add filter for furnished/unfurnished type for rented listings

crawler/main.py

@@ -6,7 +6,7 @@ import importlib
 from rec.districts import get_districts
 from data_access import Listing
 import csv_exporter
-from rec.query import ListingType
+from rec.query import ListingType, FurnishType
 
 dump_listings_module = importlib.import_module('1_dump_listings')
 dump_detail_module = importlib.import_module('2_dump_detail')
@@ -14,14 +14,6 @@ dump_images_module = importlib.import_module('3_dump_images')
 detect_floorplan_module = importlib.import_module('4_detect_floorplan')
 routing_module = importlib.import_module('5_routing')
 
-steps_to_handlers = {
-    'dump_listings': dump_listings_module.dump_listings,
-    'dump_detail': dump_detail_module.dump_detail,
-    'dump_images': dump_images_module.dump_images,
-    'detect_floorplan': detect_floorplan_module.detect_floorplan,
-    'routing': routing_module.calculate_route,
-}
-
 
 @click.group()
 @click.option(
@@ -84,6 +76,19 @@ def cli(ctx, data_dir: str):
     type=click.Choice(get_districts().keys(), case_sensitive=False),
     multiple=True,
 )
+@click.option(
+    '--furnish-types',
+    '-f',
+    help='Furnish types for rented listings',
+    type=click.Choice(
+        [
+            furnish_type.name
+            for furnish_type in FurnishType.__members__.values()
+        ],
+        case_sensitive=False,
+    ),
+    multiple=True,
+)
 @click.pass_context
 def dump_listings(
     ctx: click.core.Context,
@@ -93,6 +98,7 @@ def dump_listings(
     min_price: int,
     max_price: int,
     type: str,
+    furnish_types: list[str],
 ):
     data_dir: str = ctx.obj['data_dir']
     query_parameters = dump_listings_module.QueryParameters(
@@ -102,13 +108,16 @@ def dump_listings(
         max_bedrooms=max_bedrooms,
         min_price=min_price,
         max_price=max_price,
+        furnish_types=[
+            FurnishType[furnish_type] for furnish_type in furnish_types
+        ],
     )
     click.echo(
         f'Running dump_listings for districts {district}, data dir {data_dir} and parameters: '
-        f'{query_parameters}'
-    )
+        f'{query_parameters}')
     data_dir_path = pathlib.Path(data_dir)
-    asyncio.run(dump_listings_module.dump_listings(query_parameters, data_dir_path))
+    asyncio.run(
+        dump_listings_module.dump_listings(query_parameters, data_dir_path))
 
 
 @cli.command()