Merge pull request #34 from SvenMarcus/feature/file-cache

Cache workspaces based on last write time

ocrdbrowser/_cache.py

@@ -0,0 +1,30 @@
+from datetime import datetime
+import functools
+from pathlib import Path
+from typing import Callable, TypeVar
+
+T = TypeVar("T")
+
+
+def path_cache(fn: Callable[[Path], T]) -> Callable[[Path], T]:
+    pathcache: dict[Path, tuple[T, float]] = {}
+
+    def cache(path: Path) -> T:
+        new_value = fn(path)
+        pathcache[path] = new_value, path.stat().st_mtime
+        return new_value
+
+    @functools.wraps(fn)
+    def wrapper(path: Path) -> T:
+        if path not in pathcache:
+            return cache(path)
+
+        value, saved_timestamp = pathcache[path]
+        last_modified = path.stat().st_mtime
+
+        if last_modified > saved_timestamp:
+            return cache(path)
+
+        return value
+
+    return wrapper

ocrdbrowser/_workspace.py

@@ -2,18 +2,19 @@
 
 from pathlib import Path
 from typing import List
-from functools import lru_cache
+
+from ._cache import path_cache
 
 
 def is_valid(workspace: str) -> bool:
     return (Path(workspace) / "mets.xml").exists()
 
 
-@lru_cache(maxsize=1)
-def list_all(path: str) -> List[str]:
+@path_cache
+def list_all(path: Path) -> List[str]:
     # recursively enumerate METS file paths (excluding .backup subdirs)
     return [
         str(workspace.parent)
-        for workspace in Path(path).rglob("mets.xml")
+        for workspace in path.rglob("mets.xml")
         if not workspace.match(".backup/*/mets.xml")
     ]

tests/ocrdbrowser/test_path_cache.py

@@ -0,0 +1,54 @@
+from pathlib import Path
+from typing import Iterator
+
+import pytest
+
+from ocrdbrowser._cache import path_cache
+
+TMPDIR = Path(__file__).parent / "tmpdir"
+TMPFILE = TMPDIR / "tmp.txt"
+
+
+@pytest.fixture(autouse=True)
+def cleandir() -> Iterator[None]:
+    TMPDIR.mkdir()
+
+    yield
+
+    TMPFILE.unlink(missing_ok=True)
+    TMPDIR.rmdir()
+
+
+def test__path_cache_decorator__returns_the_same_result_without_calling_func_again() -> (
+    None
+):
+    call_count = 0
+
+    @path_cache
+    def fn(path: str | Path) -> int:
+        nonlocal call_count
+        call_count += 1
+        return call_count
+
+    first = fn(TMPDIR)
+    second = fn(TMPDIR)
+
+    assert call_count == 1
+    assert first == second
+
+
+def test__when_cached_path_changes__calls_func_again() -> None:
+    call_count = 0
+
+    @path_cache
+    def fn(path: str | Path) -> int:
+        nonlocal call_count
+        call_count += 1
+        return call_count
+
+    fn(TMPDIR)
+    TMPFILE.touch()
+
+    fn(TMPDIR)
+
+    assert call_count == 2

tests/ocrdbrowser/test_workspace.py

@@ -18,7 +18,7 @@ def test__a_workspace_without_mets_xml_is_invalid() -> None:
 
 
 def test__list_workspaces__returns_valid_workspaces() -> None:
-    assert set(workspace.list_all(str(WORKSPACES))) == {
+    assert set(workspace.list_all(WORKSPACES)) == {
         str(WORKSPACES / "a_workspace"),
         str(WORKSPACES / "another workspace"),
         str(WORKSPACES / "nested" / "workspace"),

tests/ocrdmonitor/server/test_workspace_endpoint.py

@@ -9,11 +9,9 @@
 from httpx import Response
 
 from tests.ocrdmonitor.server import scraping
-from tests.ocrdmonitor.server.decorators import use_custom_repository
 from tests.ocrdmonitor.server.fixtures.environment import (
     DevEnvironment,
     Fixture,
-    RepositoryInitializer,
 )
 from tests.ocrdmonitor.server.fixtures.settings import WORKSPACE_DIR
 from tests.testdoubles import (
@@ -60,16 +58,14 @@ def test__workspaces__shows_the_workspace_names_starting_from_workspace_root(
     assert set(texts) == {"a_workspace", "another workspace", "nested/workspace"}
 
 
-@use_custom_repository
+@pytest.mark.asyncio
 async def test__browse_workspace__passes_full_workspace_path_to_ocrdbrowser(
-    repository: RepositoryInitializer,
+    repository_fixture: Fixture,
 ) -> None:
     workspace = "a_workspace"
     full_workspace = str(WORKSPACE_DIR / workspace)
     browser = BrowserSpy()
-    fixture = (
-        Fixture().with_repository_type(repository).with_browser_type(lambda: browser)
-    )
+    fixture = repository_fixture.with_browser_type(lambda: browser)
 
     async with fixture as env:
         response = open_workspace(env.app, workspace)