Merge pull request #224 from ethanniser/add-unicode

COMPAT.md

@@ -90,11 +90,11 @@ This document describes the SQLite compatibility status of Limbo:
 | octet_length(X)              | No     |         |
 | printf(FORMAT,...)           | No     |         |
 | quote(X)                     | No     |         |
-| random()                     | Yes     |         |
+| random()                     | Yes    |         |
 | randomblob(N)                | No     |         |
 | replace(X,Y,Z)               | No     |         |
-| round(X)                     | Yes     |         |
-| round(X,Y)                   | Yes     |         |
+| round(X)                     | Yes    |         |
+| round(X,Y)                   | Yes    |         |
 | rtrim(X)                     | No     |         |
 | rtrim(X,Y)                   | No     |         |
 | sign(X)                      | No     |         |
@@ -109,12 +109,12 @@ This document describes the SQLite compatibility status of Limbo:
 | substring(X,Y,Z)             | No     |         |
 | substring(X,Y)               | No     |         |
 | total_changes()              | No     |         |
-| trim(X)                      | Yes     |         |
-| trim(X,Y)                    | Yes     |         |
+| trim(X)                      | Yes    |         |
+| trim(X,Y)                    | Yes    |         |
 | typeof(X)                    | No     |         |
 | unhex(X)                     | No     |         |
 | unhex(X,Y)                   | No     |         |
-| unicode(X)                   | No     |         |
+| unicode(X)                   | Yes    |         |
 | unlikely(X)                  | No     |         |
 | upper(X)                     | Yes    |         |
 | zeroblob(N)                  | No     |         |

core/function.rs

@@ -39,6 +39,7 @@ pub enum ScalarFunc {
     Min,
     Max,
     Date,
+    Unicode,
 }
 
 impl ToString for ScalarFunc {
@@ -56,6 +57,7 @@ impl ToString for ScalarFunc {
             ScalarFunc::Min => "min".to_string(),
             ScalarFunc::Max => "max".to_string(),
             ScalarFunc::Date => "date".to_string(),
+            ScalarFunc::Unicode => "unicode".to_string(),
         }
     }
 }
@@ -89,6 +91,7 @@ impl Func {
             "round" => Ok(Func::Scalar(ScalarFunc::Round)),
             "length" => Ok(Func::Scalar(ScalarFunc::Length)),
             "date" => Ok(Func::Scalar(ScalarFunc::Date)),
+            "unicode" => Ok(Func::Scalar(ScalarFunc::Unicode)),
             _ => Err(()),
         }
     }

core/translate/expr.rs

@@ -210,7 +210,8 @@ pub fn translate_expr(
                         ScalarFunc::Abs
                         | ScalarFunc::Lower
                         | ScalarFunc::Upper
-                        | ScalarFunc::Length => {
+                        | ScalarFunc::Length
+                        | ScalarFunc::Unicode => {
                             let args = if let Some(args) = args {
                                 if args.len() != 1 {
                                     crate::bail_parse_error!(

core/types.rs

@@ -43,7 +43,7 @@ impl Display for OwnedValue {
             OwnedValue::Integer(i) => write!(f, "{}", i),
             OwnedValue::Float(fl) => write!(f, "{:?}", fl),
             OwnedValue::Text(s) => write!(f, "{}", s),
-            OwnedValue::Blob(b) => write!(f, "{:?}", b),
+            OwnedValue::Blob(b) => write!(f, "{}", String::from_utf8_lossy(b)),
             OwnedValue::Agg(a) => match a.as_ref() {
                 AggContext::Avg(acc, _count) => write!(f, "{}", acc),
                 AggContext::Sum(acc) => write!(f, "{}", acc),

core/vdbe/mod.rs

@@ -1153,6 +1153,11 @@ impl Program {
                         }
                         state.pc += 1;
                     }
+                    ScalarFunc::Unicode => {
+                        let reg_value = state.registers[*start_reg].borrow_mut();
+                        state.registers[*dest] = exec_unicode(reg_value);
+                        state.pc += 1;
+                    }
                 },
             }
         }
@@ -1271,6 +1276,23 @@ fn exec_minmax<'a>(
     regs.into_iter().reduce(|a, b| op(a, b)).cloned()
 }
 
