String fixes (#379)

tool/src/c/runtime.h

@@ -35,11 +35,6 @@ typedef struct DiplomatWriteable {
 
 DiplomatWriteable diplomat_simple_writeable(char* buf, size_t buf_size);
 
-typedef struct DiplomatStringView {
-    const char* data;
-    size_t len;
-} DiplomatStringView;
-
 #define MAKE_SLICE_VIEW(name, c_ty) \
     typedef struct Diplomat##name##View { \
         const c_ty* data; \
@@ -60,6 +55,8 @@ MAKE_SLICE_VIEW(F32, float)
 MAKE_SLICE_VIEW(F64, double)
 MAKE_SLICE_VIEW(Bool, bool)
 MAKE_SLICE_VIEW(Char, char32_t)
+MAKE_SLICE_VIEW(String, char)
+MAKE_SLICE_VIEW(U16String, char16_t)
 
 
 #ifdef __cplusplus

tool/src/c/structs.rs

@@ -91,7 +91,7 @@ pub fn gen_method<W: fmt::Write>(
         } else if let ast::TypeName::StrReference(_, ast::StringEncoding::UnvalidatedUtf16) =
             &param.ty
         {
-            write!(out, "const uint16_t* {0}_data, size_t {0}_len", param.name)?;
+            write!(out, "const char16_t* {0}_data, size_t {0}_len", param.name)?;
         } else if let ast::TypeName::PrimitiveSlice(_, mutability, prim) = &param.ty {
             write!(
                 out,

tool/src/c2/ty.rs

@@ -250,7 +250,10 @@ impl<'ccx, 'tcx: 'ccx, 'header> TyGenContext<'ccx, 'tcx, 'header> {
                 if !is_struct =>
             {
                 vec![
-                    ("const wchar_t*".into(), format!("{param_name}_data").into()),
+                    (
+                        "const char16_t*".into(),
+                        format!("{param_name}_data").into(),
+                    ),
                     ("size_t".into(), format!("{param_name}_len").into()),
                 ]
             }
@@ -329,7 +332,7 @@ impl<'ccx, 'tcx: 'ccx, 'header> TyGenContext<'ccx, 'tcx, 'header> {
                         _,
                         hir::StringEncoding::UnvalidatedUtf8 | hir::StringEncoding::Utf8,
                     ) => "char".into(),
-                    hir::Slice::Str(_, hir::StringEncoding::UnvalidatedUtf16) => "wchar_t".into(),
+                    hir::Slice::Str(_, hir::StringEncoding::UnvalidatedUtf16) => "char16_t".into(),
                     hir::Slice::Primitive(_, prim) => self.cx.formatter.fmt_primitive_as_c(*prim),
                     &_ => unreachable!("unknown AST/HIR variant"),
                 };

tool/src/cpp/config.rs

@@ -21,6 +21,7 @@ pub struct LibraryConfig {
     pub someopt: CallableLibraryType,
     pub span: LibraryType,
     pub string_view: LibraryType,
+    pub string_view16: LibraryType,
     pub unique_ptr: LibraryType,
 }
 
@@ -53,6 +54,10 @@ impl LibraryConfig {
                 name: "std::string_view".into(),
                 expr: "std::string_view".into(),
             },
+            string_view16: LibraryType {
+                name: "std::u16string_view".into(),
+                expr: "std::u16string_view".into(),
+            },
             unique_ptr: LibraryType {
                 name: "std::unique_ptr".into(),
                 expr: "std::unique_ptr".into(),

tool/src/cpp/configs/mfbt.toml

@@ -27,6 +27,10 @@ expr = "mozilla::Span"
 name = "mozilla::Span"
 expr = "mozilla::Span<const char>"
 
+[string_view16]
+name = "mozilla::Span"
+expr = "mozilla::Span<const char16_t>"
+
 [unique_ptr]
 name = "mozilla::UniquePtr"
 expr = "mozilla::UniquePtr"

tool/src/cpp/conversions.rs

@@ -219,12 +219,12 @@ pub fn gen_rust_to_cpp<W: Write>(
         }
         ast::TypeName::StrReference(_, ast::StringEncoding::UnvalidatedUtf16) => {
             let raw_value_id = format!("diplomat_slice_raw_{path}");
-            writeln!(out, "capi::DiplomatU16View {raw_value_id} = {cpp};").unwrap();
+            writeln!(out, "capi::DiplomatU16StringView {raw_value_id} = {cpp};").unwrap();
 
             let span = &library_config.span.expr;
             writeln!(
                 out,
-                "{span}<const uint16_t> slice({raw_value_id}.data, {raw_value_id}.len);"
+                "{span}<const char16_t> slice({raw_value_id}.data, {raw_value_id}.len);"
             )
             .unwrap();
             "slice".into()

tool/src/cpp/structs.rs

@@ -199,14 +199,29 @@ fn gen_method<W: fmt::Write>(
     }
 
     if is_header {
-        gen_comment_block(
-            out,
-            &super::docs::gen_docs_and_lifetime_notes_markdown(
-                method,
-                docs_url_gen,
-                ast::MarkdownStyle::Normal,
-            ),
-        )?;
+        let mut comment = super::docs::gen_docs_and_lifetime_notes_markdown(
+            method,
+            docs_url_gen,
+            ast::MarkdownStyle::Normal,
+        );
+
+        if method.params.iter().any(|p| {
+            matches!(
+                p,
+                ast::Param {
+                    ty: ast::TypeName::StrReference(_, ast::StringEncoding::Utf8),
+                    ..
+                }
+            )
+        }) {
+            if !comment.is_empty() {
+                writeln!(comment).unwrap();
+                writeln!(comment).unwrap();
+            }
+            writeln!(comment, "Warning: Passing ill-formed UTF-8 is undefined behavior (and may be memory-unsafe).").unwrap();
+        }
+
+        gen_comment_block(out, &comment)?;
     }
     let params_to_gen = gen_method_interface(
         method,

tool/src/cpp/types.rs

@@ -164,7 +164,8 @@ fn gen_type_inner<W: fmt::Write>(
         }
 
         ast::TypeName::StrReference(_, ast::StringEncoding::UnvalidatedUtf16) => {
-            write!(out, "const {}<const uint16_t>", library_config.span.expr)?;
+            let maybe_const = if in_struct { "" } else { "const " };
+            write!(out, "{maybe_const}{}", library_config.string_view16.expr)?;
         }
 
         ast::TypeName::PrimitiveSlice(_, ast::Mutability::Mutable, prim) => {

tool/src/cpp2/formatter.rs

@@ -139,7 +139,7 @@ impl<'tcx> Cpp2Formatter<'tcx> {
 
     pub fn fmt_borrowed_utf16_str(&self) -> Cow<'static, str> {
         // TODO: This needs to change if an abstraction other than std::u16string_view is used
-        "std::wstring_view".into()
+        "std::u16string_view".into()
     }
 
     pub fn fmt_owned_str(&self) -> Cow<'static, str> {