Regex to match one or two quotes but not three in a row

Question

For the life of me I can't figure this one out.

I need to search the following text, matching only the quotes in bold:

Don't match: """This is a python docstring"""

Match: " This is a regular string "

Match: "" ← That is an empty string

How can I do this with a regular expression?

Here's what I've tried:

Doesn't work:

(?!"")"(?<!"")

Close, but doesn't match double quotes.

Doesn't work:

"(?<!""")|(?!"")"(?<!"")|(?!""")"

I naively thought that I could add the alternates that I don't want but the logic ends up reversed. This one matches everything because all quotes match at least one of the alternates.

(Please note: I'm not running the code, so solutions around using __doc__ won't help, I'm just trying to find and replace in my code editor.)

Answer 1

You can use /(?<!")"{1,2}(?!")/

DEMO

Autopsy:

• (?<!") a negative look-behind for the literal ". The match cannot have this character in front
• "{1,2} the literal " matched once or twice
• (?!") a negative look-ahead for the literal ". The match cannot have this character after

Your first try might've failed because (?!") is a negative look-ahead, and (?<!") is a negative look-behind. It makes no sense to have look-aheads before your match, or look-behinds after your match.

Answer 2

I realized that my original problem description was actually slightly wrong. That is, I need to actually only match a single quote character, unless if it's part of a group of 3 quote characters.

The difference is that this is desirable for editing so that I can find and replace with '. If I match "one or two quotes" then I can't automatically replace with a single character.

I came up with this modification to h20000000's answer that satisfies that case:

(?<!"")(?<=(?!""").)"(?!"")

In the demo, you can see that the "" are matched individually, instead of as a group.

This works very similarly to the other answer, except:

• it only matches a single "

• that leaves us with matching everything we want except it still matches the middle quotes of a """:

  

• Finally, adding the (?<=(?!""").) excludes that case specifically, by saying "look back one character, then fail the match if the next three characters are """):

  

---

I decided not to change the question because I don't want to hijack the answer, but I think this can be a useful addition.