working at least

tests/test_tag_hunter.py

@@ -0,0 +1,122 @@
+import unittest
+from pathlib import Path
+
+from tag_hunter import FetchResult, classify_fragment, classify_telegram
+
+
+class FragmentClassifierTests(unittest.TestCase):
+    def test_prefers_not_for_sale_over_nav_auction_markers(self) -> None:
+        body = """
+        <html>
+          <body>
+            <div class="tm-header-tabs">On auction For sale Sold</div>
+            <tr class="tm-row-selectable js-auction-unavail" data-username="@dsfhsjdfkfhshfkshfksj">
+              <td><div class="table-cell-desc">Not for sale</div></td>
+            </tr>
+            <div class="popup-text">The Telegram username is currently not for sale.</div>
+          </body>
+        </html>
+        """.encode("utf-8")
+        result = FetchResult(
+            url="https://fragment.com/username/dsfhsjdfkfhshfkshfksj",
+            status=200,
+            headers={},
+            body=body,
+            final_url="https://fragment.com/username/dsfhsjdfkfhshfkshfksj",
+        )
+
+        state, reason = classify_fragment("dsfhsjdfkfhshfkshfksj", result)
+
+        self.assertEqual(state, "free")
+        self.assertIn("not for sale", reason)
+
+    def test_taken_when_fragment_row_contains_sale_marker(self) -> None:
+        body = """
+        <html>
+          <body>
+            <tr data-username="@cocaine">
+              <td><div class="table-cell-desc">On auction</div></td>
+            </tr>
+          </body>
+        </html>
+        """.encode("utf-8")
+        result = FetchResult(
+            url="https://fragment.com/username/cocaine",
+            status=200,
+            headers={},
+            body=body,
+            final_url="https://fragment.com/username/cocaine",
+        )
+
+        state, reason = classify_fragment("cocaine", result)
+
+        self.assertEqual(state, "taken")
+        self.assertEqual(reason, "on auction")
+
+
+class TelegramClassifierTests(unittest.TestCase):
+    def test_existing_profile_is_taken_from_view_markers(self) -> None:
+        body = Path(__file__).resolve().parents[1].joinpath("exist.txt").read_text(encoding="utf-8")
+        result = FetchResult(
+            url="https://t.me/dicksmack",
+            status=200,
+            headers={},
+            body=body.encode("utf-8"),
+            final_url="https://t.me/dicksmack",
+        )
+
+        state, reason = classify_telegram("dicksmack", result)
+
+        self.assertEqual(state, "taken")
+        self.assertTrue(
+            any(marker in reason for marker in ("view @dicksmack", "view and join @dicksmack", "preview channel"))
+        )
+
+    def test_nonexisting_profile_is_free_from_contact_markers(self) -> None:
+        body = Path(__file__).resolve().parents[1].joinpath("nonexist.txt").read_text(encoding="utf-8")
+        result = FetchResult(
+            url="https://t.me/sahdfjkhdfjsahdkf",
+            status=200,
+            headers={},
+            body=body.encode("utf-8"),
+            final_url="https://t.me/sahdfjkhdfjsahdkf",
+        )
+
+        state, reason = classify_telegram("sahdfjkhdfjsahdkf", result)
+
+        self.assertEqual(state, "free")
+        self.assertTrue(
+            any(marker in reason for marker in ("contact @sahdfjkhdfjsahdkf", "right away"))
+        )
+
+    def test_redirect_without_explicit_body_is_unknown(self) -> None:
+        result = FetchResult(
+            url="https://t.me/doesnotexist123",
+            status=200,
+            headers={"Location": "tg://resolve?domain=doesnotexist123"},
+            body=b"<html><head><title>Telegram</title></head><body></body></html>",
+            final_url="https://t.me/doesnotexist123",
+        )
+
+        state, reason = classify_telegram("doesnotexist123", result)
+
+        self.assertEqual(state, "unknown")
+        self.assertIn("ambiguous", reason)
+
+    def test_explicit_unavailable_marker_is_free(self) -> None:
+        result = FetchResult(
+            url="https://t.me/doesnotexist123",
+            status=200,
+            headers={},
+            body=b"<html><body>Sorry, this page isn't available.</body></html>",
+            final_url="https://t.me/doesnotexist123",
+        )
+
+        state, reason = classify_telegram("doesnotexist123", result)
+
+        self.assertEqual(state, "free")
+        self.assertIn("sorry, this page isn't available", reason)
+
+
+if __name__ == "__main__":
+    unittest.main()