Delphi RegEX library and unicode characters

Question

If one uses \b for a word boundary, it seems it understands only ASCII alphabet for example the pattern

\bM\b will match aaaa M bbbbbb

but if I have

aaaaa Mädchen 

it will too, because it considers ä to be an end of word.

Are there any flags to set for this regexp lib to accept Unicode strings too? It seems very unlikely that this lib would be so primitive but it is not in the options

TRegExOption = (roNone, roIgnoreCase, roMultiLine, roExplicitCapture,
roCompiled, roSingleLine, roIgnorePatternSpace);

Answer 1

According to regular-expressions.info, Delphi regex lib is based on PCRE and the predefined character class \w in PCRE is only ASCII based, therefore \b is also only ASCII based.

Answer 2

You can use lookaround to make your own word boundaries to fit your preferred definition of a "word". E.g. if you want to match "M" as a word and treat all Unicode letters, numbers and marks as word characters, use:

(?<![\pL\pN\pM])M(?![\pL\pN\pM])