feat: validate Anwarpy publication metadata

automation/anwarpy/test_page_state.py

@@ -0,0 +1,50 @@
+import unittest
+
+from page_state import MarkerError, find_marker
+
+
+VALID = "anwarpy: status=pending | access=free | category=نماذج | queued_at=2026-10-09T15:12:00+03:00"
+
+
+class PageStateTests(unittest.TestCase):
+    def test_finds_one_exact_marker_after_unrelated_blocks(self):
+        result = find_marker([
+            {"id": "a", "text": "ملاحظات"},
+            {"id": "b", "text": VALID},
+            {"id": "c", "text": "مصدر عام"},
+        ])
+        self.assertEqual(result.block_id, "b")
+        self.assertEqual(result.meta["status"], "pending")
+
+    def test_returns_none_when_anwarpy_is_never_mentioned(self):
+        self.assertIsNone(find_marker([{"id": "a", "text": "لا توجد علامة"}]))
+
+    def test_reports_every_near_miss(self):
+        for text in (
+            VALID.replace("anwarpy:", "Anwarpy:"),
+            "prefix " + VALID,
+            "documentation mentions anwarpy but has no marker",
+            VALID.replace("anwarpy:", "anwarpy ="),
+        ):
+            with self.subTest(text=text):
+                with self.assertRaisesRegex(MarkerError, "near-miss"):
+                    find_marker([{"id": "bad", "text": text}])
+
+    def test_scans_all_blocks_and_rejects_ambiguity(self):
+        with self.assertRaisesRegex(MarkerError, "multiple"):
+            find_marker([
+                {"id": "first", "text": VALID},
+                {"id": "middle", "text": "unrelated"},
+                {"id": "last", "text": VALID.replace("category=نماذج", "category=أدوات")},
+            ])
+
+    def test_exact_plus_near_miss_is_ambiguous(self):
+        with self.assertRaisesRegex(MarkerError, "ambiguous"):
+            find_marker([
+                {"id": "good", "text": VALID},
+                {"id": "bad", "text": "Anwarpy: pending"},
+            ])
+
+
+if __name__ == "__main__":
+    unittest.main()