Changed parent_id so it could be a list

Entities need to be able to come from multiple parents so I updated the code to allow for that.

Author: ptth222

src/messes/convert/mwtab_functions.py

@@ -9,48 +9,64 @@
 
 
 def create_sample_lineages(input_json: dict, entity_table_name: str="entity", parent_key: str="parent_id") -> dict:
-    """Determine all the ancestors and siblings for each entity in the entity table.
+    """Determine all the ancestors, parents, and siblings for each entity in the entity table.
     
     The returned dictionary is of the form:
     
     {entity_id:{"ancestors":[ancestor0, ancestor1, ...],
+                "parents":[parent0, parent1, ...],
                 "siblings":[sibling0, sibling1, ...]}
      ...
     }
     
+    parents are the immediate ancestors an entity comes from. They are also included in the ancestors list.
+    
     Args:
         input_json: the dictionary where the entity table is.
         entity_table_name: the name of the entity table in input_json.
         parent_key: the field name for the field that points to the entity's parent.
         
     Returns:
         a dictionary where the keys are the entity ids and the values are a dictionary 
-        of it's ancestors and siblings.
+        of it's ancestors, parents, and siblings.
     """
 
     lineages = {}
     for entity_name, entity_attributes in input_json[entity_table_name].items():
         ancestors = []
-        while parent_name := entity_attributes.get(parent_key):
-            ancestors.append(parent_name)
-            if parent_name not in input_json[entity_table_name]:
-                print("Error: The parent entity, \"" + parent_name + "\", pulled from the entity \"" + entity_name + \
-                      "\" in the \"" + entity_table_name + "\" table is not in the \"" + entity_table_name + "\" table. " +\
-                      "Parent entities must be in the table with thier children.", file=sys.stderr)
-                sys.exit()
-            entity_attributes = input_json[entity_table_name][parent_name]
-        ancestors.reverse()
-        lineages[entity_name] = {"ancestors":ancestors}
+        immediate_parents = []
+        if parents := entity_attributes.get(parent_key):
+            parents = parents if isinstance(parents, list) else [parents]
+            immediate_parents = parents
+            next_parents = parents
+            while next_parents:
+                parents = next_parents
+                next_parents = []
+                for parent_name in parents:
+                    ancestors.append(parent_name)
+                    if parent_name not in input_json[entity_table_name]:
+                        print("Error: The parent entity, \"" + parent_name + "\", pulled from the entity \"" + entity_name + \
+                              "\" in the \"" + entity_table_name + "\" table is not in the \"" + entity_table_name + "\" table. " +\
+                              "Parent entities must be in the table with thier children.", file=sys.stderr)
+                        sys.exit()
+                    if grandparents := input_json[entity_table_name][parent_name].get(parent_key):
+                        grandparents = grandparents if isinstance(grandparents, list) else [grandparents]
+                        next_parents += grandparents
+            ancestors.reverse()
+        
+        lineages[entity_name] = {"ancestors": ancestors, "parents": immediate_parents}
 
     for entity_name in lineages:
         siblings = []
         if not lineages[entity_name]["ancestors"]:
             lineages[entity_name]["siblings"] = []
             continue
-        parent_name = lineages[entity_name]["ancestors"][-1]
+        parents = lineages[entity_name]["parents"]
         for sibling_name, entity_attributes in input_json[entity_table_name].items():
-            if (sibling_parent_name := entity_attributes.get(parent_key)) and sibling_parent_name == parent_name:
-                siblings.append(sibling_name)
+            if sibling_name != entity_name and (sibling_parents := entity_attributes.get(parent_key)):
+                sibling_parents = sibling_parents if isinstance(sibling_parents, list) else [sibling_parents]
+                if set(sibling_parents).intersection(parents):
+                    siblings.append(sibling_name)
 
