Merge pull request #101 from ProteinsWebTeam/no_cache_modifiers

No cache modifiers
ProteinsWebTeam · Feb 22, 2023 · 09bc2c7 · 09bc2c7
2 parents 209e7e7 + 73d32c0
commit 09bc2c7
Show file tree

Hide file tree

Showing 3 changed files with 65 additions and 18 deletions.
diff --git a/webfront/tests/tests_utils.py b/webfront/tests/tests_utils.py
@@ -1,6 +1,7 @@
 from django.test import TestCase
 
-from webfront.views.cache import canonical
+from webfront.views.common import map_url_to_levels
+from webfront.views.cache import canonical, get_timeout_from_path, SHOULD_NO_CACHE, FIVE_DAYS
 
 
 class CanonicalTestCase(TestCase):
@@ -54,3 +55,37 @@ def with_query_remove_unneeded_urls(self):
             "/api/entry/InterPro/?integrated=pfam",
             canonical("/api/entry/InterPro/?integrated=pfam&page_size=20"),
         )
+
+
+class CacheLifespanTestCase(TestCase):
+    def test_urls_no_cacheable(self):
+        urls = [
+            "/entry/InterPro/IPR000001/",
+            "/protein/uniprot/p99999/?extra_features",
+            "/entry/InterPro/?page",
+        ]
+        for url in urls:
+            levels = map_url_to_levels(url.split('?')[0])
+            self.assertEqual(SHOULD_NO_CACHE, get_timeout_from_path(url, levels))
+
+    def test_urls_short_life(self):
+        urls = [
+            "/entry/InterPro/?page=33",
+            "/entry/InterPro/?page_size=33",
+            "/entry/InterPro/?format",
+        ]
+        for url in urls:
+            levels = map_url_to_levels(url.split('?')[0])
+            self.assertEqual(FIVE_DAYS, get_timeout_from_path(url, levels))
+
+    def test_urls_long_life(self):
+        urls = [
+            "/entry/",
+            "/entry/InterPro/",
+            "/entry/InterPro/protein",
+            "/entry/InterPro/IPR000001/protein",
+            "/protein/uniprot/p99999/?conservation",
+        ]
+        for url in urls:
+            levels = map_url_to_levels(url.split('?')[0])
+            self.assertIsNone(get_timeout_from_path(url, levels))
diff --git a/webfront/views/cache.py b/webfront/views/cache.py
@@ -18,6 +18,29 @@
 
 names = [ep["name"].lower() for ep in endpoints]
 
+short_life_parameters = [
+    "cursor",
+    "size",
+    "go_terms",
+    "ida_ignore",
+    "ida_search",
+    "format",
+    "page_size",
+    "search",
+]
+
+no_cache_modifiers = [
+    "extra_features",
+    "residues",
+    "isoforms",
+    "ida",
+    "taxa",
+    "model:"
+    "annotation:info",
+    "subfamilies",
+    "subfamily",
+    "page_size",
+]
 
 def get_timeout_from_path(path, endpoint_levels):
     parsed = urlparse(path)
@@ -31,11 +54,9 @@ def get_timeout_from_path(path, endpoint_levels):
         # it doesn't have modifiers
         if len(query.keys()) == 0:
             return SHOULD_NO_CACHE
-        if (  # The only modifier is page_size
-            len(query.keys()) == 1
-            and "page_size" in query
-        ):
-            return SHOULD_NO_CACHE
+        for parameter in no_cache_modifiers:
+            if parameter in query:
+                return SHOULD_NO_CACHE
 
     # order querystring, lowercase keys
     query = OrderedDict(
@@ -49,17 +70,8 @@ def get_timeout_from_path(path, endpoint_levels):
                 return FIVE_DAYS
         except Exception:
             return SHOULD_NO_CACHE
-    short_life_parameters = [
-        "cursor",
-        "size",
-        "go_terms",
-        "ida_ignore",
-        "ida_search",
-        "format",
-    ]
     for parameter in short_life_parameters:
-        value = query.get(parameter)
-        if value is not None:
+        if parameter in query:
             return FIVE_DAYS
     return None
 

diff --git a/webfront/views/common.py b/webfront/views/common.py
@@ -20,7 +20,7 @@
 from webfront.views.proteome import ProteomeHandler
 from webfront.views.set import SetHandler
 from webfront.views.utils import UtilsHandler
-from webfront.views.cache import InterProCache, get_timeout_from_path, SHOULD_NO_CACHE
+from webfront.views.cache import InterProCache, get_timeout_from_path, canonical, SHOULD_NO_CACHE
 
 from webfront.models import Database
 from concurrent.futures import ThreadPoolExecutor, wait, FIRST_COMPLETED
@@ -191,7 +191,7 @@ def query(args):
                     drf_request=drf_request,
                 )
                 # Got a good response, then save it in cache
-                timeout = get_timeout_from_path(full_path, endpoint_levels)
+                timeout = get_timeout_from_path(canonical(full_path), endpoint_levels)
                 if timeout != SHOULD_NO_CACHE:
                     self._set_in_cache(caching_allowed, full_path, response, timeout)
                 # Forcing to close the connection because django is not closing it when this query is ran as future