Minor changes/clean-up

goatools/obo_parser.py

@@ -124,7 +124,6 @@ def __iter__(self):
         with open(self.obo_file) as fstream:
             rec_curr = None # Stores current GO Term
             for lnum, line in enumerate(fstream):
-                line = line.rstrip() # chomp
                 # obo lines start with any of: [Term], [Typedef], /^\S+:/, or /^\s*/
                 if line[0:6] == "[Term]":
                     rec_curr = self._init_GOTerm_ref(rec_curr, "Term", lnum)
@@ -138,6 +137,7 @@ def __iter__(self):
                 #   def: "The maintenance of ...
                 #   is_a: GO:0007005 ! mitochondrion organization
                 elif rec_curr is not None:
+                    line = line.rstrip() # chomp
                     if ":" in line:
                         self._add_to_ref(rec_curr, line, lnum)
                     elif line == "":
@@ -146,6 +146,9 @@ def __iter__(self):
                             rec_curr = None
                     else:
                         self._die("UNEXPECTED LINE CONTENT: {L}".format(L=line), lnum)
+            # Return last record, if necessary
+            if rec_curr is not None:
+                yield rec_curr
 
     def _init_GOTerm_ref(self, rec_curr, name, lnum):
         """Initialize new reference and perform checks."""
@@ -157,7 +160,7 @@ def _init_GOTerm_ref(self, rec_curr, name, lnum):
 
     def _add_to_ref(self, rec_curr, line, lnum):
         """Add new fields to the current reference."""
-        mtch = re.match(r'^(\S+):\s*(\S.*)\s*$', line)
+        mtch = re.match(r'^(\S+):\s*(\S.*)$', line)
         if mtch:
             field_name = mtch.group(1)
             field_value = mtch.group(2)

tests/test_oboreader.py

@@ -79,18 +79,18 @@ def test_run(msg, tag, use_alt=False):
 def test_oboreader_equal_(dag_fin):
     """Test that the contents of the original DAG and the alternate DAG are the same."""
     sys.stdout.write("\n\nTEST GODag EQUALITY USING {} ...\n\n".format(dag_fin))
-    # 1. Read the obo file using the original OBOReader.
-    tic = timeit.default_timer()
-    dag_orig = GODag(dag_fin)
-    sys.stdout.write("Original OBOReader Elapsed HMS: {}\n\n".format(
-        str(datetime.timedelta(seconds=(timeit.default_timer()-tic)))))
-    # 2. Read the obo file using the alternate  OBOReader.
+    # 1. Read the obo file using the alternate  OBOReader.
     tic = timeit.default_timer()
     dag_alt = GODag(dag_fin, True)
     sys.stdout.write("Alternate OBOReader Elapsed HMS: {}\n\n".format(
         str(datetime.timedelta(seconds=(timeit.default_timer()-tic)))))
+    # 2. Read the obo file using the original OBOReader.
+    tic = timeit.default_timer()
+    dag_orig = GODag(dag_fin)
+    sys.stdout.write("Original OBOReader Elapsed HMS: {}\n\n".format(
+        str(datetime.timedelta(seconds=(timeit.default_timer()-tic)))))
     # 3. Test that the contents of each GODag are the same.
-    #if len(dag_orig) != len(dag_alt): raise Exception("LENGTHES NOT THE SAME")
+    if len(dag_orig) != len(dag_alt): raise Exception("LENGTHES NOT THE SAME")
     for goid, goorig in dag_orig.items():
         goalt = dag_alt[goid]
         if goorig.id != goalt.id: raise Exception("id MISMATCH")
@@ -125,6 +125,8 @@ def test_oboreader_equal_(dag_fin):
       dag_fin))
 
 
+
+
 #################################################################
 # Tests
 #################################################################