Test that SqliteCacheRequestsDocumentLoader does not create the cache file on init

Author: anatoly-scherbakov

tests/test_sqlite_cache_requests_document_loader.py

@@ -1,6 +1,7 @@
 """Tests for SqliteCacheRequestsDocumentLoader and HTTP cache behavior."""
 
 import json
+import sqlite3
 import threading
 from http.server import BaseHTTPRequestHandler, HTTPServer
 from pathlib import Path
@@ -23,9 +24,11 @@ class _ContextHandler(BaseHTTPRequestHandler):
 
     def do_GET(self):
         type(self).request_count += 1
-        body = json.dumps({
-            '@context': {'name': 'http://example.org/name'},
-        }).encode()
+        body = json.dumps(
+            {
+                '@context': {'name': 'http://example.org/name'},
+            }
+        ).encode()
         self.send_response(200)
         self.send_header('Content-Type', 'application/ld+json')
         self.send_header('Cache-Control', 'max-age=3600')
@@ -52,33 +55,64 @@ def context_url():
 def test_requests_document_loader_accepts_custom_session():
     """RequestsDocumentLoader accepts a CachedSession via session=."""
     loader = RequestsDocumentLoader(
-        session=CachedSession(backend='memory', cache_control=True))
+        session=CachedSession(backend='memory', cache_control=True)
+    )
     assert isinstance(loader, DocumentLoader)
     assert callable(loader)
     loader.session.close()
 
 
-def test_sqlite_cache_requests_document_loader_is_document_loader():
+def test_sqlite_cache_requests_document_loader_is_document_loader(tmp_path):
     """Sqlite loader is a DocumentLoader composing RequestsDocumentLoader."""
-    loader = SqliteCacheRequestsDocumentLoader()
+    loader = SqliteCacheRequestsDocumentLoader(
+        sqlite_file_path=tmp_path / 'contexts.sqlite',
+    )
     assert isinstance(loader, DocumentLoader)
-    assert isinstance(loader._loader, RequestsDocumentLoader)
     assert callable(loader)
+    assert isinstance(loader.session, CachedSession)
+    assert isinstance(loader._loader, RequestsDocumentLoader)
+    loader.session.close()
+
+
+def test_sqlite_cache_file_is_not_created_on_init(tmp_path):
+    """Constructing the loader touches neither the cache file nor its parent."""
+    cache_path = tmp_path / 'cache' / 'contexts.sqlite'
+    SqliteCacheRequestsDocumentLoader(sqlite_file_path=cache_path)
+    assert not cache_path.exists()
+    assert not cache_path.parent.exists()
+
+
+def test_sqlite_cache_file_is_created_on_first_load(context_url, tmp_path):
+    """The cache file appears once a document is actually loaded."""
+    cache_path = tmp_path / 'contexts.sqlite'
+    loader = SqliteCacheRequestsDocumentLoader(sqlite_file_path=cache_path)
+    loader(context_url)
+    assert cache_path.exists()
     loader.session.close()
 
 
+def test_unusable_sqlite_cache_path_raises_on_first_load(context_url, tmp_path):
+    """An unopenable cache path fails the load instead of silently degrading."""
+    cache_path = tmp_path / 'contexts.sqlite'
+    cache_path.mkdir()
+    loader = SqliteCacheRequestsDocumentLoader(sqlite_file_path=cache_path)
+    with pytest.raises(sqlite3.Error):
+        loader(context_url)
+
+
 def test_sqlite_cache_requests_document_loader_rejects_relative_sqlite_file_path():
     """Relative sqlite_file_path is rejected."""
     with pytest.raises(ValueError, match='absolute path'):
-        SqliteCacheRequestsDocumentLoader(
-            sqlite_file_path=Path('relative.sqlite'))
+        SqliteCacheRequestsDocumentLoader(sqlite_file_path=Path('relative.sqlite'))
 
 
 def test_sqlite_cache_file_path_is_resolved(tmp_path):
     """Absolute sqlite_file_path is normalized to a full path."""
     sqlite_file_path = tmp_path / 'cache' / '..' / 'contexts.sqlite'
-    assert _resolve_sqlite_file_path(sqlite_file_path) == (
-        tmp_path / 'contexts.sqlite').resolve()
+    assert (
+        _resolve_sqlite_file_path(sqlite_file_path)
+        == (tmp_path / 'contexts.sqlite').resolve()
+    )
 
 
 def test_http_cache_headers_serve_from_cache_with_cache_control(context_url):
@@ -88,7 +122,8 @@ def test_http_cache_headers_serve_from_cache_with_cache_control(context_url):
             'test_memory_cache_control',
             backend='memory',
             cache_control=True,
-        ))
+        )
+    )
     loader(context_url)
     loader(context_url)
     assert _ContextHandler.request_count == 1
@@ -103,7 +138,8 @@ def test_http_cache_headers_without_cache_control_hits_server_twice(context_url)
             backend='memory',
             cache_control=False,
             expire_after=0,
-        ))
+        )
+    )
     loader(context_url)
     loader(context_url)
     assert _ContextHandler.request_count == 2