Self hosted FLOSS fitness/workout, nutrition and weight tracker
Harden and document the measurement aggregates
Roland Geider43e41d1ca231
diff --git a/wger/measurements/api/aggregates.py b/wger/measurements/api/aggregates.pyindex 13afe253f..7e8bd0edb 100644--- a/wger/measurements/api/aggregates.py+++ b/wger/measurements/api/aggregates.py@@ -37,6 +37,7 @@ from django.db.models import (Min,Sum,)+from django.db.models.fields.json import KeyTextTransformfrom django.db.models.functions import (Cast,Coalesce,@@ -92,9 +93,12 @@ def _bounds_expression(key: str, aggregate):stands for in `extra_data`, and contributes that rather than its mean, socondensing an aggregate keeps the true extremes."""+ # Read as text rather than as JSON: Postgres refuses to cast a JSON string+ # to numeric, even a numeric one, and rows predating the write validation+ # can hold `"48"`return aggregate(Coalesce(- Cast(F(f'extra_data__{key}'), DecimalField(max_digits=8, decimal_places=2)),+ Cast(KeyTextTransform(key, 'extra_data'), DecimalField(max_digits=8, decimal_places=2)),F('value'),))@@ -134,34 +138,21 @@ def bucket_rows(queryset, unit: str, tz, max_points: int = DEFAULT_MAX_POINTS) -if unit not in BUCKET_UNITS:raise InvalidBucket(f'Unknown bucket: {unit}')- rows = (+ return list(queryset.annotate(start=BUCKET_UNITS[unit]('date', tzinfo=tz),- stored_unit=F('extra_data__unit'),+ unit=F('extra_data__unit'),)- .values('category', 'start', 'stored_unit')+ .values('category', 'start', 'unit').annotate(count=Count('id'),- total=Sum('value'),- low=_bounds_expression('min', Min),- high=_bounds_expression('max', Max),+ sum=Sum('value'),+ min=_bounds_expression('min', Min),+ max=_bounds_expression('max', Max),).order_by('start'))- return [- {- 'category': row['category'],- 'start': row['start'],- 'unit': row['stored_unit'],- 'count': row['count'],- 'sum': row['total'],- 'min': row['low'],- 'max': row['high'],- }- for row in rows- ]-def value_count_rows(queryset, tz, summed_per_day: bool = False) -> list[dict]:"""@@ -175,19 +166,21 @@ def value_count_rows(queryset, tz, summed_per_day: bool = False) -> list[dict]:than single readings, for the metrics whose samples mean nothing alone."""if summed_per_day:+ # A day is one value here, so the totals are counted in python: the+ # database has already grouped them into daysdaily = (- queryset.annotate(day=TruncDay('date', tzinfo=tz), stored_unit=F('extra_data__unit'))- .values('category', 'day', 'stored_unit')+ queryset.annotate(day=TruncDay('date', tzinfo=tz), unit=F('extra_data__unit'))+ .values('category', 'day', 'unit').annotate(value=Sum('value'), newest=Max('date')))counted: dict[tuple, dict] = {}for row in daily:- key = (row['category'], row['stored_unit'], row['value'])+ key = (row['category'], row['unit'], row['value'])entry = counted.setdefault(key,{'category': row['category'],- 'unit': row['stored_unit'],+ 'unit': row['unit'],'value': row['value'],'count': 0,'newest': row['newest'],@@ -197,20 +190,9 @@ def value_count_rows(queryset, tz, summed_per_day: bool = False) -> list[dict]:entry['newest'] = max(entry['newest'], row['newest'])return sorted(counted.values(), key=lambda r: (str(r['category']), r['value']))- rows = (- queryset.annotate(stored_unit=F('extra_data__unit'))- .values('category', 'stored_unit', 'value')+ return list(+ queryset.annotate(unit=F('extra_data__unit'))+ .values('category', 'unit', 'value').annotate(count=Count('id'), newest=Max('date')).order_by('value'))-- return [- {- 'category': row['category'],- 'unit': row['stored_unit'],- 'value': row['value'],- 'count': row['count'],- 'newest': row['newest'],- }- for row in rows- ]diff --git a/wger/measurements/api/serializers.py b/wger/measurements/api/serializers.pyindex dac06486d..73c57984d 100644--- a/wger/measurements/api/serializers.py+++ b/wger/measurements/api/serializers.py@@ -269,9 +269,24 @@ class MeasurementSerializer(serializers.ModelSerializer):def validate_extra_data(self, extra_data):"""- The unit key holds the unit the value was entered in+ The unit key holds the unit the value was entered in, min and max the+ range a daily aggregate summarises.++ Those two have to be numbers: the chart aggregate casts them to a+ decimal in SQL, and Postgres refuses to cast a JSON string, even a+ numeric one. A single entry written with `"48"` would take down every+ chart read of its category."""- return validate_json_object(extra_data, 'extra_data', EXTRA_DATA_MAX_BYTES)+ validate_json_object(extra_data, 'extra_data', EXTRA_DATA_MAX_BYTES)++ for key in ('min', 'max'):+ value = (extra_data or {}).get(key)+ if value is not None and not isinstance(value, (int, float)):+ raise serializers.ValidationError(+ f'The {key} of an aggregate has to be a number, not {type(value).