From 3cea82a7344bf2ea46fca940ca006b98c3e1605d Mon Sep 17 00:00:00 2001
From: Mark007-R <markrodrigues2004@gmail.com>
Date: Sun, 24 May 2026 13:25:40 +0530
Subject: [PATCH] =?UTF-8?q?sprint:=20mark=20data-dependent=20Day-4=20tests?=
 =?UTF-8?q?=20with=20requires=5Fdata=20(Day=207=20CI=20fix)=20=E2=80=94=20?=
 =?UTF-8?q?test=5Fapi=20needs=20models/fraud=5Fmodel.pkl,=20test=5Floader?=
 =?UTF-8?q?=5Freads=5Fx=5Ftest=5Fand=5Fy=5Ftest=20needs=20data/processed/*?=
 =?UTF-8?q?.csv;=20both=20DVC-tracked,=20not=20in=20git;=20CI's=20-m=20'no?=
 =?UTF-8?q?t=20requires=5Fdata'=20now=20finds=20the=20marker=20via=20tests?=
 =?UTF-8?q?/conftest.py?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 tests/conftest.py         | 19 +++++++++++++++++++
 tests/test_api.py         |  7 +++++++
 tests/test_data_loader.py |  3 +++
 3 files changed, 29 insertions(+)
 create mode 100644 tests/conftest.py

diff --git a/tests/conftest.py b/tests/conftest.py
new file mode 100644
index 0000000..f76fedb
--- /dev/null
+++ b/tests/conftest.py
@@ -0,0 +1,19 @@
+"""Pytest configuration -- shared markers for the Sentinel test suite.
+
+Day 7 Phase 6 (2026-05-24). The CI workflow at .github/workflows/ci.yml runs
+`pytest tests/ -m "not requires_data"` so the GitHub runner skips tests that
+need the DVC-tracked data and trained model artifacts (which are not in git).
+This conftest registers the `requires_data` marker so pytest does not emit a
+PytestUnknownMarkWarning when the marker is used.
+"""
+
+from __future__ import annotations
+
+
+def pytest_configure(config) -> None:
+    config.addinivalue_line(
+        "markers",
+        "requires_data: test depends on DVC-tracked data and/or the trained "
+        "model artifact in models/. Skipped in CI; run locally after "
+        "`dvc repro train`.",
+    )
diff --git a/tests/test_api.py b/tests/test_api.py
index 1f10a86..86ebbcd 100644
--- a/tests/test_api.py
+++ b/tests/test_api.py
@@ -15,6 +15,13 @@
 from fastapi.testclient import TestClient
 
 
+# The FastAPI app boots from models/fraud_model.pkl + data/processed/X_test.csv,
+# neither of which is in git -- they come from `dvc repro train`. CI skips this
+# whole module via the `-m "not requires_data"` marker filter; run locally
+# after a successful pipeline run.
+pytestmark = pytest.mark.requires_data
+
+
 @pytest.fixture(scope="module")
 def client(tmp_path_factory: pytest.TempPathFactory) -> TestClient:
     tmp_dir = tmp_path_factory.mktemp("api-test")
diff --git a/tests/test_data_loader.py b/tests/test_data_loader.py
index fc46b19..198d6da 100644
--- a/tests/test_data_loader.py
+++ b/tests/test_data_loader.py
@@ -5,6 +5,7 @@
 from pathlib import Path
 
 import pandas as pd
+import pytest
 
 from src.data.loader import LoaderConfig, SentinelDataLoader, dvc_status
 
@@ -17,6 +18,8 @@ def test_loader_resolves_paths_from_params() -> None:
     assert paths.model.name == "fraud_model.pkl"
 
 
+# Reads data/processed/X_test.csv + y_test.csv which are not in git (DVC).
+@pytest.mark.requires_data
 def test_loader_reads_x_test_and_y_test() -> None:
     loader = SentinelDataLoader()
     X = loader.load_x_test()