some cleanups

crawler/1_dump_listings.py

@@ -5,7 +5,6 @@ from typing import Any
 from rec.query import detail_query, listing_query, QueryParameters
 from rec.districts import get_districts
 from repositories import ListingRepository
-from sqlalchemy import Engine
 from tqdm.asyncio import tqdm
 from data_access import Listing
 from models import Listing as modelListing
@@ -61,13 +60,13 @@ async def dump_listings(
             for listing in listings_without_details
             if listing.identifier not in all_listing_ids
         ],
-        desc="Fetching details",
+        desc="Fetching details (only missing)",
     )
     for listing, detail in zip(listings_without_details, listing_details):
         listing._details_object = detail
 
-    model_listings = await repository.upsert_listings_legacy(listings)  # upsert in db
     await dump_listings_to_fs(listings)
+    model_listings = await repository.upsert_listings_legacy(listings)  # upsert in db
 
     return model_listings
 

crawler/3_dump_images.py

@@ -5,7 +5,6 @@ import aiohttp
 from repositories import ListingRepository
 from tqdm.asyncio import tqdm
 
-# from data_access import Listing
 from models import Listing
 
 # Setting this too high either crashes rightmove or gets us blocked

crawler/data_access.py

@@ -263,9 +263,13 @@ class Listing:
             if self.path_detail_json().exists():
                 with open(self.path_detail_json()) as f:
                     self._details_object = json.load(f)
+                    return self._details_object  # type: ignore
             else:
-                return {}
-        return self._details_object  # type: ignore
+                raise ValueError(
+                    f"Detail object for listing {self.identifier} not found."
+                )
+        else:
+            return self._details_object
 
     @property
     def price(self) -> float:

crawler/main.py

@@ -44,9 +44,9 @@ def listing_filter_options(func):
     )
     @click.option(
         "--max-bedrooms",
-        default=5,
+        default=10,
         help="Maximum number of bedrooms",
-        type=click.IntRange(min=1),
+        type=click.IntRange(min=1, max=10),  # Right move gets unhappy with >10
     )
     @click.option(
         "--min-price",
@@ -56,9 +56,9 @@ def listing_filter_options(func):
     )
     @click.option(
         "--max-price",
-        default=1000000,
+        default=999_999,
         help="Maximum price",
-        type=click.IntRange(min=0),
+        type=click.IntRange(min=0, max=40_000),  # 40k for renting
     )
     @click.option(
         "--district",
@@ -359,7 +359,7 @@ def populate_db(
     listings = Listing.get_all_listings(
         [path for path in pathlib.Path(data_dir).glob("*/listing.json")]
     )
-    asyncio.run(repository.upsert_listings(listings))
+    asyncio.run(repository.upsert_listings_legacy(listings))
 
 
 if __name__ == "__main__":

crawler/models/listing.py

@@ -239,7 +239,7 @@ class QueryParameters:
     district_names: set[str] = dataclasses.field(default_factory=set)
     radius: float = 0
     page_size: int = 500  # items per page
-    max_days_since_added: int = 30
+    max_days_since_added: int = 14  # for buy listings
     furnish_types: list[FurnishType] | None = None
     # The values below are not supported by rightmove
     # hence we apply them after fetching

crawler/rec/query.py

@@ -84,7 +84,10 @@ async def listing_query(
             7,
             14,
         ]:
-            raise Exception("Invalid max days. Can only be", [1, 3, 7, 14])
+            raise Exception(
+                f"Invalid max days - {max_days_since_added} Can only be got",
+                [1, 3, 7, 14],
+            )
         params["maxDaysSinceAdded"] = str(max_days_since_added)
 
         if mustNewHome:

crawler/runall.sh

@@ -2,9 +2,9 @@
 
 set -euxo pipefail
 
-DATA_DIR="data/rs/test"
+DATA_DIR="data/rs"
 
-LISTING_FILTER_OPTIONS="--min-price 2000 --max-price 4000 --min-bedrooms 2 --max-bedrooms 4 -t rent --available-from $(date +%Y-%m-%d) --last-seen-days 7 --furnish-types furnished"
+LISTING_FILTER_OPTIONS="--min-price 2000 --max-price 4000 --min-bedrooms 2 -t rent --available-from $(date +%Y-%m-%d) --last-seen-days 7 --furnish-types furnished"
 #LISTING_FILTER_OPTIONS="--min-price 2000 --max-price 2500 --min-bedrooms 2 --max-bedrooms 4 -t rent --available-from $(date +%Y-%m-%d) --last-seen-days 7 --furnish-types furnished --district Islington" # DEBUG: UNCOMMENT ME WHEN TESTING
 
 

crawler/ui_exporter.py

@@ -1,8 +1,6 @@
-import dataclasses
 import json
 import pathlib
 
-from data_access import Listing
 from rec.query import QueryParameters
 from repositories.listing_repository import ListingRepository