From dd3fcad4e1b26912ae90fb320387368a51abc7ae Mon Sep 17 00:00:00 2001
From: Aleksey Dobrunov <ctapmex@ctapmex.com>
Date: Thu, 16 Jan 2025 23:04:45 +0500
Subject: [PATCH] fix read non-latin string from hrc

---
 src/colorer/strings/icu/Encodings.cpp    |  9 ++++++++-
 src/colorer/strings/icu/Encodings.h      |  1 +
 src/colorer/strings/legacy/Encodings.cpp |  6 ++++++
 src/colorer/strings/legacy/Encodings.h   |  1 +
 src/colorer/xml/libxml2/LibXmlReader.cpp | 19 ++++++++++---------
 src/colorer/xml/libxml2/LibXmlReader.h   |  2 +-
 6 files changed, 27 insertions(+), 11 deletions(-)

diff --git a/src/colorer/strings/icu/Encodings.cpp b/src/colorer/strings/icu/Encodings.cpp
index 340d5b7..b8f72c8 100644
--- a/src/colorer/strings/icu/Encodings.cpp
+++ b/src/colorer/strings/icu/Encodings.cpp
@@ -1,4 +1,5 @@
 #include "colorer/strings/icu/Encodings.h"
+#include <cstring>
 #include "colorer/Exception.h"
 #include "unicode/ucnv.h"
 #include "unicode/ustring.h"
@@ -24,7 +25,13 @@ uUnicodeString Encodings::toUnicodeString(char* data, int32_t len)
 
 uUnicodeString Encodings::fromUTF8(char* data, int32_t len)
 {
-  return std::make_unique<UnicodeString>(data , len , ENC_UTF8);
+  return std::make_unique<UnicodeString>(data, len, ENC_UTF8);
+}
+
+uUnicodeString Encodings::fromUTF8(unsigned char* data)
+{
+  const auto c = reinterpret_cast<char*>(data);
+  return fromUTF8(c,strlen(c));
 }
 
 int Encodings::toUTF8Bytes(UChar wc, byte* dest)
diff --git a/src/colorer/strings/icu/Encodings.h b/src/colorer/strings/icu/Encodings.h
index b79abc4..3f8590d 100644
--- a/src/colorer/strings/icu/Encodings.h
+++ b/src/colorer/strings/icu/Encodings.h
@@ -10,6 +10,7 @@ class Encodings
 
   static uUnicodeString toUnicodeString(char* data, int32_t len);
   static uUnicodeString fromUTF8(char* data, int32_t len);
+  static uUnicodeString fromUTF8(unsigned char* data);
   static int toUTF8Bytes(UChar, byte*);
 };
 
diff --git a/src/colorer/strings/legacy/Encodings.cpp b/src/colorer/strings/legacy/Encodings.cpp
index 23d2bc1..a9a0046 100644
--- a/src/colorer/strings/legacy/Encodings.cpp
+++ b/src/colorer/strings/legacy/Encodings.cpp
@@ -177,4 +177,10 @@ uUnicodeString Encodings::fromUTF8(char* data, int32_t len)
   return std::make_unique<UnicodeString>(data, len, Encodings::ENC_UTF8);
 }
 
+uUnicodeString Encodings::fromUTF8(unsigned char* data)
+{
+  const auto c = reinterpret_cast<char*>(data);
+  return fromUTF8(c,strlen(c));
+}
+
 
diff --git a/src/colorer/strings/legacy/Encodings.h b/src/colorer/strings/legacy/Encodings.h
index ca9c0d9..50cd3cb 100644
--- a/src/colorer/strings/legacy/Encodings.h
+++ b/src/colorer/strings/legacy/Encodings.h
@@ -46,6 +46,7 @@ class Encodings
 
   static uUnicodeString toUnicodeString(char* data, int32_t len);
   static uUnicodeString fromUTF8(char* data, int32_t len);
+  static uUnicodeString fromUTF8(unsigned char* data);
 };
 
 #endif
diff --git a/src/colorer/xml/libxml2/LibXmlReader.cpp b/src/colorer/xml/libxml2/LibXmlReader.cpp
index ee53b96..62fa729 100644
--- a/src/colorer/xml/libxml2/LibXmlReader.cpp
+++ b/src/colorer/xml/libxml2/LibXmlReader.cpp
@@ -56,8 +56,8 @@ bool LibXmlReader::populateNode(xmlNode* node, XMLNode& result)
     result.name = UnicodeString(reinterpret_cast<const char*>(node->name));
 
     const auto text_string = getElementText(node);
-    if (!text_string.isEmpty()) {
-      result.text = text_string;
+    if (text_string && !text_string->isEmpty()) {
+      result.text = UnicodeString(*text_string.get());
     }
     getChildren(node, result);
     getAttributes(node, result.attributes);
@@ -67,22 +67,22 @@ bool LibXmlReader::populateNode(xmlNode* node, XMLNode& result)
   return false;
 }
 
-UnicodeString LibXmlReader::getElementText(const xmlNode* node)
+uUnicodeString LibXmlReader::getElementText(const xmlNode* node)
 {
   for (const xmlNode* child = node->children; child != nullptr; child = child->next) {
     if (child->type == XML_CDATA_SECTION_NODE) {
-      return {reinterpret_cast<const char*>(child->content)};
+      return Encodings::fromUTF8(child->content);
     }
     if (child->type == XML_TEXT_NODE) {
-      auto temp_string = UnicodeString(reinterpret_cast<const char*>(child->content));
-      temp_string.trim();
-      if (temp_string.isEmpty()) {
+      auto temp_string = Encodings::fromUTF8(child->content);
+      temp_string->trim();
+      if (temp_string->isEmpty()) {
         continue;
       }
       return temp_string;
     }
   }
-  return {u""};
+  return nullptr;
 }
 
 void LibXmlReader::getChildren(xmlNode* node, XMLNode& result)
@@ -107,7 +107,8 @@ void LibXmlReader::getAttributes(const xmlNode* node, std::unordered_map<Unicode
 {
   for (xmlAttrPtr attr = node->properties; attr != nullptr; attr = attr->next) {
     const auto content = xmlNodeGetContent(attr->children);
-    data.try_emplace(reinterpret_cast<const char*>(attr->name), reinterpret_cast<const char*>(content));
+    auto decoded_string = Encodings::fromUTF8(content);
+    data.try_emplace(reinterpret_cast<const char*>(attr->name), *decoded_string.get());
     xmlFree(content);
   }
 }
diff --git a/src/colorer/xml/libxml2/LibXmlReader.h b/src/colorer/xml/libxml2/LibXmlReader.h
index 21a83c8..160f1cc 100644
--- a/src/colorer/xml/libxml2/LibXmlReader.h
+++ b/src/colorer/xml/libxml2/LibXmlReader.h
@@ -30,7 +30,7 @@ class LibXmlReader
   static void getAttributes(const xmlNode* node, std::unordered_map<UnicodeString, UnicodeString>& data);
   void getChildren(xmlNode* node, XMLNode& result);
   bool populateNode(xmlNode* node, XMLNode& result);
-  static UnicodeString getElementText(const xmlNode* node);
+  static uUnicodeString getElementText(const xmlNode* node);
 
   /* the name of the file that is being processed */
   static uUnicodeString current_file;