引入编码api

Cargo.lock

@@ -7,6 +7,7 @@ name = "DnfUtils"
 version = "0.1.0"
 dependencies = [
  "bytes",
+ "chardetng",
  "chrono",
  "colored",
  "cxx",
@@ -162,6 +163,17 @@ version = "1.0.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9330f8b2ff13f34540b44e946ef35111825727b38d33286ef986142615121801"
 
+[[package]]
+name = "chardetng"
+version = "0.1.17"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "14b8f0b65b7b08ae3c8187e8d77174de20cb6777864c6b832d8ad365999cf1ea"
+dependencies = [
+ "cfg-if",
+ "encoding_rs",
+ "memchr",
+]
+
 [[package]]
 name = "chrono"
 version = "0.4.43"

Cargo.toml

@@ -32,6 +32,7 @@ prost = "0.14.3"
 tokio-tungstenite = "0.28.0"
 bytes = "1.11.0"
 spin = "0.10.0"
+chardetng = "0.1.17"
 
 [build-dependencies]
 cxx-build = "1.0.192"

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();

src/utils.rs

@@ -1,5 +1,13 @@
+use chardetng::EncodingDetector;
 use encoding_rs::Encoding;
 
+pub fn guess_encoding_label(s: &cxx::CxxString) -> String {
+    let mut det = EncodingDetector::new();
+    det.feed(s.as_bytes(), true);
+    let enc = det.guess(None, true);
+    enc.name().to_string()
+}
+
 pub fn cxx_string_to_string(s: &cxx::CxxString) -> String {
     match s.to_str() {
         Ok(s) => return s.to_string(),