biocypher 0.5.17__py3-none-any.whl → 0.5.20__py3-none-any.whl

biocypher/__init__.py

@@ -13,14 +13,14 @@ BioCypher: a unifying framework for biomedical knowledge graphs.
 """
 
 __all__ = [
-    '__version__',
-    '__author__',
-    'module_data',
-    'config',
-    'logfile',
-    'log',
-    'Driver',
-    'BioCypher',
+    "__version__",
+    "__author__",
+    "module_data",
+    "config",
+    "logfile",
+    "log",
+    "Driver",
+    "BioCypher",
 ]
 
 from ._core import BioCypher
@@ -30,11 +30,10 @@ from ._metadata import __author__, __version__
 
 
 class Driver(BioCypher):
-
     # initialise parent class but log a warning
     def __init__(self, *args, **kwargs):
         logger.warning(
-            'The class `Driver` is deprecated and will be removed in a future '
-            'release. Please use `BioCypher` instead.'
+            "The class `Driver` is deprecated and will be removed in a future "
+            "release. Please use `BioCypher` instead."
         )
         super().__init__(*args, **kwargs)

biocypher/_config/__init__.py

@@ -23,10 +23,10 @@ import warnings
 import yaml
 import appdirs
 
-__all__ = ['module_data', 'module_data_path', 'read_config', 'config', 'reset']
+__all__ = ["module_data", "module_data_path", "read_config", "config", "reset"]
 
-_USER_CONFIG_DIR = appdirs.user_config_dir('biocypher', 'saezlab')
-_USER_CONFIG_FILE = os.path.join(_USER_CONFIG_DIR, 'conf.yaml')
+_USER_CONFIG_DIR = appdirs.user_config_dir("biocypher", "saezlab")
+_USER_CONFIG_FILE = os.path.join(_USER_CONFIG_DIR, "conf.yaml")
 
 
 class MyLoader(yaml.SafeLoader):
@@ -34,18 +34,18 @@ class MyLoader(yaml.SafeLoader):
         # Check if the scalar contains double quotes and an escape sequence
         value = super().construct_scalar(node)
         q = bool(node.style == '"')
-        b = bool('\\' in value.encode('unicode_escape').decode('utf-8'))
+        b = bool("\\" in value.encode("unicode_escape").decode("utf-8"))
         if q and b:
             warnings.warn(
                 (
-                    'Double quotes detected in YAML configuration scalar: '
+                    "Double quotes detected in YAML configuration scalar: "
                     f"{value.encode('unicode_escape')}. "
-                    'These allow escape sequences and may cause problems, for '
+                    "These allow escape sequences and may cause problems, for "
                     "instance with the Neo4j admin import files (e.g. '\\t'). "
-                    'Make sure you wanted to do this, and use single quotes '
-                    'whenever possible.'
+                    "Make sure you wanted to do this, and use single quotes "
+                    "whenever possible."
                 ),
-                category=UserWarning
+                category=UserWarning,
             )
         return value
 
@@ -57,7 +57,7 @@ def module_data_path(name: str) -> str:
 
     here = os.path.dirname(os.path.abspath(__file__))
 
-    return os.path.join(here, f'{name}.yaml')
+    return os.path.join(here, f"{name}.yaml")
 
 
 def module_data(name: str) -> Any:
@@ -71,11 +71,8 @@ def module_data(name: str) -> Any:
 
 
 def _read_yaml(path: str) -> Optional[dict]:
-
     if os.path.exists(path):
-
-        with open(path, 'r') as fp:
-
+        with open(path, "r") as fp:
             return yaml.load(fp.read(), Loader=MyLoader)
 
 
@@ -89,18 +86,22 @@ def read_config() -> dict:
     TODO explain path configuration
     """
 
