saper 0.5.1 → 0.5.2

data/lib/saper/items/html.rb

@@ -2,12 +2,17 @@ module Saper
   module Items
     class HTML < Item
       
+      # @todo
       def self.new(item)
         super case item
         when Nokogiri::XML::Element
           item
+        when Nokogiri::XML::Document
+          item
         when Nokogiri::HTML
           item
+        when Document
+          parse(item.body, item.uri, item.charset)
         when Text
           parse(item.to_s)
         when String
@@ -17,14 +22,23 @@ module Saper
         end
       end
       
+      # @todo
       def self.parse(string, uri = nil, charset = nil)
+        if string.empty?
+          raise(InvalidItem, string)
+        end
         begin
+          # Nokogiri adds an additional body tag for strings that don't
+          # have one. Should probably use Nokogiri::HTML.fragment in that
+          # case, but it breaks the search function:
+          # https://github.com/sparklemotion/nokogiri/issues/572
           Nokogiri::HTML.parse(string, uri, charset)
         rescue
           raise(InvalidItem, string)
         end
       end
       
+      # @todo
       def initialize(noko)
         @noko = noko
         # Force UTF-8 encoding
@@ -32,48 +46,79 @@ module Saper
         @noko.document.encoding = 'UTF-8'
       end
       
+      # @todo
+      def dup
+        HTML.new(@noko.dup)
+      end
+      
+      # @todo
+      def ==(other)
+        to_s == other.to_s
+      end
+      
+      # @todo
       def name
         @noko.name
       end
       
+      # @todo
       def [](name)
-        @noko[name]
+        @noko.has_attribute?(name) ? Text.new(@noko[name]) : nil
       end
       
+      # @todo
       def find(xpath)
         find_all(xpath).first
       end
       
+      # @todo
       def find_all(xpath)
         @noko.search(xpath).map { |element| HTML.new(element) }
       end
       
-      def remove_children_with_content(xpath)
+      # @todo
+      def remove_children_with_content!(xpath)
         @noko.search(xpath).each { |item| item.remove }; self
       end
       
-      def remove_children_preserving_content(xpath)
+      # @todo
+      def remove_children_with_content(xpath)
+        dup.remove_children_with_content!(xpath)
+      end
+      
+      # @todo
+      def remove_children_preserving_content!(xpath)
         @noko.search(xpath).each { |item| item.replace(item.children) }; self
       end
+
+      # @todo
+      def remove_children_preserving_content(xpath)
+        dup.remove_children_preserving_content!(xpath)
+      end
       
+      # @todo
       def inner_html
-        @noko.inner_html
+        Text.new @noko.inner_html.gsub("\n","")
       end
       
+      # @todo
       def inner_text
-        @noko.inner_text
+        Text.new @noko.inner_text
       end
       
+      # @todo
       def to_markdown
         Markdown.new self
       end
       
+      # @todo
       def to_native
-        inner_html
+        inner_html.to_native
       end
       
+      # @todo
       def to_s
-        inner_html
+        @noko.is_a?(Nokogiri::HTML::Document) ? to_native : @noko.to_s
       end
       
     end

data/lib/saper/items/json.rb

@@ -2,6 +2,7 @@ module Saper
   module Items
     class JSON < Item
       
+      # @todo
       def self.new(item, bypass = false)
         super case item
         when Text
@@ -15,6 +16,7 @@ module Saper
         end
       end
       
+      # @todo
       def self.parse(string)
         begin
           ::JSON.parse(string)
@@ -23,28 +25,44 @@ module Saper
         end
       end
       
+      # @todo
       def initialize(hash)
         @hash = hash
       end
       
+      # @todo
+      def ==(other)
+        @hash == other
+      end
+      
+      # @todo
       def find(xpath)
         find_all(xpath).first
       end
       
+      # @todo
       def find_all(xpath)
-        sanitize JSONSearch.find(@hash, xpath)
+        sanitize JPath.find(@hash, xpath)
       end
       
+      # @todo
       def to_s
         ::JSON.dump(@hash)
       end
       
+      # @todo
+      def to_json(*a)
+        a.empty? ? self : @hash
+      end
+      
+      # @todo
       def to_native
         to_s
       end
       
       private
       
+      # @todo
       def sanitize(item)
         case item
           when Hash

data/lib/saper/items/markdown.rb

@@ -2,33 +2,43 @@ module Saper
   module Items
     class Markdown < Item
       
+      # @todo
       def self.new(item)
         super case item
         when HTML
-          parse(item.inner_html)
+          parse(item.to_s)
         else
           raise(InvalidItem, item)
         end
       end
       
+      # @todo
       def self.parse(string)
         begin
-          ::ReverseMarkdown.convert(string).strip
+          ::ReverseMarkdown.convert(string, unknown_tags: :drop).strip
         rescue
           raise(InvalidItem, string)
         end
       end
       
+      # @todo
       def initialize(string)
         @string = string
       end
       
+      # @todo
+      def ==(other)
+        to_s == other.to_s
+      end
+      
+      # @todo
       def to_s
-        @string
+        to_native
       end
       
+      # @todo
       def to_native
-        to_s
+        @string
       end
       
     end

data/lib/saper/items/nothing.rb

@@ -2,14 +2,20 @@ module Saper
   module Items
     class Nothing < Item
       
+      # @todo
       def nil?
         true
       end
       
+      # @todo
       def to_native
         nil
       end
       
+      def to_s
+        'nil'
+      end
+      
     end
   end
 end

data/lib/saper/items/text.rb

@@ -2,6 +2,7 @@ module Saper
   module Items
     class Text < Item
       
+      # @todo
       def self.new(item)
         super case item
         when String
@@ -13,40 +14,64 @@ module Saper
         end
       end
       
