Merge pull request #11 from TheJacksonLaboratory/G3-33-geneset-by-map…

…pable-id

G3 33 geneset by mappable gene type id

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "geneweaver-api"
-version = "0.0.1a6"
+version = "0.0.1a7"
 description = "description"
 authors = ["Jax Computational Sciences <[email protected]>"]
 packages = [

src/geneweaver/api/controller/genesets.py

@@ -1,17 +1,24 @@
 """Endpoints related to genesets."""
+import json
+import os
+import time
+from tempfile import TemporaryDirectory
 from typing import Optional
 
-from fastapi import APIRouter, Depends, HTTPException, Security
+from fastapi import APIRouter, Depends, HTTPException, Query, Security
+from fastapi.responses import FileResponse
 from geneweaver.api import dependencies as deps
 from geneweaver.api.schemas.auth import UserInternal
 from geneweaver.api.services import geneset as genset_service
+from geneweaver.core.enum import GeneIdentifier
 from geneweaver.core.schema.geneset import GenesetUpload
 from geneweaver.db import geneset as db_geneset
 from geneweaver.db import geneset_value as db_geneset_value
 
 from . import message as api_message
 
 router = APIRouter(prefix="/genesets")
+gene_id_type_options = [f"{choice.name} ({choice.value})" for choice in GeneIdentifier]
 
 
 @router.get("")
@@ -29,19 +36,79 @@ def get_geneset(
     geneset_id: int,
     user: UserInternal = Security(deps.full_user),
     cursor: Optional[deps.Cursor] = Depends(deps.cursor),
+    # using int type and adding doc options here as using GeneIdentifier
+    # won't produce the right docs for gene identifier options
+    gene_id_type: Optional[int] = Query(
+        None, description=f"Options: {gene_id_type_options}"
+    ),
 ) -> dict:
-    """Get a geneset by ID."""
-    response = genset_service.get_geneset(cursor, geneset_id, user)
+    """Get a geneset by ID. Optional filter results by gene identifier type."""
+    if gene_id_type:
+        gene_identifier_type = get_gene_identifier_type(gene_id_type)
+        response = genset_service.get_geneset_w_gene_id_type(
+            cursor, geneset_id, user, gene_identifier_type
+        )
+    else:
+        response = genset_service.get_geneset(cursor, geneset_id, user)
 
     if "error" in response:
-        if response.get("message") == api_message.ACCESS_FORBIDEN:
-            raise HTTPException(status_code=403, detail=api_message.ACCESS_FORBIDEN)
+        if response.get("message") == api_message.ACCESS_FORBIDDEN:
+            raise HTTPException(status_code=403, detail=api_message.ACCESS_FORBIDDEN)
         else:
             raise HTTPException(status_code=500, detail=api_message.UNEXPECTED_ERROR)
 
     return response
 
 
+@router.get("/{geneset_id}/file", response_class=FileResponse)
+def get_export_geneset_by_id_type(
+    geneset_id: int,
+    user: UserInternal = Security(deps.full_user),
+    cursor: Optional[deps.Cursor] = Depends(deps.cursor),
+    temp_dir: TemporaryDirectory = Depends(deps.get_temp_dir),
+    # using int type and adding doc options here as using
+    # GeneIdentifier won't produce the right docs for gene identifier options
+    gene_id_type: Optional[int] = Query(
+        None, description=f"Options: {gene_id_type_options}"
+    ),
+) -> dict:
+    """Export geneset into JSON file. Search by ID and optional gene identifier type."""
+    timestr = time.strftime("%Y%m%d-%H%M%S")
+
+    # Validate gene identifier type
+    if gene_id_type:
+        gene_identifier_type = get_gene_identifier_type(gene_id_type)
+        response = genset_service.get_geneset_w_gene_id_type(
+            cursor, geneset_id, user, gene_identifier_type
+        )
+    else:
+        response = genset_service.get_geneset(cursor, geneset_id, user)
+
+    if "error" in response:
+        if response.get("message") == api_message.ACCESS_FORBIDDEN:
+            raise HTTPException(status_code=403, detail=api_message.ACCESS_FORBIDDEN)
+        else:
+            raise HTTPException(status_code=500, detail=api_message.UNEXPECTED_ERROR)
+
+    id_type = response.get("gene_identifier_type")
+    if id_type:
+        geneset_filename = f"geneset_{geneset_id}_{id_type}_{timestr}.json"
+    else:
+        geneset_filename = f"geneset_{geneset_id}_{timestr}.json"
+
+    # Write the data to temp file
+    temp_file_path = os.path.join(temp_dir, geneset_filename)
+    with open(temp_file_path, "w") as f:
+        json.dump(response, f, default=str)
+
+    # Return as a download
+    return FileResponse(
+        path=temp_file_path,
+        media_type="application/octet-stream",
+        filename=geneset_filename,
+    )
+
+
 @router.post("")
 def upload_geneset(
     geneset: GenesetUpload,
@@ -51,3 +118,20 @@ def upload_geneset(
     """Upload a geneset."""
     db_geneset_value.format_geneset_values_for_file_insert(geneset.gene_list)
     return {"geneset_id": 0}
+
+
+def get_gene_identifier_type(gene_id_type: int) -> GeneIdentifier:
+    """Get a valid GeneIdentifier object. Raise HTTP exception if invalid value.
+
+    @param gene_id_type: gene identifier type
+    @return: GeneIdentifier obj or HTTP exception if invalid id value.
+    """
+    try:
+        gene_identifier = GeneIdentifier(gene_id_type)
+    except ValueError as err:
+        raise HTTPException(
+            status_code=400,
+            detail=f"{api_message.GENE_IDENTIFIER_TYPE_VALUE_ERROR}, "
+            f"valid options= {gene_id_type_options}",
+        ) from err
+    return gene_identifier

src/geneweaver/api/controller/message.py

@@ -2,5 +2,6 @@
 
 
 ##Errors
-ACCESS_FORBIDEN = "Forbidden"
+ACCESS_FORBIDDEN = "Forbidden"
 UNEXPECTED_ERROR = "Unexpected Error"
+GENE_IDENTIFIER_TYPE_VALUE_ERROR = "Invalid gene identifier type"

src/geneweaver/api/dependencies.py

@@ -1,5 +1,6 @@
 """Dependency injection capabilities for the GeneWeaver API."""
 # ruff: noqa: B008
+from tempfile import TemporaryDirectory
 from typing import Generator
 
 import psycopg
@@ -27,11 +28,20 @@ def cursor() -> Generator:
             yield cur
 
 
-def full_user(
+async def full_user(
     cursor: Cursor = Depends(cursor),
     user: UserInternal = Depends(auth.get_user_strict),
 ) -> UserInternal:
-    """Get the full user object.""" ""
+    """Get the full user object.
+
+    Since there are external dependencies to wait for,
+    the recommendation is to use async
+    Also, Workaround FASTAPI issue, where logs hide exact place of errors
+    https://github.com/tiangolo/fastapi/discussions/8428
+    Geneweaver issue: G3-96.
+    @param cursor: DB cursor
+    @param user: GW user.
+    """
     try:
         user.id = db_user.by_sso_id_and_email(cursor, user.sso_id, user.email)[0][
             "usr_id"
@@ -49,3 +59,12 @@ def full_user(
             )
 
     yield user
+
+
+async def get_temp_dir() -> TemporaryDirectory:
+    """Get a temp directory."""
+    temp_dir = TemporaryDirectory()
+    try:
+        yield temp_dir.name
+    finally:
+        del temp_dir

src/geneweaver/api/services/geneset.py

@@ -3,17 +3,24 @@
 from fastapi.logger import logger
 from geneweaver.api.controller import message
 from geneweaver.api.schemas.auth import User
+from geneweaver.core.enum import GeneIdentifier
 from geneweaver.db import geneset as db_geneset
 from geneweaver.db import geneset_value as db_geneset_value
 from geneweaver.db.geneset import is_readable as db_is_readable
 from psycopg import Cursor
 
 
 def get_geneset(cursor: Cursor, geneset_id: int, user: User) -> dict:
-    """Get a geneset by ID."""
+    """Get a geneset by ID.
+
+    @param cursor: DB cursor
+    @param geneset_id: geneset identifier
+    @param user: GW user
+    @return: dictionary response (geneset and genset values).
+    """
     try:
         if not is_geneset_readable_by_user(cursor, geneset_id, user):
-            return {"error": True, "message": message.ACCESS_FORBIDEN}
+            return {"error": True, "message": message.ACCESS_FORBIDDEN}
 
         geneset = db_geneset.by_id(cursor, geneset_id)
         geneset_values = db_geneset_value.by_geneset_id(cursor, geneset_id)
@@ -24,8 +31,44 @@ def get_geneset(cursor: Cursor, geneset_id: int, user: User) -> dict:
         raise err
 
 
+def get_geneset_w_gene_id_type(
+    cursor: Cursor, geneset_id: int, user: User, gene_id_type: GeneIdentifier
+) -> dict:
+    """Get a geneset by ID and filter with gene identifier type.
+
+    @param cursor: DB cursor
+    @param geneset_id: geneset identifier
+    @param user: GW user
+    @param gene_id_type: gene identifier type object
+    @return: Dictionary response (geneset identifier, geneset, and genset values).
+    """
+    try:
+        if not is_geneset_readable_by_user(cursor, geneset_id, user):
+            return {"error": True, "message": message.ACCESS_FORBIDDEN}
+
+        geneset = db_geneset.by_id(cursor, geneset_id)
+        geneset_values = db_geneset_value.by_geneset_id(
+            cursor, geneset_id, gene_id_type
+        )
+        return {
+            "gene_identifier_type": gene_id_type.name,
+            "geneset": geneset,
+            "geneset_values": geneset_values,
+        }
+
+    except Exception as err:
+        logger.error(err)
+        raise err
+
+
 def is_geneset_readable_by_user(cursor: Cursor, geneset_id: int, user: User) -> bool:
-    """Check if the user can read the geneset from DB."""
+    """Check if the user can read the geneset from DB.
+
+    @param cursor: DB cursor object
+    @param geneset_id: geneset identifier
+    @param user: GW user
+    @return: True if geneset is readable by user.
+    """
     readable: bool = False
     try:
         readable = db_is_readable(cursor, user.id, geneset_id)

tests/controllers/conftest.py

@@ -1,23 +1,11 @@
 """Fixtures for the controller tests."""
-import importlib.resources
-import json
 from unittest.mock import Mock
 
 import psycopg
 import pytest
 from fastapi.testclient import TestClient
 from geneweaver.api.core.config_class import GeneweaverAPIConfig
 
