Hacked By AnonymousFox
# frozen_string_literal: false
module REXML
# If you add a method, keep in mind two things:
# (1) the first argument will always be a list of nodes from which to
# filter. In the case of context methods (such as position), the function
# should return an array with a value for each child in the array.
# (2) all method calls from XML will have "-" replaced with "_".
# Therefore, in XML, "local-name()" is identical (and actually becomes)
# "local_name()"
module Functions
@@available_functions = {}
@@context = nil
@@namespace_context = {}
@@variables = {}
INTERNAL_METHODS = [
:namespace_context,
:namespace_context=,
:variables,
:variables=,
:context=,
:get_namespace,
:send,
]
class << self
def singleton_method_added(name)
unless INTERNAL_METHODS.include?(name)
@@available_functions[name] = true
end
end
end
def Functions::namespace_context=(x) ; @@namespace_context=x ; end
def Functions::variables=(x) ; @@variables=x ; end
def Functions::namespace_context ; @@namespace_context ; end
def Functions::variables ; @@variables ; end
def Functions::context=(value); @@context = value; end
def Functions::text( )
if @@context[:node].node_type == :element
return @@context[:node].find_all{|n| n.node_type == :text}.collect{|n| n.value}
elsif @@context[:node].node_type == :text
return @@context[:node].value
else
return false
end
end
# Returns the last node of the given list of nodes.
def Functions::last( )
@@context[:size]
end
def Functions::position( )
@@context[:index]
end
# Returns the size of the given list of nodes.
def Functions::count( node_set )
node_set.size
end
# Since REXML is non-validating, this method is not implemented as it
# requires a DTD
def Functions::id( object )
end
def Functions::local_name(node_set=nil)
get_namespace(node_set) do |node|
return node.local_name
end
""
end
def Functions::namespace_uri( node_set=nil )
get_namespace( node_set ) {|node| node.namespace}
end
def Functions::name( node_set=nil )
get_namespace( node_set ) do |node|
node.expanded_name
end
end
# Helper method.
def Functions::get_namespace( node_set = nil )
if node_set == nil
yield @@context[:node] if @@context[:node].respond_to?(:namespace)
else
if node_set.respond_to? :each
result = []
node_set.each do |node|
result << yield(node) if node.respond_to?(:namespace)
end
result
elsif node_set.respond_to? :namespace
yield node_set
end
end
end
# A node-set is converted to a string by returning the string-value of the
# node in the node-set that is first in document order. If the node-set is
# empty, an empty string is returned.
#
# A number is converted to a string as follows
#
# NaN is converted to the string NaN
#
# positive zero is converted to the string 0
#
# negative zero is converted to the string 0
#
# positive infinity is converted to the string Infinity
#
# negative infinity is converted to the string -Infinity
#
# if the number is an integer, the number is represented in decimal form
# as a Number with no decimal point and no leading zeros, preceded by a
# minus sign (-) if the number is negative
#
# otherwise, the number is represented in decimal form as a Number
# including a decimal point with at least one digit before the decimal
# point and at least one digit after the decimal point, preceded by a
# minus sign (-) if the number is negative; there must be no leading zeros
# before the decimal point apart possibly from the one required digit
# immediately before the decimal point; beyond the one required digit
# after the decimal point there must be as many, but only as many, more
# digits as are needed to uniquely distinguish the number from all other
# IEEE 754 numeric values.
#
# The boolean false value is converted to the string false. The boolean
# true value is converted to the string true.
#
# An object of a type other than the four basic types is converted to a
# string in a way that is dependent on that type.
def Functions::string( object=@@context[:node] )
if object.respond_to?(:node_type)
case object.node_type
when :attribute
object.value
when :element
string_value(object)
when :document
string_value(object.root)
when :processing_instruction
object.content
else
object.to_s
end
else
case object
when Array
string(object[0])
when Float
if object.nan?
"NaN"
else
integer = object.to_i
if object == integer
"%d" % integer
else
object.to_s
end
end
else
object.to_s
end
end
end
# A node-set is converted to a string by
# returning the concatenation of the string-value
# of each of the children of the node in the
# node-set that is first in document order.
# If the node-set is empty, an empty string is returned.
def Functions::string_value( o )
rv = ""
o.children.each { |e|
if e.node_type == :text
rv << e.to_s
elsif e.node_type == :element
rv << string_value( e )
end
}
rv
end
def Functions::concat( *objects )
concatenated = ""
objects.each do |object|
concatenated << string(object)
end
concatenated
end
# Fixed by Mike Stok
def Functions::starts_with( string, test )
string(string).index(string(test)) == 0
end
# Fixed by Mike Stok
def Functions::contains( string, test )
string(string).include?(string(test))
end
# Kouhei fixed this
def Functions::substring_before( string, test )
ruby_string = string(string)
ruby_index = ruby_string.index(string(test))
if ruby_index.nil?
