Merge pull request #5254 from rspamd/vstakhov-phishing-fixes

Fix phishing symbol for the same domains

11 files changed, 152 insertions, 124 deletions

diff --git a/src/plugins/lua/antivirus.lua b/src/plugins/lua/antivirus.lua
index e39ddc5ba..5337f6666 100644
--- a/src/plugins/lua/antivirus.lua
+++ b/src/plugins/lua/antivirus.lua
@@ -258,7 +258,7 @@ if opts and type(opts) == 'table' then
             for _, p in ipairs(m['patterns']) do
               if type(p) == 'table' then
                 for sym in pairs(p) do
-                  rspamd_logger.debugm(N, rspamd_config, 'registering: %1', {
+                  lua_util.debugm(N, rspamd_config, 'registering: %1', {
                     type = 'virtual',
                     name = sym,
                     parent = m['symbol'],
@@ -292,7 +292,7 @@ if opts and type(opts) == 'table' then
             for _, p in ipairs(m['patterns_fail']) do
               if type(p) == 'table' then
                 for sym in pairs(p) do
-                  rspamd_logger.debugm(N, rspamd_config, 'registering: %1', {
+                  lua_util.debugm(N, rspamd_config, 'registering: %1', {
                     type = 'virtual',
                     name = sym,
                     parent = m['symbol'],
diff --git a/src/plugins/lua/arc.lua b/src/plugins/lua/arc.lua
index 90e254e78..fb5dd93e6 100644
--- a/src/plugins/lua/arc.lua
+++ b/src/plugins/lua/arc.lua
@@ -517,10 +517,10 @@ local function arc_sign_seal(task, params, header)
     local ar_header = task:get_header('Authentication-Results')
 
     if ar_header then
-      rspamd_logger.debugm(N, task, 'reuse authentication results header for ARC')
+      lua_util.debugm(N, task, 'reuse authentication results header for ARC')
       cur_auth_results = ar_header
     else
-      rspamd_logger.debugm(N, task, 'cannot reuse authentication results, header is missing')
+      lua_util.debugm(N, task, 'cannot reuse authentication results, header is missing')
       cur_auth_results = lua_auth_results.gen_auth_results(task, ar_settings) or ''
     end
   else
@@ -639,7 +639,7 @@ local function prepare_arc_selector(task, sel)
       ar_header = ar_header or ""
       for k, v in string.gmatch(ar_header, "(%w+)=(%w+)") do
         if k == 'arc' then
-         return v
+          return v
         end
       end
       return nil
diff --git a/src/plugins/lua/clickhouse.lua b/src/plugins/lua/clickhouse.lua
index 25eabc760..16a8ad4ec 100644
--- a/src/plugins/lua/clickhouse.lua
+++ b/src/plugins/lua/clickhouse.lua
@@ -1146,7 +1146,7 @@ local function upload_clickhouse_schema(upstream, ev_base, cfg, initial)
       errored = true
       return
     end
-    rspamd_logger.debugm(N, rspamd_config, 'uploaded clickhouse schema element %s to %s: %s',
+    lua_util.debugm(N, rspamd_config, 'uploaded clickhouse schema element %s to %s: %s',
         v, upstream:get_addr():to_string(true), reply)
   end
 
@@ -1159,7 +1159,7 @@ local function upload_clickhouse_schema(upstream, ev_base, cfg, initial)
       if initial == v[2] then
         return lua_util.template(v[1], { SCHEMA_VERSION = tostring(schema_version) })
       else
-        rspamd_logger.debugm(N, rspamd_config, 'skip clickhouse schema element %s: schema already exists',
+        lua_util.debugm(N, rspamd_config, 'skip clickhouse schema element %s: schema already exists',
             v)
       end
     end
diff --git a/src/plugins/lua/clustering.lua b/src/plugins/lua/clustering.lua
index d97bdb97e..a00ea9f43 100644
--- a/src/plugins/lua/clustering.lua
+++ b/src/plugins/lua/clustering.lua
@@ -135,7 +135,7 @@ local function clusterting_filter_cb(task, rule)
   end
 
