Utf8(None)),
229 ColumnarValue::Scalar(ScalarValue::from("ax"))
230 ],
231 Ok(None),
232 &str,
233 Utf8,
234 StringArray
235 );
236 test_function!(
237 TranslateFunc::new(),
238 vec![
239 ColumnarValue::Scalar(ScalarValue::from("12345")),
240 ColumnarValue::Scalar(ScalarValue::from("143")),
241 ColumnarValue::Scalar(ScalarValue::Utf8(None))
242 ],
243 Ok(None),
244 &str,
245 Utf8,
246 StringArray
247 );
248 test_function!(
249 TranslateFunc::new(),
250 vec![
251 ColumnarValue::Scalar(ScalarValue::from("é2íñ5")),
252 ColumnarValue::Scalar(ScalarValue::from("éñí")),
253 ColumnarValue::Scalar(ScalarValue::from("óü")),
254 ],
255 Ok(Some("ó2ü5")),
256 &str,
257 Utf8,
258 StringArray
259 );
260 #[cfg(not(feature = "unicode_expressions"))]
261 test_function!(
262 TranslateFunc::new(),
263 vec![
264 ColumnarValue::Scalar(ScalarValue::from("12345")),
265 ColumnarValue::Scalar(ScalarValue::from("143")),
266 ColumnarValue::Scalar(ScalarValue::from("ax")),
267 ],
268 internal_err!(
269 "function translate requires compilation with feature flag: unicode_expressions."
270 ),
271 &str,
272 Utf8,
273 StringArray
274 );
275
276 Ok(())
277 }
278}