Use from_fn for non-Copyable Rust types (#192)

In Rust, a struct is Copyable if it explicitly derives or implements
Copy. All fields being Copy is not enough.

Add the `derives` attribute in the IR that lists all traits that a
mapped Rust type derives. For example rustls_slice_bytes, a struct with
Copyable fields, doesn't derive Copy. So `derives` contains only
Default.

User defined types and builtin types also have their `derives` field
filled. The entire PR is built around the following check:

```cpp
    if (auto *rec = element_type->getAsRecordDecl()) {
      if (!RecordDerivesCopy(rec)) {
        return std::format("std::array::from_fn::<_, {}, _>(|_| {})",
                           size_as_string.c_str(), element_type_as_string);
      }
    }
```

Arrays of non-copyable types cannot be initialized with
`[non_copyable_default, N]`, they have to use the `std::array:from_fn`
format.

Author: lucic71

cpp2rust/converter/converter.cpp

@@ -665,6 +665,12 @@ bool Converter::RecordDerivesDefault(const clang::RecordDecl *decl) {
 }
 
 bool Converter::RecordDerivesCopy(const clang::RecordDecl *decl) {
+  auto *derives = Mapper::MappedDerives(ctx_.getCanonicalTagType(decl));
+  return derives &&
+         std::find(derives->begin(), derives->end(), "Copy") != derives->end();
+}
+
+bool Converter::RecordHasCopyableFields(const clang::RecordDecl *decl) {
   for (auto f : decl->fields()) {
     // Records that contain std::vector, std::array, std::string or anything
     // that is translated to Vec<>, do not derive Copy
@@ -751,8 +757,11 @@ void Converter::EmitRustStructOrUnion(clang::RecordDecl *decl) {
   if (EmitsReprCForRecords()) {
     StrCat("#[repr(C)]");
   }
+  auto attrs = GetStructAttributes(decl);
+  Mapper::SetDerives(ctx_.getCanonicalTagType(decl),
+                     std::vector<std::string>(attrs.begin(), attrs.end()));
   StrCat("#[derive(");
-  for (auto *attr : GetStructAttributes(decl)) {
+  for (auto *attr : attrs) {
     StrCat(attr, ',');
   }
   StrCat(")]");
@@ -3107,6 +3116,8 @@ bool Converter::VisitEnumDecl(clang::EnumDecl *decl) {
     return false;
   }
   Mapper::AddRuleForUserDefinedType(decl);
+  Mapper::SetDerives(ctx_.getCanonicalTagType(decl),
+                     {"Clone", "Copy", "PartialEq", "Debug", "Default"});
   StrCat("#[derive(Clone, Copy, PartialEq, Debug, Default)]");
   StrCat(std::format("enum {}", GetRecordName(decl)));
   StrCat('{');
@@ -3324,6 +3335,12 @@ std::string Converter::GetArrayDefaultAsString(clang::QualType qual_type) {
     auto size_as_string = GetNumAsString(array_type->getSize());
     auto element_type = array_type->getElementType();
     auto element_type_as_string = GetDefaultAsString(element_type);
+    if (auto *rec = element_type->getAsRecordDecl()) {
+      if (!RecordDerivesCopy(rec)) {
+        return std::format("std::array::from_fn::<_, {}, _>(|_| {})",
+                           size_as_string.c_str(), element_type_as_string);
+      }
+    }
     return std::format("[{}; {}]", element_type_as_string,
                        size_as_string.c_str());
   }
@@ -3487,7 +3504,7 @@ Converter::GetStructAttributes(const clang::RecordDecl *decl) {
 
   std::vector<const char *> struct_attrs;
 
-  if (RecordDerivesCopy(decl)) {
+  if (RecordHasCopyableFields(decl)) {
     struct_attrs.emplace_back("Copy");
   }
 

cpp2rust/converter/converter.h

@@ -587,6 +587,8 @@ class Converter : public clang::RecursiveASTVisitor<Converter> {
 
   bool RecordDerivesCopy(const clang::RecordDecl *decl);
 
+  bool RecordHasCopyableFields(const clang::RecordDecl *decl);
+
   bool ShouldReplaceWithMappedBody(clang::DeclRefExpr *expr) const;
 
   std::string *rs_code_;

cpp2rust/converter/mapper.cpp

@@ -460,6 +460,12 @@ void addBuiltinTypes(Model model) {
                              const std::string &initializer = {}) {
     auto plain = TranslationRule::TypeRule::Plain(rust);
     plain.initializer = initializer;
+    std::vector<std::string> derives = {"Copy",  "Clone",     "Default",
+                                        "Debug", "PartialEq", "PartialOrd"};
+    if (!(rust == "f32" || rust == "f64")) {
+      derives.insert(derives.end(), {"Eq", "Ord", "Hash"});
+    }
+    plain.type_info.derives = std::move(derives);
     AddTypeRule(cxx, TranslationRule::TypeRule(plain));
     AddTypeRule("const " + cxx, std::move(plain));
 
