From dd95313c0b65af9f2c722272f69c132f4d09c9b6 Mon Sep 17 00:00:00 2001
From: NANASE <wszhds@gmail.com>
Date: Thu, 31 Oct 2024 00:23:39 +0800
Subject: [PATCH] feat: add function CharLength & CharacterLength

---
 rust-toolchain                 |  2 +-
 src/db.rs                      |  4 ++
 src/function/char_length.rs    | 70 ++++++++++++++++++++++++++++++++++
 src/function/mod.rs            |  1 +
 tests/slt/sql_2016/E021_04.slt | 18 ++++-----
 5 files changed, 84 insertions(+), 11 deletions(-)
 create mode 100644 src/function/char_length.rs

diff --git a/rust-toolchain b/rust-toolchain
index 7df939e8..5f739953 100644
--- a/rust-toolchain
+++ b/rust-toolchain
@@ -1 +1 @@
-nightly-2024-04-27
\ No newline at end of file
+nightly-2024-10-18
\ No newline at end of file
diff --git a/src/db.rs b/src/db.rs
index a47b317c..fdfcda52 100644
--- a/src/db.rs
+++ b/src/db.rs
@@ -5,6 +5,7 @@ use crate::execution::{build_write, try_collect};
 use crate::expression::function::scala::ScalarFunctionImpl;
 use crate::expression::function::table::TableFunctionImpl;
 use crate::expression::function::FunctionSummary;
+use crate::function::char_length::CharLength;
 use crate::function::current_date::CurrentDate;
 use crate::function::lower::Lower;
 use crate::function::numbers::Numbers;
@@ -50,6 +51,9 @@ impl DataBaseBuilder {
             scala_functions: Default::default(),
             table_functions: Default::default(),
         };
+        builder = builder.register_scala_function(CharLength::new("char_length".to_lowercase()));
+        builder =
+            builder.register_scala_function(CharLength::new("character_length".to_lowercase()));
         builder = builder.register_scala_function(CurrentDate::new());
         builder = builder.register_scala_function(Lower::new());
         builder = builder.register_scala_function(Upper::new());
diff --git a/src/function/char_length.rs b/src/function/char_length.rs
new file mode 100644
index 00000000..d23cc018
--- /dev/null
+++ b/src/function/char_length.rs
@@ -0,0 +1,70 @@
+use crate::catalog::ColumnRef;
+use crate::errors::DatabaseError;
+use crate::expression::function::scala::FuncMonotonicity;
+use crate::expression::function::scala::ScalarFunctionImpl;
+use crate::expression::function::FunctionSummary;
+use crate::expression::ScalarExpression;
+use crate::types::tuple::Tuple;
+use crate::types::value::DataValue;
+use crate::types::LogicalType;
+use serde::Deserialize;
+use serde::Serialize;
+use sqlparser::ast::CharLengthUnits;
+use std::sync::Arc;
+
+#[derive(Debug, Serialize, Deserialize)]
+pub(crate) struct CharLength {
+    summary: FunctionSummary,
+}
+
+impl CharLength {
+    pub(crate) fn new(function_name: String) -> Arc<Self> {
+        let arg_types = vec![LogicalType::Varchar(None, CharLengthUnits::Characters)];
+        Arc::new(Self {
+            summary: FunctionSummary {
+                name: function_name,
+                arg_types,
+            },
+        })
+    }
+}
+
+#[typetag::serde]
+impl ScalarFunctionImpl for CharLength {
+    #[allow(unused_variables, clippy::redundant_closure_call)]
+    fn eval(
+        &self,
+        exprs: &[ScalarExpression],
+        tuples: &Tuple,
+        columns: &[ColumnRef],
+    ) -> Result<DataValue, DatabaseError> {
+        let value = exprs[0].eval(tuples, columns)?;
+        let mut value = DataValue::clone(&value);
+        if !matches!(value.logical_type(), LogicalType::Varchar(_, _)) {
+            value = DataValue::clone(&value)
+                .cast(&LogicalType::Varchar(None, CharLengthUnits::Characters))?;
+        }
+        let mut length: u64 = 0;
+        if let DataValue::Utf8 {
+            value: Some(value),
+            ty,
+            unit,
+        } = &mut value
+        {
+            length = value.len() as u64;
+        }
+        Ok(DataValue::UInt64(Some(length)))
+    }
+
+    fn monotonicity(&self) -> Option<FuncMonotonicity> {
+        todo!()
+    }
+
+    fn return_type(&self) -> &LogicalType {
+        &LogicalType::Varchar(None, CharLengthUnits::Characters)
+    }
+
+    fn summary(&self) -> &FunctionSummary {
+        &self.summary
+    }
+}
diff --git a/src/function/mod.rs b/src/function/mod.rs
index e48c7c2d..6c660c57 100644
--- a/src/function/mod.rs
+++ b/src/function/mod.rs
@@ -1,3 +1,4 @@
+pub(crate) mod char_length;
 pub(crate) mod current_date;
 pub(crate) mod lower;
 pub(crate) mod numbers;
diff --git a/tests/slt/sql_2016/E021_04.slt b/tests/slt/sql_2016/E021_04.slt
index dc7a3a52..ac401405 100644
--- a/tests/slt/sql_2016/E021_04.slt
+++ b/tests/slt/sql_2016/E021_04.slt
@@ -1,14 +1,12 @@
 # E021-04: CHARACTER_LENGTH function
 
-# TODO: CHARACTER_LENGTH()/CHAR_LENGTH()
+query I
+SELECT CHARACTER_LENGTH ( 'foo' )
+----
+3
 
-# query I
-# SELECT CHARACTER_LENGTH ( 'foo' )
-# ----
-# 3
 
-
-# query I
-# SELECT CHAR_LENGTH ( 'foo' )
-# ----
-# 3
+query I
+SELECT CHAR_LENGTH ( 'foo' )
+----
+3