   if not cluster_selector or not source_selector then
-    rspamd_logger.debugm(N, task, 'skip rule %s, selectors: source="%s", cluster="%s"',
+    lua_util.debugm(N, task, 'skip rule %s, selectors: source="%s", cluster="%s"',
         rule.name, source_selector, cluster_selector)
     return
   end
@@ -153,7 +153,7 @@ local function clusterting_filter_cb(task, rule)
       -- We have seen this element in ham mostly, so subtract average it from the size score
       final_score = math.min(1.0, size_score - cluster_score / cur_elts)
     end
-    rspamd_logger.debugm(N, task,
+    lua_util.debugm(N, task,
         'processed rule %s, selectors: source="%s", cluster="%s"; data: %s elts, %s score, %s elt score',
         rule.name, source_selector, cluster_selector, cur_elts, total_score, element_score)
     if final_score > 0.1 then
@@ -205,7 +205,7 @@ local function clusterting_idempotent_cb(task, rule)
   elseif verdict == 'junk' then
     score = rule.junk_mult
   else
-    rspamd_logger.debugm(N, task, 'skip rule %s, verdict=%s',
+    lua_util.debugm(N, task, 'skip rule %s, verdict=%s',
         rule.name, verdict)
     return
   end
@@ -218,7 +218,7 @@ local function clusterting_idempotent_cb(task, rule)
   end
 
   if not cluster_selector or not source_selector then
-    rspamd_logger.debugm(N, task, 'skip rule %s, selectors: source="%s", cluster="%s"',
+    lua_util.debugm(N, task, 'skip rule %s, selectors: source="%s", cluster="%s"',
         rule.name, source_selector, cluster_selector)
     return
   end
@@ -228,7 +228,7 @@ local function clusterting_idempotent_cb(task, rule)
       rspamd_logger.errx(task, 'got error while getting clustering keys %s: %s',
           source_selector, err)
     else
-      rspamd_logger.debugm(N, task, 'set clustering key for %s: %s{%s} = %s',
+      lua_util.debugm(N, task, 'set clustering key for %s: %s{%s} = %s',
           source_selector, "unknown error")
     end
   end
diff --git a/src/plugins/lua/elastic.lua b/src/plugins/lua/elastic.lua
index f3eb3cc4f..8bed9fcf4 100644
--- a/src/plugins/lua/elastic.lua
+++ b/src/plugins/lua/elastic.lua
@@ -72,7 +72,7 @@ local settings = {
   enabled = true,
   version = {
     autodetect_enabled = true,
-    autodetect_max_fail  = 30,
+    autodetect_max_fail = 30,
     -- override works only if autodetect is disabled
     override = {
       name = 'opensearch',
@@ -164,7 +164,7 @@ local Queue = {}
 Queue.__index = Queue
 
 function Queue:new()
-  local obj = {first = 1, last = 0, data = {}}
+  local obj = { first = 1, last = 0, data = {} }
   setmetatable(obj, self)
   return obj
 end
@@ -234,7 +234,7 @@ local buffer = {
 }
 
 local function contains(tbl, val)
-  for i=1,#tbl do
+  for i = 1, #tbl do
     if tbl[i]:lower() == val:lower() then
       return true
     end
@@ -244,7 +244,7 @@ end
 
 local function safe_get(table, ...)
   local value = table
-  for _, key in ipairs({...}) do
+  for _, key in ipairs({ ... }) do
     if value[key] == nil then
       return nil
     end
@@ -284,10 +284,10 @@ local function compare_versions(v1, v2)
   return 0 -- versions are equal
 end
 
