From 601b0cabf6b396c64c01e511f2a9e6e733892f3c Mon Sep 17 00:00:00 2001
From: Struan Donald <struan@exo.org.uk>
Date: Mon, 16 Sep 2024 15:51:23 +0200
Subject: [PATCH] Great Big Green Week 2024 import

Add a generic processing command to turn the spreadsheet of events into
one with constituencies. Add a new command for the 2024 events.

Fixes #593
---
 hub/management/commands/base_generators.py    |  6 ++-
 hub/management/commands/base_importers.py     |  5 ++-
 hub/management/commands/generate_gbgw_data.py | 28 ++++++++++++
 .../commands/import_gbgw_events_24.py         | 43 +++++++++++++++++++
 4 files changed, 79 insertions(+), 3 deletions(-)
 create mode 100644 hub/management/commands/generate_gbgw_data.py
 create mode 100644 hub/management/commands/import_gbgw_events_24.py

diff --git a/hub/management/commands/base_generators.py b/hub/management/commands/base_generators.py
index 427e193cf..e2f031d03 100644
--- a/hub/management/commands/base_generators.py
+++ b/hub/management/commands/base_generators.py
@@ -24,7 +24,7 @@
     "MTD": "STC",
     "NMD": "DIS",
     "DIS": "DIS",
-    "WMC": "WMC",
+    "WMC": "WMC23",
     "WMCF": "WMC23",
 }
 
@@ -166,12 +166,16 @@ def add_arguments(self, parser):
             "-i", "--ignore", action="store_true", help="Ignore existing data file"
         )
 
+    def _setup(self):
+        pass
+
     def handle(self, quiet=False, ignore=False, *args, **options):
         self._quiet = quiet
 
         if not self._quiet:
             self.stdout.write(self.message)
         self._ignore = ignore
+        self._setup(*args, **options)
         df = self.get_dataframe()
         out_df = self.process_data(df)
         self.save_data(out_df)
diff --git a/hub/management/commands/base_importers.py b/hub/management/commands/base_importers.py
index 1107afc7f..e0c426100 100644
--- a/hub/management/commands/base_importers.py
+++ b/hub/management/commands/base_importers.py
@@ -131,6 +131,7 @@ def add_data_sets(self, df=None):
 
     def _fill_empty_entries(self):
         for data_type in self.data_types.values():
+            datum_example = AreaData.objects.filter(data_type=data_type).first()
             if (
                 data_type.data_type
                 in [
@@ -140,8 +141,8 @@ def _fill_empty_entries(self):
                 ]
                 and data_type.data_set.table == "areadata"
                 and data_type.data_set.fill_blanks
+                and datum_example is not None
             ):
-                datum_example = AreaData.objects.filter(data_type=data_type).first()
                 if datum_example.float:
                     key = "float"
                 elif datum_example.int:
@@ -428,7 +429,7 @@ def process_data(self, df):
     def handle(self, quiet=False, *args, **options):
         self._quiet = quiet
         df = self.get_dataframe()
-        if not df:
+        if df.empty:
             if not self._quiet:
                 self.stdout.write(f"missing data for {self.message} ({self.area_type})")
             return
diff --git a/hub/management/commands/generate_gbgw_data.py b/hub/management/commands/generate_gbgw_data.py
new file mode 100644
index 000000000..937b10824
--- /dev/null
+++ b/hub/management/commands/generate_gbgw_data.py
@@ -0,0 +1,28 @@
+from django.conf import settings
+
+from .base_generators import BaseLatLonGeneratorCommand
+
+
+class Command(BaseLatLonGeneratorCommand):
+    help = "Generate CSV file of GBGW events with area name'"
+    message = "Generating a CSV of areas for GBGW events"
+
+    row_name = "Organisation"
+
+    def add_arguments(self, parser):
+        super().add_arguments(parser)
+
+        parser.add_argument(
+            "--year",
+            required=True,
+            action="store",
+            help="The last two digits of the year for the events, e.g 24 for 2024",
+        )
+
+    def _setup(self, *args, **kwargs):
+        year = kwargs["year"]
+        self.data_file = settings.BASE_DIR / "data" / f"gbgw_events_{year}.csv"
+        self.out_file = settings.BASE_DIR / "data" / f"gbgw_events_{year}_processed.csv"
+
+    def get_location_from_row(self, row):
+        return {"lat_lon": [row.Lat, row.Long]}
diff --git a/hub/management/commands/import_gbgw_events_24.py b/hub/management/commands/import_gbgw_events_24.py
new file mode 100644
index 000000000..268edbc74
--- /dev/null
+++ b/hub/management/commands/import_gbgw_events_24.py
@@ -0,0 +1,43 @@
+from django.conf import settings
+
+from hub.models import DataSet
+
+from .base_importers import BaseConstituencyCountImportCommand, MultipleAreaTypesMixin
+
+
+class Command(MultipleAreaTypesMixin, BaseConstituencyCountImportCommand):
+    help = "Import data about number of GBGW events in 2024 per constituency"
+    message = "Importing 2024 GBGW events"
+    uses_gss = False
+
+    data_file = settings.BASE_DIR / "data" / "gbgw_events_24_processed.csv"
+
+    area_types = ["WMC23", "STC", "DIS"]
+    cons_col_map = {
+        "WMC23": "WMC23",
+        "STC": "STC",
+        "DIS": "DIS",
+    }
+
+    data_sets = {
+        "constituency_gbgw_2024_event_count": {
+            "defaults": {
+                "label": "Number of Great Big Green Week 2024 events",
+                "description": "Number of public Great Big Green week 2024 events held in the area. This exludes any private events that were held.",
+                "release_date": "September 2024",
+                "data_type": "integer",
+                "category": "movement",
+                "subcategory": "events",
+                "source_label": "Data from The Climate Coalition.",
+                "source": "https://greatbiggreenweek.com/",
+                "source_type": "google sheet",
+                "data_url": "",
+                "table": "areadata",
+                "is_public": True,
+                "default_value": 10,
+                "comparators": DataSet.numerical_comparators(),
+                "unit_type": "raw",
+                "unit_distribution": "people_in_area",
+            }
+        }
+    }