uk_bin_collection 0.157.0__py3-none-any.whl → 0.158.0__py3-none-any.whl

uk_bin_collection/uk_bin_collection/councils/LondonBoroughSutton.py

@@ -1,30 +1,24 @@
-from time import sleep
-
 import requests
 from bs4 import BeautifulSoup
+from datetime import datetime
+import re
+from time import sleep
 
 from uk_bin_collection.uk_bin_collection.common import *
 from uk_bin_collection.uk_bin_collection.get_bin_data import AbstractGetBinDataClass
 
+def remove_ordinal_indicator_from_date_string(date_str):
+    return re.sub(r'(\d+)(st|nd|rd|th)', r'\1', date_str)
 
-# import the wonderful Beautiful Soup and the URL grabber
 class CouncilClass(AbstractGetBinDataClass):
-    """
-    Concrete classes have to implement all abstract operations of the
-    base class. They can also override some operations with a default
-    implementation.
-    """
 
     def parse_data(self, page: str, **kwargs) -> dict:
-
         user_uprn = kwargs.get("uprn")
-        # check_uprn(user_uprn)
         bindata = {"bins": []}
 
         URI = f"https://waste-services.sutton.gov.uk/waste/{user_uprn}"
 
         s = requests.Session()
-
         r = s.get(URI)
         while "Loading your bin days..." in r.text:
             sleep(2)
@@ -32,48 +26,65 @@ class CouncilClass(AbstractGetBinDataClass):
         r.raise_for_status()
 
         soup = BeautifulSoup(r.content, "html.parser")
-
         current_year = datetime.now().year
         next_year = current_year + 1
 
-        services = soup.find_all("h3", class_="govuk-heading-m waste-service-name")
-
+        # Find all h3 headers (bin types)
+        services = soup.find_all("h3")
         for service in services:
-            bin_type = service.get_text(
-                strip=True
-            )  # Bin type name (e.g., 'Food waste', 'Mixed recycling')
-            if bin_type == "Bulky Waste":
+            bin_type = service.get_text(strip=True)
+            if "Bulky Waste" in bin_type:
                 continue
-            service_details = service.find_next("div", class_="govuk-grid-row")
-
-            next_collection = (
-                (
-                    service_details.find("dt", string="Next collection")
-                    .find_next_sibling("dd")
-                    .get_text(strip=True)
-                )
-                .replace("(this collection has been adjusted from its usual time)", "")
-                .strip()
-            )
-
-            next_collection = datetime.strptime(
-                remove_ordinal_indicator_from_date_string(next_collection),
-                "%A, %d %B",
-            )
-
-            if (datetime.now().month == 12) and (next_collection.month == 1):
-                next_collection = next_collection.replace(year=next_year)
-            else:
-                next_collection = next_collection.replace(year=current_year)
-
-            dict_data = {
-                "type": bin_type,
-                "collectionDate": next_collection.strftime("%d/%m/%Y"),
-            }
-            bindata["bins"].append(dict_data)
-
-        bindata["bins"].sort(
-            key=lambda x: datetime.strptime(x.get("collectionDate"), "%d/%m/%Y")
-        )
 
+            # Find the next element (next sibling) which is likely a paragraph with date info
+            next_sib = service.find_next_sibling()
+            while next_sib and getattr(next_sib, 'name', None) not in [None, 'p']:
+                next_sib = next_sib.find_next_sibling()
+
+            next_coll = None
+            if next_sib:
+                text = next_sib.get_text() if hasattr(next_sib, 'get_text') else str(next_sib)
+                match = re.search(r"Next collection\s*([A-Za-z]+,? \d{1,2}(?:st|nd|rd|th)? [A-Za-z]+)", text)
+                if match:
+                    next_coll = match.group(1)
+                else:
+                    # Sometimes the text may be attached without a space after 'Next collection'
+                    match = re.search(r"Next collection([A-Za-z]+,? \d{1,2}(?:st|nd|rd|th)? [A-Za-z]+)", text)
+                    if match:
+                        next_coll = match.group(1)
+
+            # Try several siblings forward if not found
+            if not next_coll:
+                sib_try = service
+                for _ in range(3):
+                    if sib_try:
+                        sib_try = sib_try.find_next_sibling()
+                    else:
+                        break
+                    if sib_try:
+                        text = sib_try.get_text() if hasattr(sib_try, 'get_text') else str(sib_try)
+                        match = re.search(r"Next collection\s*([A-Za-z]+,? \d{1,2}(?:st|nd|rd|th)? [A-Za-z]+)", text)
+                        if match:
+                            next_coll = match.group(1)
+                            break
+
+            if next_coll:
+                next_coll = remove_ordinal_indicator_from_date_string(next_coll)
+                try:
+                    next_collection = datetime.strptime(next_coll, "%A, %d %B")
+                except ValueError:
+                    continue
+
+                if (datetime.now().month == 12 and next_collection.month == 1):
+                    next_collection = next_collection.replace(year=next_year)
+                else:
+                    next_collection = next_collection.replace(year=current_year)
+
+                dict_data = {
+                    "type": bin_type,
+                    "collectionDate": next_collection.strftime("%d/%m/%Y"),
+                }
+                bindata["bins"].append(dict_data)
+
+        bindata["bins"].sort(key=lambda x: datetime.strptime(x.get("collectionDate"), "%d/%m/%Y"))
         return bindata