__name__}'+ )++ return extra_datadef validate(self, data):"""diff --git a/wger/measurements/api/views.py b/wger/measurements/api/views.pyindex 2d081f1cf..50146aba6 100644--- a/wger/measurements/api/views.py+++ b/wger/measurements/api/views.py@@ -23,6 +23,10 @@ from django.contrib.auth.models import Userfrom django.core.exceptions import PermissionDenied# Third Party+from drf_spectacular.utils import (+ OpenApiParameter,+ extend_schema,+)from rest_framework.decorators import actionfrom rest_framework.permissions import IsAuthenticatedfrom rest_framework.response import Response@@ -130,6 +134,29 @@ class MeasurementViewSet(WgerOwnerObjectModelViewSet):except ValueError:raise InvalidBucket('max_points must be a number')+ @extend_schema(+ summary='Read the entries condensed into chart points',+ parameters=[+ OpenApiParameter(+ 'bucket',+ description='Calendar unit to condense into. The default picks the finest one '+ 'that keeps the series under max_points.',+ enum=['auto', 'hour', 'day', 'week', 'month'],+ ),+ OpenApiParameter(+ 'tz',+ description='IANA name of the zone the buckets are cut in, the server zone by '+ 'default. The column is UTC, and a reading after midnight belongs to the day '+ 'the user had it.',+ ),+ OpenApiParameter(+ 'max_points',+ type=int,+ description=f'Points the auto bucket aims for, {DEFAULT_MAX_POINTS} by default',+ ),+ ],+ responses={200: BucketSerializer(many=True)},+ )@action(detail=False, methods=['get'], serializer_class=BucketSerializer)def aggregate(self, request):"""@@ -154,6 +181,19 @@ class MeasurementViewSet(WgerOwnerObjectModelViewSet):return Response(BucketSerializer(rows, many=True).data)+ @extend_schema(+ summary='Read how often each value occurred',+ parameters=[+ OpenApiParameter('tz', description='IANA name of the zone the days are cut in'),+ OpenApiParameter(+ 'summed_per_day',+ type=bool,+ description='Count daily totals rather than single readings, for the metrics '+ 'whose samples mean nothing on their own (steps, sleep)',+ ),+ ],+ responses={200: ValueCountSerializer(many=True)},+ )@action(detail=False,url_path='value-counts',diff --git a/wger/measurements/tests/test_aggregates.py b/wger/measurements/tests/test_aggregates.pyindex 091054a55..2f655e4d4 100644--- a/wger/measurements/tests/test_aggregates.py+++ b/wger/measurements/tests/test_aggregates.py@@ -190,3 +190,18 @@ class AggregateApiTestCase(WgerTestCase):self.assertEqual(len(rows), 1)self.assertEqual(rows[0]['value'], '5000.00')self.assertEqual(rows[0]['count'], 2)++ def test_a_bound_stored_as_a_string_still_reads(self):+ # Postgres refuses to cast a JSON string to numeric, even a numeric+ # one, so a row predating the write validation would otherwise take+ # every chart read of its category down with it+ self.add(+ timezone.make_aware(datetime.datetime(2026, 5, 4, 8)),+ 70,+ extra={'min': '48', 'max': '165'},+ )++ rows = self.aggregate(bucket='day', tz='UTC')++ self.assertEqual(rows[0]['min'], '48.00')+ self.assertEqual(rows[0]['max'], '165.00')diff --git a/wger/measurements/tests/test_measurements.py b/wger/measurements/tests/test_measurements.pyindex 56959db21..32226f658 100644--- a/wger/measurements/tests/test_measurements.py+++ b/wger/measurements/tests/test_measurements.py@@ -14,6 +14,7 @@# along with Workout Manager. If not, see <http://www.gnu.org/licenses/>.# Standard Library+import jsonfrom decimal import Decimal# Django@@ -464,3 +465,45 @@ class MeasurementExtraDataApiTestCase(WgerTestCase):)self.assertEqual(response.status_code, 201)+++class AggregateBoundValidationTestCase(WgerTestCase):+ """+ The min and max of a daily aggregate have to be numbers: the chart+ aggregate casts them in SQL, and a string breaks every read of the category+ """++ category_id = 'cccccccc-cccc-cccc-cccc-000000000002'++ def setUp(self):+ super().setUp()+ self.user_login('test')++ def post(self, extra_data):+ return self.client.post(+ reverse('measurement-list'),+ json.dumps(+ {+ 'category': self.category_id,+ 'date': '2026-05-04T08:00:00Z',+ 'value': '70.00',+ 'extra_data': extra_data,+ }+ ),+ content_type='application/json',+ )++ def test_a_numeric_bound_is_accepted(self):+ response = self.post({'min': 48, 'max': 165.5})++ self.assertEqual(response.status_code, 201, response.data)++ def test_a_bound_written_as_a_string_is_refused(self):+ for extra_data in ({'min': '48'}, {'max': 'abc'}):+ response = self.post(extra_data)++ self.assertEqual(response.status_code, 400)+ self.assertIn('extra_data', response.data)++ def test_extra_data_without_bounds_is_untouched(self):+ self.assertEqual(self.post({'unit': 'kg', 'source_name': 'Withings'}).status_code, 201)