Improve sitemaps

twitch/tests/test_backup.py

@@ -5,6 +5,7 @@ import math
 import os
 import shutil
 from compression import zstd
+from pathlib import Path
 from typing import TYPE_CHECKING
 
 import pytest
@@ -21,6 +22,7 @@ from twitch.models import Game
 from twitch.models import Organization
 
 if TYPE_CHECKING:
+    import sqlite3
     from pathlib import Path
 
     from django.test import Client
@@ -42,12 +44,12 @@ class TestBackupCommand:
         # Create test data so tables exist
         Organization.objects.create(twitch_id="test000", name="Test Org")
 
-        output_dir = tmp_path / "backups"
+        output_dir: Path = tmp_path / "backups"
         output_dir.mkdir()
 
         call_command("backup_db", output_dir=str(output_dir), prefix="test")
 
-        backup_files = list(output_dir.glob("test-*.sql.zst"))
+        backup_files: list[Path] = list(output_dir.glob("test-*.sql.zst"))
         assert len(backup_files) == 1
         assert backup_files[0].exists()
         assert backup_files[0].stat().st_size > 0
@@ -55,17 +57,17 @@ class TestBackupCommand:
     def test_backup_contains_sql_content(self, tmp_path: Path) -> None:
         """Test that backup file contains valid SQL content."""
         _skip_if_pg_dump_missing()
-        output_dir = tmp_path / "backups"
+        output_dir: Path = tmp_path / "backups"
         output_dir.mkdir()
 
         # Create some test data
-        org = Organization.objects.create(twitch_id="test123", name="Test Org")
-        game = Game.objects.create(twitch_id="game456", display_name="Test Game")
+        org: Organization = Organization.objects.create(twitch_id="test123", name="Test Org")
+        game: Game = Game.objects.create(twitch_id="game456", display_name="Test Game")
         game.owners.add(org)
 
         call_command("backup_db", output_dir=str(output_dir), prefix="test")
 
-        backup_file = next(iter(output_dir.glob("test-*.sql.zst")))
+        backup_file: Path = next(iter(output_dir.glob("test-*.sql.zst")))
 
         # Decompress and read content
         with (
@@ -73,7 +75,7 @@ class TestBackupCommand:
             zstd.open(raw_handle, "r") as compressed,
             io.TextIOWrapper(compressed, encoding="utf-8") as handle,
         ):
-            content = handle.read()
+            content: str = handle.read()
 
         if connection.vendor == "postgresql":
             assert "CREATE TABLE" in content
@@ -92,19 +94,19 @@ class TestBackupCommand:
         # Create test data so tables exist
         Organization.objects.create(twitch_id="test001", name="Test Org")
 
-        output_dir = tmp_path / "backups"
+        output_dir: Path = tmp_path / "backups"
         output_dir.mkdir()
 
         call_command("backup_db", output_dir=str(output_dir), prefix="test")
 
-        backup_file = next(iter(output_dir.glob("test-*.sql.zst")))
+        backup_file: Path = next(iter(output_dir.glob("test-*.sql.zst")))
 
         with (
             backup_file.open("rb") as raw_handle,
             zstd.open(raw_handle, "r") as compressed,
             io.TextIOWrapper(compressed, encoding="utf-8") as handle,
         ):
-            content = handle.read()
+            content: str = handle.read()
 
         # Should NOT contain django admin, silk, or debug toolbar tables
         assert "django_session" not in content
@@ -121,12 +123,12 @@ class TestBackupCommand:
         # Create test data so tables exist
         Organization.objects.create(twitch_id="test002", name="Test Org")
 
-        output_dir = tmp_path / "backups"
+        output_dir: Path = tmp_path / "backups"
         output_dir.mkdir()
 
         call_command("backup_db", output_dir=str(output_dir), prefix="custom")
 
-        backup_files = list(output_dir.glob("custom-*.sql.zst"))
+        backup_files: list[Path] = list(output_dir.glob("custom-*.sql.zst"))
         assert len(backup_files) == 1
 
     def test_backup_creates_output_directory(self, tmp_path: Path) -> None:
@@ -135,7 +137,7 @@ class TestBackupCommand:
         # Create test data so tables exist
         Organization.objects.create(twitch_id="test003", name="Test Org")
 
