addressable 2.4.0 → 2.9.0

data/lib/addressable/idna.rb

@@ -1,6 +1,7 @@
-# encoding:utf-8
+# frozen_string_literal: true
+
 #--
-# Copyright (C) 2006-2015 Bob Aman
+# Copyright (C) Bob Aman
 #
 #    Licensed under the Apache License, Version 2.0 (the "License");
 #    you may not use this file except in compliance with the License.

data/lib/addressable/template.rb

@@ -1,6 +1,7 @@
-# encoding:utf-8
+# frozen_string_literal: true
+
 #--
-# Copyright (C) 2006-2015 Bob Aman
+# Copyright (C) Bob Aman
 #
 #    Licensed under the Apache License, Version 2.0 (the "License");
 #    you may not use this file except in compliance with the License.
@@ -35,9 +36,11 @@ module Addressable
       Addressable::URI::CharacterClasses::DIGIT + '_'
 
     var_char =
-      "(?:(?:[#{variable_char_class}]|%[a-fA-F0-9][a-fA-F0-9])+)"
+      "(?>(?:[#{variable_char_class}]|%[a-fA-F0-9][a-fA-F0-9])+)"
     RESERVED =
       "(?:[#{anything}]|%[a-fA-F0-9][a-fA-F0-9])"
+    RESERVED_NO_COMMA =
+      "(?:[#{anything.delete(',')}]|%[a-fA-F0-9][a-fA-F0-9])"
     UNRESERVED =
       "(?:[#{
         Addressable::URI::CharacterClasses::UNRESERVED
@@ -156,7 +159,7 @@ module Addressable
       #   the ::MatchData#[] behavior.
       #
       # @param [#to_int, nil] len
-      #   If provided, an array of values will be returend with the given
+      #   If provided, an array of values will be returned with the given
       #   parameter used as length.
       #
       # @return [Array, String, nil]
@@ -410,7 +413,7 @@ module Addressable
     #   match.captures
     #   #=> ["a", ["b", "c"]]
     def match(uri, processor=nil)
-      uri = Addressable::URI.parse(uri)
+      uri = Addressable::URI.parse(uri) unless uri.is_a?(Addressable::URI)
       mapping = {}
 
       # First, we need to process the pattern, and extract the values.
@@ -488,6 +491,8 @@ module Addressable
     # @param [Hash] mapping The mapping that corresponds to the pattern.
     # @param [#validate, #transform] processor
     #   An optional processor object may be supplied.
+    # @param [Boolean] normalize_values
+    #   Optional flag to enable/disable unicode normalization. Default: true
     #
     # The object should respond to either the <tt>validate</tt> or
     # <tt>transform</tt> messages or both. Both the <tt>validate</tt> and
@@ -518,11 +523,11 @@ module Addressable
     #     "http://example.com/{?one,two,three}/"
     #   ).partial_expand({"one" => "1", "three" => 3}).pattern
     #   #=> "http://example.com/?one=1{&two}&three=3"
-    def partial_expand(mapping, processor=nil)
+    def partial_expand(mapping, processor=nil, normalize_values=true)
       result = self.pattern.dup
       mapping = normalize_keys(mapping)
       result.gsub!( EXPRESSION ) do |capture|
-        transform_partial_capture(mapping, capture, processor)
+        transform_partial_capture(mapping, capture, processor, normalize_values)
       end
       return Addressable::Template.new(result)
     end
@@ -533,6 +538,8 @@ module Addressable
     # @param [Hash] mapping The mapping that corresponds to the pattern.
     # @param [#validate, #transform] processor
     #   An optional processor object may be supplied.
+    # @param [Boolean] normalize_values
+    #   Optional flag to enable/disable unicode normalization. Default: true
     #
     # The object should respond to either the <tt>validate</tt> or
     # <tt>transform</tt> messages or both. Both the <tt>validate</tt> and
@@ -583,11 +590,11 @@ module Addressable
     #     ExampleProcessor
     #   ).to_str
     #   #=> Addressable::Template::InvalidTemplateValueError
-    def expand(mapping, processor=nil)
+    def expand(mapping, processor=nil, normalize_values=true)
       result = self.pattern.dup
       mapping = normalize_keys(mapping)
       result.gsub!( EXPRESSION ) do |capture|
-        transform_capture(mapping, capture, processor)
+        transform_capture(mapping, capture, processor, normalize_values)
       end
       return Addressable::URI.parse(result)
     end
@@ -647,50 +654,16 @@ module Addressable
       self.to_regexp.named_captures
     end
 