-    defaults = module_data('biocypher_config')
+    defaults = module_data("biocypher_config")
     user = _read_yaml(_USER_CONFIG_FILE) or {}
     # TODO account for .yml?
-    local = _read_yaml('biocypher_config.yaml'
-                      ) or _read_yaml('config/biocypher_config.yaml') or {}
+    local = (
+        _read_yaml("biocypher_config.yaml")
+        or _read_yaml("config/biocypher_config.yaml")
+        or {}
+    )
 
     for key in defaults:
-
-        value = local[key] if key in local else user[key] if key in user else None
+        value = (
+            local[key] if key in local else user[key] if key in user else None
+        )
 
         if value is not None:
-            if type(defaults[key]) == str: # first level config (like title)
+            if type(defaults[key]) == str:  # first level config (like title)
                 defaults[key] = value
             else:
                 defaults[key].update(value)
@@ -114,20 +115,17 @@ def config(*args, **kwargs) -> Optional[Any]:
     """
 
     if args and kwargs:
-
         raise ValueError(
-            'Setting and getting values in the same call is not allowed.',
+            "Setting and getting values in the same call is not allowed.",
         )
 
     if args:
-
-        result = tuple(globals()['_config'].get(key, None) for key in args)
+        result = tuple(globals()["_config"].get(key, None) for key in args)
 
         return result[0] if len(result) == 1 else result
 
     for key, value in kwargs.items():
-
-        globals()['_config'][key].update(value)
+        globals()["_config"][key].update(value)
 
 
 def reset():
@@ -135,7 +133,7 @@ def reset():
     Reload configuration from the config files.
     """
 
-    globals()['_config'] = read_config()
+    globals()["_config"] = read_config()
 
 
 reset()

biocypher/_config/biocypher_config.yaml

@@ -102,6 +102,7 @@ postgresql:
   database_name: postgres # DB name
   user: postgres  # user name
   password: postgres  # password
+  host: localhost # host
   port: 5432 # port
 
   # PostgreSQL import batch writer settings
@@ -109,5 +110,3 @@ postgresql:
   delimiter: '\t'
   # import_call_bin_prefix: '' # path to "psql"
   # import_call_file_prefix: '/path/to/files'
-
-  

biocypher/_connect.py

@@ -13,7 +13,7 @@ BioCypher 'online' mode. Handles connection and manipulation of a running DBMS.
 """
 from ._logger import logger
 
-logger.debug(f'Loading module {__name__}.')
+logger.debug(f"Loading module {__name__}.")
 
 from typing import Optional
 from collections.abc import Iterable
@@ -27,10 +27,10 @@ from ._create import BioCypherEdge, BioCypherNode
 from ._ontology import Ontology
 from ._translate import Translator
 
-__all__ = ['_Neo4jDriver']
+__all__ = ["_Neo4jDriver"]
 
 
-class _Neo4jDriver():
+class _Neo4jDriver:
     """
     Manages a BioCypher connection to a Neo4j database using the
     ``neo4j_utils.Driver`` class.
@@ -53,11 +53,10 @@ class _Neo4jDriver():
 
         increment_version (bool): Whether to increment the version number.
 
-        ontology (Ontology): The ontology to use for mapping.
-
         translator (Translator): The translator to use for mapping.
 
     """
+
     def __init__(
         self,
         database_name: str,
@@ -65,15 +64,12 @@ class _Neo4jDriver():
         user: str,
         password: str,
         multi_db: bool,
-        ontology: Ontology,
         translator: Translator,
         wipe: bool = False,
         fetch_size: int = 1000,
         increment_version: bool = True,
     ):
-
-        self._ontology = ontology
-        self._translator = translator
+        self.translator = translator
 
         self._driver = neo4j_utils.Driver(
             db_name=database_name,
@@ -89,35 +85,30 @@ class _Neo4jDriver():
         # check for biocypher config in connected graph
 
         if wipe:
-
             self.init_db()
 
         if increment_version:
-
             # set new current version node
             self._update_meta_graph()
 
     def _update_meta_graph(self):
-
-        logger.info('Updating Neo4j meta graph.')
+        logger.info("Updating Neo4j meta graph.")
 
