RegExp - how to negate pattern

[orbs]

i admit, i suck at RegExp. ever so often i take another shot at it, fail miserably to comprehend it, and write a dozen-lines function to do the thing. but this time i'm so close to get it... or so i think.

i need to replace a substring, but only if the substring is not prefixed by alpha-numeric character or underscore.

what i mean - examples:

in the following strings, the substring ABC should be replaced:

"ABC"

" ABC"

"(ABC)"

"+ABC"

in the following strings, the substring ABC should NOT be replaced:

"_ABC"

"AABC"

"1ABC"

what i have so far:

i managed to compose a pattern that does the opposite of what i need - it returns 1 if the prefix is alphanumeric or underscore. this is the pattern:

StringRegExp('ABC','[{1}:alnum:_]ABC')

currently using StringRegExp() for testing purposes, when i get it right, i'll move on to StringRegExpReplace(). 

i need help to negate the pattern. thanks for any hints...

 

EDIT: typo.

Edited November 5, 2014 by orbs

[jchd]

Try this:

; using a negative look-behind pattern:
Local $aTest = ["ABC", " ABC", "(ABC)", "+ABC", "_ABC", "AABC", "1ABC"], $t

For $s In $aTest
    $t = StringRegExpReplace($s, "(?<!\w)(ABC)", 'xxx')
    ConsoleWrite("'" & $s & "' became '" & $t & "'" & @LF)
Next

[orbs]

that dark magic spell you cast... seems to do it! thanks!

now the hard part - me trying to figure out what exactly is that you did... 

[jchd]

Feed the pattern into http://regex101.com/ and the magic rabbit exhibits the answer:

/(?<!w)(ABC)/

• (?<!w) Negative Lookbehind - Assert that it is impossible to match the regex below
  • w match any word character [a-zA-Z0-9_]
• 1st Capturing group (ABC)
  • ABC matches the characters ABC literally (case sensitive)