DAFT-13: Add a fallback for fetching the ID (TheJokersThief#14)

Author: TheJokersThief

daft_scraper/listing.py

@@ -105,7 +105,7 @@ def post_load(self, data, **kwargs):
         data['url'] = self.get_url(data['seoFriendlyPath'])
         return data
 
-    id = fields.Int()
+    _id = fields.Int()
     title = fields.Str()
 
     seoTitle = fields.Str()
@@ -135,6 +135,8 @@ def post_load(self, data, **kwargs):
 
 class Listing(dict):
     _ad_page_info = None
+    _id = None
+    url = None
 
     def __init__(self, data: dict):
         self.__dict__ = data
@@ -147,6 +149,13 @@ def ad_page_info(self):
             self._ad_page_info = json.loads(script_text.string)
         return self._ad_page_info
 
+    @property
+    def id(self):
+        if not self._id:
+            # If we didn't get the ID on the first pass, query the listing page
+            self._id = self.ad_page_info['props']['pageProps'].get('listing', {}).get('id', None)
+        return self._id
+
     @property
     def description(self) -> str:
         return self.ad_page_info['props']['pageProps'].get('listing', {}).get('description', None)

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "daft-scraper"
-version = "1.2.4"
+version = "1.2.5"
 description = "A webscraper for Daft.ie"
 authors = ["Evan Smith <[email protected]>"]
 license = "MIT"