diff --git a/lib/oga/lexer.rl b/lib/oga/lexer.rl
index bb037e6..a922a46 100644
--- a/lib/oga/lexer.rl
+++ b/lib/oga/lexer.rl
@@ -264,7 +264,13 @@ module Oga
element_start = '<' element_name;
element_text := |*
- ^'<' => buffer_text;
+ newline => {
+ emit_text_buffer
+ t(:T_TEXT)
+ advance_line
+ };
+
+ ^('<' | newline) => buffer_text;
'<' => {
emit_text_buffer
@@ -284,7 +290,8 @@ module Oga
advance_column(2)
# Advance the column for the closing name.
- advance_column(@te - p)
+ advance_column(text.length)
+
fret;
};
diff --git a/spec/oga/lexer/documents_spec.rb b/spec/oga/lexer/documents_spec.rb
new file mode 100644
index 0000000..5cd305c
--- /dev/null
+++ b/spec/oga/lexer/documents_spec.rb
@@ -0,0 +1,54 @@
+require 'spec_helper'
+
+describe Oga::Parser do
+ context 'HTML documents' do
+ example 'lex a basic HTML document' do
+ html = <<-EOF
+
+
+
+Title
+
+
+
+ EOF
+
+ lex(html).should == [
+ [:T_DOCTYPE_START, '', 1, 15],
+ [:T_TEXT, "\n", 1, 16],
+
+ #
+ [:T_ELEM_OPEN, nil, 2, 1],
+ [:T_ELEM_NAME, 'html', 2, 2],
+ [:T_TEXT, "\n", 2, 7],
+
+ #
+ [:T_ELEM_OPEN, nil, 3, 1],
+ [:T_ELEM_NAME, 'head', 3, 2],
+ [:T_TEXT, "\n", 3, 7],
+
+ # Title
+ [:T_ELEM_OPEN, nil, 4, 1],
+ [:T_ELEM_NAME, 'title', 4, 2],
+ [:T_TEXT, 'Title', 4, 8],
+ [:T_ELEM_CLOSE, nil, 4, 13],
+ [:T_TEXT, "\n", 4, 21],
+
+ #
+ [:T_ELEM_CLOSE, nil, 5, 1],
+ [:T_TEXT, "\n", 5, 8],
+
+ #
+ [:T_ELEM_OPEN, nil, 6, 1],
+ [:T_ELEM_NAME, 'body', 6, 2],
+ [:T_ELEM_CLOSE, nil, 6, 7],
+ [:T_TEXT, "\n", 6, 14],
+
+ #
+ [:T_ELEM_CLOSE, nil, 7, 1],
+ [:T_TEXT, "\n", 7, 8]
+ ]
+ end
+ end
+end