""
else
ruby_string[ 0...ruby_index ]
end
end
# Kouhei fixed this too
def Functions::substring_after( string, test )
ruby_string = string(string)
return $1 if ruby_string =~ /#{test}(.*)/
""
end
# Take equal portions of Mike Stok and Sean Russell; mix
# vigorously, and pour into a tall, chilled glass. Serves 10,000.
def Functions::substring( string, start, length=nil )
ruby_string = string(string)
ruby_length = if length.nil?
ruby_string.length.to_f
else
number(length)
end
ruby_start = number(start)
# Handle the special cases
return '' if (
ruby_length.nan? or
ruby_start.nan? or
ruby_start.infinite?
)
infinite_length = ruby_length.infinite? == 1
ruby_length = ruby_string.length if infinite_length
# Now, get the bounds. The XPath bounds are 1..length; the ruby bounds
# are 0..length. Therefore, we have to offset the bounds by one.
ruby_start = round(ruby_start) - 1
ruby_length = round(ruby_length)
if ruby_start < 0
ruby_length += ruby_start unless infinite_length
ruby_start = 0
end
return '' if ruby_length <= 0
ruby_string[ruby_start,ruby_length]
end
# UNTESTED
def Functions::string_length( string )
string(string).length
end
# UNTESTED
def Functions::normalize_space( string=nil )
string = string(@@context[:node]) if string.nil?
if string.kind_of? Array
string.collect{|x| string.to_s.strip.gsub(/\s+/um, ' ') if string}
else
string.to_s.strip.gsub(/\s+/um, ' ')
end
end
# This is entirely Mike Stok's beast
def Functions::translate( string, tr1, tr2 )
from = string(tr1)
to = string(tr2)
# the map is our translation table.
#
# if a character occurs more than once in the
# from string then we ignore the second &
# subsequent mappings
#
# if a character maps to nil then we delete it
# in the output. This happens if the from
# string is longer than the to string
#
# there's nothing about - or ^ being special in
# http://www.w3.org/TR/xpath#function-translate
# so we don't build ranges or negated classes
map = Hash.new
0.upto(from.length - 1) { |pos|
from_char = from[pos]
unless map.has_key? from_char
map[from_char] =
if pos < to.length
to[pos]
else
nil
end
end
}
if ''.respond_to? :chars
string(string).chars.collect { |c|
if map.has_key? c then map[c] else c end
}.compact.join
else
string(string).unpack('U*').collect { |c|
if map.has_key? c then map[c] else c end
}.compact.pack('U*')
end
end
def Functions::boolean(object=@@context[:node])
case object
when true, false
object
when Float
return false if object.zero?
return false if object.nan?
true
when Numeric
not object.zero?
when String
not object.empty?
when Array
not object.empty?
else
object ? true : false
end
end
# UNTESTED
def Functions::not( object )
not boolean( object )
end
# UNTESTED
def Functions::true( )
true
end
# UNTESTED
def Functions::false( )
false
end
# UNTESTED
def Functions::lang( language )
lang = false
node = @@context[:node]
attr = nil
until node.nil?
if node.node_type == :element
attr = node.attributes["xml:lang"]
unless attr.nil?
lang = compare_language(string(language), attr)
break
else
end
end
node = node.parent
end
lang
end
def Functions::compare_language lang1, lang2
lang2.downcase.index(lang1.downcase) == 0
end
# a string that consists of optional whitespace followed by an optional
# minus sign followed by a Number followed by whitespace is converted to
# the IEEE 754 number that is nearest (according to the IEEE 754
# round-to-nearest rule) to the mathematical value represented by the
# string; any other string is converted to NaN
#
# boolean true is converted to 1; boolean false is converted to 0
#
# a node-set is first converted to a string as if by a call to the string
# function and then converted in the same way as a string argument
#
# an object of a type other than the four basic types is converted to a
# number in a way that is dependent on that type
def Functions::number(object=@@context[:node])
case object
when true
Float(1)
when false
Float(0)
when Array
number(string(object))
when Numeric
object.to_f
else
str = string(object)
case str.strip
when /\A\s*(-?(?:\d+(?:\.\d*)?|\.\d+))\s*\z/
$1.to_f
else
Float::NAN
end
end
end
def Functions::sum( nodes )
nodes = [nodes] unless nodes.kind_of? Array
nodes.inject(0) { |r,n| r + number(string(n)) }
end
def Functions::floor( number )
number(number).floor
end
def Functions::ceiling( number )
number(number).ceil
end
def Functions::round( number )
number = number(number)
begin
neg = number.negative?
number = number.abs.round
neg ? -number : number
rescue FloatDomainError
number
end
end
def Functions::processing_instruction( node )
node.node_type == :processing_instruction
end
def Functions::send(name, *args)
if @@available_functions[name.to_sym]
super
else
# TODO: Maybe, this is not XPath spec behavior.
# This behavior must be reconsidered.
XPath.match(@@context[:node], name.to_s)
end
end
end
end
Hacked By AnonymousFox1.0, Coded By AnonymousFox