Changed record filter by date

implements bp optimize-date-filtering

Change-Id: I6d4157992613a8188a94e3ad041ebd36768b0077

dashboard/decorators.py

@@ -32,24 +32,21 @@ from stackalytics import version as stackalytics_version
 LOG = logging.getLogger(__name__)
 
 
-def _get_time_filter(kwargs, ignore):
-    start_date = parameters.get_single_parameter(kwargs, 'start_date')
+def _filter_records_by_days(ignore, start_date, end_date, memory_storage_inst):
     if start_date and 'start_date' not in ignore:
         start_date = utils.date_to_timestamp_ext(start_date)
     else:
-        start_date = 0
-    end_date = parameters.get_single_parameter(kwargs, 'end_date')
+        start_date = memory_storage_inst.get_first_record_day()
     if end_date and 'end_date' not in ignore:
         end_date = utils.date_to_timestamp_ext(end_date)
     else:
         end_date = utils.date_to_timestamp_ext('now')
 
-    def time_filter(records):
-        for record in records:
-            if start_date <= record['date'] <= end_date:
-                yield record
+    start_day = utils.timestamp_to_day(start_date)
+    end_day = utils.timestamp_to_day(end_date)
 
-    return time_filter
+    return memory_storage_inst.get_record_ids_by_days(
+        six.moves.range(start_day, end_day + 1))
 
 
 def record_filter(ignore=None, use_default=True):
@@ -144,11 +141,17 @@ def record_filter(ignore=None, use_default=True):
                         memory_storage_inst.get_record_ids_by_blueprint_ids(
                             param))
 
-            time_filter = _get_time_filter(kwargs, ignore)
+            start_date = parameters.get_single_parameter(kwargs, 'start_date')
+            end_date = parameters.get_single_parameter(kwargs, 'end_date')
+
+            if (start_date and 'start_date' not in ignore) or (
+                end_date and 'end_date' not in ignore):
+                record_ids &= _filter_records_by_days(ignore,
+                                                      start_date, end_date,
+                                                      memory_storage_inst)
 
             kwargs['record_ids'] = record_ids
-            kwargs['records'] = time_filter(
-                memory_storage_inst.get_records(record_ids))
+            kwargs['records'] = memory_storage_inst.get_records(record_ids)
             return f(*args, **kwargs)
 
         return record_filter_decorated_function
@@ -217,7 +220,7 @@ def man_days_filter(result, record, param_id):
         # effort of the author (author's effort is represented in patches)
         return
 
-    day = record['date'] // (24 * 3600)
+    day = utils.timestamp_to_day(record['date'])
 
     result_by_param = result[record[param_id]]
     if 'days' not in result_by_param:

dashboard/memory_storage.py

@@ -15,6 +15,8 @@
 
 import six
 
+from stackalytics.processor import utils
+
 
 MEMORY_STORAGE_CACHED = 0
 
@@ -38,6 +40,7 @@ class CachedMemoryStorage(MemoryStorage):
         self.release_index = {}
         self.blueprint_id_index = {}
         self.company_name_mapping = {}
+        self.day_index = {}
 
         self.indexes = {
             'primary_key': self.primary_key_index,
@@ -60,6 +63,12 @@ class CachedMemoryStorage(MemoryStorage):
             else:
                 self.blueprint_id_index[bp_id] = set([record['record_id']])
 
+        record_day = utils.timestamp_to_day(record['date'])
+        if record_day in self.day_index:
+            self.day_index[record_day].add(record['record_id'])
+        else:
+            self.day_index[record_day] = set([record['record_id']])
+
     def update(self, records):
         have_updates = False
 
@@ -81,6 +90,9 @@ class CachedMemoryStorage(MemoryStorage):
         for key, index in six.iteritems(self.indexes):
             index[record[key]].remove(record['record_id'])
 
+        record_day = utils.timestamp_to_day(record['date'])
+        self.day_index[record_day].remove(record['record_id'])
+
     def _add_to_index(self, record_index, record, key):
         record_key = record[key]
         if record_key in record_index:
@@ -114,6 +126,9 @@ class CachedMemoryStorage(MemoryStorage):
         return self._get_record_ids_from_index(blueprint_ids,
                                                self.blueprint_id_index)
 
+    def get_record_ids_by_days(self, days):
+        return self._get_record_ids_from_index(days, self.day_index)
+
     def get_index_keys_by_record_ids(self, index_name, record_ids):
         return set([key
                     for key, value in six.iteritems(self.indexes[index_name])
@@ -149,6 +164,9 @@ class CachedMemoryStorage(MemoryStorage):
     def get_user_ids(self):
         return self.user_id_index.keys()
 
+    def get_first_record_day(self):
+        return min(self.day_index.keys())
+
 
 def get_memory_storage(memory_storage_type):
     if memory_storage_type == MEMORY_STORAGE_CACHED:

dashboard/web.py

@@ -541,7 +541,7 @@ def timeline(records, **kwargs):
                     (record['week'] not in weeks)):
                 continue
 
-            day = record['date'] // (24 * 3600)
+            day = utils.timestamp_to_day(record['date'])
             user = vault.get_user_from_runtime_storage(record['user_id'])
             if record['release'] == release_name:
                 release_stat[day] |= set([user['seq']])

stackalytics/processor/utils.py

@@ -68,6 +68,10 @@ def week_to_date(week):
             strftime('%Y-%m-%d %H:%M:%S'))
 
 
+def timestamp_to_day(timestamp):
+    return timestamp // (24 * 3600)
+
+
 def check_email_validity(email):
     if email:
         return re.match(r'[\w\d_\.-]+@([\w\d_\.-]+\.)+[\w]+', email)