RubyGems - oga - Versions diffs - 0.1.0 - Mend

oga 0.1.0

Files changed (46) hide show

checksums.yaml +7 -0
data/.yardopts +13 -0
data/LICENSE +19 -0
data/README.md +171 -0
data/doc/DCO.md +25 -0
data/doc/changelog.md +7 -0
data/doc/css/common.css +76 -0
data/doc/migrating_from_nokogiri.md +169 -0
data/ext/c/extconf.rb +13 -0
data/ext/c/lexer.c +1518 -0
data/ext/c/lexer.h +8 -0
data/ext/c/lexer.rl +121 -0
data/ext/c/liboga.c +6 -0
data/ext/c/liboga.h +11 -0
data/ext/java/Liboga.java +14 -0
data/ext/java/org/liboga/xml/Lexer.java +829 -0
data/ext/java/org/liboga/xml/Lexer.rl +151 -0
data/ext/ragel/base_lexer.rl +323 -0
data/lib/oga.rb +43 -0
data/lib/oga/html/parser.rb +25 -0
data/lib/oga/oga.rb +27 -0
data/lib/oga/version.rb +3 -0
data/lib/oga/xml/attribute.rb +111 -0
data/lib/oga/xml/cdata.rb +24 -0
data/lib/oga/xml/character_node.rb +39 -0
data/lib/oga/xml/comment.rb +24 -0
data/lib/oga/xml/doctype.rb +91 -0
data/lib/oga/xml/document.rb +99 -0
data/lib/oga/xml/element.rb +340 -0
data/lib/oga/xml/lexer.rb +399 -0
data/lib/oga/xml/namespace.rb +42 -0
data/lib/oga/xml/node.rb +175 -0
data/lib/oga/xml/node_set.rb +313 -0
data/lib/oga/xml/parser.rb +556 -0
data/lib/oga/xml/processing_instruction.rb +39 -0
data/lib/oga/xml/pull_parser.rb +166 -0
data/lib/oga/xml/querying.rb +32 -0
data/lib/oga/xml/text.rb +16 -0
data/lib/oga/xml/traversal.rb +48 -0
data/lib/oga/xml/xml_declaration.rb +76 -0
data/lib/oga/xpath/evaluator.rb +1748 -0
data/lib/oga/xpath/lexer.rb +2043 -0
data/lib/oga/xpath/node.rb +10 -0
data/lib/oga/xpath/parser.rb +535 -0
data/oga.gemspec +45 -0
metadata +221 -0

data/lib/oga.rb ADDED

@@ -0,0 +1,43 @@
+require 'ast'
+require 'set'
+require 'stringio'
+require_relative 'oga/version'
+require_relative 'oga/oga'
+# Load these first so that the native extensions don't have to define the
+# Oga::XML namespace.
+require_relative 'oga/xml/lexer'
+require_relative 'oga/xml/parser'
+require_relative 'oga/xml/pull_parser'
+require_relative 'liboga'
+#:nocov:
+if RUBY_PLATFORM == 'java'
+  org.liboga.Liboga.load(JRuby.runtime)
+end
+#:nocov:
+require_relative 'oga/xml/querying'
+require_relative 'oga/xml/traversal'
+require_relative 'oga/xml/node'
+require_relative 'oga/xml/document'
+require_relative 'oga/xml/character_node'
+require_relative 'oga/xml/text'
+require_relative 'oga/xml/comment'
+require_relative 'oga/xml/cdata'
+require_relative 'oga/xml/xml_declaration'
+require_relative 'oga/xml/processing_instruction'
+require_relative 'oga/xml/doctype'
+require_relative 'oga/xml/namespace'
+require_relative 'oga/xml/attribute'
+require_relative 'oga/xml/element'
+require_relative 'oga/xml/node_set'
+require_relative 'oga/html/parser'
+require_relative 'oga/xpath/node'
+require_relative 'oga/xpath/lexer'
+require_relative 'oga/xpath/parser'
+require_relative 'oga/xpath/evaluator'

data/lib/oga/html/parser.rb ADDED

@@ -0,0 +1,25 @@
+module Oga
+  module HTML
+    ##
+    # Parser for processing HTML input. This parser is a small wrapper around
+    # {Oga::XML::Parser} and takes care of setting the various options required
+    # for parsing HTML documents.
+    #
+    # A basic example:
+    #
+    #     Oga::HTML::Parser.new('<meta charset="utf-8">').parse
+    #
+    class Parser < XML::Parser
+      ##
+      # @param [String|IO] data
+      # @param [Hash] options
+      # @see [Oga::XML::Parser#initialize]
+      #
+      def initialize(data, options = {})
+        options = options.merge(:html => true)
+        super(data, options)
+      end
+    end # Parser
+  end # HTML
+end # Oga