+      # @todo
       def initialize(string)
         @string = string
       end
       
-      def split(separator, number = nil)
+      # @todo
+      def ==(other)
+        hash == other.hash
+      end
+      
+      # @todo
+      def hash
+        to_s.hash
+      end
+      
+      # @todo
+      def empty?
+        @string.empty?
+      end
+      
+      # @todo
+      def split(separator, number = -1)
         @string.split(separator, number)
       end
       
+      # @todo
       def gsub(*args, &block)
         @string.gsub(*args, &block)
       end
       
+      # @todo
       def to_xml
         XML.new(self)
       end
       
+      # @todo
       def to_html
         HTML.new(self)
       end
       
+      # @todo
       def to_json
         JSON.new(self)
       end
       
+      # @todo
       def to_time(format, tz = nil)
         Time.new(self, format, tz)
       end
       
+      # @todo
       def to_s
-        @string
+        to_native
       end
       
+      # @todo
       def to_native
-        to_s
+        @string
       end
       
     end

data/lib/saper/items/time.rb

@@ -2,17 +2,21 @@ module Saper
   module Items
     class Time < Item
       
-      def self.new(item, format, tz = nil)
+      # @todo
+      def self.new(item, format = nil, tz = nil)
         super case item
         when Text
           parse(item.to_s, format, tz)
         when String
           parse(item, format, tz)
+        when Vremya
+          item
         else
           raise(InvalidItem, item)
         end
       end
       
+      # @todo
       def self.parse(string, format, tz = nil)
         begin
           ::Vremya.parse(string, format, tz)
@@ -21,20 +25,39 @@ module Saper
         end
       end
       
+      # @todo
+      def self.now(tz = nil)
+        new ::Vremya.now(tz)
+      end
+      
+      # @todo
       def initialize(vremya)
         @vremya = vremya
       end
       
+      # @todo
+      def ==(other)
+        to_i == other.to_i
+      end
+      
+      # @todo
       def to_s
         @vremya.iso8601
       end
       
+      # @todo
       def to_native
         @vremya.to_time
       end
       
+      # @todo
+      def to_i
+        to_native.to_i
+      end
+      
+      # @todo
       def serialize
-        to_s
+        @vremya.to_utc.to_time
       end
       
     end

data/lib/saper/items/url.rb

@@ -2,6 +2,7 @@ module Saper
   module Items
     class URL < Item
       
+      # @todo
       def self.new(item)
         super case item
         when Text
@@ -13,22 +14,27 @@ module Saper
         end
       end
       
+      # @todo
       def self.parse(string)
         unless string =~ /^#{URI::regexp}$/
           raise(InvalidItem, string)
         end
+        string
       end
       
+      # @todo
       def initialize(url)
         @url = url
       end
       
+      # @todo
       def to_s
-        @url
+        to_native
       end
       
+      # @todo
       def to_native
-        to_s
+        @url
       end
       
     end

data/lib/saper/items/xml.rb

@@ -2,10 +2,15 @@ module Saper
   module Items
     class XML < Item
       
+      # @todo
       def self.new(item)
         super case item
         when Nokogiri::XML
           item
+        when Nokogiri::XML::Document
+          item
+        when Nokogiri::XML::Element
+          item
         when Text
           parse(item.to_s)
         when String
@@ -15,7 +20,11 @@ module Saper
         end
       end
       
+      # @todo
       def self.parse(string, uri = nil, charset = nil)
+        if string.empty?
+          raise(InvalidItem, string)
+        end
         begin
           Nokogiri::XML.parse(string, uri, charset)
         rescue
@@ -23,6 +32,7 @@ module Saper
         end
       end
       
+      # @todo
       def initialize(noko)
         @noko = noko
         # Force UTF-8 encoding
@@ -30,46 +40,76 @@ module Saper
         @noko.document.encoding = 'UTF-8'
       end
       
+      # @todo
+      def dup
+        XML.new(@noko.dup)
+      end
+      
+      # @todo
+      def ==(other)
+        to_s == other.to_s
+      end
+      
+      # @todo
       def name
         @noko.name
       end
       
+      # @todo
       def [](name)
-        @noko[name]
+        @noko.has_attribute?(name) ? Text.new(@noko[name]) : nil
       end
       
+      # @todo
       def find(xpath)
         find_all(xpath).first
       end
       
+      # @todo
       def find_all(xpath)
         @noko.search(xpath).map { |element| XML.new(element) }
       end
       
-      def remove_children_with_content(xpath)
+      # @todo
+      def remove_children_with_content!(xpath)
         @noko.search(xpath).each { |item| item.remove }; self
       end
       
-      def remove_children_preserving_content(xpath)
-        @noko.search(xpath).each { |item| item.replace(item.children) }; self
+      # @todo
+      def remove_children_with_content(xpath)
+        dup.remove_children_with_content!(xpath)
       end
       
-      def inner_html
-        @noko.inner_html
+      # @todo
+      def remove_children_preserving_content!(xpath)
+        @noko.search(xpath).each { |item| item.replace(item.children) }; self
+      end
+
+      # @todo
+      def remove_children_preserving_content(xpath)
+        dup.remove_children_preserving_content!(xpath)
       end
       
-      def inner_text
-        @noko.inner_text
+      # @todo
+      def inner_xml
+        Text.new @noko.inner_html
       end
       
-      def to_s
-        inner_html
+      # @todo
+      def inner_text
+        Text.new @noko.inner_text
       end
       
+      # @todo
       def to_native
-        inner_html
+        inner_xml.to_native
       end
       
+      # @todo
+      def to_s
+        @noko.is_a?(Nokogiri::XML::Document) ? to_native : @noko.to_s
+      end
+            
     end
   end
 end