ansaurus

Question

Answer 1

+4 A:

If you just need a hash, you can still use the first 12 characters from the md5 hash.

substr(md5($yourString, 0, 12)

Colin Hebert 2010-09-21 19:29:54

I was just typing the very same!

adam 2010-09-21 19:30:22

I should clarify that we still need to decode it. There could be a case where the first 12 characters of one md5 are the same as another.

tiredofcoding 2010-09-21 19:33:45

@tiredofcoding: MD5, as any other hash, is a one-way function: you *can't* decode it. What you want is a cipher, not a hash.

casablanca 2010-09-21 19:34:50

I know that. What I'm saying is that we'll maintain the values of each piece of data we encode with md5. So we'll know that value x = md5 value y.

tiredofcoding 2010-09-21 19:35:35

@tiredofcoding: Then you may consider picking out different parts of the MD5 hash to reduce chances of collision, or split the MD5 into four 8-character strings and fold them somehow into a single string.

casablanca 2010-09-21 19:37:12

"value x = md5 value y" and first12LettersFrom(value x) = first12LettersFrom(md5 value y) too. I don't really see the problem here.

Colin Hebert 2010-09-21 19:43:57

but what about the chances of the first 12 characters being the same as another?

tiredofcoding 2010-09-21 19:48:40

Every hash function has collisions. If you want to avoid absolutely collisions, you need to cypher and compress your String so it fit in 12 characters.

Colin Hebert 2010-09-21 19:56:00

@casablanca That won't improve collision odds - any well designed hash function will have good distribution over any subset.

Nick Johnson 2010-09-21 23:00:16

@tiredofcoding With 12 hexadecimal characters, you have 48 bits of hash. According to the birthday paradox, you'd expect a collision after about 2^24 hashes - 16 million of them. If you use base64, then your 12 characters encode 9 bytes, which is 72 bits, for an expected collision after about 2^36 or 68 billion hashes.

Nick Johnson 2010-09-21 23:02:48

Answer 2

A:

Try crc32() maybe?

Alex Howansky 2010-09-21 19:30:42

Answer 3

A:

this is just 7 characters long:

echo str_pad(base_convert(sprintf('%u',crc32($str)),10,36),7,'0',STR_PAD_LEFT);

it converts crc-32 long integer to base 36.

stillstanding 2010-09-21 20:04:03

Answer 4

A:

This is an addition to this answer.

The answer proposes to take the first twelve characters from a 32 character representation of md5. Thus 20 characters of information will be lost - this will result in way more possible collisions.

You can reduce the loss of information by taking the first twelve characters of a 16 character representation (the raw form):

substr(md5($string, true), 0, 12);

This will maintain 75% of the data, whereas the use of the 32 char form only maintains 37.5% of the data.

nikic 2010-09-21 20:52:17

ansaurus

tags:

views:

answers:

shorter php cipher than md5?

related questions