fix: entry removal fix

Author: alexgarel

Makefile

@@ -182,7 +182,7 @@ config_quality: ## Run quality checks on configuration files
 
 tests: backend_tests ## Run all tests
 
-backend_tests: ## Run python tests, you might provide additional arguments witr args="…"
+backend_tests: ## Run python tests, you might provide additional arguments with args="…"
 	@echo "🍜 Running python tests"
 	${DOCKER_COMPOSE_TEST} up -d neo4j
 	${DOCKER_COMPOSE_TEST}  run --rm taxonomy_api pytest /parser ${args}

backend/editor/entries.py

@@ -418,26 +418,36 @@ async def delete_node(self, label: NodeType, entry):
             // Find node to be deleted using node ID
             MATCH (deleted_node:{self.project_name}:{label.value})-[:is_before]->(next_node)
                 WHERE deleted_node.id = $id
-            MATCH (previous_node)-[:is_before]->(deleted_node)
+            MATCH (previous_node)-[r:is_before]->(deleted_node)
             // Remove node
-            DETACH (deleted_node)
+            DELETE r
             // Rebuild relationships after deletion
             CREATE (previous_node)-[:is_before]->(next_node)
         """
         await get_current_transaction().run(query, {"id": entry})
         # transfert child parent relations, and mark child nodes as modified
         query = f"""
             // Find relations to be removed using node ID
-            MATCH (child_node)-[:is_child_of]->(deleted_node:{self.project_name}:{label.value})
+            MATCH (child_node)-[r:is_child_of]->(deleted_node:{self.project_name}:{label.value})
                 WHERE deleted_node.id = $id
             MATCH (deleted_node)-[:is_child_of]->(parent_node)
-            DETACH (deleted_node)
+            DELETE r
             // transfer child
             CREATE (child_node) -[:is_child_of]->(parent_node)
             // mark modified
             SET child_node.modified = $modified
         """
         await get_current_transaction().run(query, {"id": entry, "modified": modified})
+        # or if no transfer is needed, just mark modified
+        query = f"""
+            // Find relations to be removed using node ID
+            MATCH (child_node)-[r:is_child_of]->(deleted_node:{self.project_name}:{label.value})
+                WHERE deleted_node.id = $id
+            DELETE r
+            // mark children as modified
+            SET child_node.modified = $modified
+        """
+        await get_current_transaction().run(query, {"id": entry, "modified": modified})
         # change label of node to be deleted
         query = f"""
             MATCH (deleted_node:{self.project_name}:{label.value})
@@ -447,7 +457,7 @@ async def delete_node(self, label: NodeType, entry):
             // and mark modification date also
             SET deleted_node.modified = $modified
         """
-        result = await get_current_transaction().run(query, {"id": entry})
+        result = await get_current_transaction().run(query, {"id": entry, "modified": modified})
         return await async_list(result)
 
     async def get_all_nodes(self, label: Optional[NodeType] = None, removed: bool = False):

backend/tests/data/test.txt

@@ -73,9 +73,9 @@ en: roast-beef
 
 <en: meat
 en: fake-meat
-vegan:en: yes
 # undef will stop parents from transmitting a value
 carbon_footprint_fr_foodges_value:fr: undef
+vegan:en: yes
 
 en: fake-stuff
 

backend/tests/test_export.py

@@ -322,3 +322,41 @@ async def test_change_entry_id(taxonomy_test):
     assert result == expected
     # clean files
     await background_tasks.run()
+
+
+@pytest.mark.anyio
+async def test_remove_entry(taxonomy_test):
+    async with graph_db.TransactionCtx():
+        # remove "yaourts allégés"
+        await taxonomy_test.delete_node(NodeType.ENTRY, "fr:yaourts-alleges")
+        # remove meat
+        await taxonomy_test.delete_node(NodeType.ENTRY, "en:meat")
+    background_tasks = FakeBackgroundTask()
+    file_path = taxonomy_test.dump_taxonomy(background_tasks)
+    result = list(open(file_path))
+    # expected output
+    expected = list(open("tests/data/test.txt"))
+    # entry removed
+    to_remove = [
+        32,
+        33,
+        34,  # yahourts alleges
+        62,
+        63,
+        64,
+        65,  # meat
+        66,
+        73,  # < en:meat
+    ]
+    # parent changed to "en:yogurts" for "yahourts alleges"
+    expected[45] = expected[51] = "< en:yogurts\n"
+    # parent normalized
+    expected[44] = "< en:Passion fruit yogurts\n"
+    expected[50] = "< en:lemon yogurts\n"
+    # properties reordered
+    # expected.insert(39, expected.pop(41))
+    for i in sorted(to_remove, reverse=True):
+        del expected[i]
+    assert result == expected
+    # clean files
+    await background_tasks.run()

parser/openfoodfacts_taxonomy_parser/patcher.py

@@ -20,13 +20,21 @@ def is_removed(self, node):
     def get_all_nodes(self, project_label):
         """Get modified and removed nodes, in the start line  order"""
         query = f"""
-            MATCH (n:({project_label}|REMOVED_{project_label}))
+            MATCH (n:{project_label})
             WHERE
                 // no external node
                 (n.is_external = false OR n.is_external IS NULL)
                 AND (
                     // modified nodes
-                    ((n:TEXT OR n:SYNONYMS OR n:STOPWORDS OR n:ENTRY) AND n.modified IS NOT NULL)
+                    (
+                        (
+                            n:TEXT OR n:SYNONYMS OR n:STOPWORDS OR n:ENTRY OR
+                            n:REMOVED_TEXT OR n:REMOVED_SYNONYMS OR
+                            n:REMOVED_STOPWORDS OR n:REMOVED_ENTRY
+                        )
+                        AND
+                        n.modified IS NOT NULL
+                    )
                 )
             // optional match for node might not have parents
             OPTIONAL