jekyll 4.0.0.pre.alpha1 → 4.1.1

data/lib/jekyll/drops/drop.rb

@@ -30,7 +30,6 @@ module Jekyll
       # Returns nothing
       def initialize(obj)
         @obj = obj
-        @mutations = {} # only if mutable: true
       end
 
       # Access a method in the Drop or a field in the underlying hash data.
@@ -42,8 +41,8 @@ module Jekyll
       #
       # Returns the value for the given key, or nil if none exists
       def [](key)
-        if self.class.mutable? && @mutations.key?(key)
-          @mutations[key]
+        if self.class.mutable? && mutations.key?(key)
+          mutations[key]
         elsif self.class.invokable? key
           public_send key
         else
@@ -66,11 +65,12 @@ module Jekyll
       # and the key matches a method in which case it raises a
       # DropMutationException.
       def []=(key, val)
-        if respond_to?("#{key}=")
-          public_send("#{key}=", val)
+        setter = "#{key}="
+        if respond_to?(setter)
+          public_send(setter, val)
         elsif respond_to?(key.to_s)
           if self.class.mutable?
-            @mutations[key] = val
+            mutations[key] = val
           else
             raise Errors::DropMutationException, "Key #{key} cannot be set in the drop."
           end
@@ -100,7 +100,7 @@ module Jekyll
       # Returns true if the given key is present
       def key?(key)
         return false if key.nil?
-        return true if self.class.mutable? && @mutations.key?(key)
+        return true if self.class.mutable? && mutations.key?(key)
 
         respond_to?(key) || fallback_data.key?(key)
       end
@@ -113,7 +113,7 @@ module Jekyll
       # Returns an Array of unique keys for content for the Drop.
       def keys
         (content_methods |
-          @mutations.keys |
+          mutations.keys |
           fallback_data.keys).flatten
       end
 
@@ -204,6 +204,12 @@ module Jekyll
         return yield(key) unless block.nil?
         return default unless default.nil?
       end
+
+      private
+
+      def mutations
+        @mutations ||= {}
+      end
     end
   end
 end

data/lib/jekyll/drops/site_drop.rb

@@ -8,7 +8,7 @@ module Jekyll
       mutable false
 
       def_delegator  :@obj, :site_data, :data
-      def_delegators :@obj, :time, :pages, :static_files, :documents, :tags, :categories
+      def_delegators :@obj, :time, :pages, :static_files, :tags, :categories
 
       private def_delegator :@obj, :config, :fallback_data
 
@@ -38,6 +38,16 @@ module Jekyll
         @site_collections ||= @obj.collections.values.sort_by(&:label).map(&:to_liquid)
       end
 
+      # `Site#documents` cannot be memoized so that `Site#docs_to_write` can access the
+      # latest state of the attribute.
+      #
+      # Since this method will be called after `Site#pre_render` hook, the `Site#documents`
+      # array shouldn't thereafter change and can therefore be safely memoized to prevent
+      # additional computation of `Site#documents`.
+      def documents
+        @documents ||= @obj.documents
+      end
+
       # `{{ site.related_posts }}` is how posts can get posts related to
       # them, either through LSI if it's enabled, or through the most
       # recent posts.

data/lib/jekyll/drops/url_drop.rb

@@ -35,46 +35,97 @@ module Jekyll
         category_set.to_a.join("/")
       end
 
+      # Similar to output from #categories, but each category will be downcased and
+      # all non-alphanumeric characters of the category replaced with a hyphen.
+      def slugified_categories
+        Array(@obj.data["categories"]).each_with_object(Set.new) do |category, set|
+          set << Utils.slugify(category.to_s)
+        end.to_a.join("/")
+      end
+
+      # CCYY
       def year
         @obj.date.strftime("%Y")
       end
 
+      # MM: 01..12
       def month
         @obj.date.strftime("%m")
       end
 
+      # DD: 01..31
       def day
         @obj.date.strftime("%d")
       end
 
+      # hh: 00..23
       def hour
         @obj.date.strftime("%H")
       end
 
+      # mm: 00..59
       def minute
         @obj.date.strftime("%M")
       end
 
+      # ss: 00..59
       def second
         @obj.date.strftime("%S")
       end
 
+      # D: 1..31
       def i_day
         @obj.date.strftime("%-d")
       end
 
+      # M: 1..12
       def i_month
         @obj.date.strftime("%-m")
       end
 
+      # MMM: Jan..Dec
       def short_month
         @obj.date.strftime("%b")
       end
 