         # find current version node
         db_version = self._driver.query(
-            'MATCH (v:BioCypher) '
-            'WHERE NOT (v)-[:PRECEDES]->() '
-            'RETURN v',
+            "MATCH (v:BioCypher) " "WHERE NOT (v)-[:PRECEDES]->() " "RETURN v",
         )
         # add version node
-        self.add_biocypher_nodes(self._ontology)
+        self.add_biocypher_nodes(self.translator.ontology)
 
         # connect version node to previous
         if db_version[0]:
             previous = db_version[0][0]
-            previous_id = previous['v']['id']
+            previous_id = previous["v"]["id"]
             e_meta = BioCypherEdge(
                 previous_id,
-                self._ontology.get_dict().get('node_id'),
-                'PRECEDES',
+                self.translator.ontology.get_dict().get("node_id"),
+                "PRECEDES",
             )
             self.add_biocypher_edges(e_meta)
 
@@ -132,7 +123,7 @@ class _Neo4jDriver():
                 need of the database
         """
 
-        logger.info('Initialising database.')
+        logger.info("Initialising database.")
         self._create_constraints()
 
     def _create_constraints(self):
@@ -144,17 +135,16 @@ class _Neo4jDriver():
         constraints on the id of all entities represented as nodes.
         """
 
-        logger.info('Creating constraints for node types in config.')
+        logger.info("Creating constraints for node types in config.")
 
         # get structure
-        for leaf in self._ontology.extended_schema.items():
+        for leaf in self.translator.ontology.mapping.extended_schema.items():
             label = _misc.sentencecase_to_pascalcase(leaf[0])
-            if leaf[1]['represented_as'] == 'node':
-
+            if leaf[1]["represented_as"] == "node":
                 s = (
-                    f'CREATE CONSTRAINT `{label}_id` '
-                    f'IF NOT EXISTS ON (n:`{label}`) '
-                    'ASSERT n.id IS UNIQUE'
+                    f"CREATE CONSTRAINT `{label}_id` "
+                    f"IF NOT EXISTS ON (n:`{label}`) "
+                    "ASSERT n.id IS UNIQUE"
                 )
                 self._driver.query(s)
 
@@ -178,7 +168,7 @@ class _Neo4jDriver():
                 - second entry: Neo4j summary.
         """
 
-        bn = self._translator.translate_nodes(id_type_tuples)
+        bn = self.translator.translate_nodes(id_type_tuples)
         return self.add_biocypher_nodes(bn)
 
     def add_edges(self, id_src_tar_type_tuples: Iterable[tuple]) -> tuple:
@@ -210,7 +200,7 @@ class _Neo4jDriver():
                 - second entry: Neo4j summary.
         """
 
-        bn = self._translator.translate_edges(id_src_tar_type_tuples)
+        bn = self.translator.translate_edges(id_src_tar_type_tuples)
         return self.add_biocypher_edges(bn)
 
     def add_biocypher_nodes(
@@ -246,38 +236,36 @@ class _Neo4jDriver():
         """
 
         try:
-
             nodes = _misc.to_list(nodes)
 
             entities = [node.get_dict() for node in nodes]
 
         except AttributeError:
-
-            msg = 'Nodes must have a `get_dict` method.'
+            msg = "Nodes must have a `get_dict` method."
             logger.error(msg)
 
             raise ValueError(msg)
 
-        logger.info(f'Merging {len(entities)} nodes.')
+        logger.info(f"Merging {len(entities)} nodes.")
 
         entity_query = (
-            'UNWIND $entities AS ent '
-            'CALL apoc.merge.node([ent.node_label], '
-            '{id: ent.node_id}, ent.properties, ent.properties) '
-            'YIELD node '
-            'RETURN node'
+            "UNWIND $entities AS ent "
+            "CALL apoc.merge.node([ent.node_label], "
+            "{id: ent.node_id}, ent.properties, ent.properties) "
+            "YIELD node "
+            "RETURN node"
         )
 
