diff --git a/oml/media/epub.py b/oml/media/epub.py
index 4934940..c367283 100644
--- a/oml/media/epub.py
+++ b/oml/media/epub.py
@@ -178,7 +178,9 @@ def extract_text(path):
     for f in z.filelist:
         if '/._' in f.filename or f.filename.startswith('._'):
             continue
-        if f.filename.endswith('html'):
+        if 'META-INF' in f.filename:
+            continue
+        if f.filename.split('.')[-1] in ('html', 'xml', 'htm'):
             data += z.read(f.filename).decode('utf-8', 'ignore')
     return data