feat: Add LMDB sample data retrieval API endpoint with tests

Signed-off-by: Munyoki Kilyungi <me@bonfacemunyoki.com>

8 files changed, 97 insertions, 9 deletions

diff --git a/gn3/api/lmdb_sample_data.py b/gn3/api/lmdb_sample_data.py
new file mode 100644
index 0000000..eaa71c2
--- /dev/null
+++ b/gn3/api/lmdb_sample_data.py
@@ -0,0 +1,40 @@
+"""API endpoint for retrieving sample data from LMDB storage"""
+import hashlib
+from pathlib import Path
+
+import lmdb
+from flask import Blueprint, current_app, jsonify
+
+lmdb_sample_data = Blueprint("lmdb_sample_data", __name__)
+
+
+@lmdb_sample_data.route("/sample-data/<string:dataset>/<int:trait_id>", methods=["GET"])
+def get_sample_data(dataset: str, trait_id: int):
+    """Retrieve sample data from LMDB for a given dataset and trait.
+
+    Path Parameters:
+        dataset: The name of the dataset
+        trait_id: The ID of the trait
+
+    Returns:
+        JSON object mapping sample IDs to their values
+    """
+    checksum = hashlib.md5(
+        f"{dataset}-{trait_id}".encode()
+    ).hexdigest()
+
+    db_path = Path(current_app.config["LMDB_DATA_PATH"]) / checksum
+    if not db_path.exists():
+        return jsonify(error="No data found for given dataset and trait"), 404
+    try:
+        with lmdb.open(str(db_path), max_dbs=15, readonly=True) as env:
+            data = {}
+            with env.begin(write=False) as txn:
+                cursor = txn.cursor()
+                for key, value in cursor:
+                    data[key.decode()] = float(value.decode())
+
+            return jsonify(data)
+
+    except lmdb.Error as err:
+        return jsonify(error=f"LMDB error: {str(err)}"), 500
diff --git a/gn3/app.py b/gn3/app.py
index 3841396..74bb5ab 100644
--- a/gn3/app.py
+++ b/gn3/app.py
@@ -30,6 +30,7 @@ from gn3.api.llm import gnqa
 from gn3.api.rqtl2 import rqtl2
 from gn3.api.streaming import streaming
 from gn3.case_attributes import caseattr
+from gn3.api.lmdb_sample_data import lmdb_sample_data
 
 
 class ConfigurationError(Exception):
@@ -111,6 +112,7 @@ def create_app(config: Union[Dict, str, None] = None) -> Flask:
     app.register_blueprint(gnqa, url_prefix="/api/llm")
     app.register_blueprint(rqtl2, url_prefix="/api/rqtl2")
     app.register_blueprint(streaming, url_prefix="/api/stream")
+    app.register_blueprint(lmdb_sample_data, url_prefix="/api/lmdb")
 
     register_error_handlers(app)
     return app
diff --git a/gn3/settings.py b/gn3/settings.py
index 9a3f7eb..e988106 100644
--- a/gn3/settings.py
+++ b/gn3/settings.py
@@ -19,9 +19,9 @@ SPARQL_ENDPOINT = os.environ.get(
     "SPARQL_ENDPOINT",
     "http://localhost:9082/sparql")
 
-# LMDB path
-LMDB_PATH = os.environ.get(
-    "LMDB_PATH", f"{os.environ.get('HOME')}/tmp/dataset")
+# LMDB paths
+LMDB_DATA_PATH = os.environ.get(
+    "LMDB_DATA_PATH", "/export5/lmdb-data-hashes")
 
 # SQL confs
 SQL_URI = os.environ.get(
diff --git a/tests/integration/conftest.py b/tests/integration/conftest.py
index 8e39726..bdbab09 100644
--- a/tests/integration/conftest.py
+++ b/tests/integration/conftest.py
@@ -1,4 +1,5 @@
 """Module that holds fixtures for integration tests"""
+from pathlib import Path
 import pytest
 import MySQLdb
 
@@ -6,19 +7,25 @@ from gn3.app import create_app
 from gn3.chancy import random_string
 from gn3.db_utils import parse_db_url, database_connection
 
+
 @pytest.fixture(scope="session")
 def client():
     """Create a test client fixture for tests"""
     # Do some setup
-    app = create_app()
-    app.config.update({"TESTING": True})
-    app.testing = True
+    app = create_app({
+        "TESTING": True,
+        "LMDB_DATA_PATH": str(
+            Path(__file__).parent.parent /
+            Path("test_data/lmdb-test-data")
+        ),
+        "AUTH_SERVER_URL": "http://127.0.0.1:8081",
+    })
     yield app.test_client()
     # Do some teardown/cleanup
 
