addressable 2.3.2 → 2.9.0

data/lib/addressable/idna.rb

@@ -1,6 +1,7 @@
-# encoding:utf-8
+# frozen_string_literal: true
+
 #--
-# Copyright (C) 2006-2011 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) 2006-2011 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
@@ -132,6 +135,7 @@ module Addressable
         self.template.variables
       end
       alias_method :keys, :variables
+      alias_method :names, :variables
 
       ##
       # @return [Array]
@@ -146,6 +150,64 @@ module Addressable
       end
       alias_method :captures, :values
 
+      ##
+      # Accesses captured values by name or by index.
+      #
+      # @param [String, Symbol, Fixnum] key
+      #   Capture index or name. Note that when accessing by with index
+      #   of 0, the full URI will be returned. The intention is to mimic
+      #   the ::MatchData#[] behavior.
+      #
+      # @param [#to_int, nil] len
+      #   If provided, an array of values will be returned with the given
+      #   parameter used as length.
+      #
+      # @return [Array, String, nil]
+      #   The captured value corresponding to the index or name. If the
+      #   value was not provided or the key is unknown, nil will be
+      #   returned.
+      #
+      #   If the second parameter is provided, an array of that length will
+      #   be returned instead.
+      def [](key, len = nil)
+        if len
+          to_a[key, len]
+        elsif String === key or Symbol === key
+          mapping[key.to_s]
+        else
+          to_a[key]
+        end
+      end
+
+      ##
+      # @return [Array]
+      #   Array with the matched URI as first element followed by the captured
+      #   values.
+      def to_a
+        [to_s, *values]
+      end
+
+      ##
+      # @return [String]
+      #   The matched URI as String.
+      def to_s
+        uri.to_s
+      end
+      alias_method :string, :to_s
+
+      # Returns multiple captured values at once.
+      #
+      # @param [String, Symbol, Fixnum] *indexes
+      #   Indices of the captures to be returned
+      #
+      # @return [Array]
+      #   Values corresponding to given indices.
+      #
+      # @see Addressable::Template::MatchData#[]
+      def values_at(*indexes)
+        indexes.map { |i| self[i] }
+      end
+
       ##
       # Returns a <tt>String</tt> representation of the MatchData's state.
       #
@@ -154,6 +216,15 @@ module Addressable
         sprintf("#<%s:%#0x RESULT:%s>",
           self.class.to_s, self.object_id, self.mapping.inspect)
       end
+
+      ##
+      # Dummy method for code expecting a ::MatchData instance
+      #
+      # @return [String] An empty string.
+      def pre_match
+        ""
+      end
+      alias_method :post_match, :pre_match
     end
 
     ##
@@ -166,7 +237,18 @@ module Addressable
       if !pattern.respond_to?(:to_str)
         raise TypeError, "Can't convert #{pattern.class} into String."
       end
-      @pattern = pattern.to_str.freeze
+      @pattern = pattern.to_str.dup.freeze
+    end
+
+    ##
+    # Freeze URI, initializing instance variables.
+    #
+    # @return [Addressable::URI] The frozen URI object.
+    def freeze
+      self.variables
+      self.variable_defaults
+      self.named_captures
+      super
     end
 
     ##
@@ -182,6 +264,26 @@ module Addressable
         self.class.to_s, self.object_id, self.pattern)
     end
 
+    ##
+    # Returns <code>true</code> if the Template objects are equal. This method
+    # does NOT normalize either Template before doing the comparison.
+    #
+    # @param [Object] template The Template to compare.
+    #
+    # @return [TrueClass, FalseClass]
+    #   <code>true</code> if the Templates are equivalent, <code>false</code>
+    #   otherwise.
+    def ==(template)
+      return false unless template.kind_of?(Template)
+      return self.pattern == template.pattern
+    end
+
+    ##
+    # Addressable::Template makes no distinction between `==` and `eql?`.
+    #
+    # @see #==
+    alias_method :eql?, :==
+
     ##
     # Extracts a mapping from the URI using a URI Template pattern.
     #
@@ -311,12 +413,14 @@ 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.
       expansions, expansion_regexp =
         parse_template_pattern(pattern, processor)
+
+      return nil unless uri.to_str.match(expansion_regexp)
       unparsed_values = uri.to_str.scan(expansion_regexp).flatten
 
       if uri.to_str == pattern
@@ -327,6 +431,7 @@ module Addressable
           _, operator, varlist = *expansion.match(EXPRESSION)
           varlist.split(',').each do |varspec|
             _, name, modifier = *varspec.match(VARSPEC)
+            mapping[name] ||= nil
             case operator
             when nil, '+', '#', '/', '.'
               unparsed_value = unparsed_values[index]
@@ -335,12 +440,14 @@ module Addressable
               value = value.split(JOINERS[operator]) if value && modifier == '*'
             when ';', '?', '&'
               if modifier == '*'
-                value = unparsed_values[index].split(JOINERS[operator])
-                value = value.inject({}) do |acc, v|
-                  key, val = v.split('=')
-                  val = "" if val.nil?
-                  acc[key] = val
-                  acc
+                if unparsed_values[index]
+                  value = unparsed_values[index].split(JOINERS[operator])
+                  value = value.inject({}) do |acc, v|
+                    key, val = v.split('=')
+                    val = "" if val.nil?
+                    acc[key] = val
+                    acc
+                  end
                 end
               else
                 if (unparsed_values[index])
@@ -365,10 +472,9 @@ module Addressable
                 value = Addressable::URI.unencode_component(value)
               end
             end
-            if mapping[name] == nil || mapping[name] == value
+            if !mapping.has_key?(name) || mapping[name].nil?
+              # Doesn't exist, set to value (even if value is nil)
               mapping[name] = value
-            else
-              return nil
             end
             index = index + 1
           end
@@ -385,6 +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
@@ -415,10 +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
@@ -429,6 +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
@@ -479,11 +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
@@ -499,6 +610,7 @@ module Addressable
       @variables ||= ordered_variable_defaults.map { |var, val| var }.uniq
     end
     alias_method :keys, :variables
+    alias_method :names, :variables
 
     ##
     # Returns a mapping of variables to their default values specified
@@ -510,17 +622,49 @@ module Addressable
         Hash[*ordered_variable_defaults.reject { |k, v| v.nil? }.flatten]
     end
 
+    ##
+    # Coerces a template into a `Regexp` object. This regular expression will
+    # behave very similarly to the actual template, and should match the same
+    # URI values, but it cannot fully handle, for example, values that would
+    # extract to an `Array`.
+    #
+    # @return [Regexp] A regular expression which should match the template.
+    def to_regexp
+      _, source = parse_template_pattern(pattern)
+      Regexp.new(source)
+    end
+
+    ##
+    # Returns the source of the coerced `Regexp`.
+    #
+    # @return [String] The source of the `Regexp` given by {#to_regexp}.
+    #
+    # @api private
+    def source
+      self.to_regexp.source
+    end
+
+    ##
+    # Returns the named captures of the coerced `Regexp`.
+    #
+    # @return [Hash] The named captures of the `Regexp` given by {#to_regexp}.
+    #
+    # @api private
+    def named_captures
+      self.to_regexp.named_captures
+    end
+
   private
     def ordered_variable_defaults
-      @ordered_variable_defaults ||= (
-        expansions, expansion_regexp = parse_template_pattern(pattern)
-        expansions.map do |capture|
-          _, operator, varlist = *capture.match(EXPRESSION)
+      @ordered_variable_defaults ||= begin
+        expansions, _ = parse_template_pattern(pattern)
+        expansions.flat_map do |capture|
+          _, _, varlist = *capture.match(EXPRESSION)
           varlist.split(',').map do |varspec|
-            name = varspec[VARSPEC, 1]
+            varspec[VARSPEC, 1]
           end
-        end.flatten
-      )
+        end
+      end
     end
 
 
@@ -533,6 +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
@@ -547,21 +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)
-      is_first = true
-      varlist.split(',').inject('') do |acc, varspec|
-        _, name, modifier = *varspec.match(VARSPEC)
-        value = mapping[name]
-        if value
-          operator = '&' if !is_first && operator == '?'
-          acc << transform_capture(mapping, "{#{operator}#{varspec}}", processor)
-        else
-          operator = '&' if !is_first && operator == '?'
-          acc << "{#{operator}#{varspec}}"
-        end
-        is_first = false
-        acc
+
+      vars = varlist.split(",")
+
+      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[name].nil?
+        }
+
+        vars = [first_to_expand] + vars.reject {|varspec| varspec == first_to_expand}  if first_to_expand
+      end
+
+      vars.
+        inject("".dup) do |acc, varspec|
+          _, name, _ =  *varspec.match(VARSPEC)
+          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
 
@@ -574,6 +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
@@ -588,14 +752,19 @@ 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)
         value = mapping[name]
         unless value == nil || value == {}
           allow_reserved = %w(+ #).include?(operator)
-          value = value.to_s if Numeric === value || Symbol === value
+          # Common primitives where the .to_s output is well-defined
+          if Numeric === value || Symbol === value ||
+              value == true || value == false
+            value = value.to_s
+          end
           length = modifier.gsub(':', '').to_i if modifier =~ /^:\d+/
 
           unless (Hash === value) ||
@@ -604,7 +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
@@ -667,7 +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]
@@ -698,7 +869,7 @@ module Addressable
           if v.is_a?(Array) && v.first =~ /=/
             v.join(joiner)
           elsif v.is_a?(Array)
-            v.map{|v| "#{k}=#{v}"}.join(joiner)
+            v.map{|inner_value| "#{k}=#{inner_value}"}.join(joiner)
           else
             "#{k}=#{v}"
           end
@@ -708,7 +879,7 @@ module Addressable
           if v.is_a?(Array) && v.first =~ /=/
             ';' + v.join(";")
           elsif v.is_a?(Array)
-            ';' + v.map{|v| "#{k}=#{v}"}.join(";")
+            ';' + v.map{|inner_value| "#{k}=#{inner_value}"}.join(";")
           else
             v && v != '' ?  ";#{k}=#{v}" : ";#{k}"
           end
@@ -723,25 +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
@@ -769,15 +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.
@@ -797,10 +987,13 @@ module Addressable
         _, operator, varlist = *expansion.match(EXPRESSION)
         leader = Regexp.escape(LEADERS.fetch(operator, ''))
         joiner = Regexp.escape(JOINERS.fetch(operator, ','))
-        leader + varlist.split(',').map do |varspec|
+        varspecs = varlist.split(',')
+        combined = varspecs.map do |varspec|
           _, name, modifier = *varspec.match(VARSPEC)
-          if processor != nil && processor.respond_to?(:match)
-            "(#{ processor.match(name) })"
+
+          result = processor && processor.respond_to?(:match) ? processor.match(name) : nil
+          if result
+            "(?<#{name}>#{ result })"
           else
             group = case operator
             when '+'
@@ -821,16 +1014,25 @@ module Addressable
               "#{ UNRESERVED }*?"
             end
             if modifier == '*'
-              "(#{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
-              "(#{group})?"
+              "(?<#{name}>#{group})?"
             end
           end
         end.join("#{joiner}?")
+        "(?:|#{leader}#{combined})"
       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