fix: duplicate reports when run for past years

osf/metrics/reports.py

@@ -30,6 +30,11 @@ def __init_subclass__(cls, **kwargs):
         super().__init_subclass__(**kwargs)
         assert 'report_date' in cls.UNIQUE_TOGETHER_FIELDS, f'DailyReport subclasses must have "report_date" in UNIQUE_TOGETHER_FIELDS (on {cls.__qualname__}, got {cls.UNIQUE_TOGETHER_FIELDS})'
 
+    def save(self, *args, **kwargs):
+        if self.timestamp is None:
+            self.timestamp = self.report_date
+        super().save(*args, **kwargs)
+
     class Meta:
         abstract = True
         dynamic = metrics.MetaField('strict')
@@ -41,19 +46,15 @@ def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs, format='strict_year_month')
 
     def deserialize(self, data):
-        if isinstance(data, YearMonth):
-            return data
-        elif isinstance(data, str):
-            return YearMonth.from_str(data)
-        elif isinstance(data, (datetime.datetime, datetime.date)):
-            return YearMonth.from_date(data)
-        elif isinstance(data, int):
+        if isinstance(data, int):
             # elasticsearch stores dates in milliseconds since the unix epoch
             _as_datetime = datetime.datetime.fromtimestamp(data // 1000)
             return YearMonth.from_date(_as_datetime)
         elif data is None:
             return None
-        else:
+        try:
+            return YearMonth.from_any(data)
+        except ValueError:
             raise ValueError(f'unsure how to deserialize "{data}" (of type {type(data)}) to YearMonth')
 
     def serialize(self, data):
@@ -102,6 +103,11 @@ def __init_subclass__(cls, **kwargs):
         super().__init_subclass__(**kwargs)
         assert 'report_yearmonth' in cls.UNIQUE_TOGETHER_FIELDS, f'MonthlyReport subclasses must have "report_yearmonth" in UNIQUE_TOGETHER_FIELDS (on {cls.__qualname__}, got {cls.UNIQUE_TOGETHER_FIELDS})'
 
+    def save(self, *args, **kwargs):
+        if self.timestamp is None:
+            self.timestamp = YearMonth.from_any(self.report_yearmonth).month_start()
+        super().save(*args, **kwargs)
+
 
 @receiver(metrics_pre_save)
 def set_report_id(sender, instance, **kwargs):

osf/metrics/utils.py

@@ -46,6 +46,16 @@ def from_str(cls, input_str: str) -> YearMonth:
         else:
             raise ValueError(f'expected YYYY-MM format, got "{input_str}"')
 
+    @classmethod
+    def from_any(cls, data) -> YearMonth:
+        if isinstance(data, YearMonth):
+            return data
+        elif isinstance(data, str):
+            return YearMonth.from_str(data)
+        elif isinstance(data, (datetime.datetime, datetime.date)):
+            return YearMonth.from_date(data)
+        raise ValueError(f'cannot coerce {data} into YearMonth')
+
     def __str__(self):
         """convert to string of "YYYY-MM" format"""
         return f'{self.year}-{self.month:0>2}'

osf_tests/metrics/test_yearmonth.txt

@@ -24,6 +24,24 @@ YearMonth(year=1974, month=3)
 >>> YearMonth.from_str('2000-12')
 YearMonth(year=2000, month=12)
 
+`from_any` constructor, accepts YearMonth, "YYYY-MM", or date/datetime
+>>> YearMonth.from_any('2000-12')
+YearMonth(year=2000, month=12)
+>>> YearMonth.from_any(_) is _
+True
+>>> YearMonth.from_any(datetime.date(1973, 1, 1))
+YearMonth(year=1973, month=1)
+>>> YearMonth.from_any(datetime.datetime(1974, 3, 2))
+YearMonth(year=1974, month=3)
+>>> YearMonth.from_any(None)
+Traceback (most recent call last):
+    ...
+ValueError: cannot coerce None into YearMonth
+>>> YearMonth.from_any(7)
+Traceback (most recent call last):
+    ...
+ValueError: cannot coerce 7 into YearMonth
+
 `__str__` method gives "YYYY-MM" format:
 >>> str(YearMonth(1491, 7))
 '1491-07'