2007-01-22 14:43:50 +01:00
|
|
|
require 'chunks/chunk'
|
|
|
|
|
|
|
|
# This wiki chunk matches arbitrary URIs, using patterns from the Ruby URI modules.
|
|
|
|
# It parses out a variety of fields that could be used by renderers to format
|
|
|
|
# the links in various ways (shortening domain names, hiding email addresses)
|
|
|
|
# It matches email addresses and host.com.au domains without schemes (http://)
|
|
|
|
# but adds these on as required.
|
|
|
|
#
|
|
|
|
# The heuristic used to match a URI is designed to err on the side of caution.
|
|
|
|
# That is, it is more likely to not autolink a URI than it is to accidently
|
|
|
|
# autolink something that is not a URI. The reason behind this is it is easier
|
|
|
|
# to force a URI link by prefixing 'http://' to it than it is to escape and
|
|
|
|
# incorrectly marked up non-URI.
|
|
|
|
#
|
|
|
|
# I'm using a part of the [ISO 3166-1 Standard][iso3166] for country name suffixes.
|
|
|
|
# The generic names are from www.bnoack.com/data/countrycode2.html)
|
|
|
|
# [iso3166]: http://geotags.com/iso3166/
|
|
|
|
|
|
|
|
class URIChunk < Chunk::Abstract
|
|
|
|
include URI::REGEXP::PATTERN
|
|
|
|
|
|
|
|
# this condition is to get rid of pesky warnings in tests
|
|
|
|
unless defined? URIChunk::INTERNET_URI_REGEXP
|
|
|
|
|
|
|
|
GENERIC = 'aero|biz|com|coop|edu|gov|info|int|mil|museum|name|net|org'
|
|
|
|
|
|
|
|
COUNTRY = 'ad|ae|af|ag|ai|al|am|an|ao|aq|ar|as|at|au|aw|az|ba|bb|bd|be|' +
|
|
|
|
'bf|bg|bh|bi|bj|bm|bn|bo|br|bs|bt|bv|bw|by|bz|ca|cc|cf|cd|cg|ch|ci|ck|cl|' +
|
|
|
|
'cm|cn|co|cr|cs|cu|cv|cx|cy|cz|de|dj|dk|dm|do|dz|ec|ee|eg|eh|er|es|et|fi|' +
|
|
|
|
'fj|fk|fm|fo|fr|fx|ga|gb|gd|ge|gf|gh|gi|gl|gm|gn|gp|gq|gr|gs|gt|gu|gw|gy|' +
|
|
|
|
'hk|hm|hn|hr|ht|hu|id|ie|il|in|io|iq|ir|is|it|jm|jo|jp|ke|kg|kh|ki|km|kn|' +
|
|
|
|
'kp|kr|kw|ky|kz|la|lb|lc|li|lk|lr|ls|lt|lu|lv|ly|ma|mc|md|mg|mh|mk|ml|mm|' +
|
|
|
|
'mn|mo|mp|mq|mr|ms|mt|mu|mv|mw|mx|my|mz|na|nc|ne|nf|ng|ni|nl|no|np|nr|nt|' +
|
|
|
|
'nu|nz|om|pa|pe|pf|pg|ph|pk|pl|pm|pn|pr|pt|pw|py|qa|re|ro|ru|rw|sa|sb|sc|' +
|
|
|
|
'sd|se|sg|sh|si|sj|sk|sl|sm|sn|so|sr|st|su|sv|sy|sz|tc|td|tf|tg|th|tj|tk|' +
|
|
|
|
'tm|tn|to|tp|tr|tt|tv|tw|tz|ua|ug|uk|um|us|uy|uz|va|vc|ve|vg|vi|vn|vu|wf|' +
|
|
|
|
'ws|ye|yt|yu|za|zm|zr|zw'
|
|
|
|
# These are needed otherwise HOST will match almost anything
|
|
|
|
TLDS = "(?:#{GENERIC}|#{COUNTRY})"
|
|
|
|
|
|
|
|
# Redefine USERINFO so that it must have non-zero length
|
|
|
|
USERINFO = "(?:[#{UNRESERVED};:&=+$,]|#{ESCAPED})+"
|
|
|
|
|
|
|
|
# unreserved_no_ending = alphanum | mark, but URI_ENDING [)!] excluded
|
