[Minor] Lua_scanners: Add more generic yield result

7 files changed, 62 insertions, 36 deletions

diff --git a/lualib/lua_scanners/clamav.lua b/lualib/lua_scanners/clamav.lua
index 26d5e9c81..c7dd08bfc 100644
--- a/lualib/lua_scanners/clamav.lua
+++ b/lualib/lua_scanners/clamav.lua
@@ -26,7 +26,7 @@ local rspamd_util = require "rspamd_util"
 local rspamd_logger = require "rspamd_logger"
 local common = require "lua_scanners/common"
 
-local N = "antivirus"
+local N = "clamav"
 
 local default_message = '${SCANNER}: virus found: "${VIRUS}"'
 
@@ -37,7 +37,8 @@ local function clamav_config(opts)
     scan_image_mime = false;
     default_port = 3310,
     log_clean = false,
-    timeout = 15.0, -- FIXME: this will break task_timeout!
+    timeout = 5.0, -- FIXME: this will break task_timeout!
+    detection_category = "virus",
     retransmits = 2,
     cache_expire = 3600, -- expire redis in one hour
     message = default_message,
@@ -149,7 +150,7 @@ local function clamav_check(task, content, digest, rule)
     })
   end
 
-  if common.need_av_check(task, content, rule) then
+  if common.need_av_check(task, content, rule, N) then
     if common.check_av_cache(task, digest, rule, clamav_check_uncached, N) then
       return
     else
diff --git a/lualib/lua_scanners/common.lua b/lualib/lua_scanners/common.lua
index ad99137a2..605e3bb15 100644
--- a/lualib/lua_scanners/common.lua
+++ b/lualib/lua_scanners/common.lua
@@ -22,6 +22,7 @@ limitations under the License.
 local rspamd_logger = require "rspamd_logger"
 local lua_util = require "lua_util"
 local lua_redis = require "lua_redis"
+local fun = require "fun"
 
 local exports = {}
 
@@ -46,36 +47,38 @@ local function match_patterns(default_sym, found, patterns)
   end
 end
 
-local function yield_result(task, rule, vname, N)
+local function yield_result(task, rule, vname, N, dyn_weight)
   local all_whitelisted = true
+  if not dyn_weight then dyn_weight = 1.0 end
   if type(vname) == 'string' then
-    local symname = match_patterns(rule['symbol'], vname, rule['patterns'])
-    if rule['whitelist'] and rule['whitelist']:get_key(vname) then
-      rspamd_logger.infox(task, '%s: "%s" is in whitelist', rule['type'], vname)
+    local symname = match_patterns(rule.symbol, vname, rule.patterns)
+    if rule.whitelist and rule.whitelist:get_key(vname) then
+      rspamd_logger.infox(task, '%s: "%s" is in whitelist', N, vname)
       return
     end
     task:insert_result(symname, 1.0, vname)
-    rspamd_logger.infox(task, '%s: virus found: "%s"', rule['type'], vname)
+    rspamd_logger.infox(task, '%s: %s found: "%s"', N, rule.detection_category, vname)
   elseif type(vname) == 'table' then
     for _, vn in ipairs(vname) do
-      local symname = match_patterns(rule['symbol'], vn, rule['patterns'])
-      if rule['whitelist'] and rule['whitelist']:get_key(vn) then
-        rspamd_logger.infox(task, '%s: "%s" is in whitelist', rule['type'], vn)
+      local symname = match_patterns(rule.symbol, vn, rule.patterns)
+      if rule.whitelist and rule.whitelist:get_key(vn) then
+        rspamd_logger.infox(task, '%s: "%s" is in whitelist', N, vn)
       else
         all_whitelisted = false
-        task:insert_result(symname, 1.0, vn)
-        rspamd_logger.infox(task, '%s: virus found: "%s"', rule['type'], vn)
+        task:insert_result(symname, dyn_weight, vn)
+        rspamd_logger.infox(task, '%s: %s found: "%s"',
+            N, rule.detection_category, vn)
       end
     end
   end
-  if rule['action'] then
+  if rule.action then
     if type(vname) == 'table' then
       if all_whitelisted then return end
       vname = table.concat(vname, '; ')
     end
     task:set_pre_result(rule['action'],
         lua_util.template(rule.message or 'Rejected', {
-          SCANNER = rule['type'],
+          SCANNER = N,
           VIRUS = vname,
         }), N)
   end
@@ -85,15 +88,15 @@ local function message_not_too_large(task, content, rule)
   local max_size = tonumber(rule.max_size)
   if not max_size then return true end
   if #content > max_size then
-    rspamd_logger.infox(task, "skip %s AV check as it is too large: %s (%s is allowed)",
-        rule.type, #content, max_size)
+    rspamd_logger.infox(task, "skip %s check as it is too large: %s (%s is allowed)",
+        N, #content, max_size)
     return false
   end
   return true
 end
 
