From f3eaf0056956da40ba164d46ad83e65ff8fbfb98 Mon Sep 17 00:00:00 2001
From: mutantsan <kvaqich@gmail.com>
Date: Thu, 2 May 2024 16:25:42 +0300
Subject: [PATCH] DATASHADES-321 / fix file ttl invalidaton, add more tests

---
 ckanext/charts/cache.py            | 27 +++++++++++++++----------
 ckanext/charts/tests/test_cache.py | 32 ++++++++++++++++++++++++++++--
 2 files changed, 46 insertions(+), 13 deletions(-)

diff --git a/ckanext/charts/cache.py b/ckanext/charts/cache.py
index 39b29a8..b39f5a7 100644
--- a/ckanext/charts/cache.py
+++ b/ckanext/charts/cache.py
@@ -74,9 +74,8 @@ def __init__(self):
 
     def get_data(self, key: str) -> pd.DataFrame | None:
         """Return data from cache if exists"""
-        file_path = os.path.join(
-            self.directory, f"{self.generate_unique_consistent_filename(key)}.orc"
-        )
+
+        file_path = self.make_file_path_from_key(key)
 
         if os.path.exists(file_path):
             if self.is_file_cache_expired(file_path):
@@ -89,21 +88,22 @@ def get_data(self, key: str) -> pd.DataFrame | None:
 
     def set_data(self, key: str, data: pd.DataFrame) -> None:
         """Save data to cache. The data will be stored as an ORC file."""
-        file_path = os.path.join(
-            self.directory, f"{self.generate_unique_consistent_filename(key)}.orc"
-        )
+        file_path = self.make_file_path_from_key(key)
 
         data.to_orc(file_path)
 
     def invalidate(self, key: str) -> None:
         """Remove data from cache"""
-        file_path = os.path.join(
-            self.directory, f"{self.generate_unique_consistent_filename(key)}.orc"
-        )
+        file_path = self.make_file_path_from_key(key)
 
         if os.path.exists(file_path):
             os.remove(file_path)
 
+    def make_file_path_from_key(self, key: str) -> str:
+        return os.path.join(
+            self.directory, f"{self.generate_unique_consistent_filename(key)}.orc"
+        )
+
     def generate_unique_consistent_filename(self, key: str) -> str:
         """Generate unique and consistent filename based on the key"""
         hash_object = hashlib.sha256()
@@ -112,8 +112,13 @@ def generate_unique_consistent_filename(self, key: str) -> str:
 
     @staticmethod
     def is_file_cache_expired(file_path: str) -> bool:
-        """Check if file cache is expired"""
-        return os.path.getmtime(file_path) + config.get_file_cache_ttl() < time.time()
+        """Check if file cache is expired. If TTL is 0 then cache never expires."""
+        file_ttl = config.get_file_cache_ttl()
+
+        if not file_ttl:
+            return False
+
+        return time.time() - os.path.getmtime(file_path) > file_ttl
 
 
 def get_cache_manager(cache_strategy: str | None) -> CacheStrategy:
diff --git a/ckanext/charts/tests/test_cache.py b/ckanext/charts/tests/test_cache.py
index 238b823..2c5503c 100644
--- a/ckanext/charts/tests/test_cache.py
+++ b/ckanext/charts/tests/test_cache.py
@@ -6,8 +6,10 @@
 from ckan.tests.helpers import call_action
 
 import ckanext.charts.fetchers as fetchers
+import ckanext.charts.cache as cache
 import ckanext.charts.tests.helpers as helpers
 import ckanext.charts.const as const
+import ckanext.charts.config as config
 
 
 @pytest.mark.ckan_config("ckan.plugins", "charts_view datastore")
@@ -188,8 +190,30 @@ def test_invalidate_file_cache(self):
         assert fetcher.get_cached_data() is None
 
 
-class TestCalculateFileCacheTTL:
-    def test_calculate_file_cache_ttl(self):
+from freezegun import freeze_time
+from datetime import timedelta, datetime
+
+
+@pytest.mark.usefixtures("clean_file_cache")
+@pytest.mark.ckan_config(config.CONF_FILE_CACHE_TTL, 100)
+class TestCalculateFileExpiration:
+    def test_file_is_expired(self):
+        fetcher = fetchers.FileSystemDataFetcher(
+            helpers.get_file_path("sample.csv"), cache_strategy=const.CACHE_FILE
+        )
+
+        assert fetcher.get_cached_data() is None
+
+        fetcher.fetch_data()
+
+        assert isinstance(fetcher.get_cached_data(), pd.DataFrame)
+
+        file_path = cache.FileCache().make_file_path_from_key(fetcher.make_cache_key())
+
+        with freeze_time(datetime.now() + timedelta(seconds=101)):
+            assert cache.FileCache().is_file_cache_expired(file_path)
+
+    def test_file_is_not_expired(self):
         fetcher = fetchers.FileSystemDataFetcher(
             helpers.get_file_path("sample.csv"), cache_strategy=const.CACHE_FILE
         )
@@ -199,3 +223,7 @@ def test_calculate_file_cache_ttl(self):
         fetcher.fetch_data()
 
         assert isinstance(fetcher.get_cached_data(), pd.DataFrame)
+
+        file_path = cache.FileCache().make_file_path_from_key(fetcher.make_cache_key())
+
+        assert not cache.FileCache().is_file_cache_expired(file_path)