         lineages[entity_name]["siblings"] = siblings
 

src/messes/validate/validate.py

@@ -964,43 +964,57 @@ def validate_parent_id(table: dict, table_name: str, entity_name: str, check_typ
     """
     has_errors = False
     for entity, attributes in table.items():
-        if parent_name := attributes.get("parent_id"):
+        if parents := attributes.get("parent_id"):
             ancestors = []
-            parent_attributes = attributes
-            while (parent_name := parent_attributes.get("parent_id")):
-                if parent_name in ancestors:
+            parents = parents if isinstance(parents, list) else [parents]
+            no_errors = True
+            next_parents = parents
+            while next_parents and no_errors:
+                parents = next_parents
+                next_parents = []
+                for parent_name in parents:
+                    ancestors.append(parent_name)
+                    if parent_name in table and (grandparents := table[parent_name].get("parent_id")):
+                        grandparents = grandparents if isinstance(grandparents, list) else [grandparents]
+                        for grandparent_name in grandparents:
+                            if grandparent_name in ancestors:
+                                ## Don't print the circular ancestry message if an entity has itself as a parent.
+                                ## This is specifically checked for just below this.
+                                if grandparent_name != entity:
+                                    print("Error:  The " + entity_name + ", \"" + entity + "\", in the \"" + \
+                                          table_name + "\" table has a circular ancestry, i.e., somewhere in the lineage a " + \
+                                          entity_name + " has a \"parent_id\" to a child in the lineage.", file=sys.stderr)
+                                has_errors = True
+                                no_errors = False
+                                break
+                            next_parents.append(grandparent_name)
+                    if not no_errors:
+                        break
+                                    
+            parents = attributes["parent_id"]
+            parents = parents if isinstance(parents, list) else [parents]
+            for parent_name in parents:
+                if parent_name == entity:
                     print("Error:  The " + entity_name + ", \"" + entity + "\", in the \"" + \
-                          table_name + "\" table has a circular ancestry, i.e., somewhere in the lineage a " + \
-                          entity_name + " has a \"parent_id\" to a child in the lineage.", file=sys.stderr)
+                          table_name + "\" table has itself listed in its parent_id. " + \
+                          "Records cannot be their own parents.", file=sys.stderr)
+                    has_errors = True
+                
+                if parent_name not in table:
+                    print("Error:  The parent " + entity_name + ", \"" + parent_name + \
+                          "\", for the " + entity_name + " \"" + entity + \
+                          "\" in the \"" + table_name + "\" table is not itself in the \"" + \
+                          table_name + "\" table. " +\
+                          "Parent entities must be in the table as well.", file=sys.stderr)
+                    has_errors = True
+                            
+                elif check_type and (type_to_check := attributes.get(type_keyword)) and \
+                   (parent_type := table[parent_name].get(type_keyword)) and \
+                   type_to_check != parent_type:
+                    print("Error:  The " + entity_name + ", \"" + entity + \
+                          "\", does not have the same " + type_keyword + " as its parent \"" + \
+                          parent_name + "\".", file=sys.stderr)
                     has_errors = True
-                    break
-                ancestors.append(parent_name)
-                if not parent_name in table:
-                    break
-                parent_attributes = table[parent_name]
-            
-            parent_name = attributes["parent_id"]
-            if parent_name == entity:
-                print("Error:  The " + entity_name + ", \"" + entity + "\", in the \"" + \
-                      table_name + "\" table has itself listed for its parent_id. " + \
-                      "Records cannot be their own parents.", file=sys.stderr)
-                has_errors = True
-            
-            if parent_name not in table:
-                print("Error:  The parent " + entity_name + ", \"" + parent_name + \
-                      "\", for the " + entity_name + " \"" + entity + \
-                      "\" in the \"" + table_name + "\" table is not itself in the \"" + \
-                      table_name + "\" table. " +\
-                      "Parent entities must be in the table as well.", file=sys.stderr)
-                has_errors = True
-                        
-            elif check_type and (type_to_check := attributes.get(type_keyword)) and \
-               (parent_type := table[parent_name].get(type_keyword)) and \
-               type_to_check != parent_type:
-                print("Error:  The " + entity_name + ", \"" + entity + \
-                      "\", does not have the same " + type_keyword + " as its parent \"" + \
-                      parent_name + "\".", file=sys.stderr)
-                has_errors = True
 
     return has_errors
 

src/messes/validate/validate_schema.py

@@ -87,7 +87,7 @@
                      "type":"object",
                      "properties":{
                          "id": {"type":"string", "minLength":1},
-                         "parent_id": {"type":"string"},
+                         "parent_id": {"type":["string", "array"]},
                          "protocol.id": {"type":["string", "array"], "minItems":1, "items":{"type":"string", "minLength":1}, "minLength":1},
                          "type": {"type":"string", "enum":["sample", "subject"]}
                          },

tests/test_validate/test_validate_integration.py

@@ -381,7 +381,7 @@ def test_parent_id_errors():
 
     errors = [
         'Error:  The protocol, "IC-FTMS_preparation", in the "protocol" table has itself ' +\
-        'listed for its parent_id. Records cannot be their own parents.',
+        'listed in its parent_id. Records cannot be their own parents.',
 
         'Error:  The protocol, "ICMS1", does not have the same type as its parent "IC-FTMS_preparation".',