WIP

mex/common/identity/__init__.py

@@ -0,0 +1,14 @@
+from mex.common.identity.base import BaseProvider
+from mex.common.identity.dummy import DummyIdentityProvider
+from mex.common.identity.models import Identity
+from mex.common.identity.types import IdentityProvider
+from mex.common.identity.query import assign_identity,fetch_identity
+
+__all__ = (
+    "assign_identity",
+    "BaseProvider",
+    "DummyIdentityProvider",
+    "fetch_identity",
+    "Identity",
+    "IdentityProvider",
+)

mex/common/identity/base.py

@@ -1,17 +1,9 @@
 from abc import ABCMeta, abstractmethod
-from enum import Enum
 
 from mex.common.identity.models import Identity
 from mex.common.types import Identifier, PrimarySourceID
 
 
-class IdentityProvider(Enum):
-    """Choice of available identity providers."""
-
-    BACKEND = "backend"
-    DUMMY = "dummy"
-
-
 class BaseProvider(metaclass=ABCMeta):
     """Base class to define the interface of identity providers."""
 
@@ -31,6 +23,6 @@ def fetch(
         had_primary_source: Identifier | None = None,
         identifier_in_primary_source: str | None = None,
         stable_target_id: Identifier | None = None,
-    ) -> Identity | None:  # pragma: no cover
-        """Find an Identity instance from the database if it can be found."""
+    ) -> list[Identity]:  # pragma: no cover
+        """Find Identity instances matching the given filters."""
         ...

mex/common/identity/dummy.py

@@ -1,6 +1,7 @@
 from typing import Iterable
 
 from mex.common.connector import BaseConnector
+from mex.common.exceptions import MExError
 from mex.common.identity.base import BaseProvider
 from mex.common.identity.models import Identity
 from mex.common.types import Identifier, PrimarySourceID
@@ -25,18 +26,20 @@ def assign(
         Returns:
             Newly created or updated Identity instance
         """
-        identity = self.fetch(
+        identities = self.fetch(
             had_primary_source=had_primary_source,
             identifier_in_primary_source=identifier_in_primary_source,
         )
-        if not identity:
-            identity = Identity(
-                hadPrimarySource=had_primary_source,
-                identifierInPrimarySource=identifier_in_primary_source,
-                stableTargetId=Identifier.generate(),
-                identifier=Identifier.generate(),
-            )
-            self.dummy_identity_db.append(identity)
+        if identities:
+            return identities[0]
+
+        identity = Identity(
+            hadPrimarySource=had_primary_source,
+            identifierInPrimarySource=identifier_in_primary_source,
+            stableTargetId=Identifier.generate(),
+            identifier=Identifier.generate(),
+        )
+        self.dummy_identity_db.append(identity)
         return identity
 
     def fetch(
@@ -45,18 +48,18 @@ def fetch(
         had_primary_source: Identifier | None = None,
         identifier_in_primary_source: str | None = None,
         stable_target_id: Identifier | None = None,
-    ) -> Identity | None:
-        """Find an Identity instance from the database if it can be found.
+    ) -> list[ Identity] :
+        """Find Identity instances in the dummy database.
 
         Args:
             had_primary_source: Stable target ID of primary source
             identifier_in_primary_source: Identifier in the primary source
             stable_target_id: Stable target ID of the entity
 
         Returns:
-            Optional Identity instance
+            List of Identity instances
         """
-        identities: Iterable[Identity] = self.dummy_identity_db
+        identities = iter(self.dummy_identity_db)
 
         if had_primary_source:
             identities = filter(
@@ -72,9 +75,7 @@ def fetch(
                 lambda i: i.stableTargetId == stable_target_id, identities
             )
 
-        if identities := list(identities):
-            return identities[0]
-        return None
+        return list(identities)
 
     def close(self) -> None:
         """Trash the dummy identity database."""

mex/common/identity/query.py

@@ -19,10 +19,9 @@ def assign_identity(
 
 
 def fetch_identity(
-    *,
-    had_primary_source: PrimarySourceID | None = None,
-    identifier_in_primary_source: str | None = None,
-    stable_target_id: Identifier | None = None,
+    had_primary_source: PrimarySourceID ,
+    identifier_in_primary_source: str ,
+    stable_target_id: Identifier|None
 ) -> Identity | None:
     """Find an Identity instance from the database if it can be found."""
     settings = BaseSettings.get()

mex/common/ldap/extract.py

@@ -1,7 +1,7 @@
 from collections import defaultdict
 from typing import Hashable, Iterable, cast
 