@@ -696,6 +702,16 @@ bool MapsToRefcountPointer(clang::QualType qual_type) {
   return rule && rule->type_info.is_refcount_pointer;
 }
 
+const std::vector<std::string> *MappedDerives(clang::QualType qual_type) {
+  auto rule = search(qual_type).first;
+  return rule ? &rule->type_info.derives : nullptr;
+}
+
+void SetDerives(clang::QualType qual_type, std::vector<std::string> derives) {
+  if (auto *rule = search(qual_type).first) {
+    rule->type_info.derives = std::move(derives);
+  }
+}
 bool ReturnsPointer(const clang::Expr *expr) {
   auto rule = search(expr);
   return rule && rule->return_type.is_pointer();

cpp2rust/converter/mapper.h

@@ -37,6 +37,8 @@ std::string GetParamType(const clang::Expr *expr, unsigned index);
 bool ParamIsPointer(const clang::Expr *expr, unsigned index);
 bool MapsToPointer(clang::QualType qual_type);
 bool MapsToRefcountPointer(clang::QualType qual_type);
+const std::vector<std::string> *MappedDerives(clang::QualType qual_type);
+void SetDerives(clang::QualType qual_type, std::vector<std::string> derives);
 
 enum class ScalarSugar {
   kDesugar,

cpp2rust/converter/translation_rule.cpp

@@ -25,6 +25,12 @@ TypeInfo ParseTypeInfoJSON(const llvm::json::Object &obj) {
   if (auto v = obj.getBoolean("is_unsafe_pointer"))
     info.is_unsafe_pointer = *v;
   assert(!(info.is_refcount_pointer && info.is_unsafe_pointer));
+  if (auto *arr = obj.getArray("derives")) {
+    for (const auto &elem : *arr) {
+      if (auto s = elem.getAsString())
+        info.derives.emplace_back(s->str());
+    }
+  }
   return info;
 }
 
@@ -329,6 +335,8 @@ void TypeInfo::dump() const {
     log() << " [rc_ptr]";
   if (is_unsafe_pointer)
     log() << " [unsafe_ptr]";
+  for (const auto &d : derives)
+    log() << " +" << d;
 }
 
 void TypeRule::dump() const {

cpp2rust/converter/translation_rule.h

@@ -54,6 +54,7 @@ struct MethodCallFragment {
 };
 
 struct TypeInfo {
+  std::vector<std::string> derives;
   std::string type;
   bool is_refcount_pointer = false;
   bool is_unsafe_pointer = false;
@@ -83,13 +84,13 @@ struct TypeRule {
   void dump() const;
 
   static TypeRule Plain(std::string type) {
-    return {{}, {}, {std::move(type), false, false}};
+    return {{}, {}, {{}, std::move(type), false, false}};
   }
   static TypeRule RefcountPtr(std::string type) {
-    return {{}, {}, {std::move(type), true, false}};
+    return {{}, {}, {{}, std::move(type), true, false}};
   }
   static TypeRule UnsafePtr(std::string type) {
-    return {{}, {}, {std::move(type), false, true}};
+    return {{}, {}, {{}, std::move(type), false, true}};
   }
 };
 

rule-preprocessor/src/ir.rs

@@ -44,6 +44,8 @@ pub struct TypeInfo {
     pub is_refcount_pointer: bool,
     #[serde(default, skip_serializing_if = "std::ops::Not::not")]
     pub is_unsafe_pointer: bool,
+    #[serde(default, skip_serializing_if = "Vec::is_empty")]
+    pub derives: Vec<String>,
 }
 
 #[derive(Debug, Clone, Serialize, Deserialize)]

rule-preprocessor/src/main.rs

@@ -5,9 +5,11 @@
 
 extern crate rustc_driver;
 extern crate rustc_hir;
+extern crate rustc_infer;
 extern crate rustc_interface;
 extern crate rustc_middle;
 extern crate rustc_span;
+extern crate rustc_trait_selection;
 
 mod ir;
 mod semantic;

rule-preprocessor/src/semantic.rs

@@ -90,6 +90,7 @@ fn find_rlib(deps_dir: &Path, crate_name: &str) -> Option<PathBuf> {
 struct FnDecl<'tcx> {
     source_file: String,
     name: String,
+    def_id: rustc_span::def_id::DefId,
     body: &'tcx rustc_hir::Body<'tcx>,
 }
 
