From 9a26b96cf2dfb78cbd2c7c553cad363d320b1695 Mon Sep 17 00:00:00 2001
From: Vsevolod Stakhov <vsevolod@highsecure.ru>
Date: Tue, 21 May 2019 16:58:44 +0100
Subject: [PATCH] [Feature] Add maps cache and type refinement

---
 lualib/lua_maps.lua | 69 ++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 65 insertions(+), 4 deletions(-)

diff --git a/lualib/lua_maps.lua b/lualib/lua_maps.lua
index a21a60f4f..7cbd9ae0d 100644
--- a/lualib/lua_maps.lua
+++ b/lualib/lua_maps.lua
@@ -19,8 +19,62 @@ See the License for the specific language governing permissions and
 limitations under the License.
 ]]--
 
+local rspamd_logger = require "rspamd_logger"
+
 local exports = {}
 
+local maps_cache = {}
+
+local function map_hash_key(data, mtype)
+  local hash = require "rspamd_cryptobox_hash"
+  local st = hash.create_specific('xxh64')
+  st:update(data)
+  st:update(mtype)
+
+  return st:hex()
+end
+
+local function starts(where,st)
+  return string.sub(where,1,string.len(st))==st
+end
+
+local function cut_prefix(where,st)
+  return string.sub(where,#st + 1)
+end
+
+local function maybe_adjust_type(data,mtype)
+  local function check_prefix(prefix, t)
+    if starts(data, prefix) then
+      data = cut_prefix(data, prefix)
+      mtype = t
+
+      return true
+    end
+
+    return false
+  end
+
+  local known_types = {
+    {'regexp;', 'regexp'},
+    {'re;', 'regexp'},
+    {'glob;', 'glob'},
+    {'radix;', 'radix'},
+    {'ipnet;', 'radix'},
+    {'set;', 'set'},
+    {'hash;', 'hash'},
+    {'plain;', 'hash'}
+  }
+
+  for _,t in ipairs(known_types) do
+    if check_prefix(t[1], t[2]) then
+      return data,mtype
+    end
+  end
+
+  -- No change
+  return data,mtype
+end
+
 --[[[
 -- @function lua_maps.map_add_from_ucl(opt, mtype, description)
 -- Creates a map from static data
@@ -56,6 +110,14 @@ local function rspamd_map_add_from_ucl(opt, mtype, description)
   end
 
   if type(opt) == 'string' then
+    opt,mtype = maybe_adjust_type(opt, mtype)
+    local k = map_hash_key(opt, mtype)
+    if maps_cache[k] then
+      rspamd_logger.infox(rspamd_config, 'reuse url for %s(%s)',
+          opt, mtype)
+
+      return maps_cache[k]
+    end
     -- We have a single string, so we treat it as a map
     local map = rspamd_config:add_map{
       type = mtype,
@@ -65,11 +127,14 @@ local function rspamd_map_add_from_ucl(opt, mtype, description)
 
     if map then
       ret.__data = map
+      ret.hash = k
       setmetatable(ret, ret_mt)
+      maps_cache[k] = ret
       return ret
     end
   elseif type(opt) == 'table' then
     -- it might be plain map or map of plain elements
+    -- no caching in this case (yet)
     if opt[1] then
       if mtype == 'radix' then
 
@@ -201,10 +266,6 @@ exports.map_add_from_ucl = rspamd_map_add_from_ucl
 local function rspamd_maybe_check_map(key, what)
   local fun = require "fun"
 
-  local function starts(where,st)
-    return string.sub(where,1,string.len(st))==st
-  end
-
   if type(what) == "table" then
     return fun.any(function(elt) return rspamd_maybe_check_map(key, elt) end, what)
   end
-- 
2.39.5