Fix HTML entities in GWR advance ticket dates

agenda/gwr.py

@@ -3,6 +3,7 @@
 import os
 import re
 from datetime import date, datetime
+from html import unescape
 from time import time
 
 import httpx
@@ -12,6 +13,7 @@ url = "https://www.gwr.com/your-tickets/choosing-your-ticket/advance-tickets"
 
 def parse_date_string(date_str: str) -> date:
     """Parse date string from HTML."""
+    date_str = " ".join(unescape(date_str).split())
     if not date_str[-1].isdigit():  # If the year is missing, use the current year
         date_str += f" {date.today().year}"
 

tests/test_gwr.py

@@ -2,8 +2,8 @@
 
 import os
 import tempfile
+import typing
 from datetime import date
-from typing import Any
 from unittest.mock import AsyncMock, patch
 
 import pytest
@@ -20,6 +20,16 @@ from agenda.gwr import (
 class TestParseDateString:
     """Tests for parse_date_string function."""
 
+    @pytest.mark.parametrize("space", [" ", " ", " ", "\u00a0"])
+    @pytest.mark.parametrize("include_year", [True, False])
+    def test_parse_html_whitespace(self, space: str, include_year: bool) -> None:
+        """Decode HTML spaces before parsing dates or inferring the year."""
+        date_str = f"Thursday 25{space}March"
+        if include_year:
+            date_str += " 2027"
+        result = parse_date_string(f" {space}{date_str}{space} ")
+        assert result == date(2027 if include_year else date.today().year, 3, 25)
+
     def test_parse_date_with_year(self) -> None:
         """Test parsing date string with year included."""
         result = parse_date_string("Monday 25 December 2023")
@@ -152,7 +162,7 @@ class TestAdvanceTicketsPageHtml:
 
     @pytest.mark.asyncio
     @patch("httpx.AsyncClient")
-    async def test_fetch_from_web(self, mock_client: Any) -> None:
+    async def test_fetch_from_web(self, mock_client: typing.Any) -> None:
         """Test fetching from web when cache is stale."""
         mock_response = AsyncMock()
         mock_response.text = "<html>fresh content</html>"
@@ -176,7 +186,7 @@ class TestAdvanceTicketDate:
 
     @pytest.mark.asyncio
     @patch("agenda.gwr.advance_tickets_page_html")
-    async def test_advance_ticket_date_success(self, mock_html: Any) -> None:
+    async def test_advance_ticket_date_success(self, mock_html: typing.Any) -> None:
         """Test successfully extracting advance ticket date."""
         mock_html.return_value = """
         <tr><td>Weekdays</td><td>Monday 25 December 2023</td></tr>
@@ -188,7 +198,7 @@ class TestAdvanceTicketDate:
 
     @pytest.mark.asyncio
     @patch("agenda.gwr.advance_tickets_page_html")
-    async def test_advance_ticket_date_no_match(self, mock_html: Any) -> None:
+    async def test_advance_ticket_date_no_match(self, mock_html: typing.Any) -> None:
         """Test when no weekday date can be extracted."""
         mock_html.return_value = "<p>No relevant data</p>"
 
@@ -197,7 +207,7 @@ class TestAdvanceTicketDate:
 
     @pytest.mark.asyncio
     @patch("agenda.gwr.advance_tickets_page_html")
-    async def test_advance_ticket_date_force_cache(self, mock_html: Any) -> None:
+    async def test_advance_ticket_date_force_cache(self, mock_html: typing.Any) -> None:
         """Test advance_ticket_date with force_cache parameter."""
         mock_html.return_value = """
         <tr><td>Weekdays</td><td>Tuesday 26 December 2023</td></tr>
@@ -206,3 +216,11 @@ class TestAdvanceTicketDate:
         result = await advance_ticket_date("/fake/dir", force_cache=True)
         assert result == date(2023, 12, 26)
         mock_html.assert_called_once_with("/fake/dir", force_cache=True)
+
+
+@pytest.mark.parametrize("suffix", ["", "**"])
+def test_extract_html_entity_date(suffix: str) -> None:
+    """Both extraction paths accept the date that caused the update crash."""
+    html = f"<tr><td>Weekdays</td><td>Thursday 25&nbsp;March 2027{suffix}</td></tr>"
+    assert extract_dates(html) == {"Weekdays": date(2027, 3, 25)}
+    assert extract_weekday_date(html) == date(2027, 3, 25)