diff --git a/openerp/tests/test_html_sanitize.py b/openerp/tests/test_html_sanitize.py
new file mode 100755
index 00000000000..906c78d1dcf
--- /dev/null
+++ b/openerp/tests/test_html_sanitize.py
@@ -0,0 +1,33 @@
+#!/usr/bin/env python
+import unittest
+from openerp.tools.html_sanitize import html_sanitize
+
+test_case = """
+test1
+
+test2
+test3
+test4
+test5
+test6
+
+
+test12
+google
+"""
+
+class TestSanitizer(unittest.TestCase):
+
+ def test_simple(self):
+ x = "yop"
+ self.assertEqual(x, html_sanitize(x))
+
+ def test_test_case(self):
+ res = html_sanitize(test_case)
+ print res
+
+if __name__ == '__main__':
+ unittest.main()
\ No newline at end of file
diff --git a/openerp/tools/__init__.py b/openerp/tools/__init__.py
index a6eeb14ba96..a0ca411a9df 100644
--- a/openerp/tools/__init__.py
+++ b/openerp/tools/__init__.py
@@ -33,6 +33,7 @@ from pdf_utils import *
from yaml_import import *
from sql import *
from float_utils import *
+from html_sanitize import *
#.apidoc title: Tools
diff --git a/openerp/tools/html_sanitize.py b/openerp/tools/html_sanitize.py
index 4e6fb1540b1..d25d4466bb9 100644
--- a/openerp/tools/html_sanitize.py
+++ b/openerp/tools/html_sanitize.py
@@ -1,4 +1,39 @@
+from pyquery import PyQuery as pq
def html_sanitize(x):
- return x
+ root = pq("")
+ root.html(x)
+ result = handle_element(root[0])
+ new = pq(result)
+ return new.html()
+
+def handle_element(el):
+ if type(el) == str or type(el) == unicode:
+ return [el]
+ else:
+ new = pq("<%s />" % el.tag)[0]
+ for i in children(el):
+ append_to(handle_element(i), new)
+ return [new]
+
+def children(el):
+ res = []
+ if el.text is not None:
+ res.append(el.text)
+ for i in el.getchildren():
+ res.append(i)
+ if i.tail is not None:
+ res.append(i.tail)
+ return res
+
+def append_to(new_ones, el):
+ for i in new_ones:
+ if type(i) == str or type(i) == unicode:
+ children = el.getchildren()
+ if len(children) == 0:
+ el.text = i
+ else:
+ children[-1].tail = i
+ else:
+ el.append(i)
\ No newline at end of file