-local function handle_error(action,component,limit)
+local function handle_error(action, component, limit)
   if states[component]['errors'] >= limit then
     rspamd_logger.errx(rspamd_config, 'cannot %s elastic %s, failed attempts: %s/%s, stop trying',
-      action, component:gsub('_', ' '), states[component]['errors'], limit)
+        action, component:gsub('_', ' '), states[component]['errors'], limit)
     states[component]['configured'] = true
   else
     states[component]['errors'] = states[component]['errors'] + 1
@@ -318,25 +318,25 @@ end
 local function is_empty(str)
   -- define a pattern that includes invisible unicode characters
   local str_cleared = str:gsub('[' ..
-    '\xC2\xA0'     .. -- U+00A0 non-breaking space
-    '\xE2\x80\x8B' .. -- U+200B zero width space
-    '\xEF\xBB\xBF' .. -- U+FEFF byte order mark (zero width no-break space)
-    '\xE2\x80\x8C' .. -- U+200C zero width non-joiner
-    '\xE2\x80\x8D' .. -- U+200D zero width joiner
-    '\xE2\x80\x8E' .. -- U+200E left-to-right mark
-    '\xE2\x80\x8F' .. -- U+200F right-to-left mark
-    '\xE2\x81\xA0' .. -- U+2060 word joiner
-    '\xE2\x80\xAA' .. -- U+202A left-to-right embedding
-    '\xE2\x80\xAB' .. -- U+202B right-to-left embedding
-    '\xE2\x80\xAC' .. -- U+202C pop directional formatting
-    '\xE2\x80\xAD' .. -- U+202D left-to-right override
-    '\xE2\x80\xAE' .. -- U+202E right-to-left override
-    '\xE2\x81\x9F' .. -- U+2061 function application
-    '\xE2\x81\xA1' .. -- U+2061 invisible separator
-    '\xE2\x81\xA2' .. -- U+2062 invisible times
-    '\xE2\x81\xA3' .. -- U+2063 invisible separator
-    '\xE2\x81\xA4' .. -- U+2064 invisible plus
-    ']', '') -- gsub replaces all matched characters with an empty string
+      '\xC2\xA0' .. -- U+00A0 non-breaking space
+      '\xE2\x80\x8B' .. -- U+200B zero width space
+      '\xEF\xBB\xBF' .. -- U+FEFF byte order mark (zero width no-break space)
+      '\xE2\x80\x8C' .. -- U+200C zero width non-joiner
+      '\xE2\x80\x8D' .. -- U+200D zero width joiner
+      '\xE2\x80\x8E' .. -- U+200E left-to-right mark
+      '\xE2\x80\x8F' .. -- U+200F right-to-left mark
+      '\xE2\x81\xA0' .. -- U+2060 word joiner
+      '\xE2\x80\xAA' .. -- U+202A left-to-right embedding
+      '\xE2\x80\xAB' .. -- U+202B right-to-left embedding
+      '\xE2\x80\xAC' .. -- U+202C pop directional formatting
+      '\xE2\x80\xAD' .. -- U+202D left-to-right override
+      '\xE2\x80\xAE' .. -- U+202E right-to-left override
+      '\xE2\x81\x9F' .. -- U+2061 function application
+      '\xE2\x81\xA1' .. -- U+2061 invisible separator
+      '\xE2\x81\xA2' .. -- U+2062 invisible times
+      '\xE2\x81\xA3' .. -- U+2063 invisible separator
+      '\xE2\x81\xA4' .. -- U+2064 invisible plus
+      ']', '') -- gsub replaces all matched characters with an empty string
   if str_cleared:match('[%S]') then
     return false
   else
@@ -350,7 +350,7 @@ local function fill_empty_strings(tbl, empty_value)
     if value and type(value) == 'table' then
       local nested_filtered = fill_empty_strings(value, empty_value)
       if next(nested_filtered) ~= nil then
-      filled_tbl[key] = nested_filtered
+        filled_tbl[key] = nested_filtered
       end
     elseif type(value) == 'boolean' then
       filled_tbl[key] = value
@@ -367,8 +367,8 @@ local function create_bulk_json(es_index, logs_to_send)
   local tbl = {}
   for _, row in pairs(logs_to_send) do
     local pipeline = ''
-    if settings['geoip']['enabled']then
-      pipeline = ',"pipeline":"'.. settings['geoip']['pipeline_name'] .. '"'
+    if settings['geoip']['enabled'] then
+      pipeline = ',"pipeline":"' .. settings['geoip']['pipeline_name'] .. '"'
     end
     table.insert(tbl, '{"index":{"_index":"' .. es_index .. '"' .. pipeline .. '}}')
     table.insert(tbl, ucl.to_format(row, 'json-compact'))
@@ -401,21 +401,21 @@ local function elastic_send_data(flush_all, task, cfg, ev_base)
     push_url = connect_prefix .. ip_addr .. '/' .. es_index .. '/_bulk'
 
     bulk_json = create_bulk_json(es_index, logs_to_send)
-    rspamd_logger.debugm(N, log_object, 'successfully composed payload with %s log lines', nlogs_to_send)
+    lua_util.debugm(N, log_object, 'successfully composed payload with %s log lines', nlogs_to_send)
   end
 
