Add C++ encoded strings example

3 files changed, 180 insertions, 0 deletions

diff --git a/examples/CMakeLists.txt b/examples/CMakeLists.txt
index 57355026..3e1bedce 100644
--- a/examples/CMakeLists.txt
+++ b/examples/CMakeLists.txt
@@ -1,6 +1,7 @@
 # Personal+ edition (free edition doesn't get plugins due to no API access)
 add_subdirectory(background_task)
 add_subdirectory(breakpoint)
+add_subdirectory(encoded_strings)
 add_subdirectory(x86_extension)
 add_subdirectory(workflows/inliner)
 add_subdirectory(workflows/tailcall)
diff --git a/examples/encoded_strings/CMakeLists.txt b/examples/encoded_strings/CMakeLists.txt
new file mode 100644
index 00000000..11e71c45
--- /dev/null
+++ b/examples/encoded_strings/CMakeLists.txt
@@ -0,0 +1,30 @@
+cmake_minimum_required(VERSION 3.15 FATAL_ERROR)
+
+project(encoded_strings CXX C)
+
+add_library(${PROJECT_NAME} SHARED
+    src/encoded_strings.cpp)
+
+if(NOT BN_API_BUILD_EXAMPLES AND NOT BN_INTERNAL_BUILD)
+    # Out-of-tree build
+    find_path(
+        BN_API_PATH
+        NAMES binaryninjaapi.h
+        HINTS ../.. binaryninjaapi $ENV{BN_API_PATH}
+        REQUIRED
+    )
+    add_subdirectory(${BN_API_PATH} api)
+endif()
+
+target_link_libraries(${PROJECT_NAME}
+    binaryninjaapi)
+
+set_target_properties(${PROJECT_NAME} PROPERTIES
+    CXX_STANDARD 20
+    CXX_VISIBILITY_PRESET hidden
+    CXX_STANDARD_REQUIRED ON
+    VISIBILITY_INLINES_HIDDEN ON
+    POSITION_INDEPENDENT_CODE ON
+    LIBRARY_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/out/bin)
+
+bn_install_plugin(${PROJECT_NAME})
diff --git a/examples/encoded_strings/src/encoded_strings.cpp b/examples/encoded_strings/src/encoded_strings.cpp
new file mode 100644
index 00000000..62c89bef
--- /dev/null
+++ b/examples/encoded_strings/src/encoded_strings.cpp
@@ -0,0 +1,149 @@
+#define _CRT_SECURE_NO_WARNINGS
+#include <cinttypes>
+#include <cstdio>
+#include <cstring>
+#include <map>
+#include <functional>
+#include <vector>
+#include "binaryninjaapi.h"
+#include "highlevelilinstruction.h"
+
+using namespace BinaryNinja;
+using namespace std;
+
+
+static Ref<CustomStringType> g_encodedStringType;
+
+
+class EncodedStringRecognizer : public StringRecognizer
+{
+	typedef function<uint8_t(uint8_t, uint8_t)> Decoder;
+	map<string, Decoder> m_decoders;
+
+public:
+	EncodedStringRecognizer() : StringRecognizer("encoded_strings")
+	{
+		// Initialize decoders
+		m_decoders["xor_encoded"] = [](uint8_t encoded, uint8_t key) -> uint8_t {
+			return encoded ^ key;
+		};
+		m_decoders["sub_encoded"] = [](uint8_t encoded, uint8_t key) -> uint8_t {
+			return encoded - key;
+		};
+		m_decoders["add_encoded"] = [](uint8_t encoded, uint8_t key) -> uint8_t {
+			return encoded + key;
+		};
+	}
+
+	bool IsValidForType(HighLevelILFunction*, Type* type) override
+	{
+		if (!type || type->GetClass() != PointerTypeClass)
+			return false;
+
+		auto target = type->GetChildType();
+		if (!target.GetValue())
+			return false;
+
+		// Check if any decoder attribute exists
+		for (const auto& decoder : m_decoders)
+		{
+			if (target->GetAttribute(decoder.first).has_value())
+				return true;
+		}
+
+		return false;
+	}
+
+	optional<DerivedString> RecognizeConstantPointer(
+		const HighLevelILInstruction& instr, Type* type, int64_t val) override
+	{
+		if (!type || type->GetClass() != PointerTypeClass)
+			return std::nullopt;
+
+		auto target = type->GetChildType();
+		if (!target)
+			return std::nullopt;
+
+		// Find the decoder and values
+		vector<uint8_t> values;
+		Decoder chosenDecoder;
+
+		for (const auto& [name, decoder] : m_decoders)
+		{
+			if (auto attr = target->GetAttribute(name); attr.has_value())
+			{
+				// Parse hex string
+				if (attr->length() % 2 != 0)
+					return std::nullopt;
+
+				for (size_t i = 0; i < attr->length(); i += 2)
+				{
+					string byteStr = attr->substr(i, 2);
+					try
+					{
+						values.push_back((uint8_t)stoul(byteStr, nullptr, 16));
+					}
+					catch (...)
+					{
+						return std::nullopt;
+					}
+				}
+
+				chosenDecoder = decoder;
+				break;
+			}
+		}
+
+		if (values.empty() || !chosenDecoder)
+			return std::nullopt;
+
+		bool encodedNull = target->GetAttribute("encoded_null").has_value();
+
+		// Decode the string
+		vector<uint8_t> resultBytes;
+		size_t i = 0;
+		Ref<BinaryView> view = instr.function->GetFunction()->GetView();
+
+		while (true)
+		{
+			uint8_t byte;
+			if (view->Read(&byte, val + i, 1) != 1)
+				return std::nullopt;
+
+			// Check for unencoded null terminator
+			if (!encodedNull && byte == 0)
+				break;
+
+			// Decode the byte
+			byte = chosenDecoder(byte, values[i % values.size()]);
+
+			// Check for encoded null terminator
+			if (byte == 0)
+				break;
+
+			resultBytes.push_back(byte);
+			i++;
+		}
+
+		// Create the derived string
+		DerivedStringLocation loc(DataBackedStringLocation, val, i);
+		return DerivedString(string((char*)resultBytes.data(), resultBytes.size()), loc, g_encodedStringType);
+	}
+};
+
+
+extern "C"
+{
+	BN_DECLARE_CORE_ABI_VERSION
+
+	BINARYNINJAPLUGIN void CorePluginDependencies()
+	{
+	}
+
+	BINARYNINJAPLUGIN bool CorePluginInit()
+	{
+		g_encodedStringType = CustomStringType::Register("Encoded", "", "_enc");
+		StringRecognizer::Register(new EncodedStringRecognizer());
+		return true;
+	}
+}