13

Can anyone suggest a good solution to remove duplicates from nested lists if wanting to evaluate duplicates based on first element of each nested list?

The main list looks like this:

L = [['14', '65', 76], ['2', '5', 6], ['7', '12', 33], ['14', '22', 46]]

If there is another list with the same element at first position [k][0] that had already occurred, then I'd like to remove that list and get this result:

L = [['14', '65', 76], ['2', '5', 6], ['7', '12', 33]]

Can you suggest an algorithm to achieve this goal?

p.campbell
  • 95,348
  • 63
  • 249
  • 319
elfuego1
  • 10,218
  • 9
  • 27
  • 24

5 Answers5

32

Do you care about preserving order / which duplicate is removed? If not, then:

dict((x[0], x) for x in L).values()

will do it. If you want to preserve order, and want to keep the first one you find then:

def unique_items(L):
    found = set()
    for item in L:
        if item[0] not in found:
            yield item
            found.add(item[0])

print list(unique_items(L))
Brian
  • 113,117
  • 28
  • 106
  • 111
  • your conversion to a dict was so much more elegant than mind that I stole it :) – Jiaaro Jul 17 '09 at 14:02
  • Doesn't the first one also preserve order because dicts preserve order since Python 3.7 and the keys are inserted in the order that the comprehension produces them? – xuiqzy Oct 01 '20 at 13:49
4

use a dict instead like so:

L = {'14': ['65', 76], '2': ['5', 6], '7': ['12', 33]}
L['14'] = ['22', 46]

if you are receiving the first list from some external source, convert it like so:

L = [['14', '65', 76], ['2', '5', 6], ['7', '12', 33], ['14', '22', 46]]
L_dict = dict((x[0], x[1:]) for x in L)
Jiaaro
  • 71,020
  • 40
  • 161
  • 188
2

Use Pandas : 

import pandas as pd

L = [['14', '65', 76], ['2', '5', 6], ['7', '12', 33], ['14', '22', 46],['7','a','b']]

df = pd.DataFrame(L)
df = df.drop_duplicates()

L_no_duplicates = df.values.tolist()

If you want to drop duplicates in specific columns only use instead:

df = df.drop_duplicates([1,2])
Rupert Schiessl
  • 749
  • 5
  • 10
0

i am not sure what you meant by "another list", so i assume you are saying those lists inside L

a=[]
L = [['14', '65', 76], ['2', '5', 6], ['7', '12', 33], ['14', '22', 46],['7','a','b']]
for item in L:
    if not item[0] in a:
        a.append(item[0])
        print item
ghostdog74
  • 307,646
  • 55
  • 250
  • 337
  • 1
    This would be more efficient if you used a set for 'a' - you're O(N^2) using a list like that, and amortised O(N) using a set. – RichieHindle Jul 17 '09 at 13:58
  • that has not come to mind, thanks for the info. nevertheless, that code works in older Python version that doesn't come with set. ;) – ghostdog74 Jul 17 '09 at 14:14
0

If the order does not matter, code below

print [ [k] + v for (k, v) in dict( [ [a[0], a[1:]] for a in reversed(L) ] ).items() ]

gives

[['2', '5', '6'], ['14', '65', '76'], ['7', '12', '33']]

Jinuk Kim
  • 705
  • 5
  • 5