CODEWITHSUNDEEP
pythonalgorithmsearchoperators
techshare1337
techshare1337

Reputation: 243

What algorithm is used when using the in operator in python to search a list?

When using the 'in' operator to search for an item in a list e.g.

if item in list:
  print item

What algorithm is used to search for this item. Is it a straight search of the list from beginning to end or does it use something like binary search?

Upvotes: 19

Views: 7295

Answers (2)

Removed
Removed

Reputation: 1

If list is a literal list, Python 3.2+ will take a faster approach: http://docs.python.org/dev/whatsnew/3.2.html#optimizations

Upvotes: 5

Li-aung Yip
Li-aung Yip

Reputation: 12486

lists can't be assumed to be in sorted order (or any order at all), so binary search won't work. Nor can the keys be assumed to be hashable, so unlike a dict or set a hash-table lookup can't be used to accelerate the search

At a guess it's a straight-through check of every element from first to last.

I'll try and dig up the relevant Python source code.

--

Edit: The Python list.__contains__() function, which implements the in operator, is defined in listobject.c:

   393 static int
   394 list_contains(PyListObject *a, PyObject *el)
   395 {
   396     Py_ssize_t i;
   397     int cmp;
   398 
   399     for (i = 0, cmp = 0 ; cmp == 0 && i < Py_SIZE(a); ++i)
   400         cmp = PyObject_RichCompareBool(el, PyList_GET_ITEM(a, i),
   401                                            Py_EQ);
   402     return cmp;
   403 }

It iterates over every element in the list, from the first element to the last element (or until it finds a match.) No shortcuts here.

--

Edit 2: The plot thickens. If Python detects that you're testing for membership of an element in a constant list or set, like:

if letter in ['a','e','i','o','u']:    # list version
if letter in {'a','e','i','o','u'}:    # set version

Edit 3 [@JohnMachin]:

The constant list is optimised to a constant tuple in 2.5-2.7 and 3.1-3.3.
The constant set is optimised to a (constant) frozenset in 3.3.

See also @CoryCarson's answer.

Upvotes: 23

Related Questions