Add prefix failed with percentage

Question

df = pd.DataFrame({'a':[1,4], 'b':[7,8]})
print (df)
   a  b
0  1  7
1  4  8

I try add % to columns names, so use DataFrame.add_prefix

print (df.add_prefix('%'))

TypeError: not all arguments converted during string formatting

What can be problem?

How is possible use add_prefix function with % ?

Or it is only bug?

Notice:

Possible solution is like:

df.columns = ['%' + col for col in df.columns]
print (df)
   %a  %b
0   1   7
1   4   8

but I am interested about function add_prefix.

Answer 1

NOTE:
This is likely to change in the near future as pandas will use new style string formatting in this case. When that happens:

'{}hello'.format('%')

'%hello'

adding a prefix with a single % will work just fine.

See github

---

Answer
Two percent signs! One escapes the other when using the old style of string formatting.

df.add_prefix('%%')

   %a  %b
0   1   7
1   4   8

---

The clue came from:

   2856     def add_prefix(self, prefix):
   2857         f = (str(prefix) + '%s').__mod__
-> 2858         return self.rename_axis(f, axis=0)
   2859 
   2860     def add_suffix(self, suffix):

So I tried it myself

f = ('my_prefix_' + '%s').__mod__
f('hello')

'my_prefix_hello'

And then

f = ('%' + '%s').__mod__
f('hello')

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-901-0c92e5498bbc> in <module>()
      1 f = ('%' + '%s').__mod__
----> 2 f('hello')

TypeError: not all arguments converted during string formatting

So I looked up how to escape the '%' in the old style of string formatting and found this answer

Which led to this

f = ('%%' + '%s').__mod__
f('hello')

'%hello'