I am trying to select only capital letters in polytonic Greek text using regex. The specific application is PHP, but I had trouble with it so I started playing around with it in RegExr:

https://regexr.com/6ellt

([Α-ΩΗΙΟΥΩᾼῌῼΡΆΈΉΊΌΎΏᾺῈῊῚῸῪῺἈἘἨἸὈὨᾈᾘᾨἌἜἬἼὌὬᾌᾜᾬἊἚἪἺὊὪᾊᾚᾪἎἮἾὮᾎᾞᾮἉἙἩἹὉὙὩᾉᾙᾩῬἍἝἭἽὍὝὭᾍᾝᾭἋἛἫἻὋὛὫᾋᾛᾫἏἯἿὟὯᾏᾟᾯΪΫᾹῙῩᾸῘῨ])

When the JavaScript engine is selected, the behaviour is as expected. However, if I select PCRE not only are capital letters selected, but also a bunch of seemingly random lowercase letters.

Can anyone shed some light on what is going on here? Is this a bug? Is there a way to get the desired result using the PCRE engine?

CodePudding user response：

You need to tell the PCRE regex engine the input is to be parsed as a Unicode string.

In a PCRE regex, you can prepend the pattern with a (*UTF) verb. The (*UTF)[Α-ΩΗΙΟΥΩᾼῌῼΡΆΈΉΊΌΎΏᾺῈῊῚῸῪῺἈἘἨἸὈὨᾈᾘᾨἌἜἬἼὌὬᾌᾜᾬἊἚἪἺὊὪᾊᾚᾪἎἮἾὮᾎᾞᾮἉἙἩἹὉὙὩᾉᾙᾩῬἍἝἭἽὍὝὭᾍᾝᾭἋἛἫἻὋὛὫᾋᾛᾫἏἯἿὟὯᾏᾟᾯΪΫᾹῙῩᾸῘῨ] highights the correct matches.

However, you can also make it a bit shorter with

(*UTF)(?=\p{Lu})\p{Greek}

Here,

• (*UTF) - a PCRE verb telling the PCRE engine the input is a Unicode string
• (?=\p{Lu}) - a positive lookahead requiring the next char to be an uppercase char
• \p{Greek} - a Greek char.

Note in case there is a u flag support in your PCRE implementation, it is most probably the way to go (as in PHP, /(?=\p{Lu})\p{Greek}/u).