GH-17: Load calories from Strava checkout, fix km there

geo_activity_playground/__main__.py

@@ -16,7 +16,6 @@
 from geo_activity_playground.explorer.tile_visits import compute_tile_visits
 from geo_activity_playground.explorer.video import explorer_video_main
 from geo_activity_playground.importers.directory import import_from_directory
-from geo_activity_playground.importers.strava_api import download_missing_calories
 from geo_activity_playground.importers.strava_api import import_from_strava_api
 from geo_activity_playground.webui.app import webui_main
 
@@ -113,11 +112,12 @@ def make_activity_repository(
         import_from_strava_checkout(repository)
     if "strava" in config and not skip_strava:
         import_from_strava_api(repository)
-        download_missing_calories(repository)
 
     embellish_time_series(repository)
     compute_tile_visits(repository)
     compute_tile_evolution()
+    print(repository.meta)
+
     return repository
 
 

geo_activity_playground/core/activities.py

@@ -86,7 +86,7 @@ def last_activity_date(self) -> Optional[datetime.datetime]:
 
     @property
     def activity_ids(self) -> set[int]:
-        return set(self.meta["id"])
+        return set(self.meta.index)
 
     def iter_activities(self, new_to_old=True) -> Iterator[ActivityMeta]:
         direction = -1 if new_to_old else 1

geo_activity_playground/importers/strava_api.py

@@ -151,7 +151,7 @@ def try_import_strava(repository: ActivityRepository) -> None:
                 time_series["time"] = new_time
                 time_series.to_parquet(time_series_path)
 
-            detailed_activity = get_detailed_activity(activity.id)
+            detailed_activity = get_detailed_activity(activity.id, client)
 
             if len(time_series) > 0 and "latitude" in time_series.columns:
                 repository.add_activity(
@@ -209,30 +209,3 @@ def get_detailed_activity(activity_id: int, client: Client):
         pickle.dump(detailed_activity, f)
 
     return detailed_activity
-
-
-def download_missing_calories(repository: ActivityRepository) -> None:
-    client = Client(access_token=get_current_access_token())
-
-    latest_timestamp_path = cache_dir() / "strava-detailed-activity-last.json"
-    if latest_timestamp_path.exists():
-        with open(latest_timestamp_path) as f:
-            latest_timestamp = json.load(f)
-    else:
-        latest_timestamp = "2000-01-01T00:00:00Z"
-
-    try:
-        for activity in tqdm(
-            client.get_activities(after=latest_timestamp),
-            desc="Downloading calories from Strava",
-        ):
-            if repository.has_activity(activity.id):
-                calories = get_detailed_activity(activity.id, client).calories
-                repository.meta.loc[activity.id, "calories"] = calories
-            latest_timestamp = activity.start.isoformat().replace("+00:00", "Z")
-    except RateLimitExceeded:
-        pass
-    finally:
-        repository.save()
-        with open(latest_timestamp_path, "w") as f:
-            json.dump(latest_timestamp, f)

geo_activity_playground/importers/strava_checkout.py

@@ -41,7 +41,7 @@ def import_from_strava_checkout(repository: ActivityRepository) -> None:
         table_activity_meta = {
             "calories": row["Calories"],
             "commute": row["Commute"] == "true",
-            "distance": row["Distance"],
+            "distance_km": row["Distance"],
             "elapsed_time": datetime.timedelta(seconds=int(row["Elapsed Time"])),
             "equipment": str(
                 nan_as_none(row["Activity Gear"])
@@ -59,7 +59,9 @@ def import_from_strava_checkout(repository: ActivityRepository) -> None:
         }
 
         time_series_path = activity_stream_dir / f"{activity_id}.parquet"
-        if not time_series_path.exists():
+        if time_series_path.exists():
+            time_series = pd.read_parquet(time_series_path)
+        else:
             try:
                 file_activity_meta, time_series = read_activity(activity_file)
             except ActivityParseError as e:
@@ -72,10 +74,13 @@ def import_from_strava_checkout(repository: ActivityRepository) -> None:
                 )
                 raise
 
-            if not len(time_series):
-                continue
+        if not len(time_series):
+            continue
+
+        if "latitude" not in time_series.columns:
+            continue
 
-            time_series.to_parquet(time_series_path)
+        time_series.to_parquet(time_series_path)
 
         work_tracker.mark_done(activity_id)
         repository.add_activity(table_activity_meta)