-from mex.common.identity.query import fetch_identity
+from mex.common.identity import fetch_identity
 from mex.common.ldap.models.person import LDAPPerson, LDAPPersonWithQuery
 from mex.common.models import ExtractedPrimarySource
 from mex.common.types import Identifier

mex/common/models/extracted_data.py

@@ -3,7 +3,7 @@
 from pydantic import Field, root_validator
 
 from mex.common.models.base import MExModel
-from mex.common.types import Identifier, PrimarySourceID
+from mex.common.types import PrimarySourceID
 
 MEX_PRIMARY_SOURCE_STABLE_TARGET_ID = PrimarySourceID("00000000000000")
 MEX_PRIMARY_SOURCE_IDENTIFIER_IN_PRIMARY_SOURCE = "mex"
@@ -114,7 +114,7 @@ def set_identifiers(cls, values: dict[str, Any]) -> dict[str, Any]:
             Values with identifier and provenance attributes
         """
         # break import cycle, sigh
-        from mex.common.identity.query import assign_identity
+        from mex.common.identity import assign_identity
 
         # validate ID in primary source and primary source ID
         if identifier_in_primary_source := values.get("identifierInPrimarySource"):
@@ -127,10 +127,7 @@ def set_identifiers(cls, values: dict[str, Any]) -> dict[str, Any]:
         else:
             raise ValueError("Missing value for `hadPrimarySource`.")
 
-        identity = assign_identity(
-            had_primary_source=had_primary_source,
-            identifier_in_primary_source=identifier_in_primary_source,
-        )
+        identity = assign_identity(had_primary_source, identifier_in_primary_source)
 
         # In case an identity was already found and the identifier provided to the
         # constructor do not match we raise an error because it should not be

mex/common/settings.py

@@ -9,7 +9,7 @@
 from pydantic.env_settings import DotenvType, env_file_sentinel
 from pydantic.typing import StrPath
 
-from mex.common.identity.types import IdentityProvider
+from mex.common.identity import IdentityProvider
 from mex.common.sinks import Sink
 from mex.common.transform import MExEncoder
 from mex.common.types import AssetsPath

tests/identity/test_dummy.py

@@ -0,0 +1,55 @@
+import pytest
+from pytest import MonkeyPatch
+
+from mex.common.exceptions import MExError
+from mex.common.identity import assign_identity, fetch_identity,DummyIdentityProvider
+from mex.common.settings import BaseSettings
+from mex.common.testing import Joker
+from mex.common.types import Identifier,PrimarySourceID
+
+
+def test_assign_identity() -> None:
+    had_primary_source = PrimarySourceID("0000000000")
+    identifier_in_primary_source = "thing-1"
+
+    new_identity = assign_identity( had_primary_source, identifier_in_primary_source    )
+
+    assert new_identity.dict() == dict(
+        hadPrimarySource=had_primary_source,
+        identifierInPrimarySource=identifier_in_primary_source,
+        stableTargetId=Joker(),
+        identifier=Joker(),
+    )
+
+    found_identity = assign_identity(had_primary_source, identifier_in_primary_source)
+
+    assert found_identity.dict() == dict(
+        hadPrimarySource=had_primary_source,
+        identifierInPrimarySource=identifier_in_primary_source,
+        stableTargetId=new_identity.stableTargetId,
+        identifier=new_identity.identifier
+    )
+
+
+def test_fetch_identity() -> None:
+    had_primary_source = PrimarySourceID("0000000000")
+    identifier_in_primary_source = "thing-1"
+
+    identity = fetch_identity(had_primary_source,identifier_in_primary_source)
+    assert identity is None
+
+    identity = assign_identity(had_primary_source,identifier_in_primary_source)
+
+    #
+    assert fetch_identity(
+        had_primary_source=identity.hadPrimarySource,
+        identifier_in_primary_source=identity.identifierInPrimarySource,
+        stable_target_id=identity.stableTargetId,
+    ) == identity
+
+
+def test_fetch_identity_nothing_found() -> None:
+    assert fetch_identity(
+        had_primary_source=PrimarySourceID.generate(),
+        identifier_in_primary_source="thing-xyz",
+    ) is  None

