CachingMiddleware: automatic cache pruning implemented

1 files changed, 36 insertions, 7 deletions

diff --git a/wsgitools/middlewares.py b/wsgitools/middlewares.py
index 60cab6b..b6e5d54 100644
--- a/wsgitools/middlewares.py
+++ b/wsgitools/middlewares.py
@@ -4,6 +4,7 @@ import time
 import sys
 import cgitb
 import binascii
+import collections
 from wsgitools.filters import CloseableList, CloseableIterator
 # Cannot use io module as it is broken in 2.6.
 # Writing a str to a io.StringIO results in an exception.
@@ -216,24 +217,52 @@ class CachingMiddleware:
         self.storable = storable
         self.cacheable = cacheable
         self.cache = {}
+        self.lastcached = collections.deque()
+
+    def insert_cache(self, key, obj, now=None):
+        if now is None:
+            now = time.time()
+        self.cache[key] = obj
+        self.lastcached.append((key, now))
+
+    def prune_cache(self, maxclean=16, now=None):
+        if now is None:
+            now = time.time()
+        old = now - self.maxage
+        while self.lastcached and maxclean > 0: # don't do too much work at once
+            maxclean -= 1
+            if self.lastcached[0][1] > old:
+                break
+            key, _ = self.lastcached.popleft()
+            try:
+                obj = self.cache[key]
+            except KeyError:
+                pass
+            else:
+                if obj[0] <= old:
+                    del self.cache[key]
+
     def __call__(self, environ, start_response):
         """wsgi interface
         @type environ: {str: str}
         """
         assert isinstance(environ, dict)
+        now = time.time()
+        self.prune_cache(now=now)
         if not self.storable(environ):
             return self.app(environ, start_response)
         path = environ.get("REQUEST_METHOD", "GET") + " "
         path += environ.get("SCRIPT_NAME", "/")
         path += environ.get("PATH_INFO", '')
         path += "?" + environ.get("QUERY_STRING", "")
-        if self.cacheable(environ) and path in self.cache:
-            if self.cache[path][0] + self.maxage >= time.time():
-                start_response(self.cache[path][1], self.cache[path][2])
-                return self.cache[path][3]
+        if path in self.cache and self.cacheable(environ):
+            cache_object = self.cache[path]
+            if cache_object[0] + self.maxage >= now:
+                start_response(cache_object[1], cache_object[2])
+                return cache_object[3]
             else:
                 del self.cache[path]
-        cache_object = [time.time(), "", [], []]
+        cache_object = [now, "", [], []]
         def modified_start_respesponse(status, headers, exc_info=None):
             assert isinstance(status, str)
             assert isinstance(headers, list)
@@ -252,13 +281,13 @@ class CachingMiddleware:
 
         if isinstance(ret, list):
             cache_object[3].extend(ret)
-            self.cache[path] = cache_object
+            self.insert_cache(path, cache_object, now)
             return ret
         def pass_through():
             for data in ret:
                 cache_object[3].append(data)
                 yield data
-            self.cache[path] = cache_object
+            self.insert_cache(path, cache_object, now)
         return CloseableIterator(getattr(ret, "close", None), pass_through())
 
 __all__.append("DictAuthChecker")