-        method = 'explain' if explain else 'profile' if profile else 'query'
+        method = "explain" if explain else "profile" if profile else "query"
 
         result = getattr(self._driver, method)(
             entity_query,
             parameters={
-                'entities': entities,
+                "entities": entities,
             },
         )
 
-        logger.info('Finished merging nodes.')
+        logger.info("Finished merging nodes.")
 
         return result
 
@@ -326,28 +314,23 @@ class _Neo4jDriver():
         rels = []
 
         try:
-
             for e in edges:
-
-                if hasattr(e, 'get_node'):
-
+                if hasattr(e, "get_node"):
                     nodes.append(e.get_node())
                     rels.append(e.get_source_edge().get_dict())
                     rels.append(e.get_target_edge().get_dict())
 
                 else:
-
                     rels.append(e.get_dict())
 
         except AttributeError:
-
-            msg = 'Edges and nodes must have a `get_dict` method.'
+            msg = "Edges and nodes must have a `get_dict` method."
             logger.error(msg)
 
             raise ValueError(msg)
 
         self.add_biocypher_nodes(nodes)
-        logger.info(f'Merging {len(rels)} edges.')
+        logger.info(f"Merging {len(rels)} edges.")
 
         # cypher query
 
@@ -355,41 +338,39 @@ class _Neo4jDriver():
         # properties on match and on create;
         # TODO add node labels?
         node_query = (
-            'UNWIND $rels AS r '
-            'MERGE (src {id: r.source_id}) '
-            'MERGE (tar {id: r.target_id}) '
+            "UNWIND $rels AS r "
+            "MERGE (src {id: r.source_id}) "
+            "MERGE (tar {id: r.target_id}) "
         )
 
-        self._driver.query(node_query, parameters={'rels': rels})
+        self._driver.query(node_query, parameters={"rels": rels})
 
         edge_query = (
-            'UNWIND $rels AS r '
-            'MATCH (src {id: r.source_id}) '
-            'MATCH (tar {id: r.target_id}) '
-            'WITH src, tar, r '
-            'CALL apoc.merge.relationship'
-            '(src, r.relationship_label, NULL, '
-            'r.properties, tar, r.properties) '
-            'YIELD rel '
-            'RETURN rel'
+            "UNWIND $rels AS r "
+            "MATCH (src {id: r.source_id}) "
+            "MATCH (tar {id: r.target_id}) "
+            "WITH src, tar, r "
+            "CALL apoc.merge.relationship"
+            "(src, r.relationship_label, NULL, "
+            "r.properties, tar, r.properties) "
+            "YIELD rel "
+            "RETURN rel"
         )
 
-        method = 'explain' if explain else 'profile' if profile else 'query'
+        method = "explain" if explain else "profile" if profile else "query"
 
-        result = getattr(self._driver,
-                         method)(edge_query, parameters={
-                             'rels': rels
-                         })
+        result = getattr(self._driver, method)(
+            edge_query, parameters={"rels": rels}
+        )
 
-        logger.info('Finished merging edges.')
+        logger.info("Finished merging edges.")
 
         return result
 
 
 def get_driver(
     dbms: str,
-    translator: 'Translator',
-    ontology: 'Ontology',
+    translator: "Translator",
 ):
     """
     Function to return the writer class.
@@ -400,15 +381,14 @@ def get_driver(
 
     dbms_config = _config(dbms)
 
-    if dbms == 'neo4j':
+    if dbms == "neo4j":
         return _Neo4jDriver(
-            database_name=dbms_config['database_name'],
-            wipe=dbms_config['wipe'],
-            uri=dbms_config['uri'],
-            user=dbms_config['user'],
-            password=dbms_config['password'],
-            multi_db=dbms_config['multi_db'],
-            ontology=ontology,
+            database_name=dbms_config["database_name"],
+            wipe=dbms_config["wipe"],
+            uri=dbms_config["uri"],
+            user=dbms_config["user"],
+            password=dbms_config["password"],
+            multi_db=dbms_config["multi_db"],
             translator=translator,
         )