|
|
|
UNRESERVED_NO_ENDING = "-_.~*'(#{ALNUM}"
|
|
|
|
|
|
|
|
# this ensures that query or fragment do not end with URI_ENDING
|
|
|
|
# and enable us to use a much simpler self.pattern Regexp
|
|
|
|
|
|
|
|
# uric_no_ending = reserved | unreserved_no_ending | escaped
|
|
|
|
URIC_NO_ENDING = "(?:[#{UNRESERVED_NO_ENDING}#{RESERVED}]|#{ESCAPED})"
|
|
|
|
# query = *uric
|
|
|
|
QUERY = "#{URIC_NO_ENDING}*"
|
|
|
|
# fragment = *uric
|
|
|
|
FRAGMENT = "#{URIC_NO_ENDING}*"
|
|
|
|
|
|
|
|
# DOMLABEL is defined in the ruby uri library, TLDS is defined above
|
|
|
|
INTERNET_HOSTNAME = "(?:#{DOMLABEL}\\.)+#{TLDS}"
|
|
|
|
|
|
|
|
# Correct a typo bug in ruby 1.8.x lib/uri/common.rb
|
|
|
|
PORT = '\\d*'
|
|
|
|
|
|
|
|
INTERNET_URI =
|
|
|
|
"(?:(#{SCHEME}):/{0,2})?" + # Optional scheme: (\1)
|
|
|
|
"(?:(#{USERINFO})@)?" + # Optional userinfo@ (\2)
|
|
|
|
"(#{INTERNET_HOSTNAME})" + # Mandatory hostname (\3)
|
|
|
|
"(?::(#{PORT}))?" + # Optional :port (\4)
|
|
|
|
"(#{ABS_PATH})?" + # Optional absolute path (\5)
|
|
|
|
"(?:\\?(#{QUERY}))?" + # Optional ?query (\6)
|
|
|
|
"(?:\\#(#{FRAGMENT}))?" + # Optional #fragment (\7)
|
|
|
|
'(?=\.?(?:\s|\)|\z))' # ends only with optional dot + space or ")"
|
|
|
|
# or end of the string
|
|
|
|
|
|
|
|
SUSPICIOUS_PRECEDING_CHARACTER = '(!|\"\:|\"|\\\'|\]\()?' # any of !, ":, ", ', ](
|
|
|
|
|
|
|
|
INTERNET_URI_REGEXP =
|
2008-11-24 22:53:39 +01:00
|
|
|
# Regexp.new(SUSPICIOUS_PRECEDING_CHARACTER + INTERNET_URI, Regexp::EXTENDED, 'N')
|
|
|
|
Regexp.new(SUSPICIOUS_PRECEDING_CHARACTER + INTERNET_URI, Regexp::EXTENDED)
|
2007-01-22 14:43:50 +01:00
|
|
|
|
|
|
|
end
|
|
|
|
|
|
|
|
def URIChunk.pattern
|
|
|
|
INTERNET_URI_REGEXP
|
|
|
|
end
|
|
|
|
|
|
|
|
attr_reader :user, :host, :port, :path, :query, :fragment, :link_text
|
|
|
|
|
|
|
|
def self.apply_to(content)
|
|
|
|
content.gsub!( self.pattern ) do |matched_text|
|
|
|
|
chunk = self.new($~, content)
|
|
|
|
if chunk.avoid_autolinking?
|
|
|
|
# do not substitute nor register the chunk
|
|
|
|
matched_text
|
|
|
|
else
|
|
|
|
content.add_chunk(chunk)
|
|
|
|
chunk.mask
|
|
|
|
end
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
def initialize(match_data, content)
|
|
|
|
super
|
|
|
|
@link_text = match_data[0]
|
|
|
|
@suspicious_preceding_character = match_data[1]
|
|
|
|
@original_scheme, @user, @host, @port, @path, @query, @fragment = match_data[2..-1]
|
|
|
|
treat_trailing_character
|
|
|
|
@unmask_text = "<a href=\"#{uri}\">#{link_text}</a>"
|
|
|
|
end
|
|
|
|
|
|
|
|
def avoid_autolinking?
