decouple db pipeline from data grabber

cities/modeling/model_interactions.py

@@ -3,17 +3,12 @@
 from typing import Optional
 
 import dill
-
 import torch
 
 import pyro
 import pyro.distributions as dist
-import torch
-
-from cities.modeling.modeling_utils import (
-    prep_wide_data_for_inference,
-    train_interactions_model,
-)
+from cities.modeling.modeling_utils import (prep_wide_data_for_inference,
+                                            train_interactions_model)
 from cities.utils.data_grabber import DataGrabber, find_repo_root
 
 

cities/utils/csv_to_db_pipeline.py

@@ -1,15 +1,37 @@
 import logging
 import os
 import time
+from pathlib import Path
 
 import pandas as pd
 from sqlalchemy import create_engine
 
-from cities.utils.data_grabber import find_repo_root, list_csvs
 
 logging.disable(logging.WARNING)
 
 
+# defining util functions locally to avoid circular imports and major refactor
+# TODO refactor the two functions out of data_grabber.py
+
+
+def list_csvs(csv_dir):
+    csv_names = []
+    for filename in os.listdir(csv_dir):
+        if filename.endswith(".csv"):
+            csv_names.append(filename)
+
+    assert (
+        len(csv_names) > 10
+    ), f"Expected to find more than 10 csv files in {csv_dir}, but found {len(csv_names)}"
+
+    return csv_names
+
+
+def find_repo_root() -> Path:
+    return Path(__file__).parent.parent.parent
+
+
+
 def create_database(data_dir, database_path):
     engine = create_engine(f"sqlite:///{database_path}", echo=True)
     csv_list = list_csvs(data_dir)