feat: split utf string for testing

scie-scanner/src/scanner/old/mod.rs

+pub mod scie_scanner;
+pub mod scie_onig;
+pub mod scie_error;
\ No newline at end of file

scie-scanner/src/scanner/old/scie_onig.rs

@@ -3,7 +3,7 @@
 use std::ptr::null_mut;
 use onig::{Syntax, EncodedChars};
 use std::sync::Mutex;
-use crate::scanner::scie_error::ScieOnigError;
+use crate::scanner::old::scie_error::ScieOnigError;
 
 lazy_static! {
     static ref REGEX_NEW_MUTEX: Mutex<()> = Mutex::new(());
@@ -21,9 +21,6 @@ pub struct ScieOnig {
 }
 
 impl ScieOnig {
-    pub fn new(pattern: &str) -> Result<Self, ScieOnigError> {
-
-    }
     pub fn demo_new(pattern: &str) -> Result<Self, ScieOnigError> {
         let option = ScieOnigOptions::REGEX_OPTION_NONE;
         let syntax = Syntax::default();
@@ -69,7 +66,7 @@ impl ScieOnig {
 
 #[cfg(test)]
 mod tests {
-    use crate::scanner::scie_onig::ScieOnig;
+    use crate::scanner::old::scie_onig::ScieOnig;
 
     #[test]
     fn it_works() {

scie-scanner/src/scanner/old/scie_scanner.rs

@@ -137,7 +137,7 @@ pub fn str_vec_to_string<I, T>(iter: I) -> Vec<String>
 
 #[cfg(test)]
 mod tests {
-    use crate::scanner::scie_scanner::{str_vec_to_string, ScieScanner};
+    use crate::scanner::old::scie_scanner::{ScieScanner, str_vec_to_string};
 
     #[test]
     fn should_handle_simple_regex() {

scie-scanner/src/scanner/onig_scanner.rs

@@ -7,9 +7,9 @@ pub struct OnigScanner {
 
 impl OnigScanner {
     pub fn new(pattens: Vec<&str>) -> Self {
-        let str_ptrs_arr: Vec<Pointer> = vec![];
-        let str_len_arr: Vec<i32> = vec![];
-        for x in pattens {
+        let _str_ptrs_arr: Vec<Pointer> = vec![];
+        let _str_len_arr: Vec<i32> = vec![];
+        for _x in pattens {
 
         }
         OnigScanner { _ptr: 0 }

scie-scanner/src/scanner/onig_string.rs

+use unicode_segmentation::UnicodeSegmentation;
+
 pub struct OnigString {
  pub utf16length: i32,
  pub utf8length: i32,
@@ -10,12 +12,15 @@ pub struct OnigString {
 
 impl OnigString {
     pub fn new(str: String) -> Self {
-        let utf16Length = str.len();
-        
+        let utf16length = str.len();
+        let utf8_str = str.graphemes(true).collect::<Vec<&str>>().clone();
+        println!("{:?}", utf8_str);
+        let utf8length = utf8_str.len();
+
 
         OnigString {
-            utf16length: 0,
-            utf8length: 0,
+            utf16length: utf16length as i32,
+            utf8length: utf8length as i32,
             utf16value: "".to_string(),
             utf8value: vec![],
             utf16offset_to_utf8: None,
@@ -27,12 +32,12 @@ impl OnigString {
 
 #[cfg(test)]
 mod tests {
-    use crate::scanner::onig_scanner::OnigScanner;
     use crate::scanner::onig_string::OnigString;
 
     #[test]
     fn it_show_works_works() {
-        OnigString::new(String::from(""));
-        assert!(true)
+        let onig_string = OnigString::new(String::from("a💻bYX"));
+        assert_eq!(8, onig_string.utf16length);
+        assert_eq!(5, onig_string.utf8length);
     }
 }

vscode-tests/vscode-onig/src/lib/utf-string.spec.ts

@@ -4,6 +4,11 @@ import UtfString from './utf-string';
 
 test('constructor', (t) => {
   const utfString = new UtfString("hello, world");
-  t.log(utfString);
   t.is(utfString.utf8Length, 12);
 });
+
+test('utf8 length', (t) => {
+  const utfString = new UtfString("a💻bYX");
+  console.log(utfString);
+  t.is(utfString.utf8Length, 8);
+});