Question

I need a regex in python that matches any char that is surrounded by exactly 2 underscores.
meaning, meaning

__a__ 

will match “a”,
but

___a___ 

will not match.
it needs to support overlapping matches,
such that

__a__d___b___e__c__ 

will return “ac”
because a is surrounded by double underlines, but d,e have a triple one next to them and b has a triple underline on both sides.
what I have now

(?<=[_]{2})(.)(?=[_]{2})

it solves the overlapping, but not the “exactly 2”
in the above example it returns “adbec”

Answer 1

Try the following:

(?<=(?<!_)__)([^_])(?=__(?!_))

Examples:

>>> regex = re.compile(r'(?<=(?<!_)__)([^_])(?=__(?!_))')
>>> regex.findall('__a__ ')
['a']
>>> regex.findall('___a___')
[]
>>> regex.findall('__a__d___b___e__c__')
['a', 'c']
>>> regex.findall('__a__c__')
['a', 'c']

You said you wanted overlapping matches, but if you would not like the c to match in __a__c__, use the following (which was my original answer):

(?<!_)__([^_])__(?!_)