-local function need_av_check(task, content, rule)
-  return message_not_too_large(task, content, rule)
+local function need_av_check(task, content, rule, N)
+  return message_not_too_large(task, content, rule, N)
 end
 
 local function check_av_cache(task, digest, rule, fn, N)
@@ -144,8 +147,8 @@ local function save_av_cache(task, digest, rule, to_save, N)
   local function redis_set_cb(err)
     -- Do nothing
     if err then
-      rspamd_logger.errx(task, 'failed to save virus cache for %s -> "%s": %s',
-          to_save, key, err)
+      rspamd_logger.errx(task, 'failed to save %s cache for %s -> "%s": %s',
+          rule.detection_category, to_save, key, err)
     else
       lua_util.debugm(N, task, 'saved cached result for %s: %s',
           key, to_save)
@@ -156,8 +159,8 @@ local function save_av_cache(task, digest, rule, to_save, N)
     to_save = table.concat(to_save, '\v')
   end
 
-  if rule.redis_params then
-    key = rule['prefix'] .. key
+  if rule.redis_params and rule.prefix then
+    key = rule.prefix .. key
 
     lua_redis.redis_make_request(task,
         rule.redis_params, -- connect params
@@ -165,18 +168,36 @@ local function save_av_cache(task, digest, rule, to_save, N)
         true, -- is write
         redis_set_cb, --callback
         'SETEX', -- command
-        { key, rule['cache_expire'], to_save }
+        { key, rule.cache_expire or 0, to_save }
     )
   end
 
   return false
 end
 
+local function text_parts_min_words(task, min_words)
+  local text_parts_empty = true
+  local text_parts = task:get_text_parts()
+
+  local filter_func = function(p)
+    return p:get_words_count() >= min_words
+  end
+
+  fun.each(function(p)
+    text_parts_empty = false
+  end, fun.filter(filter_func, text_parts))
+
+  return text_parts_empty
+
+end
+
+
 exports.yield_result = yield_result
 exports.match_patterns = match_patterns
 exports.need_av_check = need_av_check
 exports.check_av_cache = check_av_cache
 exports.save_av_cache = save_av_cache
+exports.text_parts_min_words = text_parts_min_words
 
 setmetatable(exports, {
   __call = function(t, override)
diff --git a/lualib/lua_scanners/dcc.lua b/lualib/lua_scanners/dcc.lua
index 43beda6ff..d34bd8425 100644
--- a/lualib/lua_scanners/dcc.lua
+++ b/lualib/lua_scanners/dcc.lua
@@ -16,7 +16,7 @@ limitations under the License.
 ]]--
 
 --[[[
--- @module fprot
+-- @module dcc
 -- This module contains dcc access functions
 --]]
 
@@ -225,7 +225,7 @@ local function dcc_check(task, content, _, rule)
       callback = dcc_callback
     })
   end
-  if common.need_av_check(task, content, rule) then
+  if common.need_av_check(task, content, rule, N) then
     dcc_check_uncached()
   end
 end
@@ -280,7 +280,7 @@ local function dcc_config(opts)
 end
 
 return {
-  type = {'dcc','bulk_scanner', 'scanner'},
+  type = {'dcc','bulk', 'hash', 'scanner'},
   description = 'dcc bulk scanner',
   configure = dcc_config,
   check = dcc_check,
diff --git a/lualib/lua_scanners/fprot.lua b/lualib/lua_scanners/fprot.lua
index 1cb21dd43..27a29a4bc 100644
--- a/lualib/lua_scanners/fprot.lua
+++ b/lualib/lua_scanners/fprot.lua
@@ -25,7 +25,7 @@ local upstream_list = require "rspamd_upstream_list"
 local rspamd_logger = require "rspamd_logger"
 local common = require "lua_scanners/common"
 
-local N = "antivirus"
+local N = "fprot"
 
 local default_message = '${SCANNER}: virus found: "${VIRUS}"'
 
