From 3b2055a30b128aa679a83332dfdfa68314271b24 Mon Sep 17 00:00:00 2001
From: Yorick Peterse <yorickpeterse@gmail.com>
Date: Wed, 4 Mar 2015 11:44:31 +0100
Subject: [PATCH] Refactored handling of literal HTML elements.

This ensures newlines can appear in <style> / <script> tags when using IOs as
input.
---
 ext/ragel/base_lexer.rl               | 27 ++++++++++++++++---------
 spec/oga/xml/lexer/html_style_spec.rb | 29 +++++++++++++++++++++++++++
 2 files changed, 47 insertions(+), 9 deletions(-)

diff --git a/ext/ragel/base_lexer.rl b/ext/ragel/base_lexer.rl
index 0b47ad3..739c15c 100644
--- a/ext/ragel/base_lexer.rl
+++ b/ext/ragel/base_lexer.rl
@@ -330,8 +330,6 @@
 
             if ( literal_html_element_p() )
             {
-                mark = ts + 1;
-
                 fnext literal_html_element;
             }
             else
@@ -404,10 +402,25 @@
     # Certain tags in HTML can contain basically anything except for the literal
     # closing tag. Two examples are script and style tags.  As a result of this
     # we can't use the regular text machine.
-    literal_html_element := |*
-        '</script>' | '</style>' => {
-            callback(id_on_text, data, encoding, mark, ts);
+    literal_html_closing_tags = '</script>' | '</style>';
+    literal_html_allowed = (any* -- literal_html_closing_tags) $count_newlines;
 
+    literal_html_element := |*
+        literal_html_allowed => {
+            callback(id_on_text, data, encoding, ts, te);
+
+            if ( lines > 0 )
+            {
+                advance_line(lines);
+
+                lines = 0;
+            }
+        };
+
+        literal_html_allowed %{ mark = p; } literal_html_closing_tags => {
+            callback(id_on_text, data, encoding, ts, mark);
+
+            p    = mark - 1;
             mark = 0;
 
             if ( lines > 0 )
@@ -417,12 +430,8 @@
                 lines = 0;
             }
 
-            callback_simple(id_on_element_end);
-
             fnext main;
         };
-
-        any $count_newlines;
     *|;
 
     # The main machine aka the entry point of Ragel.
diff --git a/spec/oga/xml/lexer/html_style_spec.rb b/spec/oga/xml/lexer/html_style_spec.rb
index cfa3907..6ac7353 100644
--- a/spec/oga/xml/lexer/html_style_spec.rb
+++ b/spec/oga/xml/lexer/html_style_spec.rb
@@ -2,6 +2,14 @@ require 'spec_helper'
 
 describe Oga::XML::Lexer do
   describe 'HTML style elements' do
+    it 'lexes an empty <style> tag' do
+      lex('<style></style>', :html => true).should == [
+        [:T_ELEM_START, nil, 1],
+        [:T_ELEM_NAME, 'style', 1],
+        [:T_ELEM_END, nil, 1]
+      ]
+    end
+
     it 'treats the content of a style tag as plain text' do
       lex('<style>foo <bar</style>', :html => true).should == [
         [:T_ELEM_START, nil, 1],
@@ -10,5 +18,26 @@ describe Oga::XML::Lexer do
         [:T_ELEM_END, nil, 1]
       ]
     end
+
+    it 'lexes a multi-line <style> tag using a String as the input' do
+      lex("<style>foo\nbar</style>", :html => true).should == [
+        [:T_ELEM_START, nil, 1],
+        [:T_ELEM_NAME, 'style', 1],
+        [:T_TEXT, "foo\nbar", 1],
+        [:T_ELEM_END, nil, 2]
+      ]
+    end
+
+    it 'lexes a multi-line <style> tag using an IO as the input' do
+      io = StringIO.new("<style>foo\nbar</style>")
+
+      lex(io, :html => true).should == [
+        [:T_ELEM_START, nil, 1],
+        [:T_ELEM_NAME, 'style', 1],
+        [:T_TEXT, "foo\n", 1],
+        [:T_TEXT, 'bar', 2],
+        [:T_ELEM_END, nil, 2]
+      ]
+    end
   end
 end