ast.literal_eval() support for set literals in Python 2.7?
In the What’s New in Python 2.7 document it says that support for set literals was back-ported from Python 3.1. However it appears that this support was not extended to the ast
module's literal_eval()
function, as illustrated below.
Was this intentional, an oversight, or something else — and what are the cleanest workarounds for creating a literal set from a string representation? (I assume the following works in Python 3.1+, right?)
import ast
a_set = {1,2,3,4,5}
print(a_set)
print(ast.literal_eval('{1,2,3,4,5}'))
Output showing error message:
set([1, 2, 3, 4, 5])
Traceback (most recent call last):
File "...\setliterals.py", line 4, in <module>
print ast.literal_eval('{1,2,3,4,5}')
File "...\Python\lib\ast.py", line 80, in lite开发者_Go百科ral_eval
return _convert(node_or_string)
File "...\Python\lib\ast.py", line 79, in _convert
raise ValueError('malformed string')
ValueError: malformed string
P.S. The only workaround I can think of is to use eval()
.
I've been using this for converting columns in a pandas DataFrame (df[col] = df[col].apply(to_set)
. Might be useful for anyone finding this question. It may not be as fast but it avoids using eval
.
def to_set(set_str):
"""
Required to get around the lack of support for sets in ast.literal_eval.
It works by converting the string to a list and then to a set.
Parameters
----------
set_str : str
A string representation of a set.
Returns
-------
set
Raises
------
ValueError
"malformed string" if the string does not start with '{' and and end
with '}'.
"""
set_str = set_str.strip()
if not (set_str.startswith('{') and set_str.endswith('}')):
raise ValueError("malformed string")
olds, news = ['{', '}'] , ['[',']']
for old, new in izip(olds, news):
set_str = set_str.replace(old, new)
return set(literal_eval(set_str))
From the bug report: http://bugs.python.org/issue10091
Raymond Hettinger says:
ast.literal_eval's docs:
'The string or node provided may only consist of the following Python literal structures: strings, numbers, tuples, lists, dicts, booleans, and None.'
I believe we can conclude from this document that the matter is not necessarily a bug, since the set literal was backported from Python 3.2 to 3.1 and 2.7. It is something that a Python 2.7 user of ast.literal
should be aware of.
精彩评论