Unfloyded/Legamunt
0
0
Fork 0
Code Activity

Enable to recognize most kinds of characters as URL paths (#4941)

Browse source
This commit is contained in:
ふぁぼ原 2017-09-15 01:03:20 +09:00 • committed by Eugen Rochko
parent b39d512ade
commit 3816943e6b
5 changed files with 96 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

42
config/initializers/twitter_regex.rb Normal file
View file

@ -0,0 +1,42 @@
module Twitter
class Regex
REGEXEN[:valid_general_url_path_chars] = /[^\p{White_Space}\(\)\?]/iou
REGEXEN[:valid_url_path_ending_chars] = /[^\p{White_Space}\(\)\?!\*';:=\,\.\$%\[\]\p{Pd}_~&\|@]|(?:#{REGEXEN[:valid_url_balanced_parens]})/iou
REGEXEN[:valid_url_balanced_parens] = /
\(
(?:
#{REGEXEN[:valid_general_url_path_chars]}+
|
# allow one nested level of balanced parentheses
(?:
#{REGEXEN[:valid_general_url_path_chars]}*
\(
#{REGEXEN[:valid_general_url_path_chars]}+
\)
#{REGEXEN[:valid_general_url_path_chars]}*
)
)
\)
/iox
REGEXEN[:valid_url_path] = /(?:
(?:
#{REGEXEN[:valid_general_url_path_chars]}*
(?:#{REGEXEN[:valid_url_balanced_parens]} #{REGEXEN[:valid_general_url_path_chars]}*)*
#{REGEXEN[:valid_url_path_ending_chars]}
)|(?:#{REGEXEN[:valid_general_url_path_chars]}+\/)
)/iox
REGEXEN[:valid_url] = %r{
( # $1 total match
(#{REGEXEN[:valid_url_preceding_chars]}) # $2 Preceeding chracter
( # $3 URL
(https?:\/\/)? # $4 Protocol (optional)
(#{REGEXEN[:valid_domain]}) # $5 Domain(s)
(?::(#{REGEXEN[:valid_port_number]}))? # $6 Port number (optional)
(/#{REGEXEN[:valid_url_path]}*)? # $7 URL Path and anchor
(\?#{REGEXEN[:valid_url_query_chars]}*#{REGEXEN[:valid_url_query_ending_chars]})? # $8 Query String
)
)
}iox
end
end