fix(importer): sanitize PostAttachment.ext to skip mangled gallery-dl URL-encoded basenames (varchar(32) overrun) — Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

tests/test_importer_attachments.py

@@ -59,3 +59,25 @@ def test_json_sidecar_is_not_attached(importer, import_layout):
         select(func.count()).select_from(PostAttachment)
     ).scalar_one()
     assert n == 0
+
+
+def test_mangled_filename_extension_is_sanitized(importer, import_layout):
+    """gallery-dl sometimes URL-encodes a query string into the basename
+    (`...https___www.patreon.com_media-u_Z0F...`). Python's Path.suffix
+    returns 50+ chars of base64-ish junk for those, which blows the
+    PostAttachment.ext varchar(32) column. Operator-flagged 2026-05-25.
+    The importer should record an empty ext rather than crash."""
+    import_root, _ = import_layout
+    f = (
+        import_root / "Alice"
+        / "79507046_media_https___www.patreon.com_media-u_Z0FBQUFBQm5q"
+    )
+    f.parent.mkdir(parents=True, exist_ok=True)
+    f.write_bytes(b"binary blob")
+    r = importer.import_one(f)
+    assert r.status == "attached"
+    att = importer.session.execute(select(PostAttachment)).scalar_one()
+    # Junk "extension" -> stored as empty string (not the 50-char garbage).
+    assert att.ext == ""
+    # original_filename is Text-typed so the full name survives intact.
+    assert att.original_filename.endswith("_Z0FBQUFBQm5q")