matcher: Add uids for all the matches

Signed-off-by: Mihai Bivol <mm.bivol@gmail.com>

json_merger/graph_builder.py

@@ -27,7 +27,6 @@
 import six
 
 from .comparator import DefaultComparator
-from .conflict import Conflict, ConflictType
 from .nothing import NOTHING
 
 FIRST = 'first'
@@ -61,14 +60,28 @@ def __init__(self, lst, root):
         self.in_result_idx = set()
         self.not_in_result_idx = set(range(len(lst)))
         self.not_in_result_root_match_idx = set()
+        self.next_root_match_uid = 0
 
-    def move_to_result(self, lst_idx):
+        # For a given index in the initial list retrieve the match uid.
+        self.match_uids = {}
+        # For a given index in the initial list retrieve root match uid.
+        self.lst_root_match_uids = {}
+        # For a given index in the root list retrieve root match uid.
+        self.root_root_match_uids = {}
+
+    def move_to_result(self, lst_idx, match_uid):
         self.in_result_idx.add(lst_idx)
         self.not_in_result_idx.remove(lst_idx)
+        self.match_uids[lst_idx] = match_uid
+
+    def add_root_match(self, lst_idx, root_idx):
+        self.lst_root_match_uids[lst_idx] = self.next_root_match_uid
+        self.root_root_match_uids[root_idx] = self.next_root_match_uid
+        self.next_root_match_uid += 1
 
-    def add_root_match(self, lst_idx):
         if lst_idx in self.in_result_idx:
             return
+
         self.not_in_result_root_match_idx.add(lst_idx)
 
     @property
@@ -127,6 +140,7 @@ def __init__(self, root, head, update, sources,
         self._update_idx_to_node = {}
 
         self._next_node_id = 0
+        self.match_uids = {}
 
     def _new_node_id(self):
         node_id = self._next_node_id
@@ -178,21 +192,26 @@ def _populate_nodes(self):
                 self._push_node(root_elem, head_elem, update_elem)
 
     def _build_stats(self):
-        for root_idx, head_idx, update_idx in self._node_src_indices.values():
+        match_uid = 0
+        for node_id, indices in self._node_src_indices.items():
+            root_idx, head_idx, update_idx = indices
+            match_uid += 1
+
             if head_idx >= 0:
-                self.head_stats.move_to_result(head_idx)
+                self.head_stats.move_to_result(head_idx, match_uid)
             if update_idx >= 0:
-                self.update_stats.move_to_result(update_idx)
+                self.update_stats.move_to_result(update_idx, match_uid)
+            self.match_uids[node_id] = match_uid
 
-        for idx in self.head_stats.not_in_result_idx:
+        for idx in range(len(self.head)):
             root_idx, root = self._get_match('root', 'head', idx)
             if root_idx >= 0:
-                self.head_stats.add_root_match(idx)
+                self.head_stats.add_root_match(idx, root_idx)
 
-        for idx in self.update_stats.not_in_result_idx:
+        for idx in range(len(self.update)):
             root_idx, root = self._get_match('root', 'update', idx)
             if root_idx >= 0:
-                self.head_stats.add_root_match(idx)
+                self.update_stats.add_root_match(idx, root_idx)
 
     def build_graph(self):
         self._populate_nodes()

json_merger/list_unify.py

@@ -33,7 +33,6 @@
     GraphBuilderError, ListMatchGraphBuilder, sort_cyclic_graph_best_effort,
     toposort
 )
-from .nothing import NOTHING
 
 _OPERATIONS = [
     'KEEP_ONLY_HEAD_ENTITIES',
@@ -87,7 +86,6 @@ def __init__(self, root, head, update, operation,
         self.update_stats = None
 
         # Wether to raise error on deleting a head entity.
-        # TODO implement this
         self.raise_on_head_delete = operation in _RAISE_ERROR_OPS
         # Sources from which to keep entities.
         self.sources = _SOURCES[operation]
@@ -96,6 +94,7 @@ def __init__(self, root, head, update, operation,
         self.pick_first = _PICK_FIRST[operation]
 
         self.unified = []
+        self.match_uids = []
 
     def unify(self):
         graph_builder = ListMatchGraphBuilder(
@@ -106,13 +105,14 @@ def unify(self):
         except GraphBuilderError as e:
             # Can't partially recover from this, just keep self.head and call.
             # For manual alignment with self.update.
+            # TODO better fallback from this
             self.unified = [(h, h, h) for h in self.head]
             raise MergeError(e.message,
                              [Conflict(ConflictType.MANUAL_MERGE, (),
                                        self.update)])
-
-        self.head_stats = graph_builder.head_stats
-        self.update_stats = graph_builder.update_stats
+        finally:
+            self.head_stats = graph_builder.head_stats
+            self.update_stats = graph_builder.update_stats
 
         try:
             node_order = toposort(graph, self.pick_first)
@@ -123,6 +123,7 @@ def unify(self):
         finally:
             for node in node_order:
                 self.unified.append(nodes[node])
+                self.match_uids.append(graph_builder.match_uids[node])
             if (self.raise_on_head_delete and
                     self.head_stats.not_in_result_not_root_match_idx):
                 removed = self.head_stats.not_in_result_not_root_match

json_merger/merger.py

@@ -53,6 +53,9 @@ def __init__(self, root, head, update,
         self.root = copy.deepcopy(root)
         self.head = copy.deepcopy(head)
         self.update = copy.deepcopy(update)
+        self.match_uids = {}
+        self.head_stats = {}
+        self.update_stats = {}
 
         self.conflicts = []
         self.merged_root = None
@@ -74,7 +77,7 @@ def _merge_objects(self, root, head, update, key_path):
             self.conflicts.extend(c.with_prefix(key_path) for c in e.content)
         return object_merger
 
-    def _merge_lists(self, root, head, update, key_path):
+    def _unify_lists(self, root, head, update, key_path):
         dotted_key_path = get_dotted_key_path(key_path, True)
 
         operation = self.list_merge_ops.get(dotted_key_path,
@@ -88,14 +91,7 @@ def _merge_lists(self, root, head, update, key_path):
         except MergeError as e:
             self.conflicts.extend(c.with_prefix(key_path) for c in e.content)
 
-        new_root_list = []
-        for idx, objects in enumerate(list_unifier.unified):
-            root_obj, head_obj, update_obj = objects
-            new_obj = self._recursive_merge(root_obj, head_obj, update_obj,
-                                            key_path + (idx, ))
-            new_root_list.append(new_obj)
-
-        return new_root_list
+        return list_unifier
 
     def _recursive_merge(self, root, head, update, key_path=()):
         dotted_key_path = get_dotted_key_path(key_path, True)
@@ -117,8 +113,20 @@ def _recursive_merge(self, root, head, update, key_path=()):
             head_l = get_obj_at_key_path(head, list_field, [])
             update_l = get_obj_at_key_path(update, list_field, [])
 
-            new_list = self._merge_lists(root_l, head_l, update_l,
-                                         absolute_key_path)
+            unifier = self._unify_lists(root_l, head_l, update_l,
+                                        absolute_key_path)
+
+            new_list = []
+            for idx, objects in enumerate(unifier.unified):
+                root_obj, head_obj, update_obj = objects
+                new_obj = self._recursive_merge(root_obj, head_obj, update_obj,
+                                                absolute_key_path + (idx, ))
+                new_list.append(new_obj)
+
+            self.head_stats[absolute_key_path] = unifier.head_stats
+            self.update_stats[absolute_key_path] = unifier.update_stats
+            self.match_uids[absolute_key_path] = unifier.match_uids
+
             if list_field == ():
                 root = new_list
             else: