Why is ag printing blank lines from this file?

Question

I want to use ag to print the classes and their methods in a python file. I thought that this would be easy using:

ag --context=0 --nocolor -os '^\s*(def|class)\s+[_A-Za-z]*' prog.py

but for reasons I don't understand this is also matching blank lines. For example, if you make prog.py the following

class MyFavouriteClass
    def __init__

    def __contains__
        blah    
class MyNextFavouriteClass
    def _repr_

    def __iter__

then it returns the full file, including the blank lines, except for the line containing blah. Of course, I can always pipe the output into something else to remove the blank lines but I'd rather get it right the first time.

I suspect that the problem has nothing to do with the regular expression and, instead, that it's a feature of ag's --context, --after and --before flags but I can't find a combination of these that does what I want.

Any ideas?

Answer 1

It's not the --context, but the \s* at the start of your regex pattern. It seems ag doesn't search line-by-line like normal grep, but looks at the whole file in one go (or at least several lines at a time). A bit like this Perl one-liner would:

perl -0777 -ne 'print "$&\n" while /^\s*(def|class)\s+[_A-Za-z]*/msg' ../prog.py

So, since \s matches any whitespace, including newlines, it matches the previous empty line, the newline, spaces in front of the next, and then the def keyword. If you add an empty line before the blah line, it's not printed, since blah doesn't fit the pattern.

To get rid the unwanted match, use /^ *... or /^[ \t]*... instead of /^\s*.... (space+asterisk in the first one)