API: Adding fallback to default latest query if record metadata.start…

… is greater than 24 hrs in future.

api/datalake_api/querier.py

@@ -20,6 +20,8 @@
 import time
 import os
 
+from datetime import datetime, timedelta
+import decimal
 import logging
 log = logging.getLogger(__name__)
 
@@ -40,6 +42,7 @@
 slow and expensive.
 '''
 DEFAULT_LOOKBACK_DAYS = 14
+LATEST_MAX_LOOKFORWARD_HOURS = 24
 
 
 _ONE_DAY_MS = 24 * 60 * 60 * 1000
@@ -347,6 +350,9 @@ def _latest_table(self):
         return self.dynamodb.Table(self.latest_table_name)
 
     def query_latest(self, what, where, lookback_days=DEFAULT_LOOKBACK_DAYS):
+        now = datetime.utcnow()
+        max_lookforward = now + timedelta(hours=LATEST_MAX_LOOKFORWARD_HOURS)
+
         if self.use_latest_table:
             log.info('inside use_latest_table=TRUE')
             response = self._latest_table.query(
@@ -359,6 +365,18 @@ def query_latest(self, what, where, lookback_days=DEFAULT_LOOKBACK_DAYS):
                 return self._default_latest(what, where, lookback_days)
 
             latest_item = items[0]
+            metadata_start = latest_item['metadata']['start']
+
+            if isinstance(metadata_start, (int, decimal.Decimal)):
+                metadata_start = datetime.utcfromtimestamp(float(metadata_start) / 1000)
+
+            elif isinstance(metadata_start, str):
+                metadata_start = datetime.strptime(metadata_start, '%Y-%m-%dT%H:%M:%S.%fZ')
+
+            if metadata_start > max_lookforward:
+                log.info(f"Record with metadata.start {metadata_start} is beyond MAX_LOOKFORWARD_HOURS. Falling back to default latest.")
+                return self._default_latest(what, where, lookback_days)
+
             return dict(url=latest_item['url'], metadata=latest_item['metadata'])
 
         else:

api/tests/test_archive_querier.py

@@ -598,3 +598,19 @@ def test_query_latest_just_latest_table(table_maker, querier, record_maker):
         _validate_latest_result(result, what='meow', where='tree')
     else:
         assert result is None
+
+
+def test_query_latest_future_record_exceeds_lookforward(table_maker, querier, record_maker):
+    future_start = (int(time.time() * 1000) + 25 * 60 * 60 * 1000)  # 25 hours ahead
+    future_end = (int(time.time() * 1000) + 26 * 60 * 60 * 1000)    # ends one hour later
+    record = record_maker(what='meow', where='tree', start=future_start, end=future_end)
+
+    default_table, latest_table = table_maker([])
+    print(default_table.__dict__, type(default_table))
+
+    default_table.put_item(Item=record[0])
+    latest_table.put_item(Item=record[0])
+
+    result = querier.query_latest('meow', 'tree')
+    assert result is None, "No result should be returned if falling back to the default query"
+