Merge pull request #74 from MJedr/authors

dict merger: dedupe only remove patches

Author: drjova

json_merger/__init__.py

@@ -119,7 +119,7 @@
 >>> m = Merger(root, head, update, DictMergerOps.FALLBACK_KEEP_HEAD,
 ...            UnifierOps.KEEP_ONLY_HEAD_ENTITIES)
 >>> m.merge() # No conflict here
->>> m.merged_root['badges'] == ['cool', 'nice']
+>>> m.merged_root['badges'] == ['cool', 'nice', 'random']
 True
 >>> m = Merger(root, head, update, DictMergerOps.FALLBACK_KEEP_HEAD,
 ...            UnifierOps.KEEP_ONLY_UPDATE_ENTITIES)
@@ -135,12 +135,12 @@
 >>> m = Merger(root, head, update, DictMergerOps.FALLBACK_KEEP_HEAD,
 ...            UnifierOps.KEEP_UPDATE_AND_HEAD_ENTITIES_HEAD_FIRST)
 >>> m.merge() # No conflict here
->>> m.merged_root['badges'] == ['cool', 'fun', 'nice', 'healthy']
+>>> m.merged_root['badges'] == ['cool', 'fun', 'nice', 'random', 'healthy']
 True
 >>> m = Merger(root, head, update, DictMergerOps.FALLBACK_KEEP_HEAD,
 ...            UnifierOps.KEEP_UPDATE_AND_HEAD_ENTITIES_UPDATE_FIRST)
 >>> m.merge()
->>> m.merged_root['badges'] == ['fun', 'cool', 'nice', 'healthy']
+>>> m.merged_root['badges'] == ['fun', 'cool', 'nice', 'healthy', 'random']
 True
 
 These options keep the order relations between the entities. For example,

json_merger/dict_merger.py

@@ -139,14 +139,14 @@ def conflicts(self):
     def _merge_base_values(self):
         if self.head == self.update:
             self.merged_root = self.head
-        elif self.head == self.root:
-            self.merged_root = self.update
-        elif self.update == self.root:
-            self.merged_root = self.head
         elif self.head == NOTHING:
             self.merged_root = self.update
         elif self.update == NOTHING:
             self.merged_root = self.head
+        elif self.head == self.root:
+            self.merged_root = self.update
+        elif self.update == self.root:
+            self.merged_root = self.head
         else:
             strategy = self._get_rule_for_field(self.key_path)
             self.merged_root, conflict = {
@@ -171,10 +171,19 @@ def _merge_dicts(self):
             self._solve_dict_conflicts(non_list_merger, e.content)
 
         self._restore_lists()
+        remove_patches = []
+        other_patches = []
+        for patch_ in non_list_merger.unified_patches:
+            if patch_[0] == 'remove':
+                remove_patches.append(patch_)
+            else:
+                other_patches.append(patch_)
+        remove_patches_deduped = dedupe_list(remove_patches)
+        unified_patches = remove_patches_deduped + other_patches
         self.merged_root = patch(
-            dedupe_list(non_list_merger.unified_patches),
-            self.root
-        )
+                unified_patches,
+                self.root
+            )
 
     def _solve_dict_conflicts(self, non_list_merger, conflicts):
         strategies = [self._get_custom_strategy(conflict)

json_merger/merger.py

@@ -33,7 +33,6 @@
 from .dict_merger import SkipListsMerger
 from .errors import MergeError
 from .list_unify import ListUnifier
-from .nothing import NOTHING
 from .utils import (
     get_conf_set_for_key_path, get_dotted_key_path, get_obj_at_key_path,
     set_obj_at_key_path
@@ -262,8 +261,7 @@ def _recursive_merge(self, root, head, update, key_path=()):
                 )
                 new_obj = self._recursive_merge(root_obj, head_obj, update_obj,
                                                 absolute_key_path + (idx, ))
-                if new_obj != NOTHING:
-                    new_list.append(new_obj)
+                new_list.append(new_obj)
 
             root = set_obj_at_key_path(root, list_field, new_list)
             self._build_aligned_lists_and_stats(unifier, absolute_key_path)

tests/unit/test_dict_merger.py

@@ -87,18 +87,6 @@ def test_merge_with_nothing():
     m.merge()
     assert m.merged_root == {'some': 'other object'}
 
-    m = SkipListsMerger({'some': 'other object'},
-                        NOTHING, {'some': 'other object'},
-                        DictMergerOps.FALLBACK_KEEP_HEAD)
-    m.merge()
-    assert m.merged_root == NOTHING
-
-    m = SkipListsMerger({'some': 'other object'},
-                        {'some': 'other object'}, NOTHING,
-                        DictMergerOps.FALLBACK_KEEP_HEAD)
-    m.merge()
-    assert m.merged_root == NOTHING
-
 
 def test_simple_conflicts_keep_head():
     r = {}

tests/unit/test_merger.py

@@ -218,38 +218,7 @@ def test_keep_head_conflict_on_new_update():
     with pytest.raises(MergeError):
         m.merge()
 
-    expected_merge = [2]
+    expected_merge = [1, 2]
     expected_conflict = [('INSERT', (0,), 3)]
     assert m.merged_root == expected_merge
     assert m.conflicts == expected_conflict
-
-
-def test_merger_doesnt_add_elements_removed_in_head_when_root_matches_update():
-    update = {
-        'ids': [
-            {'schema': 'INSPIRE ID', 'value': 'INSPIRE-00231894'},
-            {'schema': 'ORCID', 'value': '0000-0002-0165-6297'}
-        ]
-    }
-    head = {
-        'ids': [
-            {'value': 'INSPIRE-00231894', 'schema': 'INSPIRE ID'},
-            {'value': 'V.Nikolaenko.1', 'schema': 'INSPIRE BAI'}
-        ]
-    }
-    root = {
-        'ids': [
-            {'schema': 'INSPIRE ID', 'value': 'INSPIRE-00231894'},
-            {'schema': 'ORCID', 'value': '0000-0002-0165-6297'}
-        ]
-    }
-
-    m = Merger(root, head, update,
-               DictMergerOps.FALLBACK_KEEP_HEAD,
-               UnifierOps.KEEP_UPDATE_AND_HEAD_ENTITIES_UPDATE_FIRST)
-
-    m.merge()
-    assert m.merged_root['ids'] == [
-        {'value': 'INSPIRE-00231894', 'schema': 'INSPIRE ID'},
-        {'value': 'V.Nikolaenko.1', 'schema': 'INSPIRE BAI'}
-    ]