Update Moray scraper (#1789)

* add Moray Council

* fix Moray scraper

* fix docs

* run update_docu_links

* undo bin_id

* Update moray_gov_uk.py

* reformatting

---------

Co-authored-by: Dewet Diener <[email protected]>
Co-authored-by: 5ila5 <[email protected]>

custom_components/waste_collection_schedule/waste_collection_schedule/source/moray_gov_uk.py

@@ -1,32 +1,29 @@
-import logging
 from datetime import datetime
 
 import requests
 from bs4 import BeautifulSoup
 from waste_collection_schedule import Collection  # type: ignore[attr-defined]
 
-_LOGGER = logging.getLogger(__name__)
-
 TITLE = "Moray Council"
 DESCRIPTION = "Source for Moray Council, UK."
 URL = "https://moray.gov.uk"
 TEST_CASES = {
-    "test_str": {"id": "00013734"},
-    "test_int": {"id": 60216},
+    "Test_001": {"id": "00013734"},
+    "Test_002": {"id": 60216},
 }
 TEXT_MAP = {
-    "images/green_bin.png": "Refuse (Green)",
-    "images/brown_bin.png": "Garden and Kitchen Waste (Brown)",
-    "images/purple_bin.png": "Cans and Plastic (Purple)",
-    "images/blue_bin.png": "Paper and Card (Blue)",
-    "images/orange_box_glass_bag.png": "Glass (Orange)",
+    "G": "Green Refuse Bin",
+    "B": "Brown Garden and Kitchen Waste Bin",
+    "P": "Purple Cans and Plastic Bin",
+    "C": "Blue Paper and Card Bin",
+    "O": "Glass Container",
 }
 ICON_MAP = {
-    "images/green_bin.png": "mdi:trash-can",
-    "images/brown_bin.png": "mdi:compost",
-    "images/purple_bin.png": "mdi:recycle",
-    "images/blue_bin.png": "mdi:newspaper-variant-multiple",
-    "images/orange_box_glass_bag.png": "mdi:bottle-wine",
+    "G": "mdi:trash-can",
+    "B": "mdi:recycle",
+    "P": "mdi:house",
+    "C": "mdi:bulb",
+    "O": "mdi:glass",
 }
 
 
@@ -35,28 +32,38 @@ def __init__(self, id):
         self._id = str(id).zfill(8)
 
     def fetch(self):
-        response = requests.Session().get(
-            f"https://bindayfinder.moray.gov.uk/cal_2024_view.php?id={self._id}"
+        year = datetime.today().year
+        response = requests.get(
+            f"https://bindayfinder.moray.gov.uk/cal_{year}_view.php",
+            params={"id": self._id},
         )
-        response.raise_for_status()
+        if response.status_code != 200:
+            # fall back to known good calendar URL
+            response = requests.get(
+                "https://bindayfinder.moray.gov.uk/cal_2024_view.php",
+                params={"id": self._id},
+            )
         soup = BeautifulSoup(response.text, "html.parser")
 
         entries = []
 
-        for month in soup.findAll("div", class_="cal_month_box"):
-            parsed_date = None
-            for div in month.findAll("div"):
-                if "disp_day_area" in div["class"]:
-                    parsed_date = datetime.strptime(div.text, "%a %d %B %Y").date()
-                elif "disp_bins_cont" in div["class"]:
-                    for i in div.findAll("img"):
+        for month_container in soup.findAll("div", class_="month-container"):
+            for div in month_container.findAll("div"):
+                if "month-header" in div["class"]:
+                    month = div.text
+                elif div["class"] and div["class"][0] in ["B", "GPOC", "GBPOC"]:
+                    bins = div["class"][0]
+                    dom = int(div.text)
+                    parsed_date = datetime.strptime(
+                        f"{dom} {month} {year}", "%d %B %Y"
+                    ).date()
+                    for i in bins:
                         entries.append(
                             Collection(
                                 date=parsed_date,
-                                t=TEXT_MAP.get(i["src"]),
-                                icon=ICON_MAP.get(i["src"]),
+                                t=TEXT_MAP.get(i),
+                                icon=ICON_MAP.get(i),
                             )
                         )
-        if not entries:
-            _LOGGER.warning(f"No collection days found at {response.url}")
+
         return entries

doc/source/moray_gov_uk.md

@@ -9,7 +9,7 @@ waste_collection_schedule:
   sources:
   - name: moray_gov_uk
     args:
-      id: "01234567"
+      id: UNIQUE_PROPERTY_ID
 ```
 
 ### Configuration Variables
@@ -18,8 +18,8 @@ waste_collection_schedule:
 *(string) (required)*
 
 
-#### How to find your unique property `id`
-Your `id` is the 8-digit number at the end of the url when looking up your collection schedule on the [Moray Council Bin Day Finder](https://bindayfinder.moray.gov.uk/) web site.
+#### How to find your `id`
+Your `id` is the collection of numbers at the end of the url when looking up your collection schedule on the [Moray Council Bin Day Finder](https://bindayfinder.moray.gov.uk/) web site.
 
 For example:  _https://bindayfinder.moray.gov.uk/disp_bins.php?id=`00027199`_