Handle python utf-16 conversion in LangCodeItem values

api/python/PE/objects/resources/pyLangCodeItem.cpp

@@ -74,8 +74,14 @@ void create<LangCodeItem>(py::module& m) {
         "" RST_CLASS_REF(lief.PE.CODE_PAGES) " for which :attr:`~lief.PE.LangCodeItem.items` are defined")
 
     .def_property("items",
-        static_cast<std::map<std::u16string, std::u16string>& (LangCodeItem::*)(void)>(&LangCodeItem::items),
-        static_cast<setter_t<const std::map<std::u16string, std::u16string>&>>(&LangCodeItem::items))
+        [] (const LangCodeItem& item) -> py::dict {
+          py::dict output;
+          for (const auto& p : item.items()) {
+            output[u16tou8(p.first).c_str()] = py::bytes(u16tou8(p.second));
+          }
+          return output;
+        },
+        static_cast<setter_t<const LangCodeItem::items_t&>>(&LangCodeItem::items))
 
 
     .def("__eq__", &LangCodeItem::operator==)

doc/sphinx/changelog.rst

@@ -18,6 +18,8 @@ Changelog
   * Bug fix in the symbols table reconstruction (ELF)
 
 :PE:
+  * :attr:`~lief.PE.LangCodeItem.items` now returns a dictionary whose values are **bytes** (instead of
+    ``str`` object). This change is related to ``utf-16`` support.
   * :github_user:`kohnakagawa` fixed wrong enums values: :commit:`c03125045e32a9cd65c613585eb4d0385350c6d2`, :commit:`6ee808a1e4611d09c6cf0aea82a612be69584db9`, :commit:`cd05f34bae681fc8af4b5e7cc28eaef816802b6f`
   * :github_user:`kohnakagawa` fixed a bug in the PE resources parser (:commit:`a7254d1ba935783f16effbc7faddf993c57e82f7`)
   * Handle PE forwarded exports (`issues/307 <https://github.com/lief-project/LIEF/issues/307>`_)

include/LIEF/PE/resources/LangCodeItem.hpp

@@ -18,7 +18,7 @@
 #include <iostream>
 #include <sstream>
 #include <vector>
-#include <map>
+#include <unordered_map>
 
 #include "LIEF/visibility.h"
 
@@ -41,6 +41,7 @@ class LIEF_API LangCodeItem : public Object {
   friend class ResourcesManager;
 
   public:
+  using items_t = std::unordered_map<std::u16string, std::u16string>;
   LangCodeItem(void);
 
   LangCodeItem(const LangCodeItem&);
@@ -69,8 +70,8 @@ class LIEF_API LangCodeItem : public Object {
   //! @brief Sublang for which @link LangCodeItem::items items @endlink are defined
   RESOURCE_SUBLANGS sublang(void) const;
 
-  const std::map<std::u16string, std::u16string>& items(void) const;
-  std::map<std::u16string, std::u16string>&       items(void);
+  const items_t& items(void) const;
+  items_t&       items(void);
 
   void type(uint16_t type);
   void key(const std::u16string& key);
@@ -80,7 +81,7 @@ class LIEF_API LangCodeItem : public Object {
   void lang(RESOURCE_LANGS lang);
   void sublang(RESOURCE_SUBLANGS lang);
 
-  void items(const std::map<std::u16string, std::u16string>& items);
+  void items(const items_t& items);
 
 
   virtual void accept(Visitor& visitor) const override;
@@ -93,8 +94,7 @@ class LIEF_API LangCodeItem : public Object {
   private:
   uint16_t       type_;
   std::u16string key_;
-  std::map<std::u16string, std::u16string> items_;
-
+  items_t        items_;
 };
 
 

include/LIEF/PE/resources/ResourceStringFileInfo.hpp

@@ -32,7 +32,7 @@ namespace PE {
 class ResourcesManager;
 class ResourceVersion;
 
