Merge branch 'feat/import-postcode-areas'

Author: joaquimds

hub/data_imports/geocoding_config.py

@@ -483,7 +483,12 @@ async def get_postcode_data_for_area(area, loaders, steps):
     # Try a few other backup strategies (example postcode, another geocoder)
     # to get postcodes.io data
     if postcode_data is None:
-        postcode = await get_example_postcode_from_area_gss(area.gss)
+        try:
+            postcode = await get_example_postcode_from_area_gss(area.gss)
+        except Exception as e:
+            logger.error(f"Failed to get example postcode for {area.gss}: {e}")
+            postcode = None
+
         steps.append(
             {
                 "task": "postcode_from_area",

hub/graphql/types/model_types.py

@@ -590,8 +590,11 @@ def generic_data_for_hub(self, hostname: str) -> List["GenericData"]:
     async def sample_postcode(
         self, info: Info[HubDataLoaderContext]
     ) -> Optional[PostcodesIOResult]:
-        return await get_postcode_data_for_gss(self.gss)
-        # return await info.context.area_coordinate_loader.load(self.point)
+        try:
+            return await get_postcode_data_for_gss(self.gss)
+        except Exception as e:
+            logger.error(f"Failed to get sample postcode for gss {self.gss}: {e}")
+            return None
 
 
 @strawberry.type
@@ -1690,12 +1693,17 @@ def choropleth_data_for_source(
         )
 
     # Get the required data for the source
+    gss_field = (
+        "postcode_data__postcode"
+        if analytical_area_key == AnalyticalAreaType.postcode
+        else f"postcode_data__codes__{analytical_area_key.value}"
+    )
     qs = (
         external_data_source.get_import_data()
         .filter(postcode_data__codes__isnull=False)
         .annotate(
             label=F(f"postcode_data__{analytical_area_key.value}"),
-            gss=F(f"postcode_data__codes__{analytical_area_key.value}"),
+            gss=F(gss_field),
             latitude=Cast("postcode_data__latitude", output_field=FloatField()),
             longitude=Cast("postcode_data__longitude", output_field=FloatField()),
         )

hub/graphql/types/stats.py

@@ -64,6 +64,10 @@ def fk_filter(self, field_name="") -> dict[str, str]:
     AnalyticalAreaType.msoa: AreaTypeFilter(lih_area_type="MSOA"),
     AnalyticalAreaType.lsoa: AreaTypeFilter(lih_area_type="LSOA"),
     AnalyticalAreaType.output_area: AreaTypeFilter(lih_area_type="OA21"),
+    AnalyticalAreaType.postcode: AreaTypeFilter(lih_area_type="PC"),
+    AnalyticalAreaType.postcode_area: AreaTypeFilter(lih_area_type="PCA"),
+    AnalyticalAreaType.postcode_district: AreaTypeFilter(lih_area_type="PCD"),
+    AnalyticalAreaType.postcode_sector: AreaTypeFilter(lih_area_type="PCS"),
 }
 
 

hub/management/commands/export_postcode_areas_as_geojsonl.py

@@ -0,0 +1,31 @@
+import logging
+from pathlib import Path
+
+from django.conf import settings
+
+# from django postgis
+from django.core.management.base import BaseCommand
+
+from tqdm import tqdm
+
+from hub.models import Area
+
+logger = logging.getLogger(__name__)
+
+
+class Command(BaseCommand):
+    help = "Export Postcodes to GeoJSONl (for Mapbox tileset source creation)"
+
+    def handle(self, *args, **options):
+        area_type_codes = ["PCS", "PCD", "PCA"]
+        for area_type_code in area_type_codes:
+            output_file: Path = (
+                settings.BASE_DIR / "data" / f"{area_type_code}.geojsonl"
+            )
+            area_geojsons = Area.objects.filter(
+                area_type__code=area_type_code
+            ).values_list("geometry", flat=True)
+            with output_file.open("w") as f:
+                for geojson in tqdm(area_geojsons):
+                    f.write(geojson)
+                    f.write("\n")

hub/management/commands/import_msoas_and_lsoas.py

@@ -107,7 +107,7 @@ def import_area(self, area, area_type, property_prefix):
             if isinstance(polygon, Polygon):
                 polygon = MultiPolygon([polygon])
 
