Merge pull request #42 from monarch-initiative/qc_diff_unit_tests

QC diff unit tests

cat_merge/qc_diff_utils.py

@@ -111,19 +111,30 @@ def diff_type(
 
 @validate_diff_args
 def diff_dict(a: Union[Dict, None], b: Union[Dict, None], flags: Dict) -> Dict:
+    if a is None or b is None:
+        change = True
+    else:
+        change = False
+
     diff = {}
     a = {} if a is None else a
     b = {} if b is None else b
-    missing = ""
-    change = False
 
-    for key in dict.fromkeys(list(a.keys()), list(b.keys())):
+    missing = ""
+    for key in dict.fromkeys(list(a.keys()) + list(b.keys())):
         if key not in a.keys():
             missing = "-"
         elif key not in b.keys():
             missing = "+"
-        diff_value = diff_type(a.get(key), b.get(key), flags)
-        if flags["change"] or flags["show_all"]:
+
+        if a.get(key) is None and b.get(key) is None:
+            diff_value = None
+            if not (key in a.keys() and key in b.keys()):
+                flags['change'] = True
+        else:
+            diff_value = diff_type(a.get(key), b.get(key), flags)
+
+        if flags["change"] or flags["show_all"] or missing != "":
             diff[missing + key] = diff_value
         change = any([change, flags["change"]])
         flags["change"] = False
@@ -137,6 +148,12 @@ def diff_list(a: Union[List, None], b: Union[List, None], flags: Dict) -> List:
     diff = []
     a = [] if a is None else a
     b = [] if b is None else b
+
+    # Check if either list contains a list
+    if any(isinstance(n, list) and len(n) == 0 for n in a + b):
+        message = "diff_list: found list containing list, structure not supported."
+        raise NotImplementedError(message)
+
     a_as_keys = dict(zip(a, a))
     b_as_keys = dict(zip(b, b))
 
@@ -230,6 +247,11 @@ def sources_dict(a: Union[Dict, List[Dict]]) -> Dict:
         case None:
             pass
         case list():
+            if not all(isinstance(x, dict) for x in a):
+                # all list entries must be dict
+                message = "sources_dict: List contains non-dict entries, aborting"
+                raise ValueError(message)
+
             for i in a:
                 if i.get("name") is not None:
                     a_dict[i.get("name")] = i

tests/test_utils.py

@@ -20,10 +20,12 @@ def value(df: DataFrame, id: str, column: str):
 
 
 def flags_params():
-    params = {"change_T_show_T": {"change": True, "show_all": True},
-              "change_F_show_T": {"change": False, "show_all": True},
-              "change_T_show_F": {"change": True, "show_all": False},
-              "change_F_show_F": {"change": False, "show_all": False}}
+    params = {
+        # "change_T_show_T": {"change": True, "show_all": True},
+        "change_F_show_T": {"change": False, "show_all": True},
+        # "change_T_show_F": {"change": True, "show_all": False},
+        "change_F_show_F": {"change": False, "show_all": False},
+    }
     return copy.deepcopy(params)
 
 

tests/unit/test_diff_dict.py

@@ -0,0 +1,105 @@
+import pytest
+from tests.test_utils import *
+from cat_merge.qc_diff_utils import diff_dict
+from typing import Dict
+
+
+@pytest.fixture
+def dict1() -> Dict:
+    return {'item1': None, 'item7': 'item7', 'item2': 'item2'}
+
+
+def test_diff_dict_none(dict1, flags):
+    assert diff_dict(None, dict1, flags) == {'-item1': None, '-item7': '-item7', '-item2': '-item2'}
+    assert flags["change"] is True
+    flags['change'] = False
+
+    assert diff_dict(dict1, None, flags) == {'+item1': None, '+item7': '+item7', '+item2': '+item2'}
+    assert flags["change"] is True
+
+
+@pytest.fixture
+def dict1_copy() -> Dict:
+    return {'item1': None, 'item7': 'item7', 'item2': 'item2'}
+
+
+def test_diff_dict_match(dict1, dict1_copy, flags):
+    if flags["show_all"]:
+        assert diff_dict(dict1, dict1_copy, flags) == {'item1': None, 'item7': 'item7', 'item2': 'item2'}
+        assert flags["change"] is False
+        assert diff_dict(dict1_copy, dict1, flags) == {'item1': None, 'item7': 'item7', 'item2': 'item2'}
+        assert flags["change"] is False
+    else:
+        assert diff_dict(dict1, dict1_copy, flags) == dict()
+        assert flags["change"] is False
+        assert diff_dict(dict1_copy, dict1, flags) == dict()
+        assert flags["change"] is False
+
+
+@pytest.fixture
+def empty_dict() -> Dict:
+    return dict()
+
+
+def test_diff_dict_empty(empty_dict, dict1, flags):
+    assert diff_dict(None, empty_dict, flags) == dict()
+    assert flags["change"] is True
+    flags['change'] = False
+
+    assert diff_dict(empty_dict, None, flags) == dict()
+    assert flags["change"] is True
+    flags['change'] = False
+
+    assert diff_dict(empty_dict, dict1, flags) == {'-item1': None, '-item7': '-item7', '-item2': '-item2'}
+    assert flags["change"] is True
+    flags['change'] = False
+
+    assert diff_dict(dict1, empty_dict, flags) == {'+item1': None, '+item7': '+item7', '+item2': '+item2'}
+    assert flags["change"] is True
+    flags['change'] = False
+
+
+@pytest.fixture
+def dict_val_none() -> Dict:
+    return {"one": None}
+
+
+def test_diff_dict_val_none(dict_val_none, empty_dict, flags):
+    assert diff_dict(dict_val_none, empty_dict, flags) == {"+one": None}
+    assert flags['change'] is True
+    flags["change"] = False
+
+    assert diff_dict(empty_dict, dict_val_none, flags) == {"-one": None}
+    assert flags['change'] is True
+    flags["change"] = False
+
+    assert diff_dict(None, dict_val_none, flags) == {"-one": None}
+    assert flags['change'] is True
+    flags["change"] = False
+
+    assert diff_dict(dict_val_none, None, flags) == {"+one": None}
+    assert flags['change'] is True
+    flags["change"] = False
+
+
+@pytest.fixture
+def dict2() -> Dict:
+    return {'item7': 'item7', 'item1': None, 'item3': 'item3'}
+
+
+def test_diff_dict_partial_match(dict1, dict2, flags):
+    if flags["show_all"]:
+        assert diff_dict(dict1, dict2, flags) == \
+               {'item1': None, 'item7': 'item7', '+item2': '+item2', '-item3': '-item3'}
+        assert flags["change"] is True
+
+        flags["change"] = False
+        assert list(diff_dict(dict1, dict2, flags).keys()) == ['item1', 'item7', '+item2', '-item3']
+        assert flags["change"] is True
+    else:
+        assert diff_dict(dict1, dict2, flags) == {'+item2': '+item2', '-item3': '-item3'}
+        assert flags["change"] is True
+
+        flags["change"] = False
+        assert list(diff_dict(dict1, dict2, flags).keys()) == ['+item2', '-item3']
+        assert flags["change"] is True