oga 1.0.2 → 1.0.3

data/ext/java/org/liboga/xml/Lexer.rl

@@ -89,7 +89,7 @@ public class Lexer extends RubyObject
     @JRubyMethod
     public IRubyObject advance_native(ThreadContext context, RubyString rb_str)
     {
-        Boolean html_p = this.callMethod(context, "html").isTrue();
+        Boolean html_p = this.callMethod(context, "html?").isTrue();
 
         Encoding encoding = rb_str.getEncoding();
 

data/ext/ragel/base_lexer.rl

@@ -422,9 +422,35 @@
     # Characters that can be used for unquoted HTML attribute values.
     # See https://html.spec.whatwg.org/multipage/introduction.html#intro-early-example
     # for more info.
-    html_unquoted_value = ^(
-        squote | dquote | '`' | '=' | '<' | '>' | whitespace_or_newline
-    )+;
+    html_unquoted_value =
+        ^(squote | dquote | whitespace_or_newline)
+        ^('`' | '=' | '<' | '>' | whitespace_or_newline)+;
+
+    # Machine used after matching the "=" of an attribute and just before moving
+    # into the actual attribute value.
+    attribute_pre := |*
+        whitespace_or_newline $count_newlines;
+
+        any => {
+            fhold;
+
+            if ( lines > 0 )
+            {
+                advance_line(lines);
+
+                lines = 0;
+            }
+
+            if ( html_p )
+            {
+                fnext html_attribute_value;
+            }
+            else
+            {
+                fnext xml_attribute_value;
+            }
+        };
+    *|;
 
     # Machine used for processing HTML attribute values.
     html_attribute_value := |*
@@ -482,14 +508,7 @@
 
         # Attribute values.
         '=' => {
-            if ( html_p )
-            {
-                fcall html_attribute_value;
-            }
-            else
-            {
-                fcall xml_attribute_value;
-            }
+            fcall attribute_pre;
         };
 
         # We're done with the open tag of the element.

data/lib/oga/blacklist.rb

@@ -26,7 +26,7 @@ module Oga
     # @return [TrueClass|FalseClass]
     #
     def allow?(name)
-      return !names.include?(name)
+      !names.include?(name)
     end
 
     ##
@@ -34,7 +34,7 @@ module Oga
     # @return [Oga::Blacklist]
     #
     def +(other)
-      return self.class.new(names + other.names)
+      self.class.new(names + other.names)
     end
   end # Blacklist
 end # Oga

data/lib/oga/css/parser.rb

@@ -271,7 +271,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def current_element
-    return @current_element ||= s(:test, nil, '*')
+    @current_element ||= s(:test, nil, '*')
   end
 
   ##
@@ -286,7 +286,7 @@ class Parser < LL::Driver
   def parse
     reset
 
-    return super
+    super
   end
 
   ##
@@ -298,8 +298,6 @@ class Parser < LL::Driver
   #
   def on_test(namespace, name)
     @current_element = s(:test, namespace, name)
-
-    return @current_element
   end
 
   ##
@@ -310,7 +308,7 @@ class Parser < LL::Driver
   def on_pseudo_class(name, arg = nil)
     handler = "on_pseudo_class_#{name.gsub('-', '_')}"
 
-    return arg ? send(handler, arg) : send(handler)
+    arg ? send(handler, arg) : send(handler)
   end
 
   ##
@@ -319,7 +317,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_pseudo_class_root
-    return s(:call, 'not', s(:axis, 'parent', s(:test, nil, '*')))
+    s(:call, 'not', s(:axis, 'parent', s(:test, nil, '*')))
   end
 
   ##
@@ -329,7 +327,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_pseudo_class_nth_child(arg)
-    return generate_nth_child('preceding-sibling', arg)
+    generate_nth_child('preceding-sibling', arg)
   end
 
   ##
@@ -339,7 +337,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_pseudo_class_nth_last_child(arg)
-    return generate_nth_child('following-sibling', arg)
+    generate_nth_child('following-sibling', arg)
   end
 
   ##
@@ -349,7 +347,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_pseudo_class_nth_of_type(arg)
-    return generate_nth_child('preceding-sibling', arg, current_element)
+    generate_nth_child('preceding-sibling', arg, current_element)
   end
 
