It seems likely, looking at his regex, that he's insecure about the notion of "ITEM" being capitalized uniformly over his full data set. The sample he provided us is uniform, but why else would he go to all the trouble of creating alternations like (?:Item|ITEM) several times? If he's unable to depend on an all-caps "ITEM" as a delimiter, your solution won't be any more robust than his current one.