/scoring refactoring and add model-run viewer endpoints

django/src/rdwatch_scoring/api.py

@@ -1,7 +1,13 @@
 from ninja import NinjaAPI
 
-from .views import router as scores
+from .views.model_run import router as modelruns
+from .views.performer import router as performers
+from .views.region import router as regions
+from .views.scores import router as scores
 
 api = NinjaAPI(urls_namespace='scoring')
 
-api.add_router('/scores/', scores)
+api.add_router('/scoring/scores/', scores)
+api.add_router('/scoring/performers/', performers)
+api.add_router('/scoring/regions/', regions)
+api.add_router('/scoring/model-runs/', modelruns)

django/src/rdwatch_scoring/models.py

@@ -48,7 +48,7 @@ class Addrfeat(models.Model):
     rtotyp = models.CharField(max_length=1, blank=True, null=True)
     offsetl = models.CharField(max_length=1, blank=True, null=True)
     offsetr = models.CharField(max_length=1, blank=True, null=True)
-    the_geom = models.LineStringField(srid=4269, blank=True, null=True)
+    the_geom = models.LineStringField(srid=4326, blank=True, null=True)
 
     class Meta:
         managed = False
@@ -79,7 +79,7 @@ class Bg(models.Model):
     awater = models.FloatField(blank=True, null=True)
     intptlat = models.CharField(max_length=11, blank=True, null=True)
     intptlon = models.CharField(max_length=12, blank=True, null=True)
-    the_geom = models.MultiPolygonField(srid=4269, blank=True, null=True)
+    the_geom = models.MultiPolygonField(srid=4326, blank=True, null=True)
 
     class Meta:
         managed = False
@@ -106,7 +106,7 @@ class County(models.Model):
     awater = models.FloatField(blank=True, null=True)
     intptlat = models.CharField(max_length=11, blank=True, null=True)
     intptlon = models.CharField(max_length=12, blank=True, null=True)
-    the_geom = models.MultiPolygonField(srid=4269, blank=True, null=True)
+    the_geom = models.MultiPolygonField(srid=4326, blank=True, null=True)
 
     class Meta:
         managed = False
@@ -162,7 +162,7 @@ class Cousub(models.Model):
     awater = models.DecimalField(max_digits=14, decimal_places=0, blank=True, null=True)
     intptlat = models.CharField(max_length=11, blank=True, null=True)
     intptlon = models.CharField(max_length=12, blank=True, null=True)
-    the_geom = models.MultiPolygonField(srid=4269, blank=True, null=True)
+    the_geom = models.MultiPolygonField(srid=4326, blank=True, null=True)
 
     class Meta:
         managed = False
@@ -213,7 +213,7 @@ class Edges(models.Model):
     offsetr = models.CharField(max_length=1, blank=True, null=True)
     tnidf = models.DecimalField(max_digits=10, decimal_places=0, blank=True, null=True)
     tnidt = models.DecimalField(max_digits=10, decimal_places=0, blank=True, null=True)
-    the_geom = models.MultiLineStringField(srid=4269, blank=True, null=True)
+    the_geom = models.MultiLineStringField(srid=4326, blank=True, null=True)
 
     class Meta:
         managed = False
@@ -513,7 +513,7 @@ class Faces(models.Model):
     atotal = models.FloatField(blank=True, null=True)
     intptlat = models.CharField(max_length=11, blank=True, null=True)
     intptlon = models.CharField(max_length=12, blank=True, null=True)
-    the_geom = models.MultiPolygonField(srid=4269, blank=True, null=True)
+    the_geom = models.MultiPolygonField(srid=4326, blank=True, null=True)
     tractce20 = models.CharField(max_length=6, blank=True, null=True)
     blkgrpce20 = models.CharField(max_length=1, blank=True, null=True)
     blockce20 = models.CharField(max_length=4, blank=True, null=True)
@@ -754,7 +754,7 @@ class Place(models.Model):
     awater = models.BigIntegerField(blank=True, null=True)
     intptlat = models.CharField(max_length=11, blank=True, null=True)
     intptlon = models.CharField(max_length=12, blank=True, null=True)
-    the_geom = models.MultiPolygonField(srid=4269, blank=True, null=True)
+    the_geom = models.MultiPolygonField(srid=4326, blank=True, null=True)
 
     class Meta:
         managed = False
@@ -851,7 +851,7 @@ class State(models.Model):
     awater = models.BigIntegerField(blank=True, null=True)
     intptlat = models.CharField(max_length=11, blank=True, null=True)
     intptlon = models.CharField(max_length=12, blank=True, null=True)
-    the_geom = models.MultiPolygonField(srid=4269, blank=True, null=True)
+    the_geom = models.MultiPolygonField(srid=4326, blank=True, null=True)
 