-# Load test data
-# Opening JSON file
-str_json = importlib.resources.read_text("tests.data", "response_geneset_1234.json")
-# returns JSON string as a dictionary
-test_data = json.loads(str_json)
-
-response_mock = Mock()
-response_mock.status_code = 200
-response_mock.json.return_value = test_data
-
 
 # Mock dependencies
 def mock_full_user() -> Mock:

tests/controllers/test_genesets.py

@@ -2,20 +2,24 @@
 from unittest.mock import patch
 
 import pytest
+from geneweaver.api.controller import message
 
-from tests.controllers.conftest import test_data
+from tests.data import test_geneset_data
+
+geneset_by_id_resp = test_geneset_data.get("geneset_by_id_resp")
+geneset_w_gene_id_type_resp = test_geneset_data.get("geneset_w_gene_id_type_resp")
 
 
 @patch("geneweaver.api.services.geneset.get_geneset")
 @patch("geneweaver.api.services.geneset.is_geneset_readable_by_user")
 def test_get_geneset_response(mock_genset_is_readable, mock_get_genenset, client):
     """Test get geneset ID data response."""
     mock_genset_is_readable.return_value = True
-    mock_get_genenset.return_value = test_data
+    mock_get_genenset.return_value = geneset_by_id_resp
 
     response = client.get("/api/genesets/1234")
     assert response.status_code == 200
-    assert response.json() == test_data
+    assert response.json() == geneset_by_id_resp
 
 
 @patch("geneweaver.api.services.geneset.db_is_readable")
@@ -35,3 +39,44 @@ def test_get_geneset_unexpected_error(mock_genset_is_readable, client):
 
     with pytest.raises(expected_exception=Exception):
         client.get("/api/genesets/1234")
+
+
+@patch("geneweaver.api.services.geneset.get_geneset_w_gene_id_type")
+def test_get_geneset_w_gene_id_type(mock_service_get_geneset_w_gene_id_type, client):
+    """Test get geneset with gene id type response."""
+    mock_service_get_geneset_w_gene_id_type.return_value = geneset_w_gene_id_type_resp
+    response = client.get("/api/genesets/1234?gene_id_type=2")
+
+    assert response.json() == geneset_w_gene_id_type_resp
+    assert response.status_code == 200
+
+
+@patch("geneweaver.api.services.geneset.db_is_readable")
+def test_get_geneset_export_forbidden(mock_genset_is_readable, client):
+    """Test export forbidden response."""
+    mock_genset_is_readable.return_value = False
+    response = client.get("/api/genesets/1234/file?gene_id_type=2")
+
+    assert response.json() == {"detail": "Forbidden"}
+    assert response.status_code == 403
+
+
+@patch("geneweaver.api.services.geneset.get_geneset_w_gene_id_type")
+def test_export_geneset_w_gene_id_type(mock_service_get_geneset_w_gene_id_type, client):
+    """Test geneset file export."""
+    mock_service_get_geneset_w_gene_id_type.return_value = geneset_w_gene_id_type_resp
+    response = client.get("/api/genesets/1234/file?gene_id_type=2")
+
+    assert response.headers.get("content-type") == "application/octet-stream"
+    assert int(response.headers.get("content-length")) > 0
+    assert response.status_code == 200
+
+
+@patch("geneweaver.api.services.geneset.get_geneset_w_gene_id_type")
+def test_invalid_gene_type_id(mock_service_get_geneset_w_gene_id_type, client):
+    """Test geneset file export."""
+    mock_service_get_geneset_w_gene_id_type.return_value = geneset_w_gene_id_type_resp
+    response = client.get("/api/genesets/1234/file?gene_id_type=25")
+
+    assert message.GENE_IDENTIFIER_TYPE_VALUE_ERROR in response.json()["detail"]
+    assert response.status_code == 400

tests/data/__init__.py

@@ -1 +1,19 @@
 """Data package for tests."""
+
+import importlib.resources
+import json
+
+## Load test data
+# Opening JSON files
+geneset_response_json = importlib.resources.read_text(
+    "tests.data", "response_geneset_1234.json"
+)
+geneset_w_gene_id_type_json = importlib.resources.read_text(
+    "tests.data", "response_geneset_w_gene_id_type.json"
+)
+
+# returns JSON string as a dictionary
+test_geneset_data = {
+    "geneset_by_id_resp": json.loads(geneset_response_json),
+    "geneset_w_gene_id_type_resp": json.loads(geneset_w_gene_id_type_json),
+}