{uk_bin_collection-0.157.0.dist-info → uk_bin_collection-0.158.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: uk_bin_collection
-Version: 0.157.0
+Version: 0.158.0
 Summary: Python Lib to collect UK Bin Data
 Author: Robert Bradley
 Author-email: robbrad182@gmail.com

{uk_bin_collection-0.157.0.dist-info → uk_bin_collection-0.158.0.dist-info}/RECORD

@@ -182,7 +182,7 @@ uk_bin_collection/uk_bin_collection/councils/LondonBoroughLambeth.py,sha256=r9D5
 uk_bin_collection/uk_bin_collection/councils/LondonBoroughLewisham.py,sha256=d8rlJDTbY3nj-Zjg6iwvwfe-X13Gq86DGGW6QkQAUW0,5310
 uk_bin_collection/uk_bin_collection/councils/LondonBoroughOfRichmondUponThames.py,sha256=H1ej7Bw1zW3kvxCqz0W1137YIZCdG5AmhAyVK3cvp00,3987
 uk_bin_collection/uk_bin_collection/councils/LondonBoroughRedbridge.py,sha256=A_6Sis5hsF53Th04KeadHRasGbpAm6aoaWJ6X8eC4Y8,6604
-uk_bin_collection/uk_bin_collection/councils/LondonBoroughSutton.py,sha256=T0gCjpBQ9k3uNNIiJGphNiSXeh2n6bohflg9RwbL3NA,2565
+uk_bin_collection/uk_bin_collection/councils/LondonBoroughSutton.py,sha256=xixtPqXJ6Wms14e-Sc5Nq8ybbcrvZ65mzgpOhNJw5tY,3646
 uk_bin_collection/uk_bin_collection/councils/LutonBoroughCouncil.py,sha256=vScUi_R8FnBddii2_zLlZBLxuh85mKmCm8nKW3zxky0,2758
 uk_bin_collection/uk_bin_collection/councils/MaidstoneBoroughCouncil.py,sha256=T9R1_ciKitxGhGUcNu2UnSZszvUBuLVKReRA1rIlRIg,4280
 uk_bin_collection/uk_bin_collection/councils/MaldonDistrictCouncil.py,sha256=PMVt2XFggttPmbWyrBrHJ-W6R_6-0ux1BkY1kj1IKzg,1997
@@ -347,8 +347,8 @@ uk_bin_collection/uk_bin_collection/councils/YorkCouncil.py,sha256=I2kBYMlsD4bId
 uk_bin_collection/uk_bin_collection/councils/council_class_template/councilclasstemplate.py,sha256=QD4v4xpsEE0QheR_fGaNOIRMc2FatcUfKkkhAhseyVU,1159
 uk_bin_collection/uk_bin_collection/create_new_council.py,sha256=m-IhmWmeWQlFsTZC4OxuFvtw5ZtB8EAJHxJTH4O59lQ,1536
 uk_bin_collection/uk_bin_collection/get_bin_data.py,sha256=Qb76X46V0UMZJwO8zMNPvnVY7jNa-bmTlrirDi1tuJA,4553
-uk_bin_collection-0.157.0.dist-info/LICENSE,sha256=vABBUOzcrgfaTKpzeo-si9YVEun6juDkndqA8RKdKGs,1071
-uk_bin_collection-0.157.0.dist-info/METADATA,sha256=5i6McWuamdnRpr0N-CwsBKsR3RmNONT-NkirowzNiJ4,26688
-uk_bin_collection-0.157.0.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
-uk_bin_collection-0.157.0.dist-info/entry_points.txt,sha256=36WCSGMWSc916S3Hi1ZkazzDKHaJ6CD-4fCEFm5MIao,90
-uk_bin_collection-0.157.0.dist-info/RECORD,,
+uk_bin_collection-0.158.0.dist-info/LICENSE,sha256=vABBUOzcrgfaTKpzeo-si9YVEun6juDkndqA8RKdKGs,1071
+uk_bin_collection-0.158.0.dist-info/METADATA,sha256=ihuNsB2pzq5yFtT06ae1WmUvgPCkLc8gJHt-zAHNl0Q,26688
+uk_bin_collection-0.158.0.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
+uk_bin_collection-0.158.0.dist-info/entry_points.txt,sha256=36WCSGMWSc916S3Hi1ZkazzDKHaJ6CD-4fCEFm5MIao,90
+uk_bin_collection-0.158.0.dist-info/RECORD,,