|
|
|
|
not @suspicious_preceding_character.nil?
|
|
|
|
end
|
|
|
|
|
|
|
|
def treat_trailing_character
|
|
|
|
# If the last character matched by URI pattern is in ! or ), this may be part of the markup,
|
|
|
|
# not a URL. We should handle it as such. It is possible to do it by a regexp, but
|
|
|
|
# much easier to do programmatically
|
|
|
|
last_char = @link_text[-1..-1]
|
|
|
|
if last_char == ')' or last_char == '!'
|
|
|
|
@trailing_punctuation = last_char
|
|
|
|
@link_text.chop!
|
|
|
|
[@original_scheme, @user, @host, @port, @path, @query, @fragment].compact.last.chop!
|
|
|
|
else
|
|
|
|
@trailing_punctuation = nil
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
def scheme
|
|
|
|
@original_scheme or (@user ? 'mailto' : 'http')
|
|
|
|
end
|
|
|
|
|
|
|
|
def scheme_delimiter
|
|
|
|
scheme == 'mailto' ? ':' : '://'
|
|
|
|
end
|
|
|
|
|
|
|
|
def user_delimiter
|
|
|
|
'@' unless @user.nil?
|
|
|
|
end
|
|
|
|
|
|
|
|
def port_delimiter
|
|
|
|
':' unless @port.nil?
|
|
|
|
end
|
|
|
|
|
|
|
|
def query_delimiter
|
|
|
|
'?' unless @query.nil?
|
|
|
|
end
|
|
|
|
|
|
|
|
def uri
|
|
|
|
[scheme, scheme_delimiter, user, user_delimiter, host, port_delimiter, port, path,
|
|
|
|
query_delimiter, query].compact.join
|
|
|
|
end
|
|
|
|
|
|
|
|
end
|
|
|
|
|
|
|
|
# uri with mandatory scheme but less restrictive hostname, like
|
|
|
|
# http://localhost:2500/blah.html
|
|
|
|
class LocalURIChunk < URIChunk
|
|
|
|
|
|
|
|
unless defined? LocalURIChunk::LOCAL_URI_REGEXP
|
|
|
|
# hostname can be just a simple word like 'localhost'
|
|
|
|
ANY_HOSTNAME = "(?:#{DOMLABEL}\\.)*#{TOPLABEL}\\.?"
|
|
|
|
|
|
|
|
# The basic URI expression as a string
|
|
|
|
# Scheme and hostname are mandatory
|
|
|
|
LOCAL_URI =
|
|
|
|
"(?:(#{SCHEME})://)+" + # Mandatory scheme:// (\1)
|
|
|
|
"(?:(#{USERINFO})@)?" + # Optional userinfo@ (\2)
|
|
|
|
"(#{ANY_HOSTNAME})" + # Mandatory hostname (\3)
|
|
|
|
"(?::(#{PORT}))?" + # Optional :port (\4)
|
|
|
|
"(#{ABS_PATH})?" + # Optional absolute path (\5)
|
|
|
|
"(?:\\?(#{QUERY}))?" + # Optional ?query (\6)
|
|
|
|
"(?:\\#(#{FRAGMENT}))?" + # Optional #fragment (\7)
|
|
|
|
'(?=\.?(?:\s|\)|\z))' # ends only with optional dot + space or ")"
|
|
|
|
# or end of the string
|
|
|
|
|
2008-11-24 22:53:39 +01:00
|
|
|
# LOCAL_URI_REGEXP = Regexp.new(SUSPICIOUS_PRECEDING_CHARACTER + LOCAL_URI, Regexp::EXTENDED, 'N')
|
|
|
|
LOCAL_URI_REGEXP = Regexp.new(SUSPICIOUS_PRECEDING_CHARACTER + LOCAL_URI, Regexp::EXTENDED)
|
2007-01-22 14:43:50 +01:00
|
|
|
end
|
|
|
|
|
|
|
|
def LocalURIChunk.pattern
|
|
|
|
LOCAL_URI_REGEXP
|
|
|
|
end
|
|
|
|
|
|
|
|
end
|