Option for computing the hash of the output file while checking it.

Author: JeromeMartinez

Source/CLI/Global.cpp

@@ -34,6 +34,13 @@ int global::SetOutputFileName(const char* FileName)
     return 0;
 }
 
+//---------------------------------------------------------------------------
+int global::SetComputeOutputHash()
+{
+    Actions.set(Action_ComputeOutputHash, true);
+    return 0;
+}
+
 //---------------------------------------------------------------------------
 int global::SetBinName(const char* FileName)
 {
@@ -304,6 +311,8 @@ int global::SetAll(bool Value)
         return ReturnValue;
     if (int ReturnValue = SetHash(Value))
         return ReturnValue;
+    if (int ReturnValue = SetComputeOutputHash())
+        return ReturnValue;
     return 0;
 }
 
@@ -726,6 +735,12 @@ int global::ManageCommandLine(const char* argv[], int argc)
             if (Value)
                 return Value;
         }
+        else if (strcmp(argv[i], "--compute-output-hash") == 0)
+        {
+            int Value = SetComputeOutputHash();
+            if (Value)
+                return Value;
+        }
         else if (strcmp(argv[i], "--quick-check") == 0)
         {
             int Value = SetQuickCheck();

Source/CLI/Global.h

@@ -85,6 +85,7 @@ class global
     int SetVersion(const char* Value);
     int SetOption(const char* argv[], int& i, int argc);
     int SetOutputFileName(const char* FileName);
+    int SetComputeOutputHash();
     int SetBinName(const char* FileName);
     int SetLicenseKey(const char* Key, bool Add);
     int SetSubLicenseId(uint64_t Id);

Source/CLI/Help.cpp

@@ -109,6 +109,9 @@ ReturnValue Help(const char* Name)
         "              ${Input}.mkv if the input is a folder, or ${Input}.RAWcooked if\n"
         "              input is a file, such as a DPX.\n"
         "\n"
+        "       --compute-output-hash\n"
+        "              Compute the hash of the output file while checking it.\n"
+        "\n"
         "       --output-version value\n"
         "              Set the version of the output..\n"
         "              The default output value is 1.\n"

Source/CLI/rawcooked.1

@@ -67,6 +67,9 @@ Set the name of the output file or folder to \fIvalue\fR.
 .br
 The default output value is opposite to the input. Expect \fI${Input}.mkv\fR if the input is a folder, or \fI${Input}.RAWcooked\fR if input is a file, such as a DPX.
 .TP
+.B --compute-output-hash
+Compute the hash of the output file while checking it.
+.TP
 .B --output-version \fIvalue\fR
 Set the version of the output.
 .br

Source/Lib/Compressed/Matroska/Matroska.cpp

@@ -26,6 +26,10 @@
 #include <iomanip>
 #include <iostream>
 #include <thread>
+extern "C"
+{
+#include "md5.h"
+}
 //---------------------------------------------------------------------------
 
 //---------------------------------------------------------------------------
@@ -307,6 +311,11 @@ void matroska::ParseBuffer()
 
     Buffer_Offset = 0;
     Level = 0;
+    MD5_CTX MD5;
+    uint64_t MD5_Offset = 0;
+    auto ComputeOutputHash = (Actions[Action_Decode] || Actions[Action_Check]) && Actions[Action_ComputeOutputHash];
+    if (ComputeOutputHash)
+        MD5_Init(&MD5);
 
     // Progress indicator
     Cluster_Timestamp = 0;
@@ -356,8 +365,18 @@ void matroska::ParseBuffer()
             }
         }
 
+        auto MD5_Size = Buffer_Offset - MD5_Offset;
+        auto NeedRemap = Buffer_Offset - Buffer_Offset_LowerLimit >= 0x100000; // TODO: when multi-threaded frame decoding is implemented, we need to check that all thread don't need anymore memory below this value 
+        
+        // Compute MD5 of the whole file
+        if (ComputeOutputHash && (NeedRemap || MD5_Size >= 0x100000))
+        {
+            MD5_Update(&MD5, Buffer.Data() + MD5_Offset, MD5_Size);
+            MD5_Offset += MD5_Size;
+        }
+
         // Check if we can indicate the system that we'll not need anymore memory below this value, without indicating it too much
-        if (Buffer_Offset > Buffer_Offset_LowerLimit + 1024 * 1024 && Buffer_Offset < Buffer.Size()) // TODO: when multi-threaded frame decoding is implemented, we need to check that all thread don't need anymore memory below this value 
+        if (NeedRemap)
         {
             FileMap->Remap(Buffer_Offset, Buffer_Offset + 256 * 1024 * 1024);
             Buffer = *FileMap;
@@ -391,6 +410,12 @@ void matroska::ParseBuffer()
     }
 
     // Clean up
+    md5 MD5_Result;
+    if (ComputeOutputHash)
+    {
+        MD5_Update(&MD5, Buffer.Data() + MD5_Offset, Buffer_Offset - MD5_Offset);
+        MD5_Final(MD5_Result.data(), &MD5);
+    }
     if (RAWcooked_LibraryName.empty())
     {
         if (Hashes_FromRAWcooked)
@@ -413,6 +438,15 @@ void matroska::ParseBuffer()
         ProgressIndicator_Thread->join();
         delete ProgressIndicator_Thread;
     }
+
+    // Show MD5
+    if (ComputeOutputHash)
+    {
+        std::cout << "\nInfo: Output file MD5 is ";
+        for (size_t i = 0; i < MD5_Result.size(); i++)
+            std::cout << std::hex << std::setw(2) << std::setfill('0') << (int)MD5_Result[i];
+        std::cout << '.' << std::endl;
+    }
 }
 
 //---------------------------------------------------------------------------

Source/Lib/Utils/FileIO/Input_Base.h

@@ -40,6 +40,7 @@ enum action : uint8_t
     Action_Info,
     Action_FrameMd5,
     Action_FrameMd5An,
+    Action_ComputeOutputHash,
     Action_QuickCheckAfterEncode, // Internal, indicating the 2nd pass
     Action_Max
 };