diff --git a/lib/oga/lexer.rl b/lib/oga/lexer.rl
index a922a46..61d229d 100644
--- a/lib/oga/lexer.rl
+++ b/lib/oga/lexer.rl
@@ -6,6 +6,27 @@ module Oga
class Lexer
%% write data; # %
+ attr_reader :html
+
+ HTML_VOID_ELEMENTS = [
+ 'area',
+ 'base',
+ 'br',
+ 'col',
+ 'command',
+ 'embed',
+ 'hr',
+ 'img',
+ 'input',
+ 'keygen',
+ 'link',
+ 'meta',
+ 'param',
+ 'source',
+ 'track',
+ 'wbr'
+ ]
+
# Lazy way of forwarding instance method calls used internally by Ragel to
# their corresponding class methods.
private_methods.grep(/^_lexer_/).each do |name|
@@ -16,19 +37,24 @@ module Oga
private(name)
end
- def initialize
+ def initialize(options = {})
+ options.each do |key, value|
+ instance_variable_set("@#{key}", value) if respond_to?(key)
+ end
+
reset
end
def reset
- @line = 1
- @column = 1
- @data = nil
- @ts = nil
- @te = nil
- @tokens = []
- @stack = []
- @top = 0
+ @line = 1
+ @column = 1
+ @data = nil
+ @ts = nil
+ @te = nil
+ @tokens = []
+ @stack = []
+ @top = 0
+ @elements = []
@string_buffer = ''
@text_buffer = ''
@@ -49,6 +75,10 @@ module Oga
return tokens
end
+ def html?
+ return !!html
+ end
+
private
def advance_line
@@ -93,6 +123,10 @@ module Oga
@string_buffer = ''
end
+ def current_element
+ return @elements.last
+ end
+
%%{
# Use instance variables for `ts` and friends.
access @;
@@ -255,6 +289,8 @@ module Oga
advance_column
end
+ @elements << name
+
add_token(:T_ELEM_NAME, name)
fcall element;
@@ -270,7 +306,11 @@ module Oga
advance_line
};
- ^('<' | newline) => buffer_text;
+ ^('<' | newline) => {
+ @text_buffer << text
+
+ emit_text_buffer if @te == eof
+ };
'<' => {
emit_text_buffer
@@ -305,7 +345,15 @@ module Oga
# Consume the text inside the element.
'>' => {
+ # If HTML lexing is enabled and we're in a void element we'll bail
+ # out right away.
+ if html? and HTML_VOID_ELEMENTS.include?(current_element)
+ add_token(:T_ELEM_CLOSE, nil)
+ @elements.pop
+ end
+
advance_column
+
fcall element_text;
};
@@ -325,6 +373,9 @@ module Oga
# Non self-closing elements.
' {
fcall element_closing_tag;
+
+ @elements.pop
+
fret;
};
@@ -332,6 +383,9 @@ module Oga
'/>' => {
advance_column
add_token(:T_ELEM_CLOSE, nil)
+
+ @elements.pop
+
fret;
};
*|;
diff --git a/lib/oga/parser.y b/lib/oga/parser.y
index aefa1a1..6c86e5c 100644
--- a/lib/oga/parser.y
+++ b/lib/oga/parser.y
@@ -139,8 +139,8 @@ end
---- inner
- def initialize
- @lexer = Lexer.new
+ def initialize(options = {})
+ @lexer = Lexer.new(options)
end
def reset
diff --git a/spec/oga/lexer/html_void_elements_spec.rb b/spec/oga/lexer/html_void_elements_spec.rb
new file mode 100644
index 0000000..998ce4f
--- /dev/null
+++ b/spec/oga/lexer/html_void_elements_spec.rb
@@ -0,0 +1,45 @@
+require 'spec_helper'
+
+describe Oga::Lexer do
+ context 'HTML void elements' do
+ example 'lex a void element that omits the closing /' do
