Doubts in Hashing

Question

I was reading this post and I had a question regarding __cmp__().

My code

class Book(object): 

    def __init__(self, title, year):

        self.title = title
        self.year = year

    def __hash__(self):         # hash function

        print "Self = ", self
        print "Hash value of self = ", hash(str(self))
        print "Hash value of title = ", hash(self.title)
        print "Hash value of year = ", hash(self.year)
        return 0
    def __cmp__(self, other):
        return self.title == other.title

books = []

books.append(Book("ABC", 123))
print hash(books[0])

books.append(Book("DEF", 456))
print hash(books[1])


books.append(Book("ABC", 123))
print hash(books[len(books)-1])

print len(books)

print cmp(books[0],books[2])

Output

Self =  <__main__.Book object at 0x0000000001E77B00>
Hash value of self =  -1040857764
Hash value of title =  826005955
Hash value of year =  123
0
Self =  <__main__.Book object at 0x0000000001E77BA8>
Hash value of self =  -992414627
Hash value of title =  -589261154
Hash value of year =  456
0
Self =  <__main__.Book object at 0x0000000001E77BE0>
Hash value of self =  1901105233
Hash value of title =  -2015893559
Hash value of year =  789
0
Self =  <__main__.Book object at 0x0000000001E77C18>
Hash value of self =  -228580758
Hash value of title =  826005955
Hash value of year =  123
0
4
1  # How ?

== operator on strings returns a Boolean value.
In this code the cmp() compares two strings (namely the titles) so the return value of the cmp() must be a Bool.

How come I am getting an integer ? (I am looking for an explanation with respect to the comparison of objects x > y)
Even when an integer is obtained, how is its value decided ?
How can I make the 1st and the 3rd object equal ?

`__cmp__` *must* return an integer. Did you mean to use `__eq__` instead? — Martijn Pieters, Jul 09 '15 at 16:18
Also, you are using the `cmp()` *function*; it returns an integer. Do read the [function documentation](https://docs.python.org/2/library/functions.html#cmp). — Martijn Pieters, Jul 09 '15 at 16:19
I know that __cmp__ returns an integer but my doubt is how does it do so ? What does `x>y` mean ? — Kshitij Saraogi, Jul 09 '15 at 16:20
And the code you posted adds only **three** books, not four. You'd get an `IndexError` for `books[3]`. — Martijn Pieters, Jul 09 '15 at 16:21
Note that `__cmp__` doesn't call `hash()`. I'm not sure why you implemented that and what you expect it to do. Everything about the `__hash__` method is orthogonal to the rest of your question. — Martijn Pieters, Jul 09 '15 at 16:22
@MartijnPieters I thought that maybe the hash value is related to the comparisons of objects — Kshitij Saraogi, Jul 09 '15 at 16:24
@KshitijSaraogi Why? Maybe you don't know what hashing is and for what it is used for... — Bakuriu, Jul 09 '15 at 16:25
It does not. It is used when storing objects in sets or use them as keys in dictionaries, and there are restrictions on what you can return related to object equality, but object equality doesn't use hashing, no. — Martijn Pieters, Jul 09 '15 at 16:26
@Bakuriu Since I dont know how the comparison takes place, I was looking for a solution and I thought it might had been the hash values of the objects. — Kshitij Saraogi, Jul 09 '15 at 16:27
@NightShadeQueen I absolutely agree that the hash function is certainly not a good one. I was just exploring the topic — Kshitij Saraogi, Jul 09 '15 at 16:45

Ethan Furman · Accepted Answer · 2015-07-09T16:46:56.797

1

How come I am getting an integer ?

__cmp__ returns an int. Your __cmp__ is returning a bool, which is a subclass of int, so Python interprets it as an int.

Even when an integer is obtained, how is its value decided ?

-1 means the first operand is less than the second
0 means both operands are equal
+1 means the first operand is greater than the second

How can I make the 1st and the 3rd object equal?

The problem with your code is your __cmp__ returns the result of ==, which returns a bool, which is a subclass of int, and True is 1, so instead of returning 0 your __cmp__ is returning True/1 -- which is wrong.

The methods you should be using (as __cmp__ gets removed in later Pythons) are __eq__, __ne__, __le__, __lt__, __ge__, and __gt__.

Using __eq__:

def __eq__(self, other):
    if not isinstance(other, self.__class__):
        return NotImplemented
    return self.title == other.title

Note that this is only comparing by title, and not considering the year (which may be fine, or may not be, depending on how you will be using your Book class).

You can find out more about them here.

edited Jul 09 '15 at 16:46

answered Jul 09 '15 at 16:18

Ethan Furman

63,992
20
159
237

And they are using the `cmp()` *function* even. – Martijn Pieters Jul 09 '15 at 16:19
1

@EthanFurman What you said about -1 and +1 are not true! The [official API](https://docs.python.org/2/library/functions.html#cmp) only mentions negative, zero, and positive numbers. – Nayuki Jul 09 '15 at 16:31
2

@NayukiMinase: You are correct -- but the negative number is usually -1, and the positive number is usually +1. :) – Ethan Furman Jul 09 '15 at 16:32
1

@EthanFurman I agree with you. And according to this the output should be `0` rather than `1`. Don't you agree ? – Kshitij Saraogi Jul 09 '15 at 16:32
1

@KshitijSaraogi: updated my answer to address the problem you are seeing. – Ethan Furman Jul 09 '15 at 16:39
@EthanFurman I had already gone through these docs before posting the question – Kshitij Saraogi Jul 09 '15 at 16:43

Doubts in Hashing

1 Answers1