Fix bug: Pfam retriever crashes when no go_terms are available (fixes #20)

dbxref/retrieve/pfam.py

@@ -63,14 +63,15 @@ def read_annotation(entry):
             'comment': entry.find('pfam:comment', ns).text.strip()
             }
     go_terms = entry.find('pfam:go_terms', ns)
-    categories = go_terms.findall('pfam:category', ns)
-    for category in categories:
-        terms = category.findall('pfam:term', ns)
-        for term in terms:
-            annotation['terms'].append({
-                'id': term.attrib['go_id'],
-                'description': term.text
-                })
+    if go_terms:
+      categories = go_terms.findall('pfam:category', ns)
+      for category in categories:
+          terms = category.findall('pfam:term', ns)
+          for term in terms:
+              annotation['terms'].append({
+                  'id': term.attrib['go_id'],
+                  'description': term.text
+                  })
     return annotation
 
 if __name__ == "__main__":

tests/test_pfam.py

+import unittest
+from dbxref import retriever, resolver
+
+class TestPfam(unittest.TestCase):
+
+    def test_no_go_terms(self):
+      '''regression test for missing go terms in pfam entry'''
+      documents = retriever.retrieve([{'db': 'pfam', 'id': 'PF10423.8'}])
+      # this test failed due to an error due to missing None handling, 
+      # so no assertions here. Once fixed. this should suffice