   local function http_callback(err, code, body, _)
     local push_done = false
     if err then
       rspamd_logger.errx(log_object, 'cannot send logs to elastic (%s): %s; failed attempts: %s/%s',
-        push_url, err, buffer['errors'], settings['limits']['max_fail'])
+          push_url, err, buffer['errors'], settings['limits']['max_fail'])
     elseif code == 200 then
       local parser = ucl.parser()
       local res, ucl_err = parser:parse_string(body)
       if not ucl_err and res then
         local obj = parser:get_object()
         push_done = true
-        rspamd_logger.debugm(N, log_object, 'successfully sent payload with %s logs', nlogs_to_send)
+        lua_util.debugm(N, log_object, 'successfully sent payload with %s logs', nlogs_to_send)
         if obj['errors'] then
           for _, value in pairs(obj['items']) do
             if value['index'] and value['index']['status'] >= 400 then
@@ -424,20 +424,20 @@ local function elastic_send_data(flush_all, task, cfg, ev_base)
               local error_type = safe_get(value, 'index', 'error', 'type') or ''
               local error_reason = safe_get(value, 'index', 'error', 'reason') or ''
               rspamd_logger.warnx(log_object,
-                'error while pushing logs to elastic, status: %s, index: %s, type: %s, reason: %s',
-                status, index, error_type, error_reason)
+                  'error while pushing logs to elastic, status: %s, index: %s, type: %s, reason: %s',
+                  status, index, error_type, error_reason)
             end
           end
         end
       else
         rspamd_logger.errx(log_object,
-          'cannot parse response from elastic (%s): %s; failed attempts: %s/%s',
-          push_url, ucl_err, buffer['errors'], settings['limits']['max_fail'])
+            'cannot parse response from elastic (%s): %s; failed attempts: %s/%s',
+            push_url, ucl_err, buffer['errors'], settings['limits']['max_fail'])
       end
     else
       rspamd_logger.errx(log_object,
-        'cannot send logs to elastic (%s) due to bad http status code: %s, response: %s; failed attempts: %s/%s',
-        push_url, code, body, buffer['errors'], settings['limits']['max_fail'])
+          'cannot send logs to elastic (%s) due to bad http status code: %s, response: %s; failed attempts: %s/%s',
+          push_url, code, body, buffer['errors'], settings['limits']['max_fail'])
     end
     -- proccess results
     if push_done then
@@ -447,8 +447,9 @@ local function elastic_send_data(flush_all, task, cfg, ev_base)
     else
       upstream:fail()
       if buffer['errors'] >= settings['limits']['max_fail'] then
-        rspamd_logger.errx(log_object, 'failed to send %s log lines, failed attempts: %s/%s, removing failed logs from bugger',
-          nlogs_to_send, buffer['errors'], settings['limits']['max_fail'])
+        rspamd_logger.errx(log_object,
+            'failed to send %s log lines, failed attempts: %s/%s, removing failed logs from bugger',
+            nlogs_to_send, buffer['errors'], settings['limits']['max_fail'])
         buffer['logs']:pop_first(nlogs_to_send)
         buffer['errors'] = 0
       else
@@ -466,7 +467,7 @@ local function elastic_send_data(flush_all, task, cfg, ev_base)
       },
       body = bulk_json,
       method = 'post',
-      callback=http_callback,
+      callback = http_callback,
       gzip = settings.use_gzip,
       keepalive = settings.use_keepalive,
       no_ssl_verify = settings.no_ssl_verify,
@@ -564,8 +565,8 @@ local function get_general_metadata(task)
   if task:has_from('smtp') then
     local from = task:get_from({ 'smtp', 'orig' })[1]
     if from and
-      from['user'] and #from['user'] > 0 and
-      from['domain'] and #from['domain'] > 0
+        from['user'] and #from['user'] > 0 and
+        from['domain'] and #from['domain'] > 0
     then
       r.from_user = from['user']
       r.from_domain = from['domain']:lower()
@@ -577,8 +578,8 @@ local function get_general_metadata(task)
   if task:has_from('mime') then
     local mime_from = task:get_from({ 'mime', 'orig' })[1]
     if mime_from and
-      mime_from['user'] and #mime_from['user'] > 0 and
-      mime_from['domain'] and #mime_from['domain'] > 0
+        mime_from['user'] and #mime_from['user'] > 0 and
+        mime_from['domain'] and #mime_from['domain'] > 0
     then
       r.mime_from_user = mime_from['user']
       r.mime_from_domain = mime_from['domain']:lower()
@@ -612,14 +613,14 @@ local function get_general_metadata(task)
       local l = {}
       for _, h in ipairs(hdr) do
         if settings['index_template']['headers_count_ignore_above'] ~= 0 and
-          #l >= settings['index_template']['headers_count_ignore_above']
+            #l >= settings['index_template']['headers_count_ignore_above']
         then
           table.insert(l, 'ignored above...')
           break
         end
         local header
         if settings['index_template']['headers_text_ignore_above'] ~= 0 and
-          h.decoded and #h.decoded >= headers_text_ignore_above
+            h.decoded and #h.decoded >= headers_text_ignore_above
         then
           header = h.decoded:sub(1, headers_text_ignore_above) .. '...'
         elseif h.decoded and #h.decoded > 0 then
@@ -663,10 +664,10 @@ local function get_general_metadata(task)
   local lang_t = {}
   if parts then
     for _, part in ipairs(parts) do
-        local l = part:get_language()
-        if l and not contains(lang_t, l) then
-          table.insert(lang_t, l)
-        end
+      local l = part:get_language()
+      if l and not contains(lang_t, l) then
+        table.insert(lang_t, l)
+      end
     end
     if #lang_t > 0 then
       r.language = lang_t
@@ -701,15 +702,15 @@ local function elastic_collect(task)
     if buffer['logs']:length() >= settings['limits']['max_rows'] then
       buffer['logs']:pop_first(settings['limits']['max_rows'])
       rspamd_logger.errx(task,
-        'elastic distro not supported, deleting %s logs from buffer due to reaching max rows limit',
-        settings['limits']['max_rows'])
+          'elastic distro not supported, deleting %s logs from buffer due to reaching max rows limit',
+          settings['limits']['max_rows'])
     end
   end
 
