From df24c2c1b796558fef3770f8af6e3d463ef3fc76 Mon Sep 17 00:00:00 2001
From: Viktor Barzin <vbarzin@gmail.com>
Date: Sat, 17 May 2025 21:22:39 +0000
Subject: [PATCH] add cli param for querying properties to rent

example:
python main.py --data-dir data/rs2 dump-listings --max-price 3500 --min-bedrooms 2 --max-bedrooms 4 --district islington -t rent
---
 crawler/1_dump_listings.py | 12 +++++++-----
 crawler/main.py            | 13 +++++++++++++
 crawler/rec/query.py       | 37 ++++++++++++++++++++++++-------------
 3 files changed, 44 insertions(+), 18 deletions(-)

diff --git a/crawler/1_dump_listings.py b/crawler/1_dump_listings.py
index ea6162f..60b6a1b 100644
--- a/crawler/1_dump_listings.py
+++ b/crawler/1_dump_listings.py
@@ -1,12 +1,13 @@
 from dataclasses import dataclass
 import pathlib
-from rec.query import listing_query
+from rec.query import ListingType, listing_query
 from rec.districts import get_districts
 from data_access import Listing
 
 
 @dataclass(frozen=True)
 class QueryParameters:
+    listing_type: ListingType
     min_bedrooms: int
     max_bedrooms: int
     min_price: int
@@ -32,8 +33,9 @@ def dump_listings(
         print("#### District:", district)
         for i in [1, 2]:
             try:
-                d = listing_query(
+                response_json = listing_query(
                     page=i,
+                    channel=parameters.listing_type,
                     min_bedrooms=parameters.min_bedrooms,
                     max_bedrooms=parameters.max_bedrooms,
                     radius=parameters.radius,
@@ -47,12 +49,12 @@ def dump_listings(
                 print(e)
                 break
             if i == 1:
-                print("totalAvailableResults: ", d["totalAvailableResults"])
-            if len(d["properties"]) == 0:
+                print("totalAvailableResults: ", response_json["totalAvailableResults"])
+            if len(response_json["properties"]) == 0:
                 break
             print(f"page {i}", end=", ", flush=True)
 
-            for property in d["properties"]:
+            for property in response_json["properties"]:
                 identifier = property["identifier"]
 
                 listing = Listing(identifier, data_dir=data_dir)
diff --git a/crawler/main.py b/crawler/main.py
index 66e5687..be5c93c 100644
--- a/crawler/main.py
+++ b/crawler/main.py
@@ -5,6 +5,7 @@ import importlib
 from rec.districts import get_districts
 from data_access import Listing
 import csv_exporter
+from rec.query import ListingType
 
 dump_listings_module = importlib.import_module('1_dump_listings')
 dump_detail_module = importlib.import_module('2_dump_detail')
@@ -41,6 +42,16 @@ def cli(ctx, data_dir: str):
 
 
 @cli.command()
+@click.option(
+    '--type',
+    '-t',
+    help='Type of listing to scrape',
+    type=click.Choice(
+        ListingType.__members__.keys(),
+        case_sensitive=False,
+    ),
+    required=True,
+)
 @click.option(
     '--min-bedrooms',
     default=1,
@@ -80,9 +91,11 @@ def dump_listings(
     max_bedrooms: int,
     min_price: int,
     max_price: int,
+    type: str,
 ):
     data_dir: str = ctx.obj['data_dir']
     query_parameters = dump_listings_module.QueryParameters(
+        listing_type=ListingType[type],
         district_names=set(district),
         min_bedrooms=min_bedrooms,
         max_bedrooms=max_bedrooms,
diff --git a/crawler/rec/query.py b/crawler/rec/query.py
index 801cd6d..304e9b9 100644
--- a/crawler/rec/query.py
+++ b/crawler/rec/query.py
@@ -6,7 +6,11 @@ import urllib3
 
 urllib3.disable_warnings()
 
-# cache = Cache(r"_cache")
+
+class ListingType(enum.StrEnum):
+    BUY = "BUY"
+    RENT = "RENT"
+
 
 headers = {
     "Host": "api.rightmove.co.uk",
@@ -42,9 +46,9 @@ def detail_query(detail_id: int):
     return response.json()
 
 
-# @cache.memoize()
 def listing_query(
     page: int,
+    channel: ListingType,
     min_bedrooms: int,
     max_bedrooms: int,
     radius: float,
@@ -58,29 +62,38 @@ def listing_query(
 ) -> dict:
     params = {
         "locationIdentifier": location_id,
-        "channel": "BUY",
+        "channel": channel.upper(),
         "page": str(page),
         "numberOfPropertiesPerPage": str(page_size),
         "radius": str(radius),
         "sortBy": "distance",
         "includeUnavailableProperties": "false",
-        "dontShow": "sharedOwnership,retirement",
         "minPrice": str(min_price),
         "maxPrice": str(max_price),
         "minBedrooms": str(min_bedrooms),
         "maxBedrooms": str(max_bedrooms),
         "apiApplication": "ANDROID",
-        "appVersion": "3.70.0",
+        "appVersion": "4.28.0",
     }
-    if len(property_type) > 0:
-        params["propertyTypes"] = ",".join(property_type)
-    if max_days_since_added:
-        if max_days_since_added not in [1, 3, 7, 14]:
+    if channel is ListingType.BUY:
+        params["dontShow"] = "sharedOwnership,retirement",
+        if len(property_type) > 0:
+            params["propertyTypes"] = ",".join(property_type)
+        if max_days_since_added is not None and max_days_since_added not in [
+            1, 3, 7, 14
+        ]:
             raise Exception("Invalid max days. Can only be", [1, 3, 7, 14])
         params["maxDaysSinceAdded"] = max_days_since_added
 
-    if mustNewHome:
-        params["mustHave"] = "newHome"
+        if mustNewHome:
+            params["mustHave"] = "newHome"
+
+    headers = {
+        "Host": "api.rightmove.co.uk",
+        "Accept-Encoding": "gzip, deflate, br",
+        "User-Agent": "okhttp/4.12.0",
+        "Connection": "keep-alive"
+    }
     response = requests.get(
         "https://api.rightmove.co.uk/api/property-listing",
         params=params,
@@ -91,5 +104,3 @@ def listing_query(
         raise Exception("Failed due to: ", response.text)
 
     return response.json()
-
-