-        output_dir = tmp_path / "nonexistent" / "backups"
+        output_dir: Path = tmp_path / "nonexistent" / "backups"
 
         call_command("backup_db", output_dir=str(output_dir), prefix="test")
 
@@ -149,12 +151,12 @@ class TestBackupCommand:
         Organization.objects.create(twitch_id="test004", name="Test Org")
 
         monkeypatch.setattr(settings, "DATA_DIR", tmp_path)
-        datasets_dir = tmp_path / "datasets"
+        datasets_dir: Path = tmp_path / "datasets"
         datasets_dir.mkdir(exist_ok=True, parents=True)
 
         call_command("backup_db")
 
-        backup_files = list(datasets_dir.glob("ttvdrops-*.sql.zst"))
+        backup_files: list[Path] = list(datasets_dir.glob("ttvdrops-*.sql.zst"))
         assert len(backup_files) >= 1
 
 
@@ -165,7 +167,7 @@ class TestBackupHelperFunctions:
     def test_get_allowed_tables_filters_by_prefix(self) -> None:
         """Test that _get_allowed_tables returns only matching tables."""
         # Use Django's connection to access the test database
-        tables = _get_allowed_tables("twitch_")
+        tables: list[str] = _get_allowed_tables("twitch_")
 
         assert len(tables) > 0
         assert all(table.startswith("twitch_") for table in tables)
@@ -175,7 +177,7 @@ class TestBackupHelperFunctions:
     def test_get_allowed_tables_excludes_non_matching(self) -> None:
         """Test that _get_allowed_tables excludes non-matching tables."""
         # Use Django's connection to access the test database
-        tables = _get_allowed_tables("twitch_")
+        tables: list[str] = _get_allowed_tables("twitch_")
 
         # Should not include django, silk, or debug toolbar tables
         assert not any(table.startswith("django_") for table in tables)
@@ -212,25 +214,25 @@ class TestBackupHelperFunctions:
         # Create test data
         Organization.objects.create(twitch_id="test789", name="Write Test Org")
 
-        tables = _get_allowed_tables("twitch_")
+        tables: list[str] = _get_allowed_tables("twitch_")
 
         if connection.vendor == "postgresql":
             if not shutil.which("pg_dump"):
                 pytest.skip("pg_dump is not available")
-            output_path = tmp_path / "backup.sql.zst"
+            output_path: Path = tmp_path / "backup.sql.zst"
             _write_postgres_dump(output_path, tables)
             with (
                 output_path.open("rb") as raw_handle,
                 zstd.open(raw_handle, "r") as compressed,
                 io.TextIOWrapper(compressed, encoding="utf-8") as handle,
             ):
-                content = handle.read()
+                content: str = handle.read()
             assert "CREATE TABLE" in content
             assert "INSERT INTO" in content
             assert "twitch_organization" in content
             assert "Write Test Org" in content
         else:
-            db_connection = connection.connection
+            db_connection: sqlite3.Connection = connection.connection
             output = io.StringIO()
             _write_sqlite_dump(output, db_connection, tables)
             content = output.getvalue()
@@ -255,7 +257,7 @@ class TestDatasetBackupViews:
         Returns:
             Path to the created datasets directory.
         """
-        datasets_dir = tmp_path / "datasets"
+        datasets_dir: Path = tmp_path / "datasets"
         datasets_dir.mkdir()
         return datasets_dir
 
@@ -266,7 +268,7 @@ class TestDatasetBackupViews:
         Returns:
             Path to the created backup file.
         """
-        backup_file = datasets_dir / "ttvdrops-20260210-120000.sql.zst"
+        backup_file: Path = datasets_dir / "ttvdrops-20260210-120000.sql.zst"
         with (
             backup_file.open("wb") as raw_handle,
             zstd.open(raw_handle, "w") as compressed,
@@ -315,8 +317,8 @@ class TestDatasetBackupViews:
         monkeypatch.setattr(settings, "DATA_DIR", datasets_dir.parent)
 