   local now = tostring(rspamd_util.get_time() * 1000)
   local row = { ['rspamd_meta'] = get_general_metadata(task), ['@timestamp'] = now }
   buffer['logs']:push(row)
-  rspamd_logger.debugm(N, task, 'saved log to buffer')
+  lua_util.debugm(N, task, 'saved log to buffer')
 end
 
 local function periodic_send_data(cfg, ev_base)
@@ -718,7 +719,8 @@ local function periodic_send_data(cfg, ev_base)
 
   local nlogs_total = buffer['logs']:length()
   if nlogs_total >= settings['limits']['max_rows'] then
-    rspamd_logger.infox(rspamd_config, 'flushing buffer by reaching max rows: %s/%s', nlogs_total, settings['limits']['max_rows'])
+    rspamd_logger.infox(rspamd_config, 'flushing buffer by reaching max rows: %s/%s', nlogs_total,
+        settings['limits']['max_rows'])
     flush_needed = true
   else
     local first_row = buffer['logs']:get(1)
@@ -726,8 +728,9 @@ local function periodic_send_data(cfg, ev_base)
       local time_diff = now - first_row['@timestamp']
       local time_diff_sec = lua_util.round((time_diff / 1000), 1)
       if time_diff_sec > settings.limits.max_interval then
-        rspamd_logger.infox(rspamd_config, 'flushing buffer for %s by reaching max interval, oldest log in buffer written %s sec ago',
-          time_diff_sec, first_row['@timestamp'])
+        rspamd_logger.infox(rspamd_config,
+            'flushing buffer for %s by reaching max interval, oldest log in buffer written %s sec ago',
+            time_diff_sec, first_row['@timestamp'])
         flush_needed = true
       end
     end
@@ -770,8 +773,8 @@ local function configure_geoip_pipeline(cfg, ev_base)
       upstream:ok()
     else
       rspamd_logger.errx(rspamd_config,
-        'cannot configure elastic geoip pipeline (%s), status code: %s, response: %s',
-        geoip_url, code, body)
+          'cannot configure elastic geoip pipeline (%s), status code: %s, response: %s',
+          geoip_url, code, body)
       upstream:fail()
       handle_error('configure', 'geoip_pipeline', settings['limits']['max_fail'])
     end
@@ -807,7 +810,8 @@ local function put_index_policy(cfg, ev_base, upstream, host, policy_url, index_
       states['index_policy']['configured'] = true
       upstream:ok()
     else
-      rspamd_logger.errx(rspamd_config, 'cannot configure elastic index policy (%s), status code: %s, response: %s', policy_url, code, body)
+      rspamd_logger.errx(rspamd_config, 'cannot configure elastic index policy (%s), status code: %s, response: %s',
+          policy_url, code, body)
       upstream:fail()
       handle_error('configure', 'index_policy', settings['limits']['max_fail'])
     end
@@ -862,7 +866,8 @@ local function get_index_policy(cfg, ev_base, upstream, host, policy_url, index_
             local current_states = safe_get(remote_policy, 'policy', 'states')
             if not lua_util.table_cmp(our_policy['policy']['default_state'], current_default_state) then
               update_needed = true
-            elseif not lua_util.table_cmp(our_policy['policy']['ism_template'][1]['index_patterns'], current_ism_index_patterns) then
+            elseif not lua_util.table_cmp(our_policy['policy']['ism_template'][1]['index_patterns'],
+                current_ism_index_patterns) then
               update_needed = true
             elseif not lua_util.table_cmp(our_policy['policy']['states'], current_states) then
               update_needed = true
@@ -885,8 +890,8 @@ local function get_index_policy(cfg, ev_base, upstream, host, policy_url, index_
                 put_index_policy(cfg, ev_base, upstream, host, policy_url, index_policy_json)
               else
                 rspamd_logger.errx(rspamd_config,
-                  'current elastic index policy (%s) not returned correct seq_no/primary_term, policy will not be updated, response: %s',
-                  policy_url, body)
+                    'current elastic index policy (%s) not returned correct seq_no/primary_term, policy will not be updated, response: %s',
+                    policy_url, body)
                 upstream:fail()
                 handle_error('validate current', 'index_policy', settings['limits']['max_fail'])
               end
@@ -904,8 +909,8 @@ local function get_index_policy(cfg, ev_base, upstream, host, policy_url, index_
       end
     else
       rspamd_logger.errx(rspamd_config,
-        'cannot get current elastic index policy (%s), status code: %s, response: %s',
-        policy_url, code, body)
+          'cannot get current elastic index policy (%s), status code: %s, response: %s',
+          policy_url, code, body)
       handle_error('get current', 'index_policy', settings['limits']['max_fail'])
       upstream:fail()
     end
@@ -1032,7 +1037,7 @@ local function configure_index_policy(cfg, ev_base)
