From C# to Go: Achieving AES and Base64 Encoding Compatibility

A couple of weeks ago I faced an interesting problem: I had to migrate an AES encryption algorithm from C# to Go. In the Go implementation we already had an AES encryption algorithm, but it was not compatible with the C# implementation, and several tests failed because the results did not match, mainly in the last character.

The problem was that I did not have the source code of the C# implementation, only the binary, a DLL that was used in the .NET project.

I tried to get the source code of the C# implementation, but was unsuccessful. Being an old project, there was no documentation available. Fortunately, my boss was the one who developed this implementation, but I didn't remember the exact details. However, I did know that at the end of the AES encryption process a base64 encoding function was used.

With this clue, I opened the project in .NET and installed the JetBrains extension to decompile the source code, and obtained the library code that was used to encrypt the information.

Finally, I discovered that the problem was not with the AES encryption algorithm, but with the base64 encoding.

In the C# code, at the end of the AES encryption process, the following function was used for base64 encoding: HttpServerUtility.UrlTokenEncode.

The UrlTokenEncode function is a .NET function that encodes a byte array into a base64 text string for transmission in a URL. This function performs three key actions that explain the difference in results:

1. URL-safe Base64 encoding: uses a variant of Base64 that is suitable for URLs, replacing the characters with - and / with _.

2. Padding Character Removal: The function removes padding = characters that are typically used in standard Base64 encoding.

3. Adding a number to the end: The function adds a number to the end of the string to indicate how many padding characters were removed.

I discovered all this thanks to ChatGPT, not because I am an expert in base64. With this information, I was able to modify the implementation in Go to be compatible with that of C#.

In Go, after encrypting the information with AES, base64 encoding is done as follows:

encode := base64.RawURLEncoding.EncodeToString(paddedBytes)

And finally, the number of removed padding characters is added to the end of the string:

// Calcular el número de caracteres de relleno (`=`) que se habrían añadido
paddingCount := (4 - len(paddedBytes)%3) % 4

// Añadir el conteo de relleno al final de la cadena codificada (como hace UrlTokenEncode de C#)
if paddingCount > 0 {
    encoded += strconv.Itoa(paddingCount)
}

In the line paddingCount := (4 - len(paddedBytes)%3) % 4, the number of padding characters (=) is calculated and then added to the end of the base64-encoded string:

• len(paddedBytes)%3 calculates the number of bytes that have not been base64 encoded.
• (4 - len(paddedBytes)%3) % 4 calculates how many padding characters are missing so that the length is divisible by 4. If no padding is needed, the result is 0.

In short, the problem was not the AES encryption algorithm, but the base64 encoding. Thanks to the information obtained from ChatGPT, I was able to modify the implementation in Go to be compatible with that of C#. In this case, using ChatGPT was very helpful as it saved me a lot of time and headaches; Of course, I had to adjust each answer until the results of both implementations were equal.

The above is the detailed content of From C# to Go: Achieving AES and Base64 Encoding Compatibility. For more information, please follow other related articles on the PHP Chinese website!