First, md5 is a hash. The so-called hash can be considered as a feature code. Files with the same hash are not necessarily the same, but files with different hashes must be different.

If you don't love md5, you can use crc to intercept the fixed length, or write a hash function yourself. The shorter it is, the easier it is to collide with the code (that is, the hashes are the same but the contents are different).