-//! @brief Modelization of the ``StringFileInfo`` structure
+//! @brief Modelization of the ``StringFileInfo`` structure
 //!
 //! It contains version information that can be displayed for a particular language and code page.
 //!

src/PE/resources/LangCodeItem.cpp

@@ -78,12 +78,12 @@ RESOURCE_SUBLANGS LangCodeItem::sublang(void) const {
 }
 
 
-const std::map<std::u16string, std::u16string>& LangCodeItem::items(void) const {
+const LangCodeItem::items_t& LangCodeItem::items(void) const {
   return this->items_;
 }
 
-std::map<std::u16string, std::u16string>& LangCodeItem::items(void) {
-  return const_cast<std::map<std::u16string, std::u16string>&>(static_cast<const LangCodeItem*>(this)->items());
+LangCodeItem::items_t& LangCodeItem::items(void) {
+  return const_cast<LangCodeItem::items_t&>(static_cast<const LangCodeItem*>(this)->items());
 }
 
 
@@ -145,7 +145,7 @@ void LangCodeItem::sublang(RESOURCE_SUBLANGS lang) {
 }
 
 
-void LangCodeItem::items(const std::map<std::u16string, std::u16string>& items) {
+void LangCodeItem::items(const LangCodeItem::items_t& items) {
   this->items_ = items;
 }
 
@@ -176,7 +176,7 @@ std::ostream& operator<<(std::ostream& os, const LangCodeItem& item) {
      << " - "
      << std::hex << to_string(item.code_page()) << ")" << std::endl;
   os << std::setw(8) << std::setfill(' ') << "Items: " << std::endl;
-  for (const std::pair<const std::u16string, std::u16string>& p : item.items()) {
+  for (const LangCodeItem::items_t::value_type& p : item.items()) {
     os << "    " << "'" << u16tou8(p.first) << "': '" << u16tou8(p.second) << "'" << std::endl;
   }
   return os;

tests/pe/test_resources.py

@@ -157,28 +157,28 @@ class TestResource(TestCase):
         if sys.version_info < (3,):
             self.assertEqual(items['CompanyName'].encode("utf8"), 'TODO: <Nom de la société>')
         else:
-            self.assertEqual(items['CompanyName'], 'TODO: <Nom de la société>')
+            self.assertEqual(items['CompanyName'], b'TODO: <Nom de la soci\xc3\xa9t\xc3\xa9>')
 
         self.assertIn('FileVersion', items)
-        self.assertEqual(items['FileVersion'], '1.0.0.1')
+        self.assertEqual(items['FileVersion'], b'1.0.0.1')
 
         self.assertIn('InternalName', items)
-        self.assertEqual(items['InternalName'], 'Hello.exe')
+        self.assertEqual(items['InternalName'], b'Hello.exe')
 
         self.assertIn('LegalCopyright', items)
         if sys.version_info < (3,):
             self.assertEqual(items['LegalCopyright'].encode("utf8"), 'TODO: (c) <Nom de la société>.  Tous droits réservés.')
         else:
-            self.assertEqual(items['LegalCopyright'], 'TODO: (c) <Nom de la société>.  Tous droits réservés.')
+            self.assertEqual(items['LegalCopyright'].decode("utf8"), 'TODO: (c) <Nom de la société>.  Tous droits réservés.')
 
         self.assertIn('OriginalFilename', items)
-        self.assertEqual(items['OriginalFilename'], 'Hello.exe')
+        self.assertEqual(items['OriginalFilename'], b'Hello.exe')
 
         self.assertIn('ProductName', items)
-        self.assertEqual(items['ProductName'], 'TODO: <Nom du produit>')
+        self.assertEqual(items['ProductName'], b'TODO: <Nom du produit>')
 
         self.assertIn('ProductVersion', items)
-        self.assertEqual(items['ProductVersion'], '1.0.0.1')
+        self.assertEqual(items['ProductVersion'], b'1.0.0.1')
 
         # Check ResourceVarFileInfo
         self.assertEqual(var_file_info.type, 1)