-    ##
-    # Generates a route result for a given set of parameters.
-    # Should only be used by rack-mount.
-    #
-    # @param params [Hash] The set of parameters used to expand the template.
-    # @param recall [Hash] Default parameters used to expand the template.
-    # @param options [Hash] Either a `:processor` or a `:parameterize` block.
-    #
-    # @api private
-    def generate(params={}, recall={}, options={})
-      merged = recall.merge(params)
-      if options[:processor]
-        processor = options[:processor]
-      elsif options[:parameterize]
-        # TODO: This is sending me into fits trying to shoe-horn this into
-        # the existing API. I think I've got this backwards and processors
-        # should be a set of 4 optional blocks named :validate, :transform,
-        # :match, and :restore. Having to use a singleton here is a huge
-        # code smell.
-        processor = Object.new
-        class <<processor
-          attr_accessor :block
-          def transform(name, value)
-            block.call(name, value)
-          end
-        end
-        processor.block = options[:parameterize]
-      else
-        processor = nil
-      end
-      result = self.expand(merged, processor)
-      result.to_s if result
-    end
-
   private
     def ordered_variable_defaults
       @ordered_variable_defaults ||= begin
         expansions, _ = parse_template_pattern(pattern)
-        expansions.map do |capture|
+        expansions.flat_map do |capture|
           _, _, varlist = *capture.match(EXPRESSION)
           varlist.split(',').map do |varspec|
             varspec[VARSPEC, 1]
           end
-        end.flatten
+        end
       end
     end
 
@@ -704,6 +677,8 @@ module Addressable
     #   The expression to expand
     # @param [#validate, #transform] processor
     #   An optional processor object may be supplied.
+    # @param [Boolean] normalize_values
+    #   Optional flag to enable/disable unicode normalization. Default: true
     #
     # The object should respond to either the <tt>validate</tt> or
     # <tt>transform</tt> messages or both. Both the <tt>validate</tt> and
@@ -718,56 +693,36 @@ module Addressable
     # after sending the value to the transform method.
     #
     # @return [String] The expanded expression
-    def transform_partial_capture(mapping, capture, processor = nil)
+    def transform_partial_capture(mapping, capture, processor = nil,
+                                  normalize_values = true)
       _, operator, varlist = *capture.match(EXPRESSION)
 
-      vars = varlist.split(',')
+      vars = varlist.split(",")
 
-      if '?' == operator
+      if operator == "?"
         # partial expansion of form style query variables sometimes requires a
         # slight reordering of the variables to produce a valid url.
         first_to_expand = vars.find { |varspec|
           _, name, _ =  *varspec.match(VARSPEC)
-          mapping.key? name
+          mapping.key?(name) && !mapping[name].nil?
         }
 
         vars = [first_to_expand] + vars.reject {|varspec| varspec == first_to_expand}  if first_to_expand
       end
 
-      vars
-        .zip(operator_sequence(operator).take(vars.length))
-        .reduce("") do |acc, (varspec, op)|
+      vars.
+        inject("".dup) do |acc, varspec|
           _, name, _ =  *varspec.match(VARSPEC)
-
-          acc << if mapping.key? name
-                   transform_capture(mapping, "{#{op}#{varspec}}", processor)
-                 else
-                   "{#{op}#{varspec}}"
-                 end
-      end
-    end
-
-    ##
-    # Creates a lazy Enumerator of the operators that should be used to expand
-    # variables in a varlist starting with `operator`. For example, an operator
-    # `"?"` results in the sequence `"?","&","&"...`
-    #
-    # @param [String] operator from which to generate a sequence
-    #
-    # @return [Enumerator] sequence of operators
-    def operator_sequence(operator)
-      rest_operator = if "?" == operator
-                        "&"
-                      else
-                        operator
-                      end
-      head_operator = operator
-
-      Enumerator.new do |y|
-        y << head_operator.to_s
-        while true
-          y << rest_operator.to_s
-        end
+          next_val = if mapping.key? name
+                       transform_capture(mapping, "{#{operator}#{varspec}}",
+                                         processor, normalize_values)
+                     else
+                       "{#{operator}#{varspec}}"
+                     end
+          # If we've already expanded at least one '?' operator with non-empty
+          # value, change to '&'
+          operator = "&" if (operator == "?") && (next_val != "")
+          acc << next_val
       end
     end
 
@@ -780,6 +735,9 @@ module Addressable
     #   The expression to replace
     # @param [#validate, #transform] processor
     #   An optional processor object may be supplied.
+    # @param [Boolean] normalize_values
+    #   Optional flag to enable/disable unicode normalization. Default: true
+    #
     #
     # The object should respond to either the <tt>validate</tt> or
     # <tt>transform</tt> messages or both. Both the <tt>validate</tt> and
@@ -794,7 +752,8 @@ module Addressable
     # after sending the value to the transform method.
     #
     # @return [String] The expanded expression
-    def transform_capture(mapping, capture, processor=nil)
+    def transform_capture(mapping, capture, processor=nil,
+                          normalize_values=true)
       _, operator, varlist = *capture.match(EXPRESSION)
       return_value = varlist.split(',').inject([]) do |acc, varspec|
         _, name, modifier = *varspec.match(VARSPEC)
@@ -814,7 +773,7 @@ module Addressable
               "Can't convert #{value.class} into String or Array."
           end
 