-            geom["geometry"] = polygon.json
+            geom["geometry"] = json.loads(polygon.json)
 
             a.geometry = json.dumps(geom)
             a.polygon = polygon

hub/management/commands/import_output_areas.py

@@ -88,7 +88,7 @@ def import_area(self, area, area_type):
             # Transform the geometry
             polygon.transform(transform)
 
-            geom["geometry"] = polygon.json
+            geom["geometry"] = json.loads(polygon.json)
 
             a.geometry = json.dumps(geom)
             a.polygon = polygon

hub/management/commands/import_output_areas_scotland.py

@@ -78,7 +78,7 @@ def import_area(self, area, area_type):
             if isinstance(polygon, Polygon):
                 polygon = MultiPolygon([polygon])
 
-            geom["geometry"] = polygon.json
+            geom["geometry"] = json.loads(polygon.json)
             a.geometry = json.dumps(geom)
             a.polygon = polygon
             a.point = a.polygon.centroid

hub/management/commands/import_postcode_areas.py

@@ -0,0 +1,207 @@
+import json
+import logging
+import re
+from pathlib import Path
+
+from django.conf import settings
+from django.contrib.gis.db.models import Union as GisUnion
+
+# from django postgis
+from django.contrib.gis.geos import GEOSGeometry, MultiPolygon, Polygon
+from django.core.management.base import BaseCommand
+from django.db.models import F, Q
+from django.db.models.expressions import Expression, RawSQL
+from django.db.models.functions import Length, Substr
+
+from tqdm import tqdm
+
+from hub.models import Area, AreaType
+
+logger = logging.getLogger(__name__)
+
+
+class Command(BaseCommand):
+    help = "Import Postcodes from GeoJSON"
+
+    def add_arguments(self, parser):
+        parser.add_argument(
+            "-o",
+            "--only",
+            help="""
+            Pass one of {PC, PCS, PCD, PCA} to only import that level of area.
+            Note that postcode units (PC) must be imported first.
+            """,
+        )
+
+    def handle(self, only=None, *args, **options):
+        if not only or only == "PC":
+            # E.G. N14 7LU
+            self.import_postcode_units()
+        if not only or only == "PCS":
+            # E.G. N14 7
+            self.import_postcode_sectors()
+        if not only or only == "PCD":
+            # E.G. N14
+            self.import_postcode_districts()
+        if not only or only == "PCA":
+            # E.G. N
+            self.import_postcode_areas()
+
+    def import_postcode_units(self):
+        print("Importing postcode units")
+        filepaths: list[Path] = [
+            settings.BASE_DIR / "data" / f"postcodes_{i}.geojsonl" for i in range(1, 11)
+        ]
+        for filepath in filepaths:
+            if not filepath.exists():
+                print(
+                    f'Missing {filepath.name}. Download from the Mapped MinIO console, "postcodes" bucket.'
+                )
+                return
+
+            print(f"Importing postcode file {filepath.name} of 10")
+
+            data = filepath.read_text()
+            area_type, created = AreaType.objects.get_or_create(
+                name="Postcodes",
+                code="PC",
+                area_type="Postcode",
+                description="Postcodes",
+            )
+
+            for line in tqdm(re.split(r"\r?\n", data)):
+                if line.strip():
+                    area = json.loads(line)
+                    self.import_area(area, area_type)
+
+    def import_area(self, area, area_type):
+        geom = None
+        gss = area["properties"]["POSTCODE"]
+        name = gss
+
+        geom_already_loaded = Area.objects.filter(
+            gss=gss, polygon__isnull=False
+        ).exists()
+        if geom_already_loaded:
+            # Only fetch geometry data if required, to speed things up
+            # logger.debug(f"skipping geometry for {area['name']}")
+            pass
+        else:
+            geom = {
+                "type": "Feature",
+                "geometry": area["geometry"],
+                "properties": {
+                    **area["properties"],
+                    "code": gss,
+                    "name": name,
+                    "type": area_type.code,
+                },
+            }
+
+        a, created = Area.objects.update_or_create(
+            gss=gss,
+            area_type=area_type,
+            defaults={"name": name},
+        )
+
+        if geom is not None:
