|
Description:
This recipe enables the group_by functionality from Ruby on Rails, which is to be included in next version of Ruby. This code is from Ian Bicking's suggestion in discussing partition function in Ruby and Javascript by Ned Batchelder.
Original Post: http://www.nedbatchelder.com/blog/200607.html#e20060730T221504
Here is Ian Bicking's Response:
http://www.nedbatchelder.com/reactor/comment.php?entryid=e20060730T221504
Source: Text Source
def partition(iterable, func):
result = {}
for i in iterable:
result.setdefault(func(i), []).append(i)
return result
def group(seq):
'''seq is a sequence of tuple containing (item_to_be_categorized, category)'''
result = {}
for item, category in seq:
result.setdefault(category, []).append(item)
return result
def is_odd(n):
return (n%2) == 1
l = range(10)
print partition(l, is_odd)
print group( (item, is_odd(item)) for item in l)
print group( (item, item%3) for item in l )
class Article (object):
def __init__(self, title, summary, author):
self.title = title
self.summary = summary
self.author = author
articles = [ Article('Politics & Me', 'about politics', 'Dave'),
Article('Fishes & Me', 'about fishes', 'ray'),
Article('Love & Me', 'about love', 'dave'),
Article('Spams & Me', 'about spams', 'Ray'), ]
print group( (article.summary, article.author.lower()) for article in articles )
{False: [0, 2, 4, 6, 8], True: [1, 3, 5, 7, 9]}
{False: [0, 2, 4, 6, 8], True: [1, 3, 5, 7, 9]}
{0: [0, 3, 6, 9], 1: [1, 4, 7], 2: [2, 5, 8]}
{'dave': ['about politics', 'about love'], 'ray': ['about fishes', 'about spams']}
Discussion:
The group() function by Ian Bicking can be extremely useful when categorizing contents that are often returned in sequences.
Note also "(item, is_odd(item)) for item in l" returns a iterable to be only iterated within the group() function. So, even large list would be quite efficient.
As a comparison w/ Ruby's group_by method for a list of articles:
# using block
articles.group_by { |article| article.author }
# using rails enhancement to symbols
articles.group_by(&:author)
Do keep in mind categories must use immutable objects since dict requires immutable object for index. One could modify the group() function accordingly to handle categories that are given in a list instead. For a background on what I'm talking about: http://dev.rubyonrails.org/ticket/4427
|
|
Add comment
|
|
Number of comments: 2
For the one-liner lovers out there, N N, 2006/10/26
from operator import itemgetter
from itertools import groupby
dict([(k,[h[1] for h in g]) for k,g in groupby(sorted((bool(n%2),n) for n in range(10)),key=itemgetter(0))])
dict([(k,[h[1] for h in g]) for k,g in groupby(sorted((n%3,n) for n in range(10)),key=itemgetter(0))])
dict([(k,[h[1] for h in g]) for k,g in groupby(sorted((article.author.lower(),article.summary) for article in articles),key=itemgetter(0))])
These do the same thing as the examples shown. The recipe is still usefull because clarity and efficency is better than one liners in many cases.
Add comment
if you don't mind a level of indirection..., Damon McCormick, 2008/06/22
...you can also implement partition using group.
def partition (iterable, func):
return group((i, func(i)) for i in iterable)
Add comment
|
|
|
