Added a test for when all interesting chunks start inside the range and fixed a corner case

Author: Artjoms Iskovs

arctic/tickstore/tickstore.py

@@ -166,6 +166,11 @@ def _mongo_date_range_query(self, symbol, date_range):
         if date_range.start:
             assert date_range.start.tzinfo
             start = date_range.start
+            
+            # If all chunks start inside of the range, we default to capping to our
+            # range so that we don't fetch any chunks from the beginning of time
+            start_range['$gte'] = start
+            
             match = self._symbol_query(symbol)
             match.update({'s': {'$lte': start}})
 
@@ -186,18 +191,12 @@ def _mongo_date_range_query(self, symbol, date_range):
             try:
                 for candidate in result:
                     chunk = self._collection.find_one({'s': candidate['start'], 'sy': candidate['_id']}, {'e': 1})
-                    if chunk['e'].replace(tzinfo=mktz('UTC')) > start:
-                        first_dt = candidate['start'].replace(tzinfo=mktz('UTC'))
+                    if chunk['e'].replace(tzinfo=mktz('UTC')) >= start:
+                        start_range['$gte'] = candidate['start'].replace(tzinfo=mktz('UTC'))
                         break
             except StopIteration:
                 pass
 
-        if first_dt:
-            start_range['$gte'] = first_dt
-        else:
-            # If all chunks start inside of the range, make sure
-            # we don't fetch everything
-            start_range['$gte'] = start
 
         # Find the end bound
         if date_range.end:

tests/integration/tickstore/test_ts_read.py

@@ -453,6 +453,49 @@ def test_read_spanning_chunks(tickstore_lib):
         {'s': {'$gte': dt(2013, 6, 1, 12, 30, tzinfo=mktz('UTC')), '$lte': dt(2013, 6, 1, 15, 0, tzinfo=mktz('UTC'))}}
 
 
+def test_read_inside_range(tickstore_lib):
+    SYM1_DATA = [
+              {'a': 1.,
+               'b': 2.,
+               'index': dt(2013, 6, 1, 0, 00, tzinfo=mktz('UTC'))
+               },
+               {'a': 3.,
+               'b': 4.,
+               'index': dt(2013, 6, 1, 1, 00, tzinfo=mktz('UTC'))
+               },
+             # Chunk boundary here
+               {'a': 5.,
+               'b': 6.,
+               'index': dt(2013, 6, 1, 14, 00, tzinfo=mktz('UTC'))
+               }
+              ]
+    SYM2_DATA = [
+                  {'a': 7.,
+                   'b': 8.,
+                   'index': dt(2013, 6, 1, 12, 30, tzinfo=mktz('UTC'))
+                   },
+                   {'a': 9.,
+                   'b': 10.,
+                   'index': dt(2013, 6, 1, 13, 30, tzinfo=mktz('UTC'))
+                   },
+                 # Chunk boundary here
+                   {'a': 11.,
+                   'b': 12.,
+                   'index': dt(2013, 6, 1, 14, 30, tzinfo=mktz('UTC'))
+                   }
+                  ]
+    tickstore_lib._chunk_size = 2
+    tickstore_lib.write('SYM1', SYM1_DATA)
+    tickstore_lib.write('SYM2', SYM2_DATA)
+
+    # If there are no chunks spanning the range, we still cap the start range so that we don't
+    # fetch SYM1's 0am--1am chunk
+    assert tickstore_lib._mongo_date_range_query(
+                                                 ['SYM1', 'SYM2'],
+                                                 date_range=DateRange(dt(2013, 6, 1, 10, 0, tzinfo=mktz('UTC')),
+                                                                      dt(2013, 6, 1, 15, 0, tzinfo=mktz('UTC')))) == \
+        {'s': {'$gte': dt(2013, 6, 1, 10, 0, tzinfo=mktz('UTC')), '$lte': dt(2013, 6, 1, 15, 0, tzinfo=mktz('UTC'))}}
+
 def test_read_longs(tickstore_lib):
     DUMMY_DATA = [
                   {'a': 1,