Regex to match a string with an even number of quotes

Question

I came up with: ([^"]*["][^"]*["][^"]*)*

It works in all cases except against the empty string. I thought it would work because the last star matches the previous token zero or more times.

Any ideas?

Also if there's a much better way of doing this please let me know and explain it in detail.

The solution must be a regex as the place where it will be used is a hook which requires a regex.

It has to match a string without quotes as well, as zero is an even number

Answer 1

Seems like regexp isn't the right tool for the job. Edit: However, you seem be restricted to it. This doesn't answer your question given that constraint, but will work great without it.

Just iterate over your string and count. C example:

bool hasEvenNumberOfQuotes(const char *str)
{
    bool even = true;

    while(*str != '\0')
    {
        if(*str == '"')
            even = !even;

        ++str;
    }

    return even;
}

Answer 2

Try this expression:

^(?:[^"]+|"[^"]*")*$

It matches a sequence that consists of either any character other than quotes ([^"]+) or a pair of quotes with any character other than quoted between ("[^"]*"). And the * quantifier takes the empty string into account.

Answer 3

import re

def hasPairedQuotes(s):
    stripped = re.sub('[^"]', "", s)
    return len(stripped) % 2 == 0

>>> hasPairedQuotes("")
True
>>> hasPairedQuotes('""')
True
>>> hasPairedQuotes('"""')
False
>>> hasPairedQuotes('"Hello world!""')
False
>>> hasPairedQuotes('"Hello world!"')
True

---

Fine you want a regexp, here's a regexp: ^[^"]*("[^"]*")*[^"]*$... but I think the difference in legibility and maintainability speaks for itself.

>>> re.match(r'^[^"]*("[^"]*"[^"])*$', 'Hello ""')
<_sre.SRE_Match object at 0xb7cc0ce0>
>>> re.match(r'^[^"]*("[^"]*"[^"])*$', 'Hello "" "')
>>>

Answer 4

Based off your regexp:

([^"]*["][^"]*["][^"]*)*

Add line anchors:

^([^"]*["][^"]*["][^"]*)*$

Add possibility to match a non-":

^([^"]*["][^"]*["][^"]*|[^"]?)*$

This last step allows nothing to be matched, or a character to be consumed. This permits strings lacking a " to be matched. Note that the line anchors are needed, otherwise substring(s) will be matched to this.

Bonus: Prevent group backreferencing (naming/numbering groups may slow down the regexp engine a tiny bit):

^(?:[^"]*["][^"]*["][^"]*|[^"]?)*$

Answer 5

Hi,

your regex should match the completely empty string, but not e.g. a string consisting of a single space, because your regex states that if the string is not completely empty, it needs to contain at least one double quote. This is because of the ["] tokens inside the regex which are not followed by *.

The proper way to think about the needed regular expression is as follows: you want to match (string without double quotes) followed by (double quote) plus (string without double quotes) followed by (double) quote followed by (string without double quotes), and then repeat starting from the first 'followed by' ad infinitum. String without double quotes is [^"]*, so you get (whitespace added for readability):

[^"]* (" [^"]* " [^"]*)*

If you compare this with your regular expression, the first [^"]* has been moved out of the repetition.