   ##
@@ -359,7 +357,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_pseudo_class_nth_last_of_type(arg)
-    return generate_nth_child('following-sibling', arg, current_element)
+    generate_nth_child('following-sibling', arg, current_element)
   end
 
   ##
@@ -368,7 +366,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_pseudo_class_first_child
-    return generate_no_siblings('preceding-sibling')
+    generate_no_siblings('preceding-sibling')
   end
 
   ##
@@ -377,7 +375,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_pseudo_class_last_child
-    return generate_no_siblings('following-sibling')
+    generate_no_siblings('following-sibling')
   end
 
   ##
@@ -386,7 +384,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_pseudo_class_first_of_type
-    return generate_no_siblings('preceding-sibling', current_element)
+    generate_no_siblings('preceding-sibling', current_element)
   end
 
   ##
@@ -395,7 +393,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_pseudo_class_last_of_type
-    return generate_no_siblings('following-sibling', current_element)
+    generate_no_siblings('following-sibling', current_element)
   end
 
   ##
@@ -404,7 +402,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_pseudo_class_only_child
-    return s(:and, on_pseudo_class_first_child, on_pseudo_class_last_child)
+    s(:and, on_pseudo_class_first_child, on_pseudo_class_last_child)
   end
 
   ##
@@ -413,7 +411,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_pseudo_class_only_of_type
-    return s(:and, on_pseudo_class_first_of_type, on_pseudo_class_last_of_type)
+    s(:and, on_pseudo_class_first_of_type, on_pseudo_class_last_of_type)
   end
 
   ##
@@ -422,7 +420,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_pseudo_class_empty
-    return s(:call, 'not', s(:axis, 'child', s(:type_test, 'node')))
+    s(:call, 'not', s(:axis, 'child', s(:type_test, 'node')))
   end
 
   ##
@@ -433,7 +431,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_op_eq(attr, value)
-    return s(:eq, attr, value)
+    s(:eq, attr, value)
   end
 
   ##
@@ -444,7 +442,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_op_space_in(attr, value)
-    return s(
+    s(
       :call,
       'contains',
       s(:call, 'concat', s(:string, ' '), attr, s(:string, ' ')),
@@ -460,7 +458,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_op_starts_with(attr, value)
-    return s(:call, 'starts-with', attr, value)
+    s(:call, 'starts-with', attr, value)
   end
 
   ##
@@ -471,7 +469,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_op_ends_with(attr, value)
-    return s(
+    s(
       :eq,
       s(
         :call,
@@ -500,7 +498,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_op_in(attr, value)
-    return s(:call, 'contains', attr, value)
+    s(:call, 'contains', attr, value)
   end
 
   ##
@@ -511,7 +509,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def on_op_hyphen_in(attr, value)
-    return s(
+    s(
       :or,
       s(:eq, attr, value),
       s(
@@ -566,7 +564,7 @@ class Parser < LL::Driver
   # @return [AST::Node]
   #
   def generate_no_siblings(axis, test = s(:test, nil, '*'))
-    return s(:eq, s(:call, 'count', s(:axis, axis, test)), s(:int, 0))
+    s(:eq, s(:call, 'count', s(:axis, axis, test)), s(:int, 0))
   end
 
   ##
@@ -574,7 +572,7 @@ class Parser < LL::Driver
   # @return [TrueClass|FalseClass]
   #
   def int_node?(node)
-    return node.type.equal?(:int)
+    node.type.equal?(:int)
   end
 
   ##
@@ -582,7 +580,7 @@ class Parser < LL::Driver
   # @return [TrueClass|FalseClass]
   #
   def non_positive_number?(node)
-    return node.children[0] <= 0
+    node.children[0] <= 0
   end
 
   ##
@@ -590,7 +588,7 @@ class Parser < LL::Driver
   # @return [Symbol]
   #
   def step_comparison(node)
-    return node.children[0] >= 0 ? :gte : :lte
+    node.children[0] >= 0 ? :gte : :lte
   end
 
   ##
@@ -610,7 +608,7 @@ class Parser < LL::Driver
       mod_val = s(:int, 1)
     end
 
-    return mod_val
+    mod_val
   end
 
   def _rule_0(val)

data/lib/oga/entity_decoder.rb

@@ -4,7 +4,7 @@ module Oga
     # @see [decode]
     #
     def self.try_decode(input, html = false)
-      return input ? decode(input, html) : nil
+      input ? decode(input, html) : nil
     end
 
