Question

I am aware that the regex engine moves significantly faster without having to keep track of backreferences. I am also aware that I can add a ?: at the start of the inside of the brackets to prevent the regex engine from performing a backreference.

However is there anyway I can invert the behavior such that non-matching becomes the default behavior ?(rather like the U flag)

Answer 1

Short answer: no.

PHP uses the PCRE library for parsing regular expressions.

PCRE uses an NFA-based parser which keeps track of backreferences. What you are describing is a DFA-based parser or a Thompson NFA.

I’m not a PHP developer, but the PCRE library does indeed come with a “DFA mode.” Most Linux distros will come equipped with “pcretest.” If you don’t have it, it comes with the PCRE library.

In the CLI:

$ pcretest
  re> /(foo)\1/
data> foofoo
  0: foofoo
  1: foo

Now if we run this with the “-dfa” flag:

$ pcretest -dfa
  re> /(foo)\1/
data> foofoo
Error -16

You may also want to look into “possessive quantifiers” to prevent backtracking.