1 files changed, 87 insertions, 0 deletions
diff --git a/src/md/compressedinteger.h b/src/md/compressedinteger.h
new file mode 100644
index 0000000000..29a4f83dba
--- /dev/null
+++ b/src/md/compressedinteger.h
@@ -0,0 +1,87 @@
+// Licensed to the .NET Foundation under one or more agreements.
+// The .NET Foundation licenses this file to you under the MIT license.
+// See the LICENSE file in the project root for more information.
+// 
+// File: CompressedInteger.h
+// 
+
+// 
+// Class code:MetaData::CompressedInteger provides secure access to a compressed integer (as defined in CLI 
+// ECMA specification). The integer is compressed into 1, 2 or 4 bytes. See code:CompressedInteger#Format 
+// for full format description.
+// 
+// ======================================================================================
+
+#pragma once
+
+#include "external.h"
+
+namespace MetaData
+{
+
+// --------------------------------------------------------------------------------------
+// 
+// This class provides secure access to a compressed integer (as defined in CLI ECMA specification). The 
+// integer is compressed into 1, 2 or 4 bytes. See code:CompressedInteger#Format for full format description.
+// 
+class CompressedInteger
+{
+// #Format
+// 
+// The format/encoding of compressed integer is (defined in ECMA CLI specification):
+//  The encoding is 1, 2 or 4 bytes long and depends on the first byte value. If the first byte is (binary):
+//    * 0xxx xxxx ... then it's 1 byte long and the value is 0xxx xxxx.
+//    * 10xx xxxx ... then it's 2 bytes long and the value is 00xx xxxx yyyy yyyy, where yyyy yyyy is the 
+//                    second byte. Though values smaller than code:const_Max1Byte are technically invalid 
+//                    when encoded with 2 bytes.
+//    * 110x xxxx ... then it's 4 bytes long and the value is 000x xxxx yyyy yyyy zzzz zzzz wwww wwww, where 
+//                    yyyy yyyy is the 2nd byte, zzzz zzzz is the 3rd byte and wwww wwww is the 4th byte.
+//                    Though values smaller than code:const_Max2Bytes are technically invalid when encoded 
+//                    with 4 bytes.
+//    * 111x xxxx ... then it's invalid encoding.
+// 
+// Note: Some encodings are invalid, but CLR accepts them (see code:DataBlob::GetCompressedU), 
+//  e.g. 1000 0000 0000 0000 (0x8000) encodes 0 while correct/valid encoding is 0000 0000 (0x00).
+// 
+private:
+    // This class has only static methods and shouldn't be instantiated.
+    CompressedInteger() {}
+    
+public:
+    static const UINT32 const_MaxEncodingSize = 4;
+    
+    static const UINT32 const_Max1Byte  = 0x7f;
+    static const UINT32 const_Max2Bytes = 0x3fff;
+    static const UINT32 const_Max4Bytes = 0x1fffffff;
+    
+    static const UINT32 const_Max = const_Max4Bytes;
+    
+public:
+    // 
+    // Operations
+    // 
+    
+    // Returns TRUE if the value (nValue) fits into 1-byte, 2-bytes or 4-bytes encoding and fills 
+    // *pcbEncodingSize with 1, 2 or 4.
+    // Returns FALSE if the value cannot be encoded as compressed integer, doesn't fill *pcbEncodingSize 
+    // then.
+    __checkReturn 
+    __success(return)
+    static inline BOOL GetEncodingSize(
+              UINT32  nValue, 
+        __out UINT32 *pcbEncodingSize);
+    // Returns TRUE if the value (nValue) fits into 1-byte, 2-bytes or 4-bytes encoding and fills 
+    // *pcbEncodingSize with 1, 2 or 4 and *pnEncodedValue with the encoded value.
+    // Returns FALSE if the value cannot be encoded as compressed integer, doesn't fill *pcbEncodingSize 
+    // nor *pnEncodedValue then.
+    __success(return)
+    static inline BOOL Encode(
+              UINT32  nValue, 
+        __out UINT32 *pnEncodedValue, 
+        __out UINT32 *pcbEncodingSize);
+    
+};  // class CompressedInteger
+
+};  // namespace MetaData
+
+#include "compressedinteger.inl"