@@ -35,8 +35,9 @@ local function fprot_config(opts)
     scan_text_mime = false;
     scan_image_mime = false;
     default_port = 10200,
-    timeout = 15.0, -- FIXME: this will break task_timeout!
+    timeout = 5.0, -- FIXME: this will break task_timeout!
     log_clean = false,
+    detection_category = "virus",
     retransmits = 2,
     cache_expire = 3600, -- expire redis in one hour
     message = default_message,
@@ -152,7 +153,7 @@ local function fprot_check(task, content, digest, rule)
     })
   end
 
-  if common.need_av_check(task, content, rule) then
+  if common.need_av_check(task, content, rule, N) then
     if common.check_av_cache(task, digest, rule, fprot_check_uncached, N) then
       return
     else
diff --git a/lualib/lua_scanners/kaspersky_av.lua b/lualib/lua_scanners/kaspersky_av.lua
index b55b6c24c..e903467c2 100644
--- a/lualib/lua_scanners/kaspersky_av.lua
+++ b/lualib/lua_scanners/kaspersky_av.lua
@@ -26,7 +26,7 @@ local rspamd_util = require "rspamd_util"
 local rspamd_logger = require "rspamd_logger"
 local common = require "lua_scanners/common"
 
-local N = "antivirus"
+local N = "kaspersky"
 
 local default_message = '${SCANNER}: virus found: "${VIRUS}"'
 
@@ -41,6 +41,7 @@ local function kaspersky_config(opts)
     retransmits = 1, -- use local files, retransmits are useless
     cache_expire = 3600, -- expire redis in one hour
     message = default_message,
+    detection_category = "virus",
     tmpdir = '/tmp',
     prefix = 'rs_ak',
   }
@@ -170,7 +171,7 @@ local function kaspersky_check(task, content, digest, rule)
     })
   end
 
-  if common.need_av_check(task, content, rule) then
+  if common.need_av_check(task, content, rule, N) then
     if common.check_av_cache(task, digest, rule, kaspersky_check_uncached, N) then
       return
     else
diff --git a/lualib/lua_scanners/savapi.lua b/lualib/lua_scanners/savapi.lua
index 0cbe9ff48..84452e017 100644
--- a/lualib/lua_scanners/savapi.lua
+++ b/lualib/lua_scanners/savapi.lua
@@ -26,7 +26,7 @@ local rspamd_util = require "rspamd_util"
 local rspamd_logger = require "rspamd_logger"
 local common = require "lua_scanners/common"
 
-local N = "antivirus"
+local N = "savapi"
 
 local default_message = '${SCANNER}: virus found: "${VIRUS}"'
 
@@ -42,6 +42,7 @@ local function savapi_config(opts)
     retransmits = 1, -- FIXME: useless, for local files
     cache_expire = 3600, -- expire redis in one hour
     message = default_message,
+    detection_category = "virus",
     tmpdir = '/tmp',
   }
 
@@ -234,7 +235,7 @@ local function savapi_check(task, content, digest, rule)
     })
   end
 
-  if common.need_av_check(task, content, rule) then
+  if common.need_av_check(task, content, rule, N) then
     if common.check_av_cache(task, digest, rule, savapi_check_uncached, N) then
       return
     else
diff --git a/lualib/lua_scanners/sophos.lua b/lualib/lua_scanners/sophos.lua
index ef4acb3aa..c805cc56f 100644
--- a/lualib/lua_scanners/sophos.lua
+++ b/lualib/lua_scanners/sophos.lua
@@ -25,7 +25,7 @@ local upstream_list = require "rspamd_upstream_list"
 local rspamd_logger = require "rspamd_logger"
 local common = require "lua_scanners/common"
 
-local N = "antivirus"
+local N = "sophos"
 
 local default_message = '${SCANNER}: virus found: "${VIRUS}"'
 
@@ -41,6 +41,7 @@ local function sophos_config(opts)
     cache_expire = 3600, -- expire redis in one hour
     message = default_message,
     savdi_report_encrypted = false,
+    detection_category = "virus",
     savdi_report_oversize = false,
   }
 
@@ -168,7 +169,7 @@ local function sophos_check(task, content, digest, rule)
     })
   end
 
-  if common.need_av_check(task, content, rule) then
+  if common.need_av_check(task, content, rule, N) then
     if common.check_av_cache(task, digest, rule, sophos_check_uncached, N) then
       return
     else