New cache_size_kb config for SQLite, closes #304

datasette/app.py

@@ -90,6 +90,9 @@ CONFIG_OPTIONS = (
     ConfigOption("default_cache_ttl", 365 * 24 * 60 * 60, """
         Default HTTP cache TTL (used in Cache-Control: max-age= header)
     """.strip()),
+    ConfigOption("cache_size_kb", 0, """
+        SQLite cache size in KB (0 == use SQLite default)
+    """.strip()),
 )
 DEFAULT_CONFIG = {
     option.name: option.default
@@ -238,6 +241,8 @@ class Datasette:
             conn.enable_load_extension(True)
             for extension in self.sqlite_extensions:
                 conn.execute("SELECT load_extension('{}')".format(extension))
+        if self.config["cache_size_kb"]:
+            conn.execute('PRAGMA cache_size=-{}'.format(self.config["cache_size_kb"]))
         pm.hook.prepare_connection(conn=conn)
 
     def inspect(self):

docs/config.rst

@@ -110,3 +110,12 @@ Default HTTP caching max-age header in seconds, used for ``Cache-Control: max-ag
 ::
 
     datasette mydatabase.db --config default_cache_ttl:10
+
+cache_size_kb
+-------------
+
+Sets the amount of memory SQLite uses for its `per-connection cache <https://www.sqlite.org/pragma.html#pragma_cache_size>`_, in KB.
+
+::
+
+    datasette mydatabase.db --config cache_size_kb:5000

tests/fixtures.py

@@ -51,6 +51,13 @@ def app_client_returned_rows_matches_page_size():
     yield from app_client(max_returned_rows=50)
 
 
+@pytest.fixture(scope='session')
+def app_client_larger_cache_size():
+    yield from app_client(config={
+        'cache_size_kb': 2500,
+    })
+
+
 def generate_compound_rows(num):
     for a, b, c in itertools.islice(
         itertools.product(string.ascii_lowercase, repeat=3), num
@@ -114,6 +121,9 @@ METADATA = {
                 'primary_key_multiple_columns_explicit_label': {
                     'label_column': 'content2',
                 },
+            },
+            'queries': {
+                'pragma_cache_size': 'PRAGMA cache_size;'
             }
         },
     }

tests/test_api.py

@@ -1,6 +1,7 @@
 from .fixtures import ( # noqa
     app_client,
     app_client_shorter_time_limit,
+    app_client_larger_cache_size,
     app_client_returned_rows_matches_page_size,
     generate_compound_rows,
     generate_sortable_rows,
@@ -929,6 +930,7 @@ def test_config_json(app_client):
         "allow_sql": True,
         "default_cache_ttl": 365 * 24 * 60 * 60,
         "num_sql_threads": 3,
+        "cache_size_kb": 0,
     } == response.json
 
 
@@ -1175,3 +1177,10 @@ def test_json_columns(app_client, extra_args, expected):
     path += extra_args
     response = app_client.get(path, gather_request=False)
     assert expected == response.json
+
+
+def test_config_cache_size(app_client_larger_cache_size):
+    response = app_client_larger_cache_size.get(
+        '/test_tables/pragma_cache_size.json', gather_request=False
+    )
+    assert [[-2500]] == response.json['rows']