support 两, imporve the parser

magiclen · Dec 11, 2023 · b54c306 · b54c306
1 parent 0585a2e
commit b54c306
Show file tree

Hide file tree

Showing 5 changed files with 41 additions and 2 deletions.
diff --git a/Cargo.toml b/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "chinese-number"
-version = "0.7.4"
+version = "0.7.5"
 authors = ["Magic Len <len@magiclen.org>"]
 edition = "2021"
 rust-version = "1.58"

diff --git a/README.md b/README.md
@@ -37,6 +37,12 @@ assert_eq!(1000000u64, "一兆".to_number(ChineseCountMethod::Low).unwrap());
 assert_eq!(1000000000000u64, "一兆".to_number(ChineseCountMethod::TenThousand).unwrap());
 assert_eq!(10000000000000000u64, "一兆".to_number(ChineseCountMethod::Middle).unwrap());
 assert_eq!(10000000000000000u64, "一兆".to_number(ChineseCountMethod::High).unwrap());
+
+assert_eq!(120u64, "一百二".to_number(ChineseCountMethod::TenThousand).unwrap());
+assert_eq!(2300u64, "兩千三".to_number(ChineseCountMethod::TenThousand).unwrap());
+assert_eq!(34000u64, "三萬四".to_number(ChineseCountMethod::TenThousand).unwrap());
+assert_eq!(105000u64, "十萬五".to_number(ChineseCountMethod::TenThousand).unwrap());
+assert_eq!(150000000u64, "一億五".to_number(ChineseCountMethod::TenThousand).unwrap());
 ```
 
 ## No Std

diff --git a/src/chinese_characters.rs b/src/chinese_characters.rs
@@ -91,7 +91,7 @@ impl ChineseNumber {
         match character {
             '零' | '0' | '〇' => Some(Self::零),
             '一' | '壹' | '1' => Some(Self::一),
-            '二' | '貳' | '贰' | '貮' | '兩' | '2' => Some(Self::二),
+            '二' | '貳' | '贰' | '貮' | '兩' | '两' | '2' => Some(Self::二),
             '三' | '參' | '叁' | '叄' | '参' | '3' => Some(Self::三),
             '四' | '肆' | '4' => Some(Self::四),
             '五' | '伍' | '5' => Some(Self::五),

diff --git a/src/chinese_to_number/functions.rs b/src/chinese_to_number/functions.rs
@@ -226,6 +226,33 @@ pub(crate) fn chinese_to_unsigned_integer(
             exp
         },
         _ => {
+            if length >= 3 {
+                // look ahead
+
+                let left_char = chars[pointer - 1];
+
+                if let Some(ChineseNumber::零) = ChineseNumber::from_char(left_char) {
+                    // do nothing
+                } else {
+                    match ChineseExponent::from_char(left_char) {
+                        Some(exp) if exp >= ChineseExponent::百 => {
+                            let high = chinese_to_unsigned_integer(method, &chars[..pointer])?;
+
+                            let low = chinese_to_unsigned_integer_unit(
+                                method,
+                                &chars[pointer..],
+                                0,
+                                exp,
+                            )?
+                            .0 / 10;
+
+                            return high.checked_add(low).ok_or(ChineseToNumberError::Overflow);
+                        },
+                        _ => (),
+                    }
+                }
+            }
+
             pointer += 1;
             ChineseExponent::個
         },

diff --git a/src/lib.rs b/src/lib.rs
@@ -37,6 +37,12 @@ assert_eq!(1000000u64, "一兆".to_number(ChineseCountMethod::Low).unwrap());
 assert_eq!(1000000000000u64, "一兆".to_number(ChineseCountMethod::TenThousand).unwrap());
 assert_eq!(10000000000000000u64, "一兆".to_number(ChineseCountMethod::Middle).unwrap());
 assert_eq!(10000000000000000u64, "一兆".to_number(ChineseCountMethod::High).unwrap());
+
+assert_eq!(120u64, "一百二".to_number(ChineseCountMethod::TenThousand).unwrap());
+assert_eq!(2300u64, "兩千三".to_number(ChineseCountMethod::TenThousand).unwrap());
+assert_eq!(34000u64, "三萬四".to_number(ChineseCountMethod::TenThousand).unwrap());
+assert_eq!(105000u64, "十萬五".to_number(ChineseCountMethod::TenThousand).unwrap());
+assert_eq!(150000000u64, "一億五".to_number(ChineseCountMethod::TenThousand).unwrap());
 # }
 ```