more tweaks

Author: jsstevenson

pyproject.toml

@@ -44,6 +44,7 @@ dev = [
     "pre-commit>=4.2.0",
     "ipykernel",
     "fastapi[standard]",
+    "seqrepo-rest-service",  # for generating SeqRepo-based tests fixtures
 ]
 
 [project.urls]

src/anyvlm/anyvar/base_client.py

@@ -9,16 +9,23 @@ class AnyVarClientError(Exception):
     """Generic client-related exception."""
 
 
+class UnidentifiedObjectError(AnyVarClientError):
+    """Raise if input object lacks an ID property"""
+
+
 class BaseAnyVarClient(abc.ABC):
     """Interface elements for an AnyVar client"""
 
     @abc.abstractmethod
-    def put_objects(self, objects: list[VrsVariation]) -> list[VrsVariation]:
+    def put_objects(self, objects: list[VrsVariation]) -> None:
         """Register objects with AnyVar
 
+        All input objects must have a populated ID field. A validation check for this is
+        performed before any variants are registered.
+
         :param objects: variation objects to register
-        :return: completed VRS objects
         :raise AnyVarClientError: for errors relating to specifics of client interface
+        :raise UnidentifiedObjectError: if *any* provided object lacks a VRS ID
         """
 
     @abc.abstractmethod

src/anyvlm/anyvar/http_client.py

@@ -1,10 +1,18 @@
 """Provide abstraction for a VLM-to-AnyVar connection."""
 
+import logging
+
 import requests
 from anyvar.utils.types import VrsVariation
 from ga4gh.vrs import models
 
-from anyvlm.anyvar.base_client import AnyVarClientError, BaseAnyVarClient
+from anyvlm.anyvar.base_client import (
+    AnyVarClientError,
+    BaseAnyVarClient,
+    UnidentifiedObjectError,
+)
+
+_logger = logging.getLogger(__name__)
 
 
 class HttpAnyVarClient(BaseAnyVarClient):
@@ -21,36 +29,35 @@ def __init__(
         self.hostname = hostname
         self.request_timeout = request_timeout
 
-    def put_objects(self, objects: list[VrsVariation]) -> list[VrsVariation]:
+    def put_objects(self, objects: list[VrsVariation]) -> None:
         """Register objects with AnyVar
 
-        This method is intentionally naive. Future improvements could include
-        * A better way to batch requests together to mitigate HTTP latency
-        * A smarter dispatch for reconstructing variation model instances
+        All input objects must have a populated ID field. A validation check for this is
+        performed before any variants are registered.
 
         :param objects: variation objects to register
         :return: completed VRS objects
         :raise AnyVarClientError: if connection is unsuccessful during registration request
+        :raise UnidentifiedObjectError: if *any* provided object lacks a VRS ID
         """
-        results = []
+        objects_to_submit = []
         for vrs_object in objects:
+            if not vrs_object.id:
+                _logger.error("Provided variant %s has no VRS ID: %s")
+                raise UnidentifiedObjectError
+            objects_to_submit.append(
+                vrs_object.model_dump(exclude_none=True, mode="json")
+            )
+        for vrs_object in objects_to_submit:
             response = requests.put(
                 f"{self.hostname}/vrs_variation",
-                json=vrs_object.model_dump(exclude_none=True, mode="json"),
+                json=vrs_object,
                 timeout=self.request_timeout,
             )
             try:
                 response.raise_for_status()
             except requests.HTTPError as e:
                 raise AnyVarClientError from e
-            result_object = response.json()["object"]
-            if result_object.get("type") == "Allele":
-                results.append(models.Allele(**result_object))
-            else:
-                raise NotImplementedError(
-                    f"Unsupported object type: {result_object.get('type')}"
-                )
-        return results
 
     def search_by_interval(
         self, accession: str, start: int, end: int

src/anyvlm/anyvar/python_client.py

@@ -1,11 +1,15 @@
 """Implement AnyVar client interface for direct Python-based access."""
 
+import logging
+
 from anyvar import AnyVar
-from anyvar.storage.base_storage import IncompleteVrsObjectError, Storage
+from anyvar.storage.base_storage import Storage
 from anyvar.translate.translate import Translator
-from anyvar.utils.types import VrsVariation, recursive_identify
+from anyvar.utils.types import VrsVariation
+
+from anyvlm.anyvar.base_client import BaseAnyVarClient, UnidentifiedObjectError
 
-from anyvlm.anyvar.base_client import BaseAnyVarClient
+_logger = logging.getLogger(__name__)
 
 
 class PythonAnyVarClient(BaseAnyVarClient):
@@ -14,21 +18,20 @@ class PythonAnyVarClient(BaseAnyVarClient):
     def __init__(self, translator: Translator, storage: Storage) -> None:  # noqa: D107
         self.av = AnyVar(translator, storage)
 
-    def put_objects(self, objects: list[VrsVariation]) -> list[VrsVariation]:
+    def put_objects(self, objects: list[VrsVariation]) -> None:
         """Register objects with AnyVar
 
+        All input objects must have a populated ID field. A validation check for this is
+        performed before any variants are registered.
+
         :param objects: variation objects to register
-        :return: completed VRS objects
+        :raise UnidentifiedObjectError: if *any* provided object lacks a VRS ID
         """
-        complete_variations = []
-        for variation in objects:
-            try:
-                self.av.put_objects([variation])
-            except IncompleteVrsObjectError:
-                variation = recursive_identify(variation)  # noqa: PLW2901
-                self.av.put_objects([variation])
-            complete_variations.append(variation)
-        return complete_variations
+        for variant in objects:
+            if not variant.id:
+                _logger.error("Provided variant %s has no VRS ID: %s")
+                raise UnidentifiedObjectError
+        self.av.put_objects(objects)  # type: ignore[reportArgumentType]
 
     def search_by_interval(
         self, accession: str, start: int, end: int

src/anyvlm/main.py

@@ -1,5 +1,6 @@
 """Define core FastAPI app"""
 
+import logging
 from collections.abc import AsyncGenerator
 from contextlib import asynccontextmanager
 from enum import Enum
@@ -19,6 +20,8 @@
     ServiceType,
 )
 
+_logger = logging.getLogger(__name__)
+
 
 def create_anyvar_client(
     connection_string: str | None = None,
@@ -27,15 +30,20 @@ def create_anyvar_client(
 
     If given a string for connecting to an AnyVar instance via HTTP requests, then
     create an HTTP-based client. Otherwise, try to use AnyVar resource factory functions
-    for standing up a Python-based client.
+    for standing up a Python-based client. In the latter case, see the AnyVar documentation
+    for configuration info (i.e. environment variables)
 
     :param connection_string: description of connection param
     :return: client instance
     """
     if not connection_string:
         connection_string = get_config().anyvar_uri
     if connection_string.startswith("http://"):
+        _logger.info(
+            "Initializing HTTP-based AnyVar client under hostname %s", connection_string
+        )
         return HttpAnyVarClient(connection_string)
+    _logger.info("Initializing AnyVar instance directly")
     storage = create_storage()
     translator = create_translator()
     return PythonAnyVarClient(translator, storage)