Re: [Boost-users] [regex] Why pattern anchored at buffer end ($) alsomatch for (\x0c)FF, (\x0d)CR

From: John Maddock (john_at_[hidden])
Date: 2007-08-31 13:10:19

Chandan Nilange wrote:
>> Hello,
>>
>>  As per PCRE /m modifier $ will match the pattern before any newline.
>>  However boost_regex gives back matches before formfeed(\x0c) &
>>  CarriagReturn(\x0d) Same is observed for ^ i.e match after newline.

Yes, it regards any "vertical whitespace" as being suitable for a $ or ^ 
match: however if you have a \r\n sequence, then it will *not* match in the 
middle of that sequence.

I do recall extending the list of characters that will form a line boundary 
recently, but not whether it was for Perl or Unicode compatibility :-(

Ah, here we are: http://unicode.org/unicode/reports/tr18/#Line_Boundaries, 
remember that all the sequences given really do indicate a line boundary in 
most cases.

John.