3

The Spring4D library has cryptography classes, however I cannot get them to work as expected. I'm probably using them incorrectly, however lack of any examples makes it difficult.

For example on the website https://quickhash.com/hash-sha256-online, I can hash the word "test" to generate the following hash:

9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a08

Using the Spring4D library, the following code produces a different hash:

CreateSHA256.ComputeHash('test').ToString;

results in:

9EFEA1AEAC9EDA04A892885A65FDAE0E6D9BE8C9FC96DA76D31B929262E12B1D

Upper/lower case aside, it is a different hash altogether. I know must be doing something wrong, but again there's no examples of use so I'm stuck on how to do this.

Maarten Bodewes
  • 90,524
  • 13
  • 150
  • 263
Rick Wheeler
  • 1,142
  • 10
  • 22
  • 1
    It's kind of tragic that this library lulls you into believing that you can hash text. You hash binary. Presumably the two variants encode the text with different encodings. Do yourself a favour and be explicit. Encode to UTF8 bytes and hash that. – David Heffernan Dec 01 '14 at 07:15
  • 1
    I've changed the title as it is off topic to ask for examples on StackOverflow. You *didn't* but the title suggested that you did. – Maarten Bodewes Dec 01 '14 at 08:38

3 Answers3

9

Hashing algorithms operate on binary data, typically represented using byte arrays.

Unfortunately, both of the resources you have used offer the ability to hash text. In order to hash text, you first need to convert from text to binary. To do so requires a choice of encoding. And neither method makes it clear what that choice is.

When I use this Delphi code:

LowerCase(CreateSHA256.ComputeHash(TEncoding.UTF8.GetBytes('test')).ToString)

I get the same hash as appears in your question.

I urge you never to attempt to encrypt/hash text and instead regard these operations as operating on binary. Always use an explicit encoding and then encrypt/hash the array of bytes that the encoding produced.

I've picked the UTF-8 encoding here, because it is a full Unicode encoding, and tends to be efficient in terms of space. However, I don't think your online encoder uses UTF-8. In fact I've no idea what encoding it uses, it is unclear on the matter. This is of course the same old issue of text being different from binary.

In my opinion it is a design flaw of the Delphi library that you use that it allows you to hash text without an explicit choice of encoding. If this library must offer a function that hashes text, then it should require the caller to supply an extra TEncoding parameter.

David Heffernan
  • 601,492
  • 42
  • 1,072
  • 1,490
4

There is no conversion going on internally so it hashes the UnicodeString which is at least 2 bytes per character.

If you want the same result as on the page you have to use UTF8Encode or directly pass as AnsiString.

However I tried some strings that contained different unicode characters and the page returned a different result. So I am not quite sure how they treat the strings there. I guess it's a codepage thing.

Edit: If you use this page http://www.xorbin.com/tools/sha256-hash-calculator it generates the same hash as TSHA256 with UTF8Encode.

Stefan Glienke
  • 20,860
  • 2
  • 48
  • 102
0

Which type of string are you using? Do you use AnsiString or WideString (Unicode string). Delphi 2009 and Newer are using WideString by default.

Why is string type inportant? All hasging algorithm operates on raw bytes data so it is omportant if each character of your string is stored in one Byte of memory (AnsiString) or multiple Bytes of memory (WideString).

SilverWarior
  • 7,372
  • 2
  • 16
  • 22