+      # MMMM: January..December
+      def long_month
+        @obj.date.strftime("%B")
+      end
+
+      # YY: 00..99
       def short_year
         @obj.date.strftime("%y")
       end
 
+      # CCYYw, ISO week year
+      # may differ from CCYY for the first days of January and last days of December
+      def w_year
+        @obj.date.strftime("%G")
+      end
+
+      # WW: 01..53
+      # %W and %U do not comply with ISO 8601-1
+      def week
+        @obj.date.strftime("%V")
+      end
+
+      # d: 1..7 (Monday..Sunday)
+      def w_day
+        @obj.date.strftime("%u")
+      end
+
+      # dd: Mon..Sun
+      def short_day
+        @obj.date.strftime("%a")
+      end
+
+      # ddd: Monday..Sunday
+      def long_day
+        @obj.date.strftime("%A")
+      end
+
+      # DDD: 001..366
       def y_day
         @obj.date.strftime("%j")
       end
@@ -82,7 +133,7 @@ module Jekyll
       private
 
       def fallback_data
-        {}
+        @fallback_data ||= {}
       end
     end
   end

data/lib/jekyll/entry_filter.rb

@@ -3,9 +3,8 @@
 module Jekyll
   class EntryFilter
     attr_reader :site
-    SPECIAL_LEADING_CHARACTERS = [
-      ".", "_", "#", "~",
-    ].freeze
+
+    SPECIAL_LEADING_CHAR_REGEX = %r!\A#{Regexp.union([".", "_", "#", "~"])}!o.freeze
 
     def initialize(site, base_directory = nil)
       @site = site
@@ -31,13 +30,24 @@ module Jekyll
 
     def filter(entries)
       entries.reject do |e|
-        # Reject this entry if it is a symlink.
+        # Reject this entry if it is just a "dot" representation.
+        #   e.g.: '.', '..', '_movies/.', 'music/..', etc
+        next true if e.end_with?(".")
+
+        # Check if the current entry is explicitly included and cache the result
+        included = included?(e)
+
+        # Reject current entry if it is excluded but not explicitly included as well.
+        next true if excluded?(e) && !included
+
+        # Reject current entry if it is a symlink.
         next true if symlink?(e)
-        # Do not reject this entry if it is included.
-        next false if included?(e)
 
-        # Reject this entry if it is special, a backup file, or excluded.
-        special?(e) || backup?(e) || excluded?(e)
+        # Do not reject current entry if it is explicitly included.
+        next false if included
+
+        # Reject current entry if it is special or a backup file.
+        special?(e) || backup?(e)
       end
     end
 
@@ -47,12 +57,12 @@ module Jekyll
     end
 
     def special?(entry)
-      SPECIAL_LEADING_CHARACTERS.include?(entry[0..0]) ||
-        SPECIAL_LEADING_CHARACTERS.include?(File.basename(entry)[0..0])
+      SPECIAL_LEADING_CHAR_REGEX.match?(entry) ||
+        SPECIAL_LEADING_CHAR_REGEX.match?(File.basename(entry))
     end
 
     def backup?(entry)
-      entry[-1..-1] == "~"
+      entry.end_with?("~")
     end
 
     def excluded?(entry)
@@ -86,40 +96,24 @@ module Jekyll
       )
     end
 
-    # --
-    # Check if an entry matches a specific pattern and return true,false.
-    # Returns true if path matches against any glob pattern.
-    # --
-    def glob_include?(enum, entry)
-      entry_path = Pathutil.new(site.in_source_dir).join(entry)
-      enum.any? do |exp|
-        # Users who send a Regexp knows what they want to
-        # exclude, so let them send a Regexp to exclude files,
-        # we will not bother caring if it works or not, it's
-        # on them at this point.
-
-        if exp.is_a?(Regexp)
-          entry_path =~ exp
-
+    # Check if an entry matches a specific pattern.
+    # Returns true if path matches against any glob pattern, else false.
+    def glob_include?(enumerator, entry)
+      entry_with_source = PathManager.join(site.source, entry)
+      entry_is_directory = File.directory?(entry_with_source)
+
+      enumerator.any? do |pattern|
+        case pattern
+        when String
+          pattern_with_source = PathManager.join(site.source, pattern)
+
+          File.fnmatch?(pattern_with_source, entry_with_source) ||
+            entry_with_source.start_with?(pattern_with_source) ||
+            (pattern_with_source == "#{entry_with_source}/" if entry_is_directory)
+        when Regexp
+          pattern.match?(entry_with_source)
         else
-          item = Pathutil.new(site.in_source_dir).join(exp)
-
-          # If it's a directory they want to exclude, AKA
-          # ends with a "/" then we will go on to check and
-          # see if the entry falls within that path and
-          # exclude it if that's the case.
-
-          if entry.end_with?("/")
-            entry_path.in_path?(
-              item
-            )
-
-          else
-            File.fnmatch?(item, entry_path) ||
-              entry_path.to_path.start_with?(
-                item
-              )
-          end
+          false
         end
       end
     end

