Allow duplicate datasets in input

Even if two Datasets that are read in have
exactly the same information, do not treat them
as equal. This preserves more of the input file
structure (rather than combining identical datasets
into a single ID). Closes #127.

ihm/dataset.py

@@ -18,12 +18,16 @@ class Dataset(object):
     """
 
     _eq_keys = ['location']
+    _allow_duplicates = False
 
-    # Datasets compare equal iff they are the same class and have the
-    # same attributes
+    # Datasets compare equal iff they are the same class, have the
+    # same attributes, and allow_duplicates=False
     def _eq_vals(self):
-        return tuple([self.__class__]
-                     + [getattr(self, x) for x in self._eq_keys])
+        if self._allow_duplicates:
+            return id(self)
+        else:
+            return tuple([self.__class__]
+                         + [getattr(self, x) for x in self._eq_keys])
 
     def __eq__(self, other):
         return self._eq_vals() == other._eq_vals()

ihm/reader.py

@@ -1384,6 +1384,7 @@ def __call__(self, data_type, id, details):
         f = self.sysr.datasets.get_by_id(
             id, self.type_map.get(typ, ihm.dataset.Dataset))
         f.details = details
+        f._allow_duplicates = True
 
 
 class _DatasetGroupHandler(Handler):

test/test_dataset.py

@@ -40,6 +40,17 @@ def test_dataset(self):
         d.parents.append(td)
         self.assertEqual(len(d.parents), 2)
 
+    def test_dataset_allow_duplicates(self):
+        """Test Dataset base class with allow_duplicates=True"""
+        loc = ihm.location.PDBLocation('1abc', version='foo', details='bar')
+        d = ihm.dataset.Dataset(loc)
+        d._allow_duplicates = True
+        self.assertEqual(d._eq_vals(), id(d))
+
+        d2 = ihm.dataset.Dataset(loc)
+        d2._allow_duplicates = True
+        self.assertNotEqual(d, d2)
+
     def test_add_primary_no_parents(self):
         """Test add_primary() method, no parents"""
         l1 = ihm.location.PDBLocation('1abc', version='foo', details='bar')

test/test_reader.py

@@ -971,6 +971,9 @@ def test_dataset_list_handler(self):
             self.assertEqual(d1.__class__, ihm.dataset.PDBDataset)
             self.assertEqual(d2.__class__, ihm.dataset.ComparativeModelDataset)
             self.assertEqual(d3.__class__, ihm.dataset.EMMicrographsDataset)
+            self.assertTrue(d1._allow_duplicates)
+            self.assertTrue(d2._allow_duplicates)
+            self.assertTrue(d3._allow_duplicates)
             # No specified data type - use base class
             self.assertEqual(d4.__class__, ihm.dataset.Dataset)
             self.assertIsNone(d1.details)