Python-сообщество

Pysrc/Modules/operator.c:325:/* itemgetter object

import timeit
from operator import itemgetter
def bench_lambda_key(buf):
    return sorted(buf, key=lambda x: (x[2],x[1],x[0]))
def bench_lambda_auto_key(buf):
    return sorted(buf, key=lambda x: reversed(x))
def bench_itemgetter(buf):
    return sorted(buf, key=itemgetter(2,1,0))
def bench_keyfunc(buf):
    def keyfunc(x):
        return x[2],x[1],x[0]
    return sorted(buf, key=keyfunc)
def measure(func, arg, times):
    print " %s %s" % (func.__name__, timeit.timeit(lambda: func(arg),
                         number=times))
buffs = [[(i,i,i) for i in range(val,0,-1)]
         for val in (1000, 10000, 100000, 10000000)]
print [(buf[0],buf[-1]) for buf in buffs]
funcs = [func for name,func in globals().items() if name.startswith('bench_')] + [sorted]
for buf in buffs:
    assert buf[-1] < buf[0]
    assert buf[-1] == (1,1,1)
    print "Buflen:", len(buf)
    for func in funcs:       
        measure(func, buf, 3)

[((1000, 1000, 1000), (1, 1, 1)), ((10000, 10000, 10000), (1, 1, 1)), ((100000,
100000, 100000), (1, 1, 1)), ((10000000, 10000000, 10000000), (1, 1, 1))]
Buflen: 1000
 bench_itemgetter 0.00191706278265
 bench_keyfunc 0.00287790388817
 bench_lambda_key 0.00287559417398
 bench_lambda_auto_key 0.00300378331185
 sorted 0.000578968358455
Buflen: 10000
 bench_itemgetter 0.0194231565786
 bench_keyfunc 0.0218749181976
 bench_lambda_key 0.0152325651224
 bench_lambda_auto_key 0.0141616276402
 sorted 0.00305036254814
Buflen: 100000
 bench_itemgetter 0.130403383885
 bench_keyfunc 0.179477496455
 bench_lambda_key 0.178710286389
 bench_lambda_auto_key 0.152915013296
 sorted 0.040913892315
Buflen: 10000000
 bench_itemgetter 14.2914632193
 bench_keyfunc 19.6054661696
 bench_lambda_key 19.6240705325
 bench_lambda_auto_key 17.0308593215
 sorted 4.21384796845

Buflen: 1000
 bench_itemgetter 0.0267988439111
 bench_keyfunc 0.00317316235289
 bench_lambda_key 0.00303457950113
 bench_lambda_auto_key 0.00624893175718
 sorted 0.000451934077677
Buflen: 10000
 bench_itemgetter 0.0240387354469
 bench_keyfunc 0.0165294696434
 bench_lambda_key 0.0145931592425
 bench_lambda_auto_key 0.0314840991576
 sorted 0.00325438730212
Buflen: 100000
 bench_itemgetter 0.298087864608
 bench_keyfunc 0.200838888196
 bench_lambda_key 0.201263105703
 bench_lambda_auto_key 0.219333539668
 sorted 0.0355272538576
Buflen: 10000000
 bench_itemgetter 48.9225983977
 bench_keyfunc 30.2210835331
 bench_lambda_key 29.9068399876
 bench_lambda_auto_key 30.5330650985
 sorted 4.42660690666

class LinqMixin(object):
    def OrderBy(self, arg, reverse=False):
        if isinstance(arg, int):
            arg = itemgetter(arg)
        elif isinstance(arg, str):
            arg = attrgetter(arg)
        # else lambda expression
        self.sort(key=arg, reverse=reverse)
        return self
    def OrderByDescending(self, arg):
        return self.OrderBy(arg, True)
    ThenBy = OrderBy
    ThenByDescending = OrderByDescending
class CoolList(list, LinqMixin): pass
def bench_linq(buf):
    l_buf = CoolList(buf)
    return l_buf.OrderBy(2).ThenBy(1).ThenBy(0)
def bench_multiple_sort(buf):
    return sorted(sorted(sorted(buf, key=itemgetter(2)), key=itemgetter(1)), key=itemgetter(0))

 
Buflen: 100000
 bench_linq 0.184998098335
 bench_itemgetter 0.139423202773
 bench_multiple_sort 0.197778516886
 bench_lambda_auto_key 0.159128144483
 bench_keyfunc 0.184526146735
 bench_lambda_key 0.186749246648
 sorted 0.0414897810546
Buflen: 10000000
 bench_linq 19.0146251103
 bench_itemgetter 14.8408456788
 bench_multiple_sort 21.1929246525
 bench_lambda_auto_key 17.6037073223
 bench_keyfunc 20.4276331897
 bench_lambda_key 20.6035298592
 sorted 4.18729972853

from collections import namedtuple
#
Person=namedtuple('Person', 'name age city')
#
b1 = CoolList([Person("Alice",21,"Redmond"),Person("Bob",20,"Boston"),
Person("Nadine",21,"Denver")])
#
>>> b1.OrderBy('age').ThenBy(lambda person: person.name)
[Person(name='Alice', age=21, city='Redmond'), Person(name='Bob', age=20, city='
Boston'), Person(name='Nadine', age=21, city='Denver')]
>>> b1.OrderBy('age').ThenBy(2)
[Person(name='Bob', age=20, city='Boston'), Person(name='Nadine', age=21, city='
Denver'), Person(name='Alice', age=21, city='Redmond')]
>>> b1.OrderBy(1).ThenBy(lambda person: person.name)
[Person(name='Alice', age=21, city='Redmond'), Person(name='Bob', age=20, city='
Boston'), Person(name='Nadine', age=21, city='Denver')]
>>> b1.OrderBy(lambda p: p.age).ThenBy('city')
[Person(name='Bob', age=20, city='Boston'), Person(name='Nadine', age=21, city='
Denver'), Person(name='Alice', age=21, city='Redmond')]