Using hash as key for caching improvements

jeroennijhof · Sep 15, 2011 · 8c634e7 · 8c634e7
1 parent f80c1ee
commit 8c634e7
Show file tree

Hide file tree

Showing 4 changed files with 36 additions and 19 deletions.
diff --git a/CHANGELOG b/CHANGELOG
@@ -1,4 +1,5 @@
 * 2011/09/13 v1.06 Jeroen Nijhof <[email protected]>
+  Using hash as key for caching improvements
   Using structs and check for overflows
   Added websense version 1 and 4 support
   Websense keepalive needs to be answered by accept

diff --git a/cache.c b/cache.c
@@ -9,6 +9,17 @@
 #include "openufp.h"
 #define DATABASE "/var/cache/openufp/cache.db"
 
+void get_hash(const char *s, char hash[10]) {
+    unsigned int _hash = 0;
+    int c;
+
+    while((c = *s++)) {
+        /* hash = hash * 33 ^ c */
+        _hash = ((_hash << 5) + _hash) ^ c;
+    }
+    snprintf(hash, sizeof(hash), "%u", _hash);
+}
+
 DB *open_cache() {
     DB *dbp;
     int ret;
@@ -34,7 +45,7 @@ int close_cache(DB *dbp, int debug) {
     return dbp->close(dbp, 0);
 }
 
-int in_cache(DB *dbp, char url[URL_SIZE], int expire_sec, int debug) {
+int in_cache(DB *dbp, char hash[10], int expire_sec, int debug) {
     DBT key, data;
     int ret;
     char sec[15];
@@ -48,28 +59,28 @@ int in_cache(DB *dbp, char url[URL_SIZE], int expire_sec, int debug) {
 
     bzero(&key, sizeof(key));
     bzero(&data, sizeof(data));
-    key.data = url;
-    key.size = strlen(url)+1;
+    key.data = hash;
+    key.size = strlen(hash)+1;
     data.data = sec;
     data.size = strlen(sec)+1;
 
     if ((ret = dbp->get(dbp, NULL, &key, &data, 0)) == 0) {
         if (strcmp((char *)data.data, sec) > 0) {
             if (debug > 0)
-                syslog(LOG_INFO, "cache: url %s retrieved, time %s expire at %s.",
+                syslog(LOG_INFO, "cache: hash %s retrieved, time %s expire at %s.",
                         (char *)key.data, (char *)data.data, sec);
             return 1;
         }
         if (debug > 0)
-            syslog(LOG_INFO, "cache: url in cache expired.");
-        rm_cache(dbp, url, debug);
+            syslog(LOG_INFO, "cache: hash in cache expired.");
+        rm_cache(dbp, hash, debug);
     }
     if (debug > 0)
-        syslog(LOG_INFO, "cache: url not in cache.");
+        syslog(LOG_INFO, "cache: hash not in cache.");
     return 0;
 }
 
-int add_cache(DB *dbp, char url[URL_SIZE], int debug) {
+int add_cache(DB *dbp, char hash[10], int debug) {
     DBT key, data;
     int ret;
     char sec[15];
@@ -83,8 +94,8 @@ int add_cache(DB *dbp, char url[URL_SIZE], int debug) {
 
     bzero(&key, sizeof(key));
     bzero(&data, sizeof(data));
-    key.data = url;
-    key.size = strlen(url)+1;
+    key.data = hash;
+    key.size = sizeof(hash);
     data.data = sec;
     data.size = strlen(sec)+1;
 
@@ -98,7 +109,7 @@ int add_cache(DB *dbp, char url[URL_SIZE], int debug) {
     return -1;
 }
 
-int rm_cache(DB *dbp, char url[URL_SIZE], int debug) {
+int rm_cache(DB *dbp, char hash[10], int debug) {
     DBT key;
     int ret;
 
@@ -111,8 +122,8 @@ int rm_cache(DB *dbp, char url[URL_SIZE], int debug) {
     }
 
     bzero(&key, sizeof(key));
-    key.data = url;
-    key.size = strlen(url)+1;
+    key.data = hash;
+    key.size = sizeof(hash);
 
     if ((ret = dbp->del(dbp, NULL, &key, 0)) == 0) {
         dbp->sync(dbp, 0);

diff --git a/cache.h b/cache.h
@@ -8,9 +8,10 @@
 
 #include <db.h>
 
+extern void get_hash(const char *s, char hash[10]);
 extern DB *open_cache();
 extern int close_cache(DB *dbp, int debug);
-extern int in_cache(DB *dbp, char url[URL_SIZE], int expire_sec, int debug);
-extern int add_cache(DB *dbp, char url[URL_SIZE], int debug);
-extern int rm_cache(DB *dbp, char url[URL_SIZE], int debug);
+extern int in_cache(DB *dbp, char hash[10], int expire_sec, int debug);
+extern int add_cache(DB *dbp, char hash[10], int debug);
+extern int rm_cache(DB *dbp, char hash[10], int debug);
 
diff --git a/openufp.c b/openufp.c
@@ -66,6 +66,7 @@ int main(int argc, char**argv) {
     int c;
     while ((c = getopt(argc, argv, "l:r:c:C:d:nwp:f:g")) != -1) {
         char *p;
+        char hash[10];
         DB *cachedb;
         int ret = 0;
         switch(c) {
@@ -80,7 +81,8 @@ int main(int argc, char**argv) {
                 break;
             case 'C':
                 cachedb = open_cache();
-                if (rm_cache(cachedb, optarg, 255) == -1)
+                get_hash(optarg, hash);
+                if (rm_cache(cachedb, hash, 255) == -1)
                     ret = 1;
                 close_cache(cachedb, 0);
                 exit(ret);
@@ -199,6 +201,7 @@ int main(int argc, char**argv) {
                     syslog(LOG_INFO, "caching disabled.");
 
                 int cached = 0;
+                char hash[10];
                 struct websns_req *websns_request = NULL;
                 struct n2h2_req *n2h2_request = NULL;
                 for(;;) {
@@ -248,7 +251,8 @@ int main(int argc, char**argv) {
                             syslog(LOG_INFO, "received url request.");
 
                         // check if cached
-                        cached = in_cache(cachedb, request.url, cache_exp_secs, debug);
+                        get_hash(request.url, hash);
+                        cached = in_cache(cachedb, hash, cache_exp_secs, debug);
                         if (cached == -1) // Happens when there is a cache problem
                             cached = 0;
 
@@ -283,7 +287,7 @@ int main(int argc, char**argv) {
                                 websns_accept(cli_fd, websns_request);
                             }
                             if (!cached)
-                                add_cache(cachedb, request.url, debug);
+                                add_cache(cachedb, hash, debug);
                             if (debug > 0)
                                 syslog(LOG_INFO, "url accepted: srcip %s, dstip %s, url %s.",
                                                  inet_ntoa(request.srcip), inet_ntoa(request.dstip), request.url);