 
 @pytest.fixture(scope="session")
-def db_conn(client): # pylint: disable=[redefined-outer-name]
+def db_conn(client):  # pylint: disable=[redefined-outer-name]
     """Create a db connection fixture for tests"""
     # 01) Generate random string to append to all test db artifacts for the session
     live_db_uri = client.application.config["SQL_URI"]
diff --git a/tests/integration/test_lmdb_sample_data.py b/tests/integration/test_lmdb_sample_data.py
new file mode 100644
index 0000000..30a23f4
--- /dev/null
+++ b/tests/integration/test_lmdb_sample_data.py
@@ -0,0 +1,31 @@
+"""Tests for the LMDB sample data API endpoint"""
+import pytest
+
+
+@pytest.mark.unit_test
+def test_nonexistent_data(client):
+    """Test endpoint returns 404 when data doesn't exist"""
+    response = client.get("/api/lmdb/sample-data/nonexistent/123")
+    assert response.status_code == 404
+    assert response.json["error"] == "No data found for given dataset and trait"
+
+
+@pytest.mark.unit_test
+def test_successful_retrieval(client):
+    """Test successful data retrieval using test LMDB data"""
+    # Use known test data hash: 7308efbd84b33ad3d69d14b5b1f19ccc
+    response = client.get("/api/lmdb/sample-data/BXDPublish/10007")
+    assert response.status_code == 200
+
+    data = response.json
+    assert len(data) == 31
+    # Verify some known values from the test database
+    assert data["BXD1"] == 18.700001
+    assert data["BXD11"] == 18.9
+
+
+@pytest.mark.unit_test
+def test_invalid_trait_id(client):
+    """Test endpoint handles invalid trait IDs appropriately"""
+    response = client.get("/api/lmdb/sample-data/BXDPublish/999999")
+    assert response.status_code == 404
diff --git a/tests/test_data/lmdb-test-data/7308efbd84b33ad3d69d14b5b1f19ccc/data.mdb b/tests/test_data/lmdb-test-data/7308efbd84b33ad3d69d14b5b1f19ccc/data.mdb
new file mode 100755
index 0000000..5fa213b
--- /dev/null
+++ b/tests/test_data/lmdb-test-data/7308efbd84b33ad3d69d14b5b1f19ccc/data.mdb
Binary files differdiff --git a/tests/test_data/lmdb-test-data/7308efbd84b33ad3d69d14b5b1f19ccc/lock.mdb b/tests/test_data/lmdb-test-data/7308efbd84b33ad3d69d14b5b1f19ccc/lock.mdb
new file mode 100755
index 0000000..116d824
--- /dev/null
+++ b/tests/test_data/lmdb-test-data/7308efbd84b33ad3d69d14b5b1f19ccc/lock.mdb
Binary files differdiff --git a/tests/unit/conftest.py b/tests/unit/conftest.py
index d9d5492..5526d16 100644
--- a/tests/unit/conftest.py
+++ b/tests/unit/conftest.py
@@ -7,6 +7,7 @@ import pytest
 
 from gn3.app import create_app
 
+
 @pytest.fixture(scope="session")
 def fxtr_app():
     """Fixture: setup the test app"""
@@ -16,6 +17,11 @@ def fxtr_app():
             f'testdb_{datetime.now().strftime("%Y%m%dT%H%M%S")}')
         app = create_app({
             "TESTING": True,
+            "LMDB_DATA_PATH": str(
+                Path(__file__).parent.parent /
+                Path("test_data/lmdb-test-data")
+            ),
+            "AUTH_SERVER_URL": "http://127.0.0.1:8081",
             "OAUTH2_ACCESS_TOKEN_GENERATOR": "tests.unit.auth.test_token.gen_token"
         })
         app.testing = True
@@ -23,13 +29,15 @@ def fxtr_app():
         # Clean up after ourselves
         testdb.unlink(missing_ok=True)
 
+
 @pytest.fixture(scope="session")
-def client(fxtr_app): # pylint: disable=redefined-outer-name
+def client(fxtr_app):  # pylint: disable=redefined-outer-name
     """Create a test client fixture for tests"""
     with fxtr_app.app_context():
         yield fxtr_app.test_client()
 
+
 @pytest.fixture(scope="session")
-def fxtr_app_config(client): # pylint: disable=redefined-outer-name
+def fxtr_app_config(client):  # pylint: disable=redefined-outer-name
     """Return the test application's configuration object"""
     return client.application.config