I was using this pattern to match single quoted strings in parser:
"'.+?'"
But I need regex that can find single quoted string with postgres like escape of single qoutes (doubling single qoutes). Need to match something like this:
"'first', 'sec''ond', 't''hi''rd'"
I want to find shortest matches for strings that start and end with single single quotes, so the string above would mean 3 substrings:
'first'
'sec''ond'
't''hi''rd'
Certainly, '(?:[^']|'')*' is the working regex for this: it matches a ' followed with zero or more characters other than ' or double 's followed with a trailing '.
However, to make it more efficient, you can unroll it using the unroll-the-loop technique.
'[^']*(?:''[^']*)*'
See the regex demo and pay attention how many steps it takes for the regexps to find all matches.
The regex can be read as
' - match a '[^']* - then zero or more characters other than '(?:''[^']*)* - then zero or more sequences of '' followed with zero or more characters other than '' - and then match the trailing '.This regex has a linear pattern involving as little backtracking as possible.
Just a note: you can still make your regex work for the current scenario if you add a lookahead checking if there is a , or the end of string after the trailing ':
'.+?'(?=,|$)
^^^^^^^
See the regex demo. However, it is context dependent and less efficient than the unrolled regex.