python lib for custom hash function

Question

I need a hash function to take a sequence of decimal numbers and return a decimal number as hash value.

for example:

>> def my_simple_hash(*args):
    return reduce(lambda x1, x2: 2*x1 + x2,  args) 

>>> my_simple_hash(1,3,4)
14
>>> my_simple_hash(1,4,3)
15
>>> my_simple_hash(4,3,1)
23

My questions are:

does python has a built-in lib to do this more efficiently?
how could I make the output hash value in a relative small range?

Question 2 explanation:

because 1, 3, 4 has six different combinations as following:

1,3,4
1,4,3
3,1,4
3,4,1
4,1,3
4,3,1

the corresponding output is [14, 15, 18, 21, 21, 23], and I expect the hash values of the six output would be something like [1,2,3,4,6](a small range)

any suggestions would be appreciated. thanks in advance :-)

I would generally advise not to roll your own, and instead go with a hash that’s virtually guaranteed not to have collisions (hashlib.sha256()). — Jed Fox
– Jed Fox, Commented Sep 3, 2016 at 10:38
That strongly depends on what you want - sha256 is a cryptographic hash function that is relatively slow to calculate, takes bytes-input and produces bytes output. good non-cryptographic hashes are fast to compute, have an even distribution of resulting values and not too many collisions. — janbrohl
– janbrohl, Commented Sep 3, 2016 at 11:08
What do you intend to do with this hash? How large is the domain of the input integers, what's a typical size of an input sequence, and how large should the range of the output be? How important is collision avoidance? Your needs may be satisfied by a variant of format-preserving encryption, either a full cryptographic-strength implementation, or a faster, simpler version. — PM 2Ring
– PM 2Ring, Commented Sep 3, 2016 at 11:08
@PM2Ring, I want to use a single decimal num to identify a sequence of numbers. — Fujiao Liu
– Fujiao Liu, Commented Sep 3, 2016 at 11:36

janbrohl · Accepted Answer · 2016-09-03 11:11:55Z

1

If you just want to hash a number sequence you can do

def my_hash(*args):
    return hash(args)

which returns the hash (for the current run of the program) of the args-tuple (hash is fast and well tested for builtin-types) - but this is still most often a large number.

For getting a smaller value you can take the modulo like

def my_hash(*args):
    return hash(args)%10 # or whatever number you like

Actually you could also use

def my_hash(*args):
    return sum(args)%10 # or whatever number you like

which doesnt change between runs of the program but sum does not distribute the results evenly at all.

Warning: These are not cryptographical hashes

edited Sep 3, 2016 at 11:11

answered Sep 3, 2016 at 11:01

janbrohl

2,6541 gold badge19 silver badges15 bronze badges

Sign up to request clarification or add additional context in comments.

5 Comments

Fujiao Liu Over a year ago

hi, @janbrohl, does python2 have a similar function like hash?

janbrohl Over a year ago

Yes (same name) - It possibly returns different values but it is essentially the same

janbrohl Over a year ago

Most stuff didnt change (much) from Python 2.7 to 3.x and simple programs work on both requiring ony minimal changes (or none at all)

Fujiao Liu Over a year ago

I marked the answer as accepted, but for the second question, the output will be in range 0 to 9 if I use %10 and thus cause collisions. I will use a dict to map every hash values to my desire numbers.

janbrohl Over a year ago

for less collisions ditribute the hashes over a larger range of numbers by using bigger values for %-ing than 10

Collectives™ on Stack Overflow

python lib for custom hash function

1 Answer 1

5 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

5 Comments

Your Answer

Sign up or log in

Post as a guest

Related