chore: re-org python deps for provider map

.github/resources/process_providers.py

@@ -0,0 +1,45 @@
+import json
+import pandas as pd
+
+providers_file = 'src/metadata/providers/providers.csv'
+
+df = pd.read_csv(providers_file)
+city_lookup = pd.read_csv('src/metadata/cities_to_county.csv')
+city_to_county = dict(zip(city_lookup['City'], city_lookup['County']))
+
+lookup_records = df[df['counties_served'].isna()]['ntd_id']
+
+# Fill in the null values for counties served with the HQ county
+for record in lookup_records:
+  city = df[df['ntd_id'] == record]['hq_city'].values[0]
+
+  try:
+    county = city_to_county[city] or city_to_county[f'City of {city}']
+    df.loc[df['ntd_id'] == record, 'hq_county'] = county
+    df.loc[df['ntd_id'] == record, 'counties_served'] = county
+
+  except KeyError:
+    print("No county found for city: ", city)
+
+df.to_csv(providers_file)
+
+# Do a group by for the counties served
+county_counts = df['counties_served'].str.split(';') \
+  .explode('counties_served') \
+  .value_counts()
+
+geojson_file = 'src/metadata/providers/counties.geojson'
+geojson = json.load(open(geojson_file))
+
+# Add the county counts to the geojson
+for feature in geojson['features']:
+  county_name = feature['properties']['county']
+
+  if county_name in county_counts:
+    feature['properties']['num_providers'] = int(county_counts[county_name])
+  else:
+    feature['properties']['num_providers'] = 0
+
+# Write the geojson back to the file
+with open(geojson_file, 'w') as f:
+  json.dump(geojson, f, indent=2)

.github/workflows/provider-map-jobs.yml

@@ -35,68 +35,20 @@ jobs:
               "SELECT * FROM \`mart_transit_database.dim_mobility_mart_providers\`" \
           > src/metadata/providers/providers.csv
 
-      - name: Fix our CSV file
-        run: |
-          # Workaround because of...
-          #   https://github.com/google-github-actions/setup-gcloud/issues/666
-          sed -i -n -e '/agency_name/,$p' src/metadata/providers/providers.csv
+#      - name: Fix our CSV file
+#        run: |
+#          # Workaround because of...
+#          #   https://github.com/google-github-actions/setup-gcloud/issues/666
+#          sed -i -n -e '/agency_name/,$p' src/metadata/providers/providers.csv
 
       - uses: actions/setup-python@v4
         with:
           python-version: '3.10'
 
-      - name: Install pandas
-        run: |
-          pip install pandas
-
-      - name: Update Counties GeoJSON file
-        shell: python
+      - name: Install dependencies, process providers, and update GeoJSON
         run: |
-          import json
-          import pandas as pd
-
-          providers_file = 'src/metadata/providers/providers.csv'
-
-          df = pd.read_csv(providers_file)
-          city_lookup = pd.read_csv('src/metadata/cities_to_county.csv')
-          city_to_county = dict(zip(city_lookup['City'], city_lookup['County']))
-
-          lookup_records = df[df['counties_served'].isna()]['ntd_id']
-
-          # Fill in the null values for counties served with the HQ county
-          for record in lookup_records:
-            city = df[df['ntd_id'] == record]['hq_city'].values[0]
-
-            try:
-              county = city_to_county[city] or city_to_county[f'City of {city}']
-              df.loc[df['ntd_id'] == record, 'hq_county'] = county
-              df.loc[df['ntd_id'] == record, 'counties_served'] = county
-
-            except KeyError:
-              print("No county found for city: ", city)
-
-          df.to_csv(providers_file)
-
-          # Do a group by for the counties served
-          county_counts = df['counties_served'].str.split(';') \
-            .explode('counties_served') \
-            .value_counts()
-
-          geojson_file = 'src/metadata/providers/counties.geojson'
-          geojson = json.load(open(geojson_file))
-
-          # Add the county counts to the geojson
-          for feature in geojson['features']:
-            county_name = feature['properties']['county']
-
-            if county_name in county_counts:
-              feature['properties']['num_providers'] = int(county_counts[county_name])
-            else:
-              feature['properties']['num_providers'] = 0
-
-          # Write the geojson back to the file
-          with open(geojson_file, 'w') as f:
-            json.dump(geojson, f, indent=2)
+          pip install -r requirements.txt
+          python .github/resources/process_providers.py
 
       - name: Create Pull Request
         uses: peter-evans/create-pull-request@v5

.gitignore

@@ -6,4 +6,5 @@ _site/
 data/providers/*.json
 node_modules/
 vendor/
+venv/
 **/.DS_Store

requirements.txt

@@ -0,0 +1,9 @@
+distlib==0.3.8
+filelock==3.13.1
+numpy==1.26.4
+pandas==2.2.0
+platformdirs==4.2.0
+python-dateutil==2.8.2
+pytz==2024.1
+six==1.16.0
+tzdata==2024.1