         # Create multiple backup files with different timestamps
-        older_backup = datasets_dir / "ttvdrops-20260210-100000.sql.zst"
-        newer_backup = datasets_dir / "ttvdrops-20260210-140000.sql.zst"
+        older_backup: Path = datasets_dir / "ttvdrops-20260210-100000.sql.zst"
+        newer_backup: Path = datasets_dir / "ttvdrops-20260210-140000.sql.zst"
 
         for backup in [older_backup, newer_backup]:
             with (
@@ -334,9 +336,9 @@ class TestDatasetBackupViews:
 
         response: _MonkeyPatchedWSGIResponse = client.get(reverse("twitch:dataset_backups"))
 
-        content = response.content.decode()
-        newer_pos = content.find("20260210-140000")
-        older_pos = content.find("20260210-100000")
+        content: str = response.content.decode()
+        newer_pos: int = content.find("20260210-140000")
+        older_pos: int = content.find("20260210-100000")
 
         # Newer backup should appear first (sorted descending)
         assert 0 < newer_pos < older_pos
@@ -370,7 +372,9 @@ class TestDatasetBackupViews:
         monkeypatch.setattr(settings, "DATA_DIR", datasets_dir.parent)
 
         # Attempt path traversal
-        response = client.get(reverse("twitch:dataset_backup_download", args=["../../../etc/passwd"]))
+        response: _MonkeyPatchedWSGIResponse = client.get(
+            reverse("twitch:dataset_backup_download", args=["../../../etc/passwd"]),
+        )
         assert response.status_code == 404
 
     def test_dataset_download_rejects_invalid_extensions(
@@ -383,10 +387,12 @@ class TestDatasetBackupViews:
         monkeypatch.setattr(settings, "DATA_DIR", datasets_dir.parent)
 
         # Create a file with invalid extension
-        invalid_file = datasets_dir / "malicious.exe"
-        invalid_file.write_text("not a backup")
+        invalid_file: Path = datasets_dir / "malicious.exe"
+        invalid_file.write_text("not a backup", encoding="utf-8")
 
-        response = client.get(reverse("twitch:dataset_backup_download", args=["malicious.exe"]))
+        response: _MonkeyPatchedWSGIResponse = client.get(
+            reverse("twitch:dataset_backup_download", args=["malicious.exe"]),
+        )
         assert response.status_code == 404
 
     def test_dataset_download_file_not_found(
@@ -398,7 +404,9 @@ class TestDatasetBackupViews:
         """Test download returns 404 for non-existent file."""
         monkeypatch.setattr(settings, "DATA_DIR", datasets_dir.parent)
 
-        response = client.get(reverse("twitch:dataset_backup_download", args=["nonexistent.sql.zst"]))
+        response: _MonkeyPatchedWSGIResponse = client.get(
+            reverse("twitch:dataset_backup_download", args=["nonexistent.sql.zst"]),
+        )
         assert response.status_code == 404
 
     def test_dataset_list_view_shows_file_sizes(
@@ -414,8 +422,9 @@ class TestDatasetBackupViews:
         response: _MonkeyPatchedWSGIResponse = client.get(reverse("twitch:dataset_backups"))
 
         assert response.status_code == 200
+
         # Should contain size information (bytes, KB, MB, or GB)
-        content = response.content.decode()
+        content: str = response.content.decode()
         assert any(unit in content for unit in ["bytes", "KB", "MB", "GB"])
 
     def test_dataset_list_ignores_non_zst_files(
@@ -434,7 +443,7 @@ class TestDatasetBackupViews:
 
         response: _MonkeyPatchedWSGIResponse = client.get(reverse("twitch:dataset_backups"))
 
-        content = response.content.decode()
+        content: str = response.content.decode()
         assert "backup.sql.zst" in content
         assert "readme.txt" not in content
         assert "old_backup.gz" not in content
@@ -449,7 +458,7 @@ class TestDatasetBackupViews:
         monkeypatch.setattr(settings, "DATA_DIR", datasets_dir.parent)
 
         # Create subdirectory with backup
-        subdir = datasets_dir / "2026" / "02"
+        subdir: Path = datasets_dir / "2026" / "02"
         subdir.mkdir(parents=True)
         backup_file = subdir / "backup.sql.zst"
         with (