引入编码api

src/lib.rs

@@ -7,11 +7,11 @@ use encoding_rs::Encoding;
 use futures_util::SinkExt;
 use prost::Message as WebMessage;
 use spin::Mutex;
+use std::sync::OnceLock;
 use std::sync::mpsc;
 use tokio::net::TcpStream;
 use tokio::runtime::Runtime;
 use tokio_tungstenite::tungstenite::Message;
-use std::sync::OnceLock;
 
 static LOGGER_SENDER: OnceLock<mpsc::Sender<Starpoles>> = OnceLock::new();
 
@@ -28,6 +28,15 @@ mod ffi {
 
         fn http_get(url: &CxxString) -> Result<String>;
 
+        /// 尝试将未知编码的字符串转换为UTF-8
+        fn to_utf8(string: &CxxString) -> String;
+
+        /// UTF-16LE转换为UTF-8
+        fn unicode_to_utf_8(string: &CxxString) -> String;
+
+        /// 猜测编码
+        fn guess_encoding(string: &CxxString) -> String;
+
         fn init_log(is_debug: bool, ws_uel: &CxxString);
 
         fn log_trace(msg: &CxxString);
@@ -42,6 +51,28 @@ mod ffi {
     }
 }
 
+fn unicode_to_utf_8(string: &CxxString) -> String {
+    // 约定：传入的内容按 UTF-16LE 字节序解释
+    let mut bytes = string.as_bytes();
+
+    // UTF-16 必须是偶数长度；若是奇数长度，丢弃最后 1 个字节避免越界/误解码
+    if (bytes.len() & 1) == 1 {
+        bytes = &bytes[..bytes.len() - 1];
+    }
+
+    // 用 encoding_rs 直接把 UTF-16LE 转成 Rust 的 UTF-8 String
+    let (cow, _actual_used, _had_errors) = encoding_rs::UTF_16LE.decode(bytes);
+    cow.into_owned()
+}
+
+fn guess_encoding(string: &CxxString) -> String {
+    utils::guess_encoding_label(string)
+}
+
+fn to_utf8(string: &CxxString) -> String {
+    cxx_string_to_string(string).to_string()
+}
+
 fn log_error(msg: &cxx::CxxString) {
     let msg = cxx_string_to_string(msg);
     let lock = LOGGER_SENDER.get().unwrap();