     class Meta:
         managed = False
@@ -898,7 +898,7 @@ class Tabblock(models.Model):
     awater = models.FloatField(blank=True, null=True)
     intptlat = models.CharField(max_length=11, blank=True, null=True)
     intptlon = models.CharField(max_length=12, blank=True, null=True)
-    the_geom = models.MultiPolygonField(srid=4269, blank=True, null=True)
+    the_geom = models.MultiPolygonField(srid=4326, blank=True, null=True)
 
     class Meta:
         managed = False
@@ -922,7 +922,7 @@ class Tabblock20(models.Model):
     awater = models.FloatField(blank=True, null=True)
     intptlat = models.CharField(max_length=11, blank=True, null=True)
     intptlon = models.CharField(max_length=12, blank=True, null=True)
-    the_geom = models.MultiPolygonField(srid=4269, blank=True, null=True)
+    the_geom = models.MultiPolygonField(srid=4326, blank=True, null=True)
     housing = models.FloatField(blank=True, null=True)
     pop = models.FloatField(blank=True, null=True)
 
@@ -958,7 +958,7 @@ class Tract(models.Model):
     awater = models.FloatField(blank=True, null=True)
     intptlat = models.CharField(max_length=11, blank=True, null=True)
     intptlon = models.CharField(max_length=12, blank=True, null=True)
-    the_geom = models.MultiPolygonField(srid=4269, blank=True, null=True)
+    the_geom = models.MultiPolygonField(srid=4326, blank=True, null=True)
 
     class Meta:
         managed = False
@@ -978,7 +978,7 @@ class Zcta5(models.Model):
     intptlat = models.CharField(max_length=11, blank=True, null=True)
     intptlon = models.CharField(max_length=12, blank=True, null=True)
     partflg = models.CharField(max_length=1, blank=True, null=True)
-    the_geom = models.MultiPolygonField(srid=4269, blank=True, null=True)
+    the_geom = models.MultiPolygonField(srid=4326, blank=True, null=True)
 
     class Meta:
         managed = False

django/src/rdwatch_scoring/views/model_run.py

@@ -0,0 +1,162 @@
+from ninja import FilterSchema, Query, Schema
+from ninja.errors import ValidationError
+from ninja.pagination import RouterPaginated
+
+from django.contrib.gis.db.models.fields import PolygonField
+from django.contrib.gis.db.models.functions import AsGeoJSON
+from django.db.models import Aggregate, CharField, Count, F, JSONField, Max, Min, Value
+from django.db.models.functions import Cast, Concat, JSONObject, NullIf, Upper
+from django.http import HttpRequest
+
+from rdwatch.db.functions import AggregateArraySubquery, ExtractEpoch
+from rdwatch.schemas.common import TimeRangeSchema
+from rdwatch.views.performer import PerformerSchema
+from rdwatch.views.region import RegionSchema
+from rdwatch_scoring.models import EvaluationRun
+
+router = RouterPaginated()
+
+
+class TimeRangeJSON(NullIf):
+    """Represents the min/max time of a field as JSON"""
+
+    def __init__(self, min, max, performer, site_originator):
+        json = JSONObject(
+            min=ExtractEpoch(Min(min, filter=F(performer) == F(site_originator))),
+            max=ExtractEpoch(Max(max, filter=F(performer) == F(site_originator))),
+        )
+        null = Value({'min': None, 'max': None}, output_field=JSONField())
+        return super().__init__(json, null)
+
+
+class ModelRunFilterSchema(FilterSchema):
+    performer: str | None
+    region: str | None
+
+
+class BoundingBoxPolygon(Aggregate):
+    """Gets the WGS-84 bounding box of a geometry stored in Web Mercator coordinates"""
+
+    template = 'ST_Extent(%(expressions)s)'
+    arity = 1
+    output_field = PolygonField()  # type: ignore
+
+
+class BoundingBoxGeoJSON(Cast):
+    """Gets the GeoJSON bounding box of a geometry in Web Mercator coordinates"""
+
+    def __init__(self, field):
+        bbox = BoundingBoxPolygon(field)
+        json_str = AsGeoJSON(bbox)
+        return super().__init__(json_str, JSONField())
+
+
+class HyperParametersDetailSchema(Schema):
+    id: str
+    title: str
+    region: RegionSchema | None = None
+    performer: PerformerSchema
+    # parameters: dict
+    numsites: int
+    # downloading: int | None = None
+    score: float | None = None
+    timestamp: int | None = None
+    timerange: TimeRangeSchema | None = None
+    bbox: dict | None
+    ground_truth: bool
+    # created: datetime
+    expiration_time: str | None = None
+    evaluation: int | None = None
+    evaluation_run: int | None = None
+
+
+class EvaluationListSchema(Schema):
+    id: int
+    siteNumber: int
+    region: RegionSchema | None
+
+
+class HyperParametersListSchema(Schema):
+    count: int
+    timerange: TimeRangeSchema | None = None
+    bbox: dict | None = None
+    results: list[HyperParametersDetailSchema]
+
+
+def get_queryset():
+    return (
+        EvaluationRun.objects.select_related('site', 'observation')
+        .order_by(
+            'start_datetime',
+        )
+        .annotate(
+            json=JSONObject(
+                id='pk',
+                title=Concat(
+                    F('performer'),
+                    Value('_'),
+                    F('region'),
+                    Value('_'),
+                    F('evaluation_number'),
+                    Value('_'),
+                    F('evaluation_run_number'),
+                    output_field=CharField(),
+                ),
+                performer=JSONObject(
+                    id=Value(-1), team_name='performer', short_code=Upper('performer')
+                ),
+                region=JSONObject(id=Value(-1), name='region'),
+                score=None,
+                numsites=Count(
+                    'site__site_id',
+                    filter=F('performer') == F('site__originator'),
+                    distinct=True,
+                ),
+                evaluation='evaluation_number',
+                evaluation_run='evaluation_run_number',
+                timerange=TimeRangeJSON(
+                    'site__start_date', 'site__end_date', 'performer', 'site_originator'
+                ),
+                timestamp=ExtractEpoch('start_datetime'),
+                ground_truth=False,
+                # timerange=TimeRangeJSON('evaluations__observations__timestamp'),
+                bbox=BoundingBoxGeoJSON('site__union_geometry'),
+            )
+        )
+    )
+
+
+@router.get('/', response={200: HyperParametersListSchema})
+def list_model_runs(
+    request: HttpRequest,
+    filters: ModelRunFilterSchema = Query(...),  # noqa: B008
+    limit: int = 25,
+    page: int = 1,
+):
+    queryset = get_queryset()
+    queryset = filters.filter(queryset=queryset)
+
+    if page < 1 or (not limit and page != 1):
+        raise ValidationError(f"Invalid page '{page}'")
+
+    # Calculate total number of model runs prior to paginating queryset
+    total_model_run_count = queryset.count()
+
+    subquery = queryset[(page - 1) * limit : page * limit] if limit else queryset
+    aggregate = queryset.defer('json').aggregate(
+        timerange=TimeRangeJSON(
+            'site__start_date', 'site__end_date', 'performer', 'site_originator'
+        ),
+        results=AggregateArraySubquery(subquery.values('json')),
+    )
+
+    aggregate['count'] = total_model_run_count
+
+    if filters.region is not None:
+        aggregate |= queryset.defer('json').aggregate(
+            # Use the region polygon for the bbox if it exists.
+            # Otherwise, fall back on the site polygon.
+            bbox=BoundingBoxGeoJSON('site__union_geometry'),
+        )
+
+    return 200, aggregate