       }
       index_policy['policy']['phases']['delete'] = delete_obj
     end
-  -- opensearch state policy with hot state
+    -- opensearch state policy with hot state
   elseif detected_distro['name'] == 'opensearch' then
     local retry = {
       count = 3,
@@ -1376,7 +1381,7 @@ local function configure_index_template(cfg, ev_base)
       upstream:ok()
     else
       rspamd_logger.errx(rspamd_config, 'cannot configure elastic index template (%s), status code: %s, response: %s',
-        template_url, code, body)
+          template_url, code, body)
       upstream:fail()
       handle_error('configure', 'index_template', settings['limits']['max_fail'])
     end
@@ -1419,7 +1424,8 @@ local function verify_distro(manual)
     local supported_distro_info = supported_distro[detected_distro_name]
     -- check that detected_distro_version is valid
     if not detected_distro_version or type(detected_distro_version) ~= 'string' then
-      rspamd_logger.errx(rspamd_config, 'elastic version should be a string, but we received: %s', type(detected_distro_version))
+      rspamd_logger.errx(rspamd_config, 'elastic version should be a string, but we received: %s',
+          type(detected_distro_version))
       valid = false
     elseif detected_distro_version == '' then
       rspamd_logger.errx(rspamd_config, 'unsupported elastic version: empty string')
@@ -1429,18 +1435,20 @@ local function verify_distro(manual)
       local cmp_from = compare_versions(detected_distro_version, supported_distro_info['from'])
       if cmp_from == -1 then
         rspamd_logger.errx(rspamd_config, 'unsupported elastic version: %s, minimal supported version of %s is %s',
-          detected_distro_version, detected_distro_name, supported_distro_info['from'])
+            detected_distro_version, detected_distro_name, supported_distro_info['from'])
         valid = false
       else
         local cmp_till = compare_versions(detected_distro_version, supported_distro_info['till'])
         if (cmp_till >= 0) and not supported_distro_info['till_unknown'] then
-          rspamd_logger.errx(rspamd_config, 'unsupported elastic version: %s, maximum supported version of %s is less than %s',
-            detected_distro_version, detected_distro_name, supported_distro_info['till'])
+          rspamd_logger.errx(rspamd_config,
+              'unsupported elastic version: %s, maximum supported version of %s is less than %s',
+              detected_distro_version, detected_distro_name, supported_distro_info['till'])
           valid = false
         elseif (cmp_till >= 0) and supported_distro_info['till_unknown'] then
           rspamd_logger.warnx(rspamd_config,
-            'compatibility of elastic version: %s is unknown, maximum known supported version of %s is less than %s, use at your own risk',
-            detected_distro_version, detected_distro_name, supported_distro_info['till'])
+              'compatibility of elastic version: %s is unknown, maximum known supported version of %s is less than %s,' ..
+                  'use at your own risk',
+              detected_distro_version, detected_distro_name, supported_distro_info['till'])
           valid_unknown = true
         end
       end
@@ -1452,14 +1460,14 @@ local function verify_distro(manual)
   else
     if valid and manual then
       rspamd_logger.infox(
-        rspamd_config, 'assuming elastic distro: %s, version: %s', detected_distro_name, detected_distro_version)
+          rspamd_config, 'assuming elastic distro: %s, version: %s', detected_distro_name, detected_distro_version)
       detected_distro['supported'] = true
     elseif valid and not manual then
       rspamd_logger.infox(rspamd_config, 'successfully connected to elastic distro: %s, version: %s',
-        detected_distro_name, detected_distro_version)
+          detected_distro_name, detected_distro_version)
       detected_distro['supported'] = true
     else
-      handle_error('configure','distro',settings['version']['autodetect_max_fail'])
+      handle_error('configure', 'distro', settings['version']['autodetect_max_fail'])
     end
   end
 end
@@ -1468,7 +1476,8 @@ local function configure_distro(cfg, ev_base)
   if not settings['version']['autodetect_enabled'] then
     detected_distro['name'] = settings['version']['override']['name']
     detected_distro['version'] = settings['version']['override']['version']
-    rspamd_logger.infox(rspamd_config, 'automatic detection of elastic distro and version is disabled, taking configuration from settings')
+    rspamd_logger.infox(rspamd_config,
+        'automatic detection of elastic distro and version is disabled, taking configuration from settings')
     verify_distro(true)
   end
 
