PHP to python pack('H')

I'm translating an authentication library, written in PHP to Python. It's all legacy code, the original devs long gone. They used PHP's 'pack' command to transform a string into hex using the 'H' flag. PHP's documentation describes this as 'Hex string, high nibble first'. I read another question (Python equivalent of php pack) which suggested using binascii.unhexlify(), but this complains whenever I pass in a non-hex character.

So my question is what does the PHP pack function do with non-hex characters? Does it throw them away, or is there an extra step that performs a translation. Is there a better method in Python than binascii.unhexlify?

So pack'ing 'H*'

php -r 'print pack("H*", md5("Dummy String"));'

Returns

??????=?PW??

In python:

secret = binascii.unhexlify( "Dummy String" )
TypeError: Non-hexadecimal digit found

Thanks for the help.

[EDIT]

So DJV was fundamentally right. I needed to convert the value into md5 first, however that's where it's interesting. In python the md5 library returns binary data via the 'digest' method.

In my case I could skip all the binascii calls and just use

md5.md5('Dummy String').digest()

Which is the same in PHP as:

pack("H*", md5("Dummy String"));

Fun stuff. Good to know.

Solution

I think you need it the other way around. "Dummy String" is not a valid number in hex. You can hexlify it:

>>> binascii.hexlify('Dummy String')
'44756d6d7920537472696e67'

but not unhexlify it. unhexlify takes a string in hex and turns it into it's ASCII representation:

>>> binascii.unhexlify('44756d6d7920537472696e67')
'Dummy String'

What you need is to md5 the string ("Dummy String" in our case) and unhexlify it's hash:

import binascii
import hashlib

the_hash = hashlib.md5('Dummy String').hexdigest()
print the_hash
the_unhex = binascii.unhexlify(the_hash)
print the_unhex

Which yields the hash, and the unhexlified hash:

ec041da9f891c09b3d1617ba5057b3f5
ЛLЬ-ю?=¦PWЁУ

Note: although the output doesn't look exactly like yours - "??????=?PW??", the "PW" and "=" in both, makes me pretty certain it's correct.