add option to filter for min sqm per listing

crawler/data_access.py

@@ -158,7 +158,7 @@ class Listing:
             json.dump(objs, f)
 
     @property
-    def sqm_model(self, recalculate=True):
+    def sqm_model(self, recalculate=True) -> float:
         if not self.path_floorplan_model_json().exists() or recalculate:
             self.calculate_sqm_model()
 
@@ -193,7 +193,7 @@ class Listing:
         with open(self.path_floorplan_ocr_json(), "w") as f:
             json.dump(objs, f)
 
-    async def sqm_ocr(self, recalculate=False):
+    async def sqm_ocr(self, recalculate=False) -> float | None:
         if not self.path_floorplan_ocr_json().exists() or recalculate:
             await self.calculate_sqm_ocr()
 

crawler/main.py

@@ -88,6 +88,12 @@ def listing_filter_options(func):
         default=14,
         type=int,
     )
+    @click.option(
+        "--min-sqm",
+        help="Minimum square meters for the listing",
+        default=None,
+        type=int,
+    )
     @wraps(func)
     def wrapper(*args, **kwargs):
         return func(*args, **kwargs)
@@ -128,6 +134,7 @@ def dump_listings(
     furnish_types: list[str],
     available_from: datetime | None,
     last_seen_days: int,
+    min_sqm: int | None = None,
 ):
     data_dir: str = ctx.obj["data_dir"]
     query_parameters = QueryParameters(
@@ -140,6 +147,7 @@ def dump_listings(
         furnish_types=[FurnishType[furnish_type] for furnish_type in furnish_types],
         let_date_available_from=available_from,
         last_seen_days=last_seen_days,
+        min_sqm=min_sqm,
     )
     click.echo(
         f"Running dump_listings for districts {district}, data dir {data_dir} and parameters: "
@@ -288,6 +296,7 @@ def export_immoweb(
     furnish_types: list[str],
     available_from: datetime | None,
     last_seen_days: int,
+    min_sqm: int | None = None,
 ):
     query_parameters = QueryParameters(
         listing_type=ListingType[type],
@@ -299,6 +308,7 @@ def export_immoweb(
         furnish_types=[FurnishType[furnish_type] for furnish_type in furnish_types],
         let_date_available_from=available_from,
         last_seen_days=last_seen_days,
+        min_sqm=min_sqm,
     )
     click.echo(
         f"Exporting data to {output_file} that matches the query parameters: {query_parameters}"

crawler/rec/query.py

@@ -36,6 +36,7 @@ class QueryParameters:
     # available from; council tax
     let_date_available_from: datetime | None = None
     last_seen_days: int | None = None
+    min_sqm: int | None = None
 
 
 async def filter_listings(
@@ -68,6 +69,9 @@ async def filter_listings(
             and listing.letDateAvailable < query_parameters.let_date_available_from
         ):
             continue
+        sqm_ocr = await listing.sqm_ocr() or 0
+        if query_parameters.min_sqm is not None and sqm_ocr < query_parameters.min_sqm:
+            continue
         filtered_listings.append(listing)
 
     return filtered_listings