django/src/rdwatch_scoring/views/performer.py

@@ -0,0 +1,31 @@
+from ninja import Schema
+from ninja.pagination import RouterPaginated
+
+from django.http import HttpRequest
+
+from rdwatch_scoring.models import EvaluationRun
+
+
+class PerformerSchema(Schema):
+    id: int
+    team_name: str
+    short_code: str
+
+
+router = RouterPaginated()
+
+
+@router.get('/', response=list[PerformerSchema])
+def list_performers(request: HttpRequest):
+    unique_performers = (
+        EvaluationRun.objects.order_by().values_list('performer', flat=True).distinct()
+    )
+    performers_list = [
+        {
+            'id': idx + 1,
+            'team_name': performer,
+            'short_code': performer.lower().replace(' ', '_'),
+        }
+        for idx, performer in enumerate(unique_performers)
+    ]
+    return performers_list

django/src/rdwatch_scoring/views/region.py

@@ -0,0 +1,27 @@
+from ninja import Schema
+from ninja.pagination import RouterPaginated
+
+from django.http import HttpRequest
+
+from rdwatch_scoring.models import EvaluationRun
+
+
+class RegionSchema(Schema):
+    id: int
+    name: str
+
+
+router = RouterPaginated()
+
+
+@router.get('/', response=list[RegionSchema])
+def list_regions(request: HttpRequest):
+    unique_regions = (
+        EvaluationRun.objects.order_by().values_list('region', flat=True).distinct()
+    )
+
+    region_list = [
+        {'id': idx + 1, 'name': region} for idx, region in enumerate(unique_regions)
+    ]
+
+    return region_list

django/src/rdwatch_scoring/views.py → django/src/rdwatch_scoring/views/scores.py

@@ -11,7 +11,7 @@
 from rdwatch.models import HyperParameters, Region
 from rdwatch.views.region import RegionSchema
 
-from .models import (
+from ..models import (
     EvaluationActivityClassificationTemporalIou,
     EvaluationBroadAreaSearchDetection,
     EvaluationRun,