complex sorting in python

Question

I am using Python 3.2 and would like to sort a list of tuples based on a configuration file:

formats=CCC;aaa;BBB
providers=yy;QQ;TT

Each tuple contains this information:

( title, size, format, provider )

I would like this group of tuples to first be sorted by the providers list. All yy's come before QQ's and TT's.

Then, keeping this result order, move onto formats. All CCC's before aaa's before BBB's.

Finally, the third criteria would be to sort by size (float), in decending order.

It is critical that each step uses a stable sort so that the secondary sort keeps the ordering of the first sort and so on.

How can I do this in a pythonic way? Thanks.

EDIT 1

This is what I tried, my it will obviously not work because of sorted(mydata). mydata can't be a list in this context.

providers="yy;QQ;TT"
formats="CCC;aaa;BBB"

p_dict = {}
f_dict = {}

for k,v in enumerate(providers.split(';')):
    p_dict[k] = v

for k,v in enumerate(formats.split(';')):
    f_dict[k] = v

mydata = (
                ('title1', 423.4, 'QQ', 'aaa'),
                ('title2', 523.2, 'TT', 'CCC'),
                ('title3', 389.0, 'yy', 'aaa'),
                ('title4', 503.2, 'QQ', 'BBB') )

sort1 = sorted( mydata, key=p_dict.__getitem__)
print(sort1)

If you ask about a pythonic way, what's the way you've come up with so far? What code do you use to split the input? What individual sort function have you tried? — cfi
– cfi, Commented Jul 10, 2012 at 13:54

Winston Ewert · Accepted Answer · 2012-07-10 17:27:11Z

3

def sort_key(data):
    title, size, format, provider = data
    return p_dict[provider], f_dict[format], -size 

print sorted( mydata, key = sort_key)

Basically, devise a key function that produces tuples that will sort in the desired order.

answered Jul 10, 2012 at 17:27

Winston Ewert

45.2k10 gold badges70 silver badges86 bronze badges

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

2 Comments

jftuga Over a year ago

I think this is very close to working. I changed my code to "p_dict[v] = k" and "f_dict[v] = k"

jftuga Over a year ago

I was able to get this working within my program. Thanks again!

steveha · Accepted Answer · 2012-07-10 17:54:44Z

2

This is basically @Winston Ewert's answer, but it's complete working code. Note that you can create a dictionary by passing an iterable to dict(); here we make a generator expression to swap around the result we get from enumerate().

providers="yy;QQ;TT"
formats="CCC;aaa;BBB"

d_providers = dict((k, v) for v, k in enumerate(providers.split(';')))
d_formats = dict((k, v) for v, k in enumerate(formats.split(';')))

def key_mydata(m):
    return (d_providers[m[2]], d_formats[m[3]], -m[1], m[0])

mydata = (
                ('title1', 423.4, 'QQ', 'aaa'),
                ('title2', 523.2, 'TT', 'CCC'),
                ('title3', 389.0, 'yy', 'aaa'),
                ('title4', 503.2, 'QQ', 'BBB') )

sort1 = sorted(mydata, key=key_mydata)
print(sort1)

answered Jul 10, 2012 at 17:54

steveha

77.1k21 gold badges94 silver badges119 bronze badges

2 Comments

jftuga Over a year ago

Thanks, this is a great answer, but feel compelled to give Winston the check mark. Great idea about the dict.

steveha Over a year ago

No worries; he did answer it first, and besides I have plenty of points.

Ignacio Vazquez-Abrams · Accepted Answer · 2012-07-10 14:18:59Z

1

Create sequences of the formats and providers, then use a compound key (read: tuple) that looks up the indexes of the values in the current element.

answered Jul 10, 2012 at 14:18

Ignacio Vazquez-Abrams

804k160 gold badges1.4k silver badges1.4k bronze badges

1 Comment

jftuga Over a year ago

Can you please elaborate? I am just not following this. Thanks.

ecatmur · Accepted Answer · 2012-07-10 17:38:02Z

0

Use a key function that converts each part to an index:

providers = providers.split(';')
formats = formats.split(';')
def sort_key(item):
    title, size, format, provider = item
    return (providers.index(provider), formats.index(format), -size, title)
print(sorted(mydata, key=sort_key))

Note that Python sorting is stable, so you could also e.g. sort by size descending first, then by format, then by provider.

edited Jul 10, 2012 at 17:38

answered Jul 10, 2012 at 17:28

ecatmur

158k28 gold badges311 silver badges387 bronze badges

2 Comments

Winston Ewert Over a year ago

Doesn't sort_key has to take a single parameter, not four?

steveha Over a year ago

What I don't like about this: list.index() is O(N) where N is the length of the list. For short lists it would be okay, but for longer lists it would be better to build a dict() that maps each value onto its index value.

Collectives™ on Stack Overflow

complex sorting in python

4 Answers 4

2 Comments

2 Comments

1 Comment

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

4 Answers 4

2 Comments

2 Comments

1 Comment

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related