Remove encoding_rs again.

bindgen-integration/cpp/Test.h

@@ -20,7 +20,7 @@
         a
 //#define TESTMACRO_INVALID("string") // A conforming preprocessor rejects this
 #define TESTMACRO_STRING_EXPR ("string")
-#define TESTMACRO_STRING_FUNC_NON_UTF8(x) (x "ÿÿ") /* invalid UTF-8 on purpose */
+#define TESTMACRO_STRING_FUNC_NON_UTF8(x) (x "ÿÿ") /* invalid UTF-8 on purpose */
 
 enum {
   MY_ANNOYING_MACRO =

bindgen/Cargo.toml

@@ -29,7 +29,6 @@ annotate-snippets = { version = "0.9.1", features = ["color"], optional = true }
 bitflags = "2.2.1"
 cexpr = "0.6"
 clang-sys = { version = "1", features = ["clang_6_0"] }
-encoding_rs = "0.8.33"
 itertools = { version = ">=0.10,<0.13", default-features = false }
 lazy_static = "1"
 lazycell = "1"

bindgen/clang.rs

@@ -1038,12 +1038,9 @@ pub(crate) struct ClangToken {
 }
 
 impl ClangToken {
-    /// Get the token spelling, without being converted to utf-8.
-    pub(crate) fn spelling(&self) -> &[u8] {
-        let c_str = unsafe {
-            CStr::from_ptr(clang_getCString(self.spelling) as *const _)
-        };
-        c_str.to_bytes()
+    /// Returns the token spelling.
+    pub(crate) fn spelling(&self) -> &CStr {
+        unsafe { CStr::from_ptr(clang_getCString(self.spelling) as *const _) }
     }
 
     /// Converts a ClangToken to a `cexpr` token if possible.

bindgen/ir/context.rs

@@ -2159,8 +2159,8 @@ If you encounter an error missing from this list, please file an issue or a PR!"
         let mut kind = ModuleKind::Normal;
         let mut looking_for_name = false;
         for token in cursor.tokens().iter() {
-            match token.spelling() {
-                b"inline" => {
+            match token.spelling().to_str().unwrap() {
+                "inline" => {
                     debug_assert!(
                         kind != ModuleKind::Inline,
                         "Multiple inline keywords?"
@@ -2177,20 +2177,18 @@ If you encounter an error missing from this list, please file an issue or a PR!"
                 // but the tokenization of the second begins with the double
                 // colon. That's ok, so we only need to handle the weird
                 // tokenization here.
-                b"namespace" | b"::" => {
+                "namespace" | b"::" => {
                     looking_for_name = true;
                 }
-                b"{" => {
+                "{" => {
                     // This should be an anonymous namespace.
                     assert!(looking_for_name);
                     break;
                 }
                 name => {
                     if looking_for_name {
                         if module_name.is_none() {
-                            module_name = Some(
-                                String::from_utf8_lossy(name).into_owned(),
-                            );
+                            module_name = Some(name.into_owned());
                         }
                         break;
                     } else {
@@ -2205,7 +2203,7 @@ If you encounter an error missing from this list, please file an issue or a PR!"
                         // See also https://github.com/rust-lang/rust-bindgen/issues/1676.
                         warn!(
                             "Ignored unknown namespace prefix '{}' at {:?} in {:?}",
-                            String::from_utf8_lossy(name),
+                            name,
                             token,
                             cursor
                         );

bindgen/ir/var.rs

@@ -164,16 +164,14 @@ fn handle_function_macro(
 ) {
     let is_closing_paren = |t: &ClangToken| {
         // Test cheap token kind before comparing exact spellings.
-        t.kind == clang_sys::CXToken_Punctuation && t.spelling() == b")"
+        t.kind == clang_sys::CXToken_Punctuation &&
+            t.spelling().to_bytes() == b")"
     };
     let tokens: Vec<_> = cursor.tokens().iter().collect();
     if let Some(boundary) = tokens.iter().position(is_closing_paren) {
         let mut tokens = tokens
             .iter()
-            .map(|token| {
-                let s = token.spelling();
-                encoding_rs::mem::decode_latin1(s)
-            })
+            .map(|token| token.spelling().to_str().unwrap())
             .collect::<Vec<_>>();
 
         let name = tokens.remove(0);
@@ -185,11 +183,8 @@ fn handle_function_macro(
             .collect();
         let body = tokens;
 
-        let args = args.iter().map(|s| s.as_ref()).collect::<Vec<_>>();
-        let body = body.iter().map(|s| s.as_ref()).collect::<Vec<_>>();
-
         let info = FnMacroInfo {
-            name: &name,
+            name: name,
             args: &args,
             body: &body,
         };