wip: add create_anyvlm_storage + sanitized_url

Author: korikuzma

docs/source/configuration/index.rst

@@ -3,7 +3,7 @@ Configuration
 
 This section details AnyVLM configuration. It is broken down into the following subsections:
 
-* :doc:`Object Storage <storage>`: define database connection, alter table names, and set parameters for bulk processing
+* :doc:`Object Storage <storage>`: define database connection
 * :doc:`Example .env file <dotenv_example>`: use a ``.env`` file to declare environment variables when running REST API service
 * :doc:`Docker Compose <docker_compose>`: edit the provided Docker Compose file to tailor it to your needs
 

pyproject.toml

@@ -19,7 +19,7 @@ classifiers = [
 dependencies = [
     "ga4gh.vrs>=2.2.0,<3.0",
     "ga4gh.va_spec~=0.4.2",
-    "biocommons.anyvar@git+https://github.com/biocommons/anyvar.git@main",
+    "biocommons.anyvar@git+https://github.com/biocommons/anyvar.git@0d3ab56fe936b27235a1ce136da4641ea81c0bbf",
     "fastapi>=0.95.0",
     "python-multipart",  # required for fastapi file uploads
     "uvicorn",

src/anyvlm/main.py

@@ -3,6 +3,8 @@
 import logging
 from collections.abc import AsyncGenerator
 from contextlib import asynccontextmanager
+from os import environ
+from urllib.parse import urlparse
 
 from anyvar.anyvar import create_storage, create_translator
 from fastapi import FastAPI
@@ -18,6 +20,8 @@
     ServiceOrganization,
     ServiceType,
 )
+from anyvlm.storage import DEFAULT_STORAGE_URI
+from anyvlm.storage.base_storage import Storage
 from anyvlm.utils.types import (
     EndpointTag,
 )
@@ -51,6 +55,34 @@ def create_anyvar_client(
     return PythonAnyVarClient(translator, storage)
 
 
+def create_anyvlm_storage(uri: str | None = None) -> Storage:
+    """Provide factory to create storage based on `uri`, the ANYVLM_STORAGE_URI
+    environment value, or the default value if neither is provided.
+
+    The URI format is as follows:
+
+    `postgresql://[username]:[password]@[domain]/[database]`
+
+    :param uri: AnyVLM storage URI
+    :raises ValueError: if the URI scheme is not supported
+    :return: AnyVLM storage instance
+    """
+    if not uri:
+        uri = environ.get("ANYVLM_STORAGE_URI", DEFAULT_STORAGE_URI)
+
+    parsed_uri = urlparse(uri)
+    if parsed_uri.scheme == "postgresql":
+        from anyvlm.storage.postgres import PostgresObjectStore  # noqa: PLC0415
+
+        storage = PostgresObjectStore(uri)
+    else:
+        msg = f"URI scheme {parsed_uri.scheme} is not implemented"
+        raise ValueError(msg)
+
+    _logger.debug("create_storage: %s → %s}", storage.sanitized_url, storage)
+    return storage
+
+
 @asynccontextmanager
 async def lifespan(app: FastAPI) -> AsyncGenerator:
     """Configure FastAPI instance lifespan.
@@ -59,8 +91,10 @@ async def lifespan(app: FastAPI) -> AsyncGenerator:
     :return: async context handler
     """
     app.state.anyvar_client = create_anyvar_client()
+    app.state.anyvlm_storage = create_anyvlm_storage()
     yield
     app.state.anyvar_client.close()
+    app.state.anyvlm_storage.close()
 
 
 app = FastAPI(

src/anyvlm/storage/__init__.py

@@ -2,4 +2,6 @@
 
 from .base_storage import Storage
 
-__all__ = ["Storage"]
+DEFAULT_STORAGE_URI = "postgresql://postgres@localhost:5432/anyvlm"
+
+__all__ = ["DEFAULT_STORAGE_URI", "Storage"]

src/anyvlm/storage/base_storage.py

@@ -28,5 +28,14 @@ def wipe_db(self) -> None:
     def add_allele_frequencies(self, caf: CohortAlleleFrequencyStudyResult) -> None:
         """Add allele frequency data to the database. Will skip conflicts.
 
