diff --git a/npybabel.py b/npybabel.py
index d2a5553fc80..0c79e5d7d1e 100755
--- a/npybabel.py
+++ b/npybabel.py
@@ -3,12 +3,15 @@
 __requires__ = 'Babel==0.9.6'
 import sys
 from pkg_resources import load_entry_point
+import re
 
 if __name__ == '__main__':
     sys.exit(
         load_entry_point('Babel==0.9.6', 'console_scripts', 'pybabel')()
     )
     
+QWEB_EXPR = re.compile(r"""(?:\< *t\-tr *\>(.*?)\< *\/t\-tr *\>)|(?:\_t *\( *((?:\".*?\")|(?:\'.*?\')) *\))""")
+    
 def extract_qweb(fileobj, keywords, comment_tags, options):
     """Extract messages from XXX files.
     :param fileobj: the file-like object the messages should be extracted
@@ -23,4 +26,15 @@ def extract_qweb(fileobj, keywords, comment_tags, options):
     :rtype: ``iterator``
     """
     content = fileobj.read()
-    return []
+    found = QWEB_EXPR.finditer(content)
+    result = []
+    index = 0
+    line_nbr = 0
+    for f in found:
+        group = 1 if f.group(1) else 2
+        while index < f.start():
+            if content[index] == "\n":
+                line_nbr += 1
+            index += 1
+        result.append((line_nbr, None, f.group(group), ""))
+    return result