+fn exec_unicode(reg: &OwnedValue) -> OwnedValue {
+    match reg {
+        OwnedValue::Text(_)
+        | OwnedValue::Integer(_)
+        | OwnedValue::Float(_)
+        | OwnedValue::Blob(_) => {
+            let text = reg.to_string();
+            if let Some(first_char) = text.chars().next() {
+                OwnedValue::Integer(first_char as u32 as i64)
+            } else {
+                OwnedValue::Null
+            }
+        }
+        _ => OwnedValue::Null,
+    }
+}
+
 fn exec_round(reg: &OwnedValue, precision: Option<OwnedValue>) -> OwnedValue {
     let precision = match precision {
         Some(OwnedValue::Text(x)) => x.parse().unwrap_or(0.0),
@@ -1327,7 +1349,7 @@ fn exec_if(reg: &OwnedValue, null_reg: &OwnedValue, not: bool) -> bool {
 mod tests {
     use super::{
         exec_abs, exec_if, exec_length, exec_like, exec_lower, exec_minmax, exec_random,
-        exec_round, exec_trim, exec_upper, OwnedValue,
+        exec_round, exec_trim, exec_unicode, exec_upper, OwnedValue,
     };
     use std::rc::Rc;
 
@@ -1345,12 +1367,48 @@ mod tests {
         let expected_len = OwnedValue::Integer(7);
         assert_eq!(exec_length(&input_float), expected_len);
 
-        // Expected byte array for "example"
-        let expected_blob = OwnedValue::Blob(Rc::new(vec![101, 120, 97, 109, 112, 108, 101]));
+        let expected_blob = OwnedValue::Blob(Rc::new("example".as_bytes().to_vec()));
         let expected_len = OwnedValue::Integer(7);
         assert_eq!(exec_length(&expected_blob), expected_len);
     }
 
+    #[test]
+    fn test_unicode() {
+        assert_eq!(
+            exec_unicode(&OwnedValue::Text(Rc::new("a".to_string()))),
+            OwnedValue::Integer(97)
+        );
+        assert_eq!(
+            exec_unicode(&OwnedValue::Text(Rc::new("😊".to_string()))),
+            OwnedValue::Integer(128522)
+        );
+        assert_eq!(
+            exec_unicode(&OwnedValue::Text(Rc::new("".to_string()))),
+            OwnedValue::Null
+        );
+        assert_eq!(
+            exec_unicode(&OwnedValue::Integer(23)),
+            OwnedValue::Integer(50)
+        );
+        assert_eq!(
+            exec_unicode(&OwnedValue::Integer(0)),
+            OwnedValue::Integer(48)
+        );
+        assert_eq!(
+            exec_unicode(&OwnedValue::Float(0.0)),
+            OwnedValue::Integer(48)
+        );
+        assert_eq!(
+            exec_unicode(&OwnedValue::Float(23.45)),
+            OwnedValue::Integer(50)
+        );
+        assert_eq!(exec_unicode(&OwnedValue::Null), OwnedValue::Null);
+        assert_eq!(
+            exec_unicode(&OwnedValue::Blob(Rc::new("example".as_bytes().to_vec()))),
+            OwnedValue::Integer(101)
+        );
+    }
+
     #[test]
     fn test_minmax() {
         let min_fn = |a, b| if a < b { a } else { b };

testing/scalar-functions.test

@@ -243,3 +243,26 @@ do_execsql_test date-with-invalid-timezone {
   SELECT date('2023-05-18 15:30:45+25:00');
 } {{}}
 
+do_execsql_test unicode-a {
+  SELECT unicode('a');
+} {97}
+
+do_execsql_test unicode-emoji {
+  SELECT unicode('😊');
+} {128522}
+
+do_execsql_test unicode-empty {
+  SELECT unicode('');
+} {}
+
+do_execsql_test unicode-number {
+  SELECT unicode(23);
+} {50}
+
+do_execsql_test unicode-float {
+  SELECT unicode(23.45);
+} {50}
+
+do_execsql_test unicode-null {
+  SELECT unicode(NULL);
+} {}