Merge pull request #444 from kohnakagawa/feature/parse_resource_html

Fix to parse html resource

api/python/PE/objects/pyResourcesManager.cpp

@@ -113,6 +113,14 @@ void create<ResourcesManager>(py::module& m) {
       &ResourcesManager::string_table,
       "Return list of " RST_CLASS_REF(lief.PE.ResourceStringTable) " present in the resource")
 
+    .def_property_readonly("has_html",
+      &ResourcesManager::has_html,
+      "``True`` if resources contain HTML resource")
+
+    .def_property_readonly("html",
+      &ResourcesManager::html,
+      "HTML resource as the list of ``string``")
+
     .def("get_node_type",
       static_cast<no_const_func<ResourceNode&, RESOURCE_TYPES>>(&ResourcesManager::get_node_type),
       "Return " RST_CLASS_REF(lief.PE.ResourceNode) " with "

include/LIEF/PE/ResourcesManager.hpp

@@ -127,6 +127,15 @@ class LIEF_API ResourcesManager : public Object {
   //! @brief Return the list of the string table in the resource
   std::vector<ResourceStringTable> string_table(void) const;
 
+  // HTML
+  // ====
+
+  //! @brief ``true`` if resources contain html
+  bool has_html(void) const;
+
+  //! @brief Return the list of the html resource
+  std::vector<std::string> html(void) const;
+
   // Print
   // =====
 

src/PE/ResourcesManager.cpp

@@ -353,7 +353,6 @@ std::string ResourcesManager::manifest(void) const {
   }
 
   it_childs childs_l2 = childs_l1[0].childs();
-
   if (childs_l2.size() < 1) {
     throw not_found("Manifest corrupted");
   }
@@ -1248,6 +1247,46 @@ bool ResourcesManager::has_string_table(void) const {
   return this->has_type(RESOURCE_TYPES::STRING);
 }
 
+std::vector<std::string> ResourcesManager::html(void) const {
+  it_childs nodes = this->resources_->childs();
+  auto&& it_html = std::find_if(
+    std::begin(nodes),
+    std::end(nodes),
+    [] (const ResourceNode& node) {
+      return static_cast<RESOURCE_TYPES>(node.id()) == RESOURCE_TYPES::HTML;
+    }
+  );
+
+  if (it_html == std::end(nodes)) {
+    throw not_found(std::string("Missing '") + to_string(RESOURCE_TYPES::HTML) + "' entry");
+  }
+
+  std::vector<std::string> html;
+  for (const ResourceNode& child_l1 : it_html->childs()) {
+    for (const ResourceNode& child_l2 : child_l1.childs()) {
+      const ResourceData* html_node = dynamic_cast<const ResourceData*>(&child_l2);
+      if (!html_node) {
+        LOG(ERROR) << "html node is null";
+        continue;
+      }
+
+      const std::vector<uint8_t>& content = html_node->content();
+      if (content.empty()) {
+        LOG(ERROR) << "html content is empty";
+        continue;
+      }
+
+      html.push_back(std::string{std::begin(content), std::end(content)});
+    }
+  }
+
+  return html;
+}
+
+bool ResourcesManager::has_html(void) const {
+  return this->has_type(RESOURCE_TYPES::HTML);
+}
+
 // Prints
 // ======
 

src/PE/json.cpp

@@ -507,6 +507,14 @@ void JsonVisitor::visit(const ResourcesManager& resources_manager) {
     }
   }
 
+  if (resources_manager.has_html()) {
+    try {
+      this->node_["html"] = resources_manager.html();
+    } catch (const LIEF::exception& e) {
+      LOG(WARNING) << e.what();
+    }
+  }
+
   if (resources_manager.has_version()) {
     JsonVisitor version_visitor;
     try {