✨ #993 ADD celery task 1: every day check that extracts ever… (#1018)

src/sdg/core/tests/test_management_commands.py

@@ -1,6 +1,7 @@
 import os
 from datetime import date, datetime
 from io import StringIO
+from unittest.mock import MagicMock, patch
 
 from django.conf import settings
 from django.core import mail
@@ -23,6 +24,9 @@
 from sdg.core.tests.factories.logius import ThemaFactory, UniformeProductnaamFactory
 from sdg.organisaties.models import LokaleOverheid
 from sdg.organisaties.tests.factories.overheid import BevoegdeOrganisatieFactory
+from sdg.producten.management.commands.check_broken_links import (
+    Command as CheckBrokenLinksCommand,
+)
 from sdg.producten.tests.factories.product import (
     GeneriekProductFactory,
     ReferentieProductFactory,
@@ -503,3 +507,54 @@ def test_user_flag_with_eligable_user(self):
 
         self.assertIn("No eligable users found.", out)
         self.assertEqual(len(mail.outbox), 0)
+
+
+class TestCommandCheckBrokenLinks(CommandTestCase):
+    def setUp(self):
+        super().setUp()
+        self.command = CheckBrokenLinksCommand
+
+    def test_handle_response_code_successful(self):
+        out = self.call_command("check_broken_links")
+        self.assertIn("Deleted 0 old BrokenLinks.", out)
+
+    def test_request_head_redirect_handling(self):
+        def test_case(url, equals_status_code):
+            response = self.command().request_head(url=url)
+            self.assertEqual(response.status_code, equals_status_code)
+
+        # Test case 1 - Successful request to Example
+        test_case("https://google.com/", 200)
+        # Test case 2 - Successful request to Google
+        test_case("https://example.com/", 200)
+        # Test case 3 - Multiple redirects, final status 200
+        test_case("https://digid.nl/inloggen", 200)
+        # Test case 4 - 301 Redirect
+        test_case("https://httpbin.org/status/301", 200)
+        # Test case 5 - 302 Redirect
+        test_case("https://httpbin.org/status/302", 200)
+        # Test case 6 - 303 Redirect
+        test_case("https://httpbin.org/status/303", 200)
+        # Test case 7 - 308 Redirect
+        test_case(
+            "https://www.zoetermeer.nl/verhuizen-naar-het-buitenland-emigreren/", 200
+        )
+        # Test case 8 - 403 Client error
+        test_case("https://httpbin.org/status/403", 403)
+        # Test case 9 - 404 Not Found
+        test_case("https://google.com/404", 404)
+        # Test case 10 - URL without protocol (should raise an error or handle gracefully)
+        test_case("www.google.com", 200)
+        # Test case 11 - 500 Server error
+        test_case("https://httpbin.org/status/500", 500)
+
+    def test_clean_up_removed_urls(self):
+        removed_links_count = 0
+        out = self.call_command("check_broken_links")
+        self.assertIn(f"Deleted {removed_links_count} old BrokenLinks.", str(out))
+
+    def test_reset_all_broken_links(self):
+        out = self.call_command("check_broken_links", "--reset")
+        self.assertIn(
+            "Successfully cleared the error_count of every BrokenLink.", str(out)
+        )

src/sdg/fixtures/django_celery_beat.json

@@ -79,6 +79,18 @@
         "timezone": "UTC"
     }
 },
+{
+    "model": "django_celery_beat.crontabschedule",
+    "pk": 7,
+    "fields": {
+        "minute": "0",
+        "hour": "9",
+        "day_of_week": "*",
+        "day_of_month": "1",
+        "month_of_year": "*",
+        "timezone": "UTC"
+    }
+},
 {
     "model": "django_celery_beat.periodictasks",
     "pk": 1,
@@ -281,5 +293,61 @@
         "date_changed": "2023-04-06T12:22:36.304",
         "description": "Send notification emails about recent reference product updates."
     }
+},
+{
+    "model": "django_celery_beat.periodictask",
+    "pk": 8,
+    "fields": {
+        "name": "Check Broken Links",
+        "task": "sdg.producten.tasks.check_broken_links",
+        "interval": null,
+        "crontab": 1,
+        "solar": null,
+        "clocked": null,
+        "args": "[]",
+        "kwargs": "{}",
+        "queue": null,
+        "exchange": null,
+        "routing_key": null,
+        "headers": "{}",
+        "priority": null,
+        "expires": null,
+        "expire_seconds": null,
+        "one_off": false,
+        "start_time": null,
+        "enabled": false,
+        "last_run_at": null,
+        "total_run_count": 0,
+        "date_changed": "2024-11-04T12:00:00.000",
+        "description": "Send notification emails about broken links detected in product fields."
+    }
+},
+{
+    "model": "django_celery_beat.periodictask",
+    "pk": 9,
+    "fields": {
+        "name": "Send monthly broken links report",
+        "task": "sdg.producten.tasks.send_monthly_broken_links_report",
+        "interval": null,
+        "crontab": 7,
+        "solar": null,
+        "clocked": null,
+        "args": "[]",
+        "kwargs": "{}",
+        "queue": null,
+        "exchange": null,
+        "routing_key": null,
+        "headers": "{}",
+        "priority": null,
+        "expires": null,
+        "expire_seconds": null,
+        "one_off": false,
+        "start_time": null,
+        "enabled": false,
+        "last_run_at": null,
+        "total_run_count": 0,
+        "date_changed": "2024-11-04T14:00:00.000",
+        "description": "Send monthly broken links report to the all redactors of the content."
+    }
 }
 ]