Improve the xhtml_to_poezio_colors sanitization

add some tests, and remove the test_datetime_tuple because
of python time module misbehavior with timezones set manually.
(potentially due to http://bugs.python.org/issue6478)

3 files changed, 12 insertions, 15 deletions

diff --git a/src/xhtml.py b/src/xhtml.py
index 5b71823c..b84ce943 100644
--- a/src/xhtml.py
+++ b/src/xhtml.py
@@ -183,7 +183,8 @@ whitespace_re = re.compile(r'\s+')
 
 xhtml_attr_re = re.compile(r'\x19-?\d[^}]*}|\x19[buaio]')
 xhtml_data_re = re.compile(r'data:image/([a-z]+);base64,(.+)')
-poezio_colors_trim = re.compile(r'\x19o(\x19\d+}|\x19\d|\x19[buaio])*\x19o')
+poezio_color_double = re.compile(r'(?:\x19\d+}|\x19\d)+(\x19\d|\x19\d+})')
+poezio_format_trim = re.compile(r'(\x19\d+}|\x19\d|\x19[buaio]|\x19o)+\x19o')
 
 xhtml_simple_attr_re = re.compile(r'\x19\d')
 
@@ -304,7 +305,8 @@ class XHTMLHandler(sax.ContentHandler):
 
     @property
     def result(self):
-        return re.sub(poezio_colors_trim, '\x19o', ''.join(self.builder).strip())
+        sanitized = re.sub(poezio_color_double, r'\1', ''.join(self.builder).strip())
+        return re.sub(poezio_format_trim, '\x19o', sanitized)
 
     def append_formatting(self, formatting):
         self.formatting.append(formatting)
diff --git a/test/test_common.py b/test/test_common.py
index 50643733..0b2bd279 100644
--- a/test/test_common.py
+++ b/test/test_common.py
@@ -14,19 +14,6 @@ from common import (datetime_tuple, get_utc_time, get_local_time, shell_split,
                     find_argument_quoted, find_argument_unquoted,
                     parse_str_to_secs, parse_secs_to_str, safeJID)
 
-def test_datetime_tuple():
-    time.timezone = 0
-    time.altzone = 0
-
-    assert datetime_tuple('20130226T06:23:12') == datetime.datetime(2013, 2, 26, 6, 23, 12)
-    assert datetime_tuple('2013-02-26T06:23:12+02:00') == datetime.datetime(2013, 2, 26, 4, 23, 12)
-
-    time.timezone = -3600
-    time.altzone = -3600
-
-    assert datetime_tuple('20130226T07:23:12') == datetime.datetime(2013, 2, 26, 8, 23, 12)
-    assert datetime_tuple('2013-02-26T07:23:12+02:00') == datetime.datetime(2013, 2, 26, 6, 23, 12)
-
 def test_utc_time():
     delta = timedelta(seconds=-3600)
     d = datetime.datetime.now()
diff --git a/test/test_xhtml.py b/test/test_xhtml.py
index 58857d67..5afd08ff 100644
--- a/test/test_xhtml.py
+++ b/test/test_xhtml.py
@@ -38,6 +38,14 @@ def test_xhtml_to_poezio_colors():
     xhtml = start + b'<a href="http://perdu.com">http://perdu.com</a>' + end
     assert xhtml_to_poezio_colors(xhtml) == '\x19uhttp://perdu.com\x19o'
 
+    xhtml = b'<div style="font-weight:bold">Allo <div style="color:red">test <div style="color: blue">test2</div></div></div>'
+    assert xhtml_to_poezio_colors(xhtml, force=True) == '\x19bAllo \x19196}test \x1921}test2\x19o'
+
+    xhtml = (b'<div style="color:blue"><div style="color:yellow">'
+             b'<div style="color:blue">Allo <div style="color:red">'
+             b'test <div style="color: blue">test2</div></div></div></div></div>')
+    assert xhtml_to_poezio_colors(xhtml, force=True) == '\x1921}Allo \x19196}test \x1921}test2\x19o'
+
     with pytest.raises(xml.sax._exceptions.SAXParseException):
         xhtml_to_poezio_colors(b'<p>Invalid xml')