Added date filter

eyra · Dec 6, 2023 · 6db584d · 6db584d
1 parent 6a561d9
commit 6db584d
Show file tree

Hide file tree

Showing 4 changed files with 20 additions and 0 deletions.
diff --git a/public/port-0.0.0-py3-none-any.whl b/public/port-0.0.0-py3-none-any.whl
diff --git a/src/framework/processing/py/dist/port-0.0.0-py3-none-any.whl b/src/framework/processing/py/dist/port-0.0.0-py3-none-any.whl
diff --git a/src/framework/processing/py/port/script.py b/src/framework/processing/py/port/script.py
@@ -10,6 +10,7 @@
 import zipfile
 
 ExtractionResult = namedtuple("ExtractionResult", ["id", "title", "data_frame"])
+filter_start_date = datetime(2017, 1, 1)
 
 
 def get_in(dct, *key_path):
@@ -37,6 +38,9 @@ def parse_json_to_dataframe(parsed_dict):
             start_timestamp_str[:-1]
         )  # remove the 'Z'
 
+        if start_timestamp < filter_start_date:
+            continue
+
         if meters := get_in(segment, "waypointPath", "distanceMeters"):
             distance_meters = meters
         elif meters := get_in(segment, "simplifiedRawPath", "distanceMeters"):

diff --git a/src/framework/processing/py/tests/script_test.py b/src/framework/processing/py/tests/script_test.py
@@ -106,6 +106,22 @@ def test_parse_json_to_dataframe_skips_non_walking_or_cycling():
     assert "DRIVING" not in df.activityType.values
 
 
+def test_parse_json_to_dataframe_skips_entries_before_filter_date():
+    parsed_dict= {
+        "timelineObjects": [
+            {
+                "activitySegment": {
+                    "duration": {"startTimestamp": "2016-12-31T19:13:27.023Z"},
+                    "activityType": "CYCLING",
+                    "waypointPath": {"distanceMeters": 3600.33},
+                }
+            }
+        ]
+    }
+    df = parse_json_to_dataframe(parsed_dict)
+    assert len(df) == 0
+
+
 def test_aggregate_distance_by_day_activity(sample_data):
     df = parse_json_to_dataframe(sample_data)
     aggregated_df = aggregate_distance_by_day_activity(df)