@@ -124,11 +125,17 @@ struct MethodResolver {
 }
 
 impl MethodResolver {
-    fn resolve_fn_decl<'tcx>(&mut self, tcx: rustc_middle::ty::TyCtxt<'tcx>, f: &FnDecl<'tcx>) {
-        if let Some(file_ir) = self.ir.all_ir.get_mut(&f.source_file)
-            && let Some(RuleIr::Fn(fn_ir)) = file_ir.get_mut(&f.name)
-        {
-            f.resolve_unknowns(tcx, fn_ir);
+    fn resolve_rule<'tcx>(&mut self, tcx: rustc_middle::ty::TyCtxt<'tcx>, f: &FnDecl<'tcx>) {
+        let Some(file_ir) = self.ir.all_ir.get_mut(&f.source_file) else {
+            return;
+        };
+        match file_ir.get_mut(&f.name) {
+            Some(RuleIr::Fn(fn_ir)) => f.resolve_unknowns(tcx, fn_ir),
+            Some(RuleIr::Type(type_ir)) => {
+                let ret_ty = tcx.fn_sig(f.def_id).skip_binder().output().skip_binder();
+                type_ir.type_info.derives = type_derives(tcx, ret_ty);
+            }
+            None => {}
         }
     }
 
@@ -154,7 +161,7 @@ impl rustc_driver::Callbacks for MethodResolver {
         tcx: rustc_middle::ty::TyCtxt<'_>,
     ) -> rustc_driver::Compilation {
         for f in iter_fn_decls(tcx) {
-            self.resolve_fn_decl(tcx, &f);
+            self.resolve_rule(tcx, &f);
         }
 
         rustc_driver::Compilation::Stop
@@ -181,6 +188,7 @@ fn iter_fn_decls<'tcx>(tcx: rustc_middle::ty::TyCtxt<'tcx>) -> Vec<FnDecl<'tcx>>
         result.push(FnDecl {
             source_file,
             name: ident.name.as_str().to_string(),
+            def_id: decl_id.owner_id.to_def_id(),
             body: tcx.hir_body(body_id),
         });
     }
@@ -205,6 +213,44 @@ fn decl_source_file(
     )
 }
 
+fn type_derives<'tcx>(
+    tcx: rustc_middle::ty::TyCtxt<'tcx>,
+    ty: rustc_middle::ty::Ty<'tcx>,
+) -> Vec<String> {
+    use rustc_infer::infer::TyCtxtInferExt;
+    use rustc_span::sym;
+    use rustc_trait_selection::infer::InferCtxtExt;
+
+    let lang = tcx.lang_items();
+    let derivable = [
+        lang.copy_trait(),
+        lang.clone_trait(),
+        tcx.get_diagnostic_item(sym::Debug),
+        tcx.get_diagnostic_item(sym::Default),
+        tcx.get_diagnostic_item(sym::PartialEq),
+        tcx.get_diagnostic_item(sym::Eq),
+        tcx.get_diagnostic_item(sym::PartialOrd),
+        tcx.get_diagnostic_item(sym::Ord),
+        tcx.get_diagnostic_item(sym::Hash),
+    ];
+
+    let infcx = tcx
+        .infer_ctxt()
+        .build(rustc_middle::ty::TypingMode::non_body_analysis());
+
+    derivable
+        .into_iter()
+        .flatten()
+        .filter(|&trait_def_id| {
+            let args = vec![ty; tcx.generics_of(trait_def_id).count()];
+            infcx
+                .type_implements_trait(trait_def_id, args, rustc_middle::ty::ParamEnv::empty())
+                .must_apply_modulo_regions()
+        })
+        .map(|trait_def_id| tcx.item_name(trait_def_id).to_string())
+        .collect()
+}
+
 struct AstVisitor<'a, 'tcx> {
     tcx: rustc_middle::ty::TyCtxt<'tcx>,
     param_names: Vec<String>,

rule-preprocessor/src/syntactic.rs

@@ -322,6 +322,7 @@ impl<'a> FnIrBuilder<'a> {
                 ty: ty_str,
                 is_refcount_pointer,
                 is_unsafe_pointer,
+                derives: Vec::new(),
             })
         }
     }
@@ -482,6 +483,7 @@ impl<'a> FnIrBuilder<'a> {
                         ty: p.ty.clone(),
                         is_refcount_pointer: p.is_refcount_pointer,
                         is_unsafe_pointer: p.is_unsafe_pointer,
+                        derives: Vec::new(),
                     },
                 )
             })
@@ -560,6 +562,7 @@ impl<'a> TypeIrBuilder<'a> {
                 ty: ty.syntax().text().to_string(),
                 is_refcount_pointer,
                 is_unsafe_pointer,
+                derives: Vec::new(),
             },
         }
     }