If you have the string "'a','b'" -> as a list("'a','b'") it gets converted to ["'", 'a', "'", ',', "'", 'b', "'"] however I want it to be like: ['a', 'b']
for example:
a = 'a','b' -> list(a) -> ['a', 'b']
-
Where does the string come from (user input, file, literal)?georg– georg2013-01-08 11:04:51 +00:00Commented Jan 8, 2013 at 11:04
Add a comment
|
2 Answers
You could replace and then split on the ,:
>>> "'a','b'".replace("'", "").split(",")
['a', 'b']
Or use a regular expression:
>>> import re
>>> re.findall(r"\w", "'a','b'")
['a', 'b']
Or use evaluation:
>>> import ast
>>> list(ast.literal_eval("'a','b'"))
['a', 'b']
It really depends on your use-case. You could also utilize the CSV module, as @abarnert suggested:
>>> import csv
>>> import io
>>> next(csv.reader(io.StringIO(u"'a','b'"), quotechar="'"))
['a', 'b']
(Note that what works when best also depends on the size of your strings and lists - I cannot link to comments, but please read @abarnert's comments below to get an impression.)
Some benchmarks:
>>> from timeit import timeit
>>> def crude(): "'a','b'".replace("'", "").split(",")
>>> def regex(): re.findall(r"\w", "'a','b'")
>>> def _eval(): list(ast.literal_eval("'a','b'"))
>>> def csvio(): next(csv.reader(io.StringIO(u"'a','b'"), quotechar="'"))
>>> timeit(crude)
1.2010400295257568
>>> timeit(regex)
2.745753049850464
>>> timeit(_eval)
17.016868114471436
>>> timeit(csvio)
3.7680318355560303
4 Comments
abarnert
@miku: +1. But if you want to show everything, what about
next(csv.reader(io.StringIO("'a','b'"), quotechar="'"))? (For the OP's given test case, this is even sillier than regex… but if his real use case can have commas inside the quoted strings, it may not be so bad.)abarnert
@miku: You probably shouldn't call that function
eval. (Then again, maybe you should, since it prevents people from writing Himanshu's answer, making Lattyware happy…)
miku
@abarnert, yes there are more solutions, but what is best really depends on OP's use case. Thanks for the hints: Added your CSV method. Renamed
eval.abarnert
@miku: Yes, it definitely depends on the use case. For example, with lots of strings, csvio is faster than crude; with very long strings, crude pulls way ahead of everything else. (At 5000 1-char strings vs. 2 2500-char strings, with
number=10000, crude: 2.52 / 0.11, regex: 15.31 / 11.72, _eval: (killed after >4 min both times), csvio: 2.02 / 0.79. And that's not even taking into account that the real use case could be more complex, in which case either csvio or regex may be much easier to adapt…Or perhaps just this :-
print eval("'a', 'b'")
This prints :-
('a', 'b')
2 Comments
Gareth Latty
-1. This is a terrible idea, it's unsafe, slow, and not designed for the job. People need to stop recommending this rubbish every time.
Himanshu
Sure. I'll leave it so readers know its not recommended. Thanks :)