TypeError: 'float' object is not iterable on a list in built in max function

Question

I am trying to find the closest match to an approximate movie title given an actual movie title using the max function and its key argument. If I define a sample list and test the function it works...

from difflib import SequenceMatcher as SM
movies = ['fake movie title', 'faker movie title', 'shaun died']
approx_title = 'Shaun of the Dead.'
max(movies, key = lambda title: SM(None, approx_title, title).ratio())
'shaun died'

But I am trying to match to an entire column in a separate dataframe, so I tried converting that Pandas Series to a list and running the same function, but instead I get a type_error, even though I've checked the datatype of both movies & movie_lst are lists.

Old id  New id  Title   Year    Critics Score   Audience Score  Rating
NaN     21736.0 Peter Pan   1999.0  NaN 70.0    PG nothing objectionable
NaN     771471359.0 Dragonheart Battle for the Heartfire    2017.0  NaN 50.0    PG13
NaN     770725090.0 The Nude Vampire Vampire nue, La    1974.0  NaN 24.0    NR
2281.0  19887.0 Beyond the Clouds   1995.0  65.0    67.0    NR
10913.0 11286.0 Wild America    1997.0  27.0    59.0    PG violence

movie_lst = rt_info['Title'].tolist()
 ['Peter Pan',
 'Dragonheart Battle for the Heartfire',
 'The Nude Vampire Vampire nue, La',
 'Beyond the Clouds',
 'Wild America',
 'Sexual Dependency',
 'Body Slam',
 'Hatchet II',
 'Lion of the Desert Omar Mukhtar',
 'Imagine That',
 'Harold',
 'A United Kingdom',
 'Violent City The FamilyCitt violenta',
 'Ratchet  Clank',
 'Wes Craven Presents Carnival of Souls',
 'The Adventures of Ociee Nash',
 'Blackfish',
 'For Petes Sake',
 'Daybreakers',
 'The Big One',
 'Godzilla vs Megaguirus',
 'In a Lonely Place',
 'Case 39', ...
]

max(movie_lst, key = lambda title: SM(None, approx_title, title).ratio())

TypeError                                 Traceback (most recent call last)
<ipython-input-88-0022a3c1bdb9> in <module>()
----> 1 max(movie_lst, key = lambda title: SM(None, approx_title, title).ratio())

<ipython-input-88-0022a3c1bdb9> in <lambda>(title)
----> 1 max(movie_lst, key = lambda title: SM(None, approx_title, title).ratio())

/usr/lib/python3.4/difflib.py in __init__(self, isjunk, a, b, autojunk)
    211         self.a = self.b = None
    212         self.autojunk = autojunk
--> 213         self.set_seqs(a, b)
    214 
    215     def set_seqs(self, a, b):

/usr/lib/python3.4/difflib.py in set_seqs(self, a, b)
    223 
    224         self.set_seq1(a)
--> 225         self.set_seq2(b)
    226 
    227     def set_seq1(self, a):

/usr/lib/python3.4/difflib.py in set_seq2(self, b)
    277         self.matching_blocks = self.opcodes = None
    278         self.fullbcount = None
--> 279         self.__chain_b()
    280 
    281     # For each element x in b, set b2j[x] to a list of the indices in

/usr/lib/python3.4/difflib.py in __chain_b(self)
    309         self.b2j = b2j = {}
    310 
--> 311         for i, elt in enumerate(b):
    312             indices = b2j.setdefault(elt, [])
    313             indices.append(i)

TypeError: 'float' object is not iterable

I'm stumped as to why - any help would be appreciated!

Answer 1

Not a pandas expert and cannot reproduce but depending on how the file is read, since there are titles (like the french movie 11.6 for instance) which match a float, it's possible that some data are floats instead of strings (well your issue proves that it is possible :))

A good workaround would be to force data as string like this:

movie_lst = [str(x) for x in movie_lst]

It doesn't create copies of the strings if they are already strings (Should I avoid converting to a string if a value is already a string?) so it's efficient, and you are sure to get only strings.

note that you can find the offenders by printing:

[x for x in movie_lst if not isinstance(x,str)]