+        NOTE: For now, this will only insert a single caf record into the database.
+        Single insertion is used to do a simple test of the storage backend.
+        Issue-34 will support batch insertion of caf records.
+
         :param caf: Cohort allele frequency study result object to insert into the DB
         """
+
+    @property
+    @abstractmethod
+    def sanitized_url(self) -> str:
+        """Return a sanitized URL (password masked) of the database connection string."""

src/anyvlm/storage/mappers.py

@@ -41,7 +41,7 @@ def from_db_entity(
         homozygotes = db_entity.ac_hom
         heterozygotes = db_entity.ac_het
         hemizygotes = db_entity.ac_hemi
-        ac = sum((homozygotes, heterozygotes, hemizygotes))
+        ac = sum((homozygotes or 0, heterozygotes or 0, hemizygotes or 0))
         an = db_entity.an
 
         return CohortAlleleFrequencyStudyResult(
@@ -56,7 +56,7 @@ def from_db_entity(
                 "hemizygotes": hemizygotes,
             },
             cohort=StudyGroup(name=db_entity.cohort),  # type: ignore
-        )  # type: ignore
+        )
 
     def to_db_entity(
         self, va_model: CohortAlleleFrequencyStudyResult

src/anyvlm/storage/postgres.py

@@ -1,5 +1,7 @@
 """Provide PostgreSQL-based storage implementation."""
 
+from urllib.parse import urlparse
+
 from ga4gh.va_spec.base import CohortAlleleFrequencyStudyResult
 from sqlalchemy import create_engine, delete
 from sqlalchemy.dialects.postgresql import insert
@@ -36,9 +38,29 @@ def wipe_db(self) -> None:
         with self.session_factory() as session, session.begin():
             session.execute(delete(orm.AlleleFrequencyData))
 
+    @property
+    def sanitized_url(self) -> str:
+        """Return a sanitized URL (password masked) of the database connection string."""
+        parsed = urlparse(self.db_url)
+        netloc = ""
+        if parsed.username:
+            netloc += parsed.username
+            if parsed.password:
+                netloc += ":****"
+            netloc += "@"
+        if parsed.hostname:
+            netloc += f"{parsed.hostname}"
+        if parsed.port:
+            netloc += f":{parsed.port}"
+        return f"{parsed.scheme}://{netloc}{parsed.path}"
+
     def add_allele_frequencies(self, caf: CohortAlleleFrequencyStudyResult) -> None:
         """Add allele frequency data to the database. Will skip conflicts.
 
+        NOTE: For now, this will only insert a single caf record into the database.
+        Single insertion is used to do a simple test of the storage backend.
+        Issue-34 will support batch insertion of caf records.
+
         :param caf: Cohort allele frequency study result object to insert into the DB
         """
         db_entity = mapper_registry.to_db_entity(caf)

tests/unit/storage/test_postgres_unit.py

@@ -32,6 +32,28 @@ def caf_empty_cohort(caf_iri: CohortAlleleFrequencyStudyResult):
     return caf
 
 
+@pytest.mark.parametrize(
+    ("db_url", "sanitized_db_url"),
+    [
+        (
+            "postgresql://postgres:postgres@localhost:5432/anyvlm_test",
+            "postgresql://postgres:****@localhost:5432/anyvlm_test",
+        ),
+        (
+            "postgresql://postgres@localhost:5432/anyvlm_test",
+            "postgresql://postgres@localhost:5432/anyvlm_test",
+        ),
+    ],
+)
+def test_sanitized_url(monkeypatch, db_url: str, sanitized_db_url: str):
+    """Test that sanitized_url method works correctly"""
+    monkeypatch.setattr(PostgresObjectStore, "__init__", lambda *_: None)
+    object_store = PostgresObjectStore("")
+    monkeypatch.setattr(object_store, "db_url", db_url, raising=False)
+    assert object_store.db_url == db_url
+    assert object_store.sanitized_url == sanitized_db_url
+
+
 @pytest.mark.parametrize("caf_fixture_name", ["caf_iri", "caf_allele"])
 def test_add_allele_frequencies(
     request, caf_fixture_name: str, postgres_storage: PostgresObjectStore