data/lib/jekyll/excerpt.rb

@@ -10,7 +10,7 @@ module Jekyll
 
     def_delegators :@doc,
                    :site, :name, :ext, :extname,
-                   :collection, :related_posts,
+                   :collection, :related_posts, :type,
                    :coffeescript_file?, :yaml_file?,
                    :url, :next_doc, :previous_doc
 
@@ -56,7 +56,7 @@ module Jekyll
     #
     # Returns true if the string passed in
     def include?(something)
-      (output&.include?(something)) || content.include?(something)
+      output&.include?(something) || content.include?(something)
     end
 
     # The UID for this doc (useful in feeds).
@@ -155,7 +155,7 @@ module Jekyll
       tag_names.flatten!
       tag_names.reverse_each do |tag_name|
         next unless liquid_block?(tag_name)
-        next if head =~ endtag_regex_stash(tag_name)
+        next if endtag_regex_stash(tag_name).match?(head)
 
         modified = true
         head << "\n{% end#{tag_name} %}"

data/lib/jekyll/filters.rb

@@ -121,8 +121,20 @@ module Jekyll
     # input - The String on which to operate.
     #
     # Returns the Integer word count.
-    def number_of_words(input)
-      input.split.length
+    def number_of_words(input, mode = nil)
+      cjk_charset = '\p{Han}\p{Katakana}\p{Hiragana}\p{Hangul}'
+      cjk_regex = %r![#{cjk_charset}]!o
+      word_regex = %r![^#{cjk_charset}\s]+!o
+
+      case mode
+      when "cjk"
+        input.scan(cjk_regex).length + input.scan(word_regex).length
+      when "auto"
+        cjk_count = input.scan(cjk_regex).length
+        cjk_count.zero? ? input.split.length : cjk_count + input.scan(word_regex).length
+      else
+        input.split.length
+      end
     end
 
     # Join an array of things into a string by separating with commas and the
@@ -161,13 +173,15 @@ module Jekyll
 
     # Filter an array of objects
     #
-    # input - the object array
-    # property - property within each object to filter by
-    # value - desired value
+    # input    - the object array.
+    # property - the property within each object to filter by.
+    # value    - the desired value.
+    #            Cannot be an instance of Array nor Hash since calling #to_s on them returns
+    #            their `#inspect` string object.
     #
     # Returns the filtered array of objects
     def where(input, property, value)
-      return input if property.nil? || value.nil?
+      return input if !property || value.is_a?(Array) || value.is_a?(Hash)
       return input unless input.respond_to?(:select)
 
       input    = input.values if input.is_a?(Hash)
@@ -182,8 +196,8 @@ module Jekyll
       # stash or retrive results to return
       @where_filter_cache[input_id][property][value] ||= begin
         input.select do |object|
-          Array(item_property(object, property)).map!(&:to_s).include?(value.to_s)
-        end || []
+          compare_property_vs_target(item_property(object, property), value)
+        end.to_a
       end
     end
 