data/lib/oga/oga.rb ADDED

@@ -0,0 +1,27 @@
+module Oga
+  ##
+  # Parses the given XML document.
+  #
+  # @example
+  #  document = Oga.parse_xml('<root>Hello</root>')
+  #
+  # @param [String|IO] xml The XML input to parse.
+  # @return [Oga::XML::Document]
+  #
+  def self.parse_xml(xml)
+    return XML::Parser.new(xml).parse
+  end
+  ##
+  # Parses the given HTML document.
+  #
+  # @example
+  #  document = Oga.parse_html('<html>...</html>')
+  #
+  # @param [String|IO] html The HTML input to parse.
+  # @return [Oga::XML::Document]
+  #
+  def self.parse_html(html)
+    return HTML::Parser.new(html).parse
+  end
+end # Oga

data/lib/oga/version.rb ADDED

@@ -0,0 +1,3 @@
+module Oga
+  VERSION = '0.1.0'
+end # Oga

data/lib/oga/xml/attribute.rb ADDED

@@ -0,0 +1,111 @@
+module Oga
+  module XML
+    ##
+    # Class for storing information about a single XML attribute.
+    #
+    # @!attribute [rw] name
+    #  The name of the attribute.
+    #  @return [String]
+    #
+    # @!attribute [rw] namespace_name
+    #  @return [String]
+    #
+    # @!attribute [rw] value
+    #  The value of the attribute.
+    #  @return [String]
+    #
+    # @!attribute [r] element
+    #  The element this attribute belongs to.
+    #  @return [Oga::XML::Element]
+    #
+    class Attribute
+      attr_accessor :name, :namespace_name, :element, :value
+      ##
+      # The default namespace available to all attributes. This namespace can
+      # not be modified.
+      #
+      # @return [Oga::XML::Namespace]
+      #
+      DEFAULT_NAMESPACE = Namespace.new(
+        :name => 'xml',
+        :uri  => 'http://www.w3.org/XML/1998/namespace'
+      ).freeze
+      ##
+      # @param [Hash] options
+      #
+      # @option options [String] :name
+      # @option options [String] :namespace_name
+      # @option options [String] :value
+      # @option options [Oga::XML::Element] :element
+      #
+      def initialize(options = {})
+        @name    = options[:name]
+        @value   = options[:value]
+        @element = options[:element]
+        @namespace_name = options[:namespace_name]
+      end
+      ##
+      # Returns the {Oga::XML::Namespace} instance for the current namespace
+      # name.
+      #
+      # @return [Oga::XML::Namespace]
+      #
+      def namespace
+        unless @namespace
+          if namespace_name == DEFAULT_NAMESPACE.name
+            @namespace = DEFAULT_NAMESPACE
+          else
+            @namespace = element.available_namespaces[namespace_name]
+          end
+        end
+        return @namespace
+      end
+      ##
+      # Returns the value of the attribute.
+      #
+      # @return [String]
+      #
+      def text
+        return value.to_s
+      end
+      alias_method :to_s, :text
+      ##
+      # @return [String]
+      #
+      def to_xml
+        if namespace_name
+          full_name = "#{namespace.name}:#{name}"
+        else
+          full_name = name
+        end
+        return %Q(#{full_name}="#{value}")
+      end
+      ##
+      # @return [String]
+      #
+      def inspect
+        segments = []
+        [:name, :namespace, :value].each do |attr|
+          value = send(attr)
+          if value
+            segments << "#{attr}: #{value.inspect}"
+          end
+        end
+        return "Attribute(#{segments.join(' ')})"
+      end
+    end # Attribute
+  end # XML
+end # Oga

data/lib/oga/xml/cdata.rb ADDED

@@ -0,0 +1,24 @@
+module Oga
+  module XML
+    ##
+    # Class used for storing information about CDATA tags.
+    #
+    class Cdata < CharacterNode
+      ##
+      # Converts the node back to XML.
+      #
+      # @return [String]
+      #
+      def to_xml
+        return "<![CDATA[#{text}]]>"
+      end
+      ##
+      # @return [Symbol]
+      #
+      def node_type
+        return :cdata
+      end
+    end # Cdata
+  end # XML
+end # Oga

data/lib/oga/xml/character_node.rb ADDED

@@ -0,0 +1,39 @@
+module Oga
+  module XML
+    ##
+    # Base class for nodes that represent a text-like value such as Text and
+    # Comment nodes.
+    #
+    # @!attribute [rw] text
+    #  @return [String]
+    #
+    class CharacterNode < Node
+      attr_accessor :text
+      ##
+      # @param [Hash] options
+      #
+      # @option options [String] :text The text of the node.
+      #
+      def initialize(options = {})
+        super
+        @text = options[:text]
+      end
+      ##
+      # @return [String]
+      #
+      def to_xml
+        return text.to_s
+      end
+      ##
+      # @return [String]
+      #
+      def inspect
+        return "#{self.class.to_s.split('::').last}(#{text.inspect})"
+      end
+    end # CharacterNode
+  end # XML
+end # Oga

data/lib/oga/xml/comment.rb ADDED

@@ -0,0 +1,24 @@
+module Oga
+  module XML
+    ##
+    # Class used for storing information about XML comments.
+    #
+    class Comment < CharacterNode
+      ##
+      # Converts the node back to XML.
+      #
+      # @return [String]
+      #
+      def to_xml
+        return "<!--#{text}-->"
+      end
+      ##
+      # @return [Symbol]
+      #
+      def node_type
+        return :comment
+      end
+    end # Comment
+  end # XML
+end # Oga