-          value = normalize_value(value)
+          value = normalize_value(value) if normalize_values
 
           if processor == nil || !processor.respond_to?(:transform)
             # Handle percent escaping
@@ -877,7 +836,9 @@ module Addressable
             end
             if processor.respond_to?(:transform)
               transformed_value = processor.transform(name, value)
-              transformed_value = normalize_value(transformed_value)
+              if normalize_values
+                transformed_value = normalize_value(transformed_value)
+              end
             end
           end
           acc << [name, transformed_value]
@@ -933,25 +894,24 @@ module Addressable
     # operator.
     #
     # @param [Hash, Array, String] value
-    #   Normalizes keys and values with IDNA#unicode_normalize_kc
+    #   Normalizes unicode keys and values with String#unicode_normalize (NFC)
     #
     # @return [Hash, Array, String] The normalized values
     def normalize_value(value)
-      unless value.is_a?(Hash)
-        value = value.respond_to?(:to_ary) ? value.to_ary : value.to_str
-      end
-
       # Handle unicode normalization
-      if value.kind_of?(Array)
-        value.map! { |val| Addressable::IDNA.unicode_normalize_kc(val) }
+      if value.respond_to?(:to_ary)
+        value.to_ary.map! { |val| normalize_value(val) }
       elsif value.kind_of?(Hash)
         value = value.inject({}) { |acc, (k, v)|
-          acc[Addressable::IDNA.unicode_normalize_kc(k)] =
-            Addressable::IDNA.unicode_normalize_kc(v)
+          acc[normalize_value(k)] = normalize_value(v)
           acc
         }
       else
-        value = Addressable::IDNA.unicode_normalize_kc(value)
+        value = value.to_s if !value.kind_of?(String)
+        if value.encoding != Encoding::UTF_8
+          value = value.dup.force_encoding(Encoding::UTF_8)
+        end
+        value = value.unicode_normalize(:nfc)
       end
       value
     end
@@ -979,15 +939,35 @@ module Addressable
       end
     end
 
+    ##
+    # Generates the <tt>Regexp</tt> that parses a template pattern. Memoizes the
+    # value if template processor not set (processors may not be deterministic)
+    #
+    # @param [String] pattern The URI template pattern.
+    # @param [#match] processor The template processor to use.
+    #
+    # @return [Array, Regexp]
+    #   An array of expansion variables nad a regular expression which may be
+    #   used to parse a template pattern
+    def parse_template_pattern(pattern, processor = nil)
+      if processor.nil? && pattern == @pattern
+        @cached_template_parse ||=
+          parse_new_template_pattern(pattern, processor)
+      else
+        parse_new_template_pattern(pattern, processor)
+      end
+    end
+
     ##
     # Generates the <tt>Regexp</tt> that parses a template pattern.
     #
     # @param [String] pattern The URI template pattern.
     # @param [#match] processor The template processor to use.
     #
-    # @return [Regexp]
-    #   A regular expression which may be used to parse a template pattern.
-    def parse_template_pattern(pattern, processor=nil)
+    # @return [Array, Regexp]
+    #   An array of expansion variables nad a regular expression which may be
+    #   used to parse a template pattern
+    def parse_new_template_pattern(pattern, processor = nil)
       # Escape the pattern. The two gsubs restore the escaped curly braces
       # back to their original form. Basically, escape everything that isn't
       # within an expansion.
@@ -1007,7 +987,8 @@ module Addressable
         _, operator, varlist = *expansion.match(EXPRESSION)
         leader = Regexp.escape(LEADERS.fetch(operator, ''))
         joiner = Regexp.escape(JOINERS.fetch(operator, ','))
-        combined = varlist.split(',').map do |varspec|
+        varspecs = varlist.split(',')
+        combined = varspecs.map do |varspec|
           _, name, modifier = *varspec.match(VARSPEC)
 
           result = processor && processor.respond_to?(:match) ? processor.match(name) : nil
@@ -1033,7 +1014,15 @@ module Addressable
               "#{ UNRESERVED }*?"
             end
             if modifier == '*'
-              "(?<#{name}>#{group}(?:#{joiner}?#{group})*)?"
+              seg = case operator
+                    when '+', '#' then "#{RESERVED_NO_COMMA}*+"
+                    else group
+                    end
+              joiner_pattern = operator.nil? ? joiner : "#{joiner}?"
+              "(?<#{name}>#{seg}(?:#{joiner_pattern}#{seg})*)?"
+            elsif varspecs.size > 1 && (operator == '+' || operator == '#') &&
+                  varspec != varspecs.last
+              "(?<#{name}>#{RESERVED_NO_COMMA}*+)?"
             else
               "(?<#{name}>#{group})?"
             end
@@ -1043,7 +1032,7 @@ module Addressable
       end
 
       # Ensure that the regular expression matches the whole URI.
-      regexp_string = "^#{regexp_string}$"
+      regexp_string = "\\A#{regexp_string}\\z"
       return expansions, Regexp.new(regexp_string)
     end