datafusion_functions/string/
upper.rs1use crate::string::common::to_upper;
19use crate::utils::utf8_to_str_type;
20use arrow::datatypes::DataType;
21use datafusion_common::types::logical_string;
22use datafusion_common::Result;
23use datafusion_expr::{
24 Coercion, ColumnarValue, Documentation, ScalarFunctionArgs, ScalarUDFImpl, Signature,
25 TypeSignatureClass, Volatility,
26};
27use datafusion_macros::user_doc;
28use std::any::Any;
29
30#[user_doc(
31 doc_section(label = "String Functions"),
32 description = "Converts a string to upper-case.",
33 syntax_example = "upper(str)",
34 sql_example = r#"```sql
35> select upper('dataFusion');
36+---------------------------+
37| upper(Utf8("dataFusion")) |
38+---------------------------+
39| DATAFUSION |
40+---------------------------+
41```"#,
42 standard_argument(name = "str", prefix = "String"),
43 related_udf(name = "initcap"),
44 related_udf(name = "lower")
45)]
46#[derive(Debug, PartialEq, Eq, Hash)]
47pub struct UpperFunc {
48 signature: Signature,
49}
50
51impl Default for UpperFunc {
52 fn default() -> Self {
53 Self::new()
54 }
55}
56
57impl UpperFunc {
58 pub fn new() -> Self {
59 Self {
60 signature: Signature::coercible(
61 vec![Coercion::new_exact(TypeSignatureClass::Native(
62 logical_string(),
63 ))],
64 Volatility::Immutable,
65 ),
66 }
67 }
68}
69
70impl ScalarUDFImpl for UpperFunc {
71 fn as_any(&self) -> &dyn Any {
72 self
73 }
74
75 fn name(&self) -> &str {
76 "upper"
77 }
78
79 fn signature(&self) -> &Signature {
80 &self.signature
81 }
82
83 fn return_type(&self, arg_types: &[DataType]) -> Result<DataType> {
84 utf8_to_str_type(&arg_types[0], "upper")
85 }
86
87 fn invoke_with_args(&self, args: ScalarFunctionArgs) -> Result<ColumnarValue> {
88 to_upper(&args.args, "upper")
89 }
90
91 fn documentation(&self) -> Option<&Documentation> {
92 self.doc()
93 }
94}
95
96#[cfg(test)]
97mod tests {
98 use super::*;
99 use arrow::array::{Array, ArrayRef, StringArray};
100 use arrow::datatypes::DataType::Utf8;
101 use arrow::datatypes::Field;
102 use datafusion_common::config::ConfigOptions;
103 use std::sync::Arc;
104
105 fn to_upper(input: ArrayRef, expected: ArrayRef) -> Result<()> {
106 let func = UpperFunc::new();
107
108 let arg_field = Field::new("a", input.data_type().clone(), true).into();
109 let args = ScalarFunctionArgs {
110 number_rows: input.len(),
111 args: vec![ColumnarValue::Array(input)],
112 arg_fields: vec![arg_field],
113 return_field: Field::new("f", Utf8, true).into(),
114 config_options: Arc::new(ConfigOptions::default()),
115 lambdas: None,
116 };
117
118 let result = match func.invoke_with_args(args)? {
119 ColumnarValue::Array(result) => result,
120 _ => unreachable!("upper"),
121 };
122 assert_eq!(&expected, &result);
123 Ok(())
124 }
125
126 #[test]
127 fn upper_maybe_optimization() -> Result<()> {
128 let input = Arc::new(StringArray::from(vec![
129 Some("农历新年"),
130 None,
131 Some("datafusion"),
132 Some("0123456789"),
133 Some(""),
134 ])) as ArrayRef;
135
136 let expected = Arc::new(StringArray::from(vec![
137 Some("农历新年"),
138 None,
139 Some("DATAFUSION"),
140 Some("0123456789"),
141 Some(""),
142 ])) as ArrayRef;
143
144 to_upper(input, expected)
145 }
146
147 #[test]
148 fn upper_full_optimization() -> Result<()> {
149 let input = Arc::new(StringArray::from(vec![
150 Some("arrow"),
151 None,
152 Some("datafusion"),
153 Some("0123456789"),
154 Some(""),
155 ])) as ArrayRef;
156
157 let expected = Arc::new(StringArray::from(vec![
158 Some("ARROW"),
159 None,
160 Some("DATAFUSION"),
161 Some("0123456789"),
162 Some(""),
163 ])) as ArrayRef;
164
165 to_upper(input, expected)
166 }
167
168 #[test]
169 fn upper_partial_optimization() -> Result<()> {
170 let input = Arc::new(StringArray::from(vec![
171 Some("arrow"),
172 None,
173 Some("datafusion"),
174 Some("@_"),
175 Some("0123456789"),
176 Some(""),
177 Some("\t\n"),
178 Some("ὀδυσσεύς"),
179 Some("tschüß"),
180 Some("ⱦ"), Some("农历新年"),
182 ])) as ArrayRef;
183
184 let expected = Arc::new(StringArray::from(vec![
185 Some("ARROW"),
186 None,
187 Some("DATAFUSION"),
188 Some("@_"),
189 Some("0123456789"),
190 Some(""),
191 Some("\t\n"),
192 Some("ὈΔΥΣΣΕΎΣ"),
193 Some("TSCHÜSS"),
194 Some("Ⱦ"),
195 Some("农历新年"),
196 ])) as ArrayRef;
197
198 to_upper(input, expected)
199 }
200}