fix: Dask pulling of latest data

patilsuraj767 · patilsuraj767 · commit 208171942d39 · 2025-04-07T10:04:18.000+05:30
Signed-off-by: Suraj Patil &lt;patilsuraj767@gmail.com&gt;
diff --git a/sdk/python/feast/infra/offline_stores/dask.py b/sdk/python/feast/infra/offline_stores/dask.py
@@ -361,7 +361,7 @@ def evaluate_offline_job():
 
             source_df = source_df[
                 (source_df[timestamp_field] >= start_date)
-                & (source_df[timestamp_field] < end_date)
+                & (source_df[timestamp_field] <= end_date)
             ]
 
             source_df = source_df.persist()
diff --git a/sdk/python/tests/utils/feature_records.py b/sdk/python/tests/utils/feature_records.py
@@ -519,7 +519,7 @@ def get_last_feature_row(df: pd.DataFrame, driver_id, max_date: datetime):
     """Manually extract last feature value from a dataframe for a given driver_id with up to `max_date` date"""
     filtered = df[
         (df["driver_id"] == driver_id)
-        & (df["event_timestamp"] < max_date.replace(tzinfo=timezone.utc))
+        & (df["event_timestamp"] <= max_date.replace(tzinfo=timezone.utc))
     ]
     max_ts = filtered.loc[filtered["event_timestamp"].idxmax()]["event_timestamp"]
     filtered_by_ts = filtered[filtered["event_timestamp"] == max_ts]

Original file line number	Diff line number	Diff line change
`@@ -361,7 +361,7 @@ def evaluate_offline_job():`
`361`	`361`
`362`	`362`	`source_df = source_df[`
`363`	`363`	`(source_df[timestamp_field] >= start_date)`
`364`		`- & (source_df[timestamp_field] < end_date)`
	`364`	`+ & (source_df[timestamp_field] <= end_date)`
`365`	`365`	`]`
`366`	`366`
`367`	`367`	`source_df = source_df.persist()`
Original file line number	Diff line number	Diff line change
`@@ -519,7 +519,7 @@ def get_last_feature_row(df: pd.DataFrame, driver_id, max_date: datetime):`
`519`	`519`	"""Manually extract last feature value from a dataframe for a given driver_id with up to `max_date` date"""
`520`	`520`	`filtered = df[`
`521`	`521`	`(df["driver_id"] == driver_id)`
`522`		`- & (df["event_timestamp"] < max_date.replace(tzinfo=timezone.utc))`
	`522`	`+ & (df["event_timestamp"] <= max_date.replace(tzinfo=timezone.utc))`
`523`	`523`	`]`
`524`	`524`	`max_ts = filtered.loc[filtered["event_timestamp"].idxmax()]["event_timestamp"]`
`525`	`525`	`filtered_by_ts = filtered[filtered["event_timestamp"] == max_ts]`