Workflow refactoring (#58)

* separate out etl processes

* move naming logic into etl module

* Move prompts

* remove unused code from data

* simplifications

* robust testing

* add app target to Makefile for running FastAPI application

Author: niklasvm

Makefile

@@ -42,5 +42,8 @@ off:
 	sudo systemctl stop strava.service
 	sudo systemctl stop nginx
 
+app:
+	fastapi run ./src/app/main.py
+
 logs:
 	journalctl -u strava.service -f -n 300

src/tasks/data.py

@@ -1,52 +1,10 @@
 import polyline
 from stravalib.model import SummaryActivity
-from typing import Any
 from shapely.geometry import Polygon
 
 from src.database.models import Activity
 
 
-def process_activity(activity: SummaryActivity) -> dict[str, Any]:
-    activity_dict = activity.model_dump()
-
-    # process columns
-    if activity_dict["start_latlng"]:
-        activity_dict["start_lat"] = activity_dict["start_latlng"][0]
-        activity_dict["start_lng"] = activity_dict["start_latlng"][1]
-    del activity_dict["start_latlng"]
-
-    if activity_dict["end_latlng"]:
-        activity_dict["end_lat"] = activity_dict["end_latlng"][0]
-        activity_dict["end_lng"] = activity_dict["end_latlng"][1]
-    del activity_dict["end_latlng"]
-
-    activity_dict["date"] = activity_dict["start_date_local"].date()
-    activity_dict["time"] = activity_dict["start_date_local"].time()
-    activity_dict["day_of_week"] = activity_dict["start_date_local"].strftime("%A")
-    activity_dict["moving_time_minutes"] = activity_dict["moving_time"] / 60
-    activity_dict["distance_km"] = activity_dict["distance"] / 1000
-
-    try:
-        activity_dict["pace_min_per_km"] = (
-            1.0 * activity_dict["moving_time_minutes"] / activity_dict["distance_km"]
-        )
-    except ZeroDivisionError:
-        activity_dict["pace_min_per_km"] = None
-
-    decoded = polyline.decode(activity_dict["map"]["summary_polyline"])
-    if len(decoded) > 0:
-        poly = Polygon(decoded)
-        centroid = poly.centroid
-        centroid_lat = centroid.x
-        centroid_lon = centroid.y
-        area = poly.area
-
-        activity_dict["map_centroid_lat"] = centroid_lat
-        activity_dict["map_centroid_lon"] = centroid_lon
-        activity_dict["map_area"] = area
-    return activity_dict
-
-
 def summary_activity_to_activity_model(summary_activity: SummaryActivity) -> Activity:
     activity_dict = summary_activity.model_dump()
 

src/tasks/etl.py

@@ -0,0 +1,5 @@
+from src.tasks.etl.auth_etl import AuthETL  # noqa: F401
+from src.tasks.etl.single_activity_etl import SingleActivityETL  # noqa: F401
+from src.tasks.etl.user_etl import UserETL  # noqa: F401
+from src.tasks.etl.historic_activities_etl import ActivitiesETL  # noqa: F401
+from src.tasks.etl.naming_etl import run_name_activity_etl  # noqa: F401

src/tasks/etl/__init__.py

@@ -0,0 +1,33 @@
+import uuid
+
+from stravalib import Client
+
+from src.app.config import Settings
+from src.database.models import Auth
+from src.tasks.etl.base import ETL
+
+
+class AuthETL(ETL):
+    def __init__(self, code: str, scope: str, settings: Settings):
+        super().__init__(settings=settings)
+        self.code = code
+        self.scope = scope
+
+    def extract(self):
+        client = Client()
+        self._token_response = client.exchange_code_for_token(
+            client_id=self.settings.strava_client_id,
+            client_secret=self.settings.strava_client_secret,
+            code=self.code,
+        )
+
+    def load(self):
+        auth = Auth(
+            uuid=uuid.uuid4(),
+            access_token=self._token_response["access_token"],
+            refresh_token=self._token_response["refresh_token"],
+            expires_at=self._token_response["expires_at"],
+            scope=self.scope,
+        )
+        self.db.add_auth(auth)
+        return auth.uuid