Publish curated VacatureRadar source snapshot

tests/unit/test_canonicalize.py

@@ -0,0 +1,13 @@
+from apps.sources.services.canonicalize import canonicalize_url, domain_matches
+
+
+def test_canonicalize_url_removes_tracking_and_fragment():
+    value = canonicalize_url(
+        "HTTPS://Jobs.Example.org:443/jobs/../jobs/42/?utm_source=mail&b=2&a=1#apply"
+    )
+    assert value == "https://jobs.example.org/jobs/42/?a=1&b=2"
+
+
+def test_domain_matches_subdomains_but_not_suffix_attack():
+    assert domain_matches("www.linkedin.com", "linkedin.com")
+    assert not domain_matches("linkedin.com.evil.example", "linkedin.com")