+ lex('', :html => true).should == [
+ [:T_ELEM_OPEN, nil, 1, 1],
+ [:T_ELEM_NAME, 'link', 1, 2],
+ [:T_ELEM_CLOSE, nil, 1, 6]
+ ]
+ end
+
+ example 'lex text after a void element' do
+ lex('foo', :html => true).should == [
+ [:T_ELEM_OPEN, nil, 1, 1],
+ [:T_ELEM_NAME, 'link', 1, 2],
+ [:T_ELEM_CLOSE, nil, 1, 6],
+ [:T_TEXT, 'foo', 1, 7]
+ ]
+ end
+
+ example 'lex a void element inside another element' do
+ lex('', :html => true).should == [
+ [:T_ELEM_OPEN, nil, 1, 1],
+ [:T_ELEM_NAME, 'head', 1, 2],
+ [:T_ELEM_OPEN, nil, 1, 7],
+ [:T_ELEM_NAME, 'link', 1, 8],
+ [:T_ELEM_CLOSE, nil, 1, 12],
+ [:T_ELEM_CLOSE, nil, 1, 13]
+ ]
+ end
+
+ example 'lex a void element inside another element with whitespace' do
+ lex("\n", :html => true).should == [
+ [:T_ELEM_OPEN, nil, 1, 1],
+ [:T_ELEM_NAME, 'head', 1, 2],
+ [:T_ELEM_OPEN, nil, 1, 7],
+ [:T_ELEM_NAME, 'link', 1, 8],
+ [:T_ELEM_CLOSE, nil, 1, 12],
+ [:T_TEXT, "\n", 1, 13],
+ [:T_ELEM_CLOSE, nil, 2, 1]
+ ]
+ end
+ end
+end
diff --git a/spec/oga/parser/html_void_elements_spec.rb b/spec/oga/parser/html_void_elements_spec.rb
new file mode 100644
index 0000000..666007c
--- /dev/null
+++ b/spec/oga/parser/html_void_elements_spec.rb
@@ -0,0 +1,64 @@
+require 'spec_helper'
+
+describe Oga::Parser do
+ context 'HTML void elements' do
+ example 'parse a void element that omits the closing /' do
+ parse_html('').should == s(
+ :document,
+ s(:element, nil, 'link', nil, nil)
+ )
+ end
+
+ example 'parse a void element inside another element' do
+ parse_html('').should == s(
+ :document,
+ s(:element, nil, 'head', nil, s(:element, nil, 'link', nil, nil))
+ )
+ end
+
+ example 'parse a void element with attributes inside another element' do
+ parse_html('').should == s(
+ :document,
+ s(
+ :element,
+ nil,
+ 'head',
+ nil,
+ s(
+ :element,
+ nil,
+ 'link',
+ s(:attributes, s(:attribute, 'href', 'foo.css')),
+ nil
+ )
+ )
+ )
+ end
+
+ example 'parse a void element and a non void element in the same parent' do
+ parse_html('Foo').should == s(
+ :document,
+ s(
+ :element,
+ nil,
+ 'head',
+ nil,
+ s(
+ :element,
+ nil,
+ 'link',
+ nil,
+ nil
+ ),
+ s(
+ :element,
+ nil,
+ 'title',
+ nil,
+ s(:text, 'Foo')
+ )
+ )
+ )
+ end
+ end
+end
diff --git a/spec/support/parsing.rb b/spec/support/parsing.rb
index 1488add..6a0fa5a 100644
--- a/spec/support/parsing.rb
+++ b/spec/support/parsing.rb
@@ -15,10 +15,11 @@ module Oga
# Lexes a string and returns the tokens.
#
# @param [String] input
+ # @param [Hash] options
# @return [Array]
#
- def lex(input)
- return Oga::Lexer.new.lex(input)
+ def lex(input, options = {})
+ return Oga::Lexer.new(options).lex(input)
end
##
@@ -28,7 +29,7 @@ module Oga
# @return [Oga::AST::Node]
#
def parse_html(input)
- return Oga::Parser.new.parse(input)
+ return Oga::Parser.new(:html => true).parse(input)
end
end # ParsingHelpers
end # Oga