From 0800654c962c20fb139a389245359bca9952dcd1 Mon Sep 17 00:00:00 2001
From: Yorick Peterse <yorickpeterse@gmail.com>
Date: Fri, 3 Apr 2015 00:46:37 +0200
Subject: [PATCH] Support lexing or carriage returns

Fixes #89.
---
 ext/ragel/base_lexer.rl             |  2 +-
 spec/oga/xml/lexer/elements_spec.rb |  8 ++++++++
 spec/oga/xml/lexer/general_spec.rb  | 10 +++++++++-
 3 files changed, 18 insertions(+), 2 deletions(-)

diff --git a/ext/ragel/base_lexer.rl b/ext/ragel/base_lexer.rl
index 739c15c..a04140f 100644
--- a/ext/ragel/base_lexer.rl
+++ b/ext/ragel/base_lexer.rl
@@ -46,7 +46,7 @@
     # stack.
     #
 
-    newline = '\n' | '\r\n';
+    newline = '\r\n' | '\n' | '\r';
 
     action count_newlines {
         if ( fc == '\n' ) lines++;
diff --git a/spec/oga/xml/lexer/elements_spec.rb b/spec/oga/xml/lexer/elements_spec.rb
index 59a76ec..2450232 100644
--- a/spec/oga/xml/lexer/elements_spec.rb
+++ b/spec/oga/xml/lexer/elements_spec.rb
@@ -41,6 +41,14 @@ describe Oga::XML::Lexer do
         [:T_ELEM_END, nil, 2]
       ]
     end
+
+    it 'lexes an element with a carriage return in the open tag' do
+      lex("<p\r></p>").should == [
+        [:T_ELEM_START, nil, 1],
+        [:T_ELEM_NAME, 'p', 1],
+        [:T_ELEM_END, nil, 2]
+      ]
+    end
   end
 
   describe 'elements with attributes' do
diff --git a/spec/oga/xml/lexer/general_spec.rb b/spec/oga/xml/lexer/general_spec.rb
index f9b3105..bc02a11 100644
--- a/spec/oga/xml/lexer/general_spec.rb
+++ b/spec/oga/xml/lexer/general_spec.rb
@@ -10,10 +10,18 @@ describe Oga::XML::Lexer do
       lex(' ').should == [[:T_TEXT, ' ', 1]]
     end
 
-    it 'lexes a newline' do
+    it 'lexes a Unix newline' do
       lex("\n").should == [[:T_TEXT, "\n", 1]]
     end
 
+    it 'lexes a Windows newline' do
+      lex("\r\n").should == [[:T_TEXT, "\r\n", 1]]
+    end
+
+    it 'lexes a carriage return' do
+      lex("\r").should == [[:T_TEXT, "\r", 1]]
+    end
+
     it 'lexes text followed by a newline' do
       lex("foo\n").should == [[:T_TEXT, "foo\n", 1]]
     end