@@ -208,6 +222,66 @@ module Jekyll
       end || []
     end
 
+    # Search an array of objects and returns the first object that has the queried attribute
+    # with the given value or returns nil otherwise.
+    #
+    # input    - the object array.
+    # property - the property within each object to search by.
+    # value    - the desired value.
+    #            Cannot be an instance of Array nor Hash since calling #to_s on them returns
+    #            their `#inspect` string object.
+    #
+    # Returns the found object or nil
+    #
+    # rubocop:disable Metrics/CyclomaticComplexity
+    def find(input, property, value)
+      return input if !property || value.is_a?(Array) || value.is_a?(Hash)
+      return input unless input.respond_to?(:find)
+
+      input    = input.values if input.is_a?(Hash)
+      input_id = input.hash
+
+      # implement a hash based on method parameters to cache the end-result for given parameters.
+      @find_filter_cache ||= {}
+      @find_filter_cache[input_id] ||= {}
+      @find_filter_cache[input_id][property] ||= {}
+
+      # stash or retrive results to return
+      # Since `enum.find` can return nil or false, we use a placeholder string "<__NO MATCH__>"
+      #   to validate caching.
+      result = @find_filter_cache[input_id][property][value] ||= begin
+        input.find do |object|
+          compare_property_vs_target(item_property(object, property), value)
+        end || "<__NO MATCH__>"
+      end
+      return nil if result == "<__NO MATCH__>"
+
+      result
+    end
+    # rubocop:enable Metrics/CyclomaticComplexity
+
+    # Searches an array of objects against an expression and returns the first object for which
+    # the expression evaluates to true, or returns nil otherwise.
+    #
+    # input - the object array
+    # variable - the variable to assign each item to in the expression
+    # expression - a Liquid comparison expression passed in as a string
+    #
+    # Returns the found object or nil
+    def find_exp(input, variable, expression)
+      return input unless input.respond_to?(:find)
+
+      input = input.values if input.is_a?(Hash)
+
+      condition = parse_condition(expression)
+      @context.stack do
+        input.find do |object|
+          @context[variable] = object
+          condition.evaluate(@context)
+        end
+      end
+    end
+
     # Convert the input into integer
     #
     # input - the object string
@@ -323,26 +397,66 @@ module Jekyll
         .map!(&:last)
     end
 
+    # `where` filter helper
+    #
+    # rubocop:disable Metrics/PerceivedComplexity
+    def compare_property_vs_target(property, target)
+      case target
+      when NilClass
+        return true if property.nil?
+      when Liquid::Expression::MethodLiteral # `empty` or `blank`
+        target = target.to_s
+        return true if property == target || Array(property).join == target
+      else
+        target = target.to_s
+        if property.is_a? String
+          return true if property == target
+        else
+          Array(property).each do |prop|
+            return true if prop.to_s == target
+          end
+        end
+      end
+
+      false
+    end
+
+    # rubocop:enable Metrics/PerceivedComplexity
+
     def item_property(item, property)
       @item_property_cache ||= {}
       @item_property_cache[property] ||= {}
       @item_property_cache[property][item] ||= begin
-        if item.respond_to?(:to_liquid)
-          property.to_s.split(".").reduce(item.to_liquid) do |subvalue, attribute|
-            parse_sort_input(subvalue[attribute])
-          end
-        elsif item.respond_to?(:data)
-          parse_sort_input(item.data[property.to_s])
-        else
-          parse_sort_input(item[property.to_s])
-        end
+        property = property.to_s
+        property = if item.respond_to?(:to_liquid)
+                     read_liquid_attribute(item.to_liquid, property)
+                   elsif item.respond_to?(:data)
+                     item.data[property]
+                   else
+                     item[property]
+                   end
+
+        parse_sort_input(property)
+      end
+    end
+
+    def read_liquid_attribute(liquid_data, property)
+      return liquid_data[property] unless property.include?(".")
+
+      property.split(".").reduce(liquid_data) do |data, key|
+        data.respond_to?(:[]) && data[key]
       end
     end
 
+    FLOAT_LIKE   = %r!\A\s*-?(?:\d+\.?\d*|\.\d+)\s*\Z!.freeze
+    INTEGER_LIKE = %r!\A\s*-?\d+\s*\Z!.freeze
+    private_constant :FLOAT_LIKE, :INTEGER_LIKE
+
     # return numeric values as numbers for proper sorting
     def parse_sort_input(property)
-      number_like = %r!\A\s*-?(?:\d+\.?\d*|\.\d+)\s*\Z!
-      return property.to_f if property =~ number_like
+      stringified = property.to_s
+      return property.to_i if INTEGER_LIKE.match?(stringified)
+      return property.to_f if FLOAT_LIKE.match?(stringified)
 
       property
     end
@@ -389,10 +503,14 @@ module Jekyll
     #
     # Returns an instance of Liquid::Condition
     def parse_binary_comparison(parser)
-      parse_comparison(parser).tap do |condition|
-        binary_operator = parser.id?("and") || parser.id?("or")
-        condition.send(binary_operator, parse_comparison(parser)) if binary_operator
+      condition = parse_comparison(parser)
+      first_condition = condition
+      while (binary_operator = parser.id?("and") || parser.id?("or"))
+        child_condition = parse_comparison(parser)
+        condition.send(binary_operator, child_condition)
+        condition = child_condition
       end
+      first_condition
     end
 
     # Generates a Liquid::Condition object from a Liquid::Parser object based on whether the parsed