+            geos = json.dumps(geom["geometry"])
+            polygon = GEOSGeometry(geos)
+            if isinstance(polygon, Polygon):
+                polygon = MultiPolygon([polygon])
+
+            geom["geometry"] = json.loads(polygon.json)
+
+            a.geometry = json.dumps(geom)
+            a.polygon = polygon
+            a.point = a.polygon.centroid
+            a.save()
+
+    def import_postcode_sectors(self):
+        print("Importing postcode sectors")
+        area_type, created = AreaType.objects.get_or_create(
+            name="Postcode Sectors",
+            code="PCS",
+            area_type="Postcode Sector",
+            description="Postcode sectors",
+        )
+        self.import_postcode_grouping(
+            area_type, prefix_expression=Substr(F("gss"), 1, Length(F("gss")) - 2)
+        )
+
+    def import_postcode_districts(self):
+        print("Importing postcode districts")
+        area_type, created = AreaType.objects.get_or_create(
+            name="Postcode Districts",
+            code="PCD",
+            area_type="Postcode District",
+            description="Postcode districts",
+        )
+        self.import_postcode_grouping(
+            area_type, prefix_expression=Substr(F("gss"), 1, Length(F("gss")) - 4)
+        )
+
+    def import_postcode_areas(self):
+        print("Importing postcode areas")
+        area_type, created = AreaType.objects.get_or_create(
+            name="Postcode Areas",
+            code="PCA",
+            area_type="Postcode Area",
+            description="Postcode areas",
+        )
+        self.import_postcode_grouping(
+            area_type,
+            prefix_expression=RawSQL("SUBSTRING(gss FROM '[a-zA-Z]+')", tuple()),
+        )
+
+    def import_postcode_grouping(
+        self, area_type: AreaType, prefix_expression: Expression
+    ):
+        """
+        Import postcodes grouped by prefix, calculated by the provided expression, into
+        the provided area type.
+        """
+        # Only consider postcodes that contain a space
+        # The others are "vertical streets" which are not relevant
+        prefixes = (
+            Area.objects.filter(Q(gss__contains=" ") & Q(area_type__code="PC"))
+            .annotate(prefix=prefix_expression)
+            .values_list("prefix", flat=True)
+            .distinct()
+        )
+
+        for prefix in tqdm(prefixes):
+            geom_already_loaded = Area.objects.filter(
+                gss=prefix, polygon__isnull=False
+            ).exists()
+            if geom_already_loaded:
+                continue
+
+            postcodes = Area.objects.annotate(prefix=prefix_expression).filter(
+                area_type__code="PC", prefix=prefix
+            )
+            polygon = postcodes.aggregate(union=GisUnion("polygon"))["union"]
+
+            if isinstance(polygon, Polygon):
+                polygon = MultiPolygon([polygon])
+
+            geom = {
+                "type": "Feature",
+                "geometry": json.loads(polygon.json),
+                "properties": {
+                    "code": prefix,
+                    "name": prefix,
+                    "type": area_type.code,
+                },
+            }
+
+            Area.objects.update_or_create(
+                gss=prefix,
+                area_type=area_type,
+                defaults={
+                    "name": prefix,
+                    "geometry": json.dumps(geom),
+                    "polygon": polygon,
+                    "point": polygon.centroid,
+                },
+            )

nextjs/src/__generated__/graphql.ts

@@ -54,6 +54,7 @@ import { useRouter } from 'next/navigation'
 import pluralize from 'pluralize'
 import { useEffect, useState } from 'react'
 import { toast } from 'sonner'
+import toSpaceCase from 'to-space-case'
 import { CREATE_MAP_REPORT } from '../../../reports/ReportList/CreateReportCard'
 import ExternalDataSourceBadCredentials from './ExternalDataSourceBadCredentials'
 import { ManageSourceSharing } from './ManageSourceSharing'
@@ -304,7 +305,7 @@ export default function InspectExternalDataSource({
             {dataType === DataSourceType.Member
               ? 'Membership list'
               : dataType
-                ? pluralize(dataType.toLowerCase())
+                ? pluralize(toSpaceCase(dataType.toLowerCase()))
                 : 'Data source'}
             <span>&nbsp;&#x2022;&nbsp;</span>
             {crmInfo?.name || crmType}