     ##
@@ -15,7 +15,7 @@ module Oga
     def self.decode(input, html = false)
       decoder = html ? HTML::Entities : XML::Entities
 
-      return decoder.decode(input)
+      decoder.decode(input)
     end
   end # EntityDecoder
 end # Oga

data/lib/oga/html/entities.rb

@@ -2143,7 +2143,7 @@ module Oga
       # @see [decode]
       #
       def self.decode(input)
-        return XML::Entities.decode(input, DECODE_MAPPING)
+        XML::Entities.decode(input, DECODE_MAPPING)
       end
     end # Entities
   end # HTML

data/lib/oga/lru.rb

@@ -50,7 +50,7 @@ module Oga
     # @return [Fixnum]
     #
     def maximum
-      return synchronize { @maximum }
+      synchronize { @maximum }
     end
 
     ##
@@ -60,7 +60,7 @@ module Oga
     # @return [Mixed]
     #
     def [](key)
-      return synchronize { @cache[key] }
+      synchronize { @cache[key] }
     end
 
     ##
@@ -90,14 +90,14 @@ module Oga
     # @return [Mixed]
     #
     def get_or_set(key)
-      return synchronize { self[key] ||= yield }
+      synchronize { self[key] ||= yield }
     end
 
     ##
     # @return [Array]
     #
     def keys
-      return synchronize { @keys }
+      synchronize { @keys }
     end
 
     ##
@@ -105,7 +105,7 @@ module Oga
     # @return [TrueClass|FalseClass]
     #
     def key?(key)
-      return synchronize { @cache.key?(key) }
+      synchronize { @cache.key?(key) }
     end
 
     ##
@@ -122,7 +122,7 @@ module Oga
     # @return [Fixnum]
     #
     def size
-      return synchronize { @cache.size }
+      synchronize { @cache.size }
     end
 
     alias_method :length, :size

data/lib/oga/oga.rb

@@ -5,11 +5,12 @@ module Oga
   # @example
   #  document = Oga.parse_xml('<root>Hello</root>')
   #
-  # @param [String|IO] xml The XML input to parse.
+  # @see [Oga::XML::Lexer#initialize]
+  #
   # @return [Oga::XML::Document]
   #
-  def self.parse_xml(xml)
-    return XML::Parser.new(xml).parse
+  def self.parse_xml(xml, options = {})
+    XML::Parser.new(xml, options).parse
   end
 
   ##
@@ -18,11 +19,12 @@ module Oga
   # @example
   #  document = Oga.parse_html('<html>...</html>')
   #
-  # @param [String|IO] html The HTML input to parse.
+  # @see [Oga::XML::Lexer#initialize]
+  #
   # @return [Oga::XML::Document]
   #
-  def self.parse_html(html)
-    return HTML::Parser.new(html).parse
+  def self.parse_html(html, options = {})
+    HTML::Parser.new(html, options).parse
   end
 
   ##
@@ -33,11 +35,10 @@ module Oga
   #
   #  Oga.sax_parse_html(handler, '<root>Hello</root>')
   #
-  # @param [Object] handler The SAX handler for the parser.
-  # @param [String|IO] xml The XML to parse.
+  # @see [Oga::XML::SaxParser#initialize]
   #
-  def self.sax_parse_xml(handler, xml)
-    XML::SaxParser.new(handler, xml).parse
+  def self.sax_parse_xml(handler, xml, options = {})
+    XML::SaxParser.new(handler, xml, options).parse
   end
 
   ##
@@ -48,10 +49,9 @@ module Oga
   #
   #  Oga.sax_parse_html(handler, '<script>foo()</script>')
   #
-  # @param [Object] handler The SAX handler for the parser.
-  # @param [String|IO] html The HTML to parse.
+  # @see [Oga::XML::SaxParser#initialize]
   #
-  def self.sax_parse_html(handler, html)
-    HTML::SaxParser.new(handler, html).parse
+  def self.sax_parse_html(handler, html, options = {})
+    HTML::SaxParser.new(handler, html, options).parse
   end
 end # Oga