feat: validate Anwarpy publication metadata

2026-10-09 15:52:08 +03:00
parent 00f8e4aee7
commit 4aeb747d56
8 changed files with 457 additions and 0 deletions
@@ -0,0 +1,79 @@
+import unittest
+
+from publication_meta import (
+    clean,
+    mark_blocked,
+    mark_published,
+    mark_publishing,
+    parse_meta,
+    same_meta,
+)
+
+
+PENDING = "anwarpy: status=pending | access=vip | category=أدوات | queued_at=2026-10-09T15:12:00+03:00"
+
+
+class PublicationMetaTests(unittest.TestCase):
+    def test_parses_pending_and_normalizes_formatting_controls(self):
+        raw = "\ufeffanwarpy:\u00a0status=pending | access=vip | category=أدوات | queued_at=2026-10-09T15:12:00+03:00\u200f"
+        meta = parse_meta(raw)
+        self.assertEqual(meta["status"], "pending")
+        self.assertEqual(meta["category"], "أدوات")
+
+    def test_rejects_invalid_values_and_dates(self):
+        for raw in (
+            PENDING.replace("access=vip", "access=paid"),
+            PENDING.replace("category=أدوات", "category=عام"),
+            PENDING.replace("2026-10-09", "2026-13-45"),
+            PENDING.replace("+03:00", ""),
+        ):
+            with self.subTest(raw=raw):
+                with self.assertRaises(ValueError):
+                    parse_meta(raw)
+
+    def test_rejects_invalid_state_shapes(self):
+        with self.assertRaisesRegex(ValueError, "slug"):
+            parse_meta(PENDING.replace("status=pending", "status=publishing"))
+        with self.assertRaisesRegex(ValueError, "blocked_reason"):
+            parse_meta(PENDING.replace("status=pending", "status=blocked"))
+        with self.assertRaisesRegex(ValueError, "not allowed"):
+            parse_meta(PENDING + " | url=https://anwarpy.ai/posts/example")
+
+    def test_transitions_reparse_raw_input(self):
+        publishing = mark_publishing(PENDING, "arabic-ai-guide")
+        self.assertEqual(parse_meta(publishing)["status"], "publishing")
+        published = mark_published(
+            publishing,
+            "2026-10-09T12:20:11+00:00",
+            "https://anwarpy.ai/posts/arabic-ai-guide",
+        )
+        self.assertEqual(parse_meta(published)["status"], "published")
+
+    def test_blocked_is_terminal_and_carries_a_bounded_reason(self):
+        blocked = mark_blocked(PENDING, "cover")
+        self.assertEqual(parse_meta(blocked)["blocked_reason"], "cover")
+        with self.assertRaises(ValueError):
+            mark_blocked(PENDING, "network")
+
+    def test_canonical_url_and_slug_are_exact(self):
+        publishing = mark_publishing(PENDING, "arabic-ai-guide")
+        for url in (
+            "http://anwarpy.ai/posts/arabic-ai-guide",
+            "https://www.anwarpy.ai/posts/arabic-ai-guide",
+            "https://anwarpy.ai/posts/arabic-ai-guide/",
+            "https://anwarpy.ai/posts/other",
+        ):
+            with self.subTest(url=url):
+                with self.assertRaises(ValueError):
+                    mark_published(publishing, "2026-10-09T15:20:11+03:00", url)
+        with self.assertRaises(ValueError):
+            mark_publishing(PENDING, "عنوان-عربي")
+
+    def test_readback_comparison_is_semantic_after_cleaning(self):
+        formatted = "\u200fanwarpy:  status=pending|access=vip|category=أدوات|queued_at=2026-10-09T15:12:00+03:00"
+        self.assertTrue(same_meta(PENDING, formatted))
+        self.assertEqual(clean("\ufeff أداة\u00a0"), "أداة")
+
+
+if __name__ == "__main__":
+    unittest.main()