data/lib/oga/xml/doctype.rb ADDED

@@ -0,0 +1,91 @@
+module Oga
+  module XML
+    ##
+    # Class used for storing information about Doctypes.
+    #
+    # @!attribute [rw] name
+    #  The name of the doctype (e.g. "HTML").
+    #  @return [String]
+    #
+    # @!attribute [rw] type
+    #  The type of the doctype (e.g. "PUBLIC").
+    #  @return [String]
+    #
+    # @!attribute [rw] public_id
+    #  The public ID of the doctype.
+    #  @return [String]
+    #
+    # @!attribute [rw] system_id
+    #  The system ID of the doctype.
+    #  @return [String]
+    #
+    # @!attribute [rw] inline_rules
+    #  The inline doctype rules.
+    #  @return [String]
+    #
+    class Doctype
+      attr_accessor :name, :type, :public_id, :system_id, :inline_rules
+      ##
+      # @example
+      #  dtd = Doctype.new(:name => 'html', :type => 'PUBLIC')
+      #
+      # @param [Hash] options
+      #
+      # @option options [String] :name
+      # @option options [String] :type
+      # @option options [String] :public_id
+      # @option options [String] :system_id
+      #
+      def initialize(options = {})
+        @name         = options[:name]
+        @type         = options[:type]
+        @public_id    = options[:public_id]
+        @system_id    = options[:system_id]
+        @inline_rules = options[:inline_rules]
+      end
+      ##
+      # Converts the doctype back to XML.
+      #
+      # @return [String]
+      #
+      def to_xml
+        segments = "<!DOCTYPE #{name}"
+        segments << " #{type}" if type
+        segments << %Q{ "#{public_id}"} if public_id
+        segments << %Q{ "#{system_id}"} if system_id
+        segments << " [#{inline_rules}]" if inline_rules
+        return segments + '>'
+      end
+      ##
+      # Inspects the doctype.
+      #
+      # @return [String]
+      #
+      def inspect
+        segments = []
+        [:name, :type, :public_id, :system_id, :inline_rules].each do |attr|
+          value = send(attr)
+          if value and !value.empty?
+            segments << "#{attr}: #{value.inspect}"
+          end
+        end
+        return "Doctype(#{segments.join(' ')})"
+      end
+      ##
+      # @return [Symbol]
+      #
+      def node_type
+        return :doctype
+      end
+    end # Doctype
+  end # XML
+end # Oga

data/lib/oga/xml/document.rb ADDED

@@ -0,0 +1,99 @@
+module Oga
+  module XML
+    ##
+    # Class used for storing information about an entire XML document. This
+    # includes the doctype, XML declaration, child nodes and more.
+    #
+    # @!attribute [rw] doctype
+    #  The doctype of the document.
+    #  @return [Oga::XML::Doctype]
+    #
+    # @!attribute [rw] xml_declaration
+    #  The XML declaration of the document.
+    #  @return [Oga::XML::XmlDeclaration]
+    #
+    class Document
+      include Querying
+      include Traversal
+      attr_accessor :doctype, :xml_declaration
+      ##
+      # @param [Hash] options
+      #
+      # @option options [Oga::XML::NodeSet] :children
+      # @option options [Oga::XML::Doctype] :doctype
+      # @option options [Oga::XML::XmlDeclaration] :xml_declaration
+      #
+      def initialize(options = {})
+        @doctype         = options[:doctype]
+        @xml_declaration = options[:xml_declaration]
+        self.children = options[:children] if options[:children]
+      end
+      ##
+      # @return [Oga::XML::NodeSet]
+      #
+      def children
+        return @children ||= NodeSet.new([], self)
+      end
+      ##
+      # Sets the child nodes of the document.
+      #
+      # @param [Oga::XML::NodeSet|Array] nodes
+      #
+      def children=(nodes)
+        if nodes.is_a?(NodeSet)
+          @children = nodes
+        else
+          @children = NodeSet.new(nodes, self)
+        end
+      end
+      ##
+      # Converts the document and its child nodes to XML.
+      #
+      # @return [String]
+      #
+      def to_xml
+        xml = children.map(&:to_xml).join('')
+        if doctype
+          xml = doctype.to_xml + "\n" + xml.strip
+        end
+        if xml_declaration
+          xml = xml_declaration.to_xml + "\n" + xml.strip
+        end
+        return xml
+      end
+      ##
+      # Inspects the document and its child nodes. Child nodes are indented for
+      # each nesting level.
+      #
+      # @return [String]
+      #
+      def inspect
+        segments = []
+        [:doctype, :xml_declaration, :children].each do |attr|
+          value = send(attr)
+          if value
+            segments << "#{attr}: #{value.inspect}"
+          end
+        end
+        return <<-EOF.strip
+Document(
+  #{segments.join("\n  ")}
+)
+        EOF
+      end
+    end # Document
+  end # XML
+end # Oga