tests/identity/test_query.py

@@ -2,76 +2,44 @@
 from pytest import MonkeyPatch
 
 from mex.common.exceptions import MExError
-from mex.common.identity.query import assign_identity, fetch_identity
+from mex.common.identity import assign_identity, fetch_identity, IdentityProvider
 from mex.common.settings import BaseSettings
 from mex.common.testing import Joker
-from mex.common.types.identifier import Identifier
-
-
-def test_assign_identity() -> None:
-    system_a_id = Identifier.generate()
-    old_target_id = Identifier.generate()
-    new_target_id = Identifier.generate()
-    identity_after_insert = assign_identity(
-        system_a_id, "thing-1", old_target_id, "type-x"
-    )
-
-    assert identity_after_insert.dict() == dict(
-        hadPrimarySource=system_a_id,
-        identifierInPrimarySource="thing-1",
-        stableTargetId=old_target_id,
-        entityType="type-x",
-        identifier=Joker(),
-    )
-
-    identity_after_update = assign_identity(
-        system_a_id, "thing-1", new_target_id, "type-x"
-    )
-
-    assert identity_after_update.dict() == dict(
-        hadPrimarySource=system_a_id,
-        identifierInPrimarySource="thing-1",
-        stableTargetId=new_target_id,
-        entityType="type-x",
-        identifier=Joker(),
-    )
+from mex.common.types import Identifier, PrimarySourceID
 
 
 def test_unsupported_providers(monkeypatch: MonkeyPatch) -> None:
     settings = BaseSettings.get()
     monkeypatch.setattr(BaseSettings, "__setattr__", object.__setattr__)
     monkeypatch.setattr(settings, "identity_provider", "bogus-provider")
 
+    had_primary_source = PrimarySourceID("0000000000")
+    identifier_in_primary_source = "thing-1"
+
     with pytest.raises(MExError, match="Cannot fetch identity from bogus-provider."):
-        fetch_identity()
+        fetch_identity(had_primary_source, identifier_in_primary_source)
 
     with pytest.raises(MExError, match="Cannot assign identity to bogus-provider."):
-        assign_identity(
-            Identifier.generate(), "thing-1", Identifier.generate(), "type-x"
-        )
+        assign_identity(had_primary_source, identifier_in_primary_source)
 
 
-def test_fetch_identity() -> None:
-    system_a_id = Identifier.generate()
-    entity_1 = Identifier.generate()
+def test_under_specified_fetch()->None:
+    with pytest.raises(MExError, match="Need either `stable_target_id` or both"):
+        fetch_identity()
+    with pytest.raises(MExError, match="Need either `stable_target_id` or both"):
+        fetch_identity(had_primary_source=PrimarySourceID("0000000000"))
+    with pytest.raises(MExError, match="Need either `stable_target_id` or both"):
+        fetch_identity(identifier_in_primary_source="thing-1")
 
-    result = fetch_identity()
-    assert result is None
 
-    identity = assign_identity(system_a_id, "thing-1", entity_1, "type-x")
+def test_dummy_provider() -> None:
+    settings = BaseSettings.get()
+    assert settings.identity_provider == IdentityProvider.DUMMY
 
-    result = fetch_identity(
-        had_primary_source=identity.hadPrimarySource,
-        identifier_in_primary_source=identity.identifierInPrimarySource,
-        stable_target_id=identity.stableTargetId,
-    )
-    assert result == identity
+    had_primary_source = PrimarySourceID("0000000000")
+    identifier_in_primary_source = "thing-1"
 
+    assigned = assign_identity(had_primary_source, identifier_in_primary_source)
+    fetched = fetch_identity(had_primary_source, identifier_in_primary_source)
 
-def test_fetch_identity_nothing_found() -> None:
-    result = fetch_identity(
-        had_primary_source=Identifier.generate(),
-        identifier_in_primary_source=Identifier.generate(),
-        stable_target_id=Identifier.generate(),
-    )
-    assert result is None
+    assert assigned == fetched

tests/models/test_extracted_data.py

@@ -4,7 +4,7 @@
 import pytest
 from pydantic import ValidationError
 
-from mex.common.identity.query import fetch_identity
+from mex.common.identity import fetch_identity
 from mex.common.models import BaseModel, ExtractedData
 from mex.common.types import Identifier