From 6835c339d76db84bce6cbf77db8a8ab101e568d7 Mon Sep 17 00:00:00 2001
From: Akash Mozumdar <akashmozumdar@gmail.com>
Date: Fri, 3 Aug 2018 01:48:57 -0400
Subject: [PATCH] implement utf8

---
 GUI/mainwindow.cpp             |  2 +-
 GUI/misc.cpp                   |  7 ++++++-
 texthook/textthread.cc         |  6 ++++++
 vnrhook/include/const.h        |  4 +---
 vnrhook/src/engine/engine.cc   | 13 ++++++++++++-
 vnrhook/src/hijack/texthook.cc | 15 ---------------
 6 files changed, 26 insertions(+), 21 deletions(-)

diff --git a/GUI/mainwindow.cpp b/GUI/mainwindow.cpp
index 8bb472a..a2c6b38 100644
--- a/GUI/mainwindow.cpp
+++ b/GUI/mainwindow.cpp
@@ -180,7 +180,7 @@ void MainWindow::on_hookButton_clicked()
 {
 	bool ok;
 	QString hookCode = QInputDialog::getText(this, "Add Hook",
-		"Enter hook code\r\n/H{A|B|W|S|Q}[N]data_offset[*drdo][:sub_offset[*drso]]@addr[:module]",
+		"Enter hook code\r\n/H{A|B|W|S|Q|V}[N]data_offset[*drdo][:sub_offset[*drso]]@addr[:module]",
 		QLineEdit::Normal, "/H", &ok
 	);
 	if (!ok) return;
diff --git a/GUI/misc.cpp b/GUI/misc.cpp
index 01580ff..fc4336f 100644
--- a/GUI/misc.cpp
+++ b/GUI/misc.cpp
@@ -65,6 +65,9 @@ HookParam ParseHCode(QString HCode)
 		hp.type |= USING_UNICODE;
 		hp.length_offset = 1;
 		break;
+	case L'V':
+		hp.type |= USING_STRING | USING_UTF8;
+		break;
 	default:
 		return {};
 	}
@@ -129,7 +132,9 @@ QString GenerateHCode(HookParam hp, DWORD processId)
 	}
 	else
 	{
-		if (hp.type & USING_STRING)
+		if (hp.type & USING_UTF8)
+			code += "V";
+		else if (hp.type & USING_STRING)
 			code += "S";
 		else if (hp.type & BIG_ENDIAN)
 			code += "A";
diff --git a/texthook/textthread.cc b/texthook/textthread.cc
index 4da5160..2c00329 100644
--- a/texthook/textthread.cc
+++ b/texthook/textthread.cc
@@ -54,6 +54,12 @@ void TextThread::AddSentence()
 	{
 		sentence = std::wstring((wchar_t*)sentenceBuffer.data(), sentenceBuffer.size() / 2);
 	}
+	else if (status & USING_UTF8)
+	{
+		wchar_t* converted = new wchar_t[sentenceBuffer.size()];
+		sentence = std::wstring(converted, MultiByteToWideChar(CP_UTF8, 0, sentenceBuffer.data(), sentenceBuffer.size(), converted, sentenceBuffer.size()));
+		delete[] converted;
+	}
 	else
 	{
 		wchar_t* converted = new wchar_t[sentenceBuffer.size()];
diff --git a/vnrhook/include/const.h b/vnrhook/include/const.h
index 1edabd5..0966e3c 100644
--- a/vnrhook/include/const.h
+++ b/vnrhook/include/const.h
@@ -91,7 +91,6 @@ enum HostNotificationType {
 // - 0x100
 enum HookParamType : unsigned long {
   USING_STRING      = 0x1     // type(data) is char* or wchar_t* and has length
-  , USING_UTF8 = USING_STRING // jichi 10/21/2014: temporarily handled the same way as USING_STRING
   , USING_UNICODE     = 0x2     // type(data) is wchar_t or wchar_t*
   , BIG_ENDIAN        = 0x4     // type(data) is char
   , DATA_INDIRECT     = 0x8
@@ -99,8 +98,7 @@ enum HookParamType : unsigned long {
   , SPLIT_INDIRECT    = 0x20
   , MODULE_OFFSET     = 0x40    // do hash module, and the address is relative to module
   , FUNCTION_OFFSET   = 0x80    // do hash function, and the address is relative to funccion
-  , PRINT_DWORD       = 0x100 // jichi 12/7/2014: Removed
-  , NO_ASCII          = 0x100   // jichi 1l/22/2015: Skip ascii characters
+  , USING_UTF8        = 0x100
   , STRING_LAST_CHAR  = 0x200
   , NO_CONTEXT        = 0x400
   , HOOK_EMPTY        = 0x800
diff --git a/vnrhook/src/engine/engine.cc b/vnrhook/src/engine/engine.cc
index efed52f..65e7da9 100644
--- a/vnrhook/src/engine/engine.cc
+++ b/vnrhook/src/engine/engine.cc
@@ -16464,6 +16464,16 @@ static void SpecialHookMonoString(DWORD esp_base, HookParam *hp, BYTE, DWORD *da
   }
 }
 
+bool NoAsciiFilter(LPVOID data, DWORD *size, HookParam *, BYTE)
+{
+	auto text = reinterpret_cast<LPBYTE>(data);
+	if (text)
+		for (size_t i = 0; i < *size; i++)
+			if (text[i] > 127)
+				return true;
+	return false;
+}
+
 bool InsertMonoHooks()
 {
   HMODULE h = ::GetModuleHandleA("mono.dll");
@@ -16481,7 +16491,8 @@ bool InsertMonoHooks()
     const auto &it = funcs[i];
     if (FARPROC addr = ::GetProcAddress(h, it.functionName)) {
       hp.address = (DWORD)addr;
-      hp.type = it.hookType|NO_ASCII; // 11/27/2015: Disable ascii string
+	  hp.type = it.hookType;
+	  hp.filter_fun = NoAsciiFilter;
       hp.offset = it.textIndex * 4;
       hp.length_offset = it.lengthIndex * 4;
       hp.text_fun = (HookParam::text_fun_t)it.text_fun;
diff --git a/vnrhook/src/hijack/texthook.cc b/vnrhook/src/hijack/texthook.cc
index 6dd86a7..16ff21e 100644
--- a/vnrhook/src/hijack/texthook.cc
+++ b/vnrhook/src/hijack/texthook.cc
@@ -292,17 +292,6 @@ int ProcessHook(DWORD dwDataBase, DWORD dwRetn, TextHook *hook) // Use SEH to en
 }
 #endif // 1
 
-// Return false if all text are ascii
-bool NoAsciiFilter(LPVOID data, DWORD *size, HookParam *, BYTE)
-{
-  auto text = reinterpret_cast<LPBYTE>(data);
-  if (text)
-    for (size_t i = 0; i < *size; i++)
-      if (text[i] > 127)
-        return true;
-  return false;
-}
-
 } // unnamed namespace
 
 // - TextHook methods -
@@ -339,11 +328,7 @@ DWORD TextHook::UnsafeSend(DWORD dwDataBase, DWORD dwRetn)
   BYTE *pbData,
        pbSmallBuff[SMALL_BUFF_SIZE];
   DWORD dwType = hp.type;
-  //if ((dwType & NO_CONTEXT) == 0 && HookFilter(dwRetn))
-  //  return 0;
 
-  if ((dwType & NO_ASCII) && !hp.filter_fun)
-    hp.filter_fun = NoAsciiFilter;
 
   // jichi 10/24/2014: Skip GDI functions 
   // Artikash 6/3/2018: ^ why??