fix(data): improve management of timeagg and timestep in dataset

src/d3tools/data/dataset.py

@@ -64,10 +64,12 @@ def __init__(self, **kwargs):
         if 'time_signature' in kwargs:
             self.time_signature = kwargs.pop('time_signature')
 
+        if 'aggregation' in kwargs:
+            self.agg = TimeWindow.from_str(kwargs.pop('aggregation'))
+
         if 'timestep' in kwargs:
             self.timestep = TimeStep.from_unit(kwargs.pop('timestep'))
-            if 'aggregation' in kwargs:
-                self.agg = TimeWindow.from_str(kwargs.pop('agg'))
+            if hasattr(self, 'agg'):
                 self.timestep = self.timestep.with_agg(self.agg)
 
         if 'notification' in kwargs:
@@ -114,8 +116,10 @@ def update(self, in_place = False, **kwargs):
                 new_dataset._tile_names = self._tile_names
 
             new_dataset.time_signature = self.time_signature
-            if hasattr(self, 'timestep'):
+            if hasattr(self, 'timestep') and self.timestep is not None:
                 new_dataset.timestep = self.timestep
+            if hasattr(self, 'agg'):
+                new_dataset.agg = self.agg
 
             new_tags = self.tags.copy()
             new_tags.update(kwargs)
@@ -303,7 +307,7 @@ def get_last_date(self, now = None, n = 1, **kwargs) -> dt.datetime|list[dt.date
         while len(last_date) < n:
             this_month_times = self.get_times(this_month, **kwargs)
             if len(this_month_times) > 0:
-                valid_time = [t for t in this_month_times if t <= now]
+                valid_time = [t for t in this_month_times if t < now]
                 valid_time.sort(reverse = True)
                 last_date.extend(valid_time)
             elif this_month.start.year < 1900:
@@ -324,7 +328,7 @@ def get_last_ts(self, **kwargs) -> TimeStep:
         if last_date is None:
             return None
 
-        if hasattr(self, 'timestep'):
+        if hasattr(self, 'timestep') and self.timestep is not None:
             timestep = self.timestep
         else:
             other_dates = self.get_last_date(now = last_date, n = 8, **kwargs)
@@ -338,13 +342,13 @@ def get_last_ts(self, **kwargs) -> TimeStep:
             return timestep.from_date(last_date)
 
     def estimate_timestep(self, date_sample = None, **kwargs) -> TimeStep:
-        if hasattr(self, 'timestep'):
+        if hasattr(self, 'timestep') and self.timestep is not None:
             return self.timestep
 
-        if date_sample is None:
+        if date_sample is None or len(date_sample) == 0:
             date_sample = self.get_last_date(n = 8, **kwargs)
         elif len(date_sample) < 5:
-            other_dates = self.get_last_date(n = 8 - len(date_sample), now = min(date_sample), **kwargs)
+            other_dates = self.get_last_date(n = 8 - len(date_sample), now = min(date_sample), **kwargs)  or []
             date_sample = other_dates + date_sample
 
         timestep = estimate_timestep(date_sample)
@@ -405,7 +409,7 @@ def get_first_ts(self, **kwargs) -> TimeStep:
         if first_date is None:
             return None
 
-        if hasattr(self, 'timestep'):
+        if hasattr(self, 'timestep') and self.timestep is not None:
             timestep = self.timestep
         else:
             other_dates = self.get_first_date(start = first_date, n = 8, **kwargs)
@@ -457,7 +461,7 @@ def get_time_signature(self, timestep: Optional[TimeStep | dt.datetime]) -> dt.d
             time = timestep
             # calculating the length in this way is not perfect,
             # but should work given that timesteps are always requested in order
-            if hasattr(self, 'timestep'): 
+            if hasattr(self, 'timestep') and self.timestep is not None: 
                 length = self.timestep.from_date(time).get_length()
             elif hasattr(self, 'previous_requested_time'):
                 length = (time - self.previous_requested_time).days

src/d3tools/timestepping/timestep.py

@@ -127,6 +127,9 @@ def estimate_timestep(sample) -> TimeStep:
 
         def mode(arr: list): return max(set(arr), key = arr.count)
 
+        if len(sample) < 2:
+            return None
+
         sample.sort()
         all_diff = [(sample[i+1] - sample[i]).days for i in range(len(sample)-1)]
         step_length = mode(all_diff)
@@ -140,8 +143,11 @@ def mode(arr: list): return max(set(arr), key = arr.count)
         if np.isclose(step_length, 1):
             return Day
         elif np.isclose(step_length, 8):
-            return ViirsModisTimeStep
-        elif np.isclose(step_length, 10):
+            if sample[-1].month == 2 and sample[-1].day == 28:
+                return Dekad
+            else:
+                return ViirsModisTimeStep
+        elif 9 <= step_length <= 11:
             return Dekad
         elif 30 <= step_length <= 31:
             return Month