@@ -1481,7 +1490,8 @@ local function configure_distro(cfg, ev_base)
       rspamd_logger.errx(rspamd_config, 'cannot connect to elastic (%s): %s', root_url, err)
       upstream:fail()
     elseif code ~= 200 then
-      rspamd_logger.errx(rspamd_config, 'cannot connect to elastic (%s), status code: %s, response: %s', root_url, code, body)
+      rspamd_logger.errx(rspamd_config, 'cannot connect to elastic (%s), status code: %s, response: %s', root_url, code,
+          body)
       upstream:fail()
     else
       local parser = ucl.parser()
@@ -1492,10 +1502,10 @@ local function configure_distro(cfg, ev_base)
       else
         local obj = parser:get_object()
         if obj['tagline'] == "The OpenSearch Project: https://opensearch.org/" then
-            detected_distro['name'] = 'opensearch'
+          detected_distro['name'] = 'opensearch'
         end
         if obj['tagline'] == "You Know, for Search" then
-            detected_distro['name'] = 'elastic'
+          detected_distro['name'] = 'elastic'
         end
         if obj['version'] then
           if obj['version']['number'] then
@@ -1537,7 +1547,7 @@ end
 local opts = rspamd_config:get_all_opt('elastic')
 
 if opts then
-  for k,v in pairs(opts) do
+  for k, v in pairs(opts) do
     settings[k] = v
   end
 
@@ -1574,7 +1584,7 @@ if opts then
     rspamd_config:register_finish_script(function(task)
       local nlogs_total = buffer['logs']:length()
       if nlogs_total > 0 then
-        rspamd_logger.debugm(N, task, 'flushing buffer on shutdown, buffer size: %s', nlogs_total)
+        lua_util.debugm(N, task, 'flushing buffer on shutdown, buffer size: %s', nlogs_total)
         elastic_send_data(true, task)
       end
     end)
diff --git a/src/plugins/lua/external_services.lua b/src/plugins/lua/external_services.lua
index e299d9faa..307218d94 100644
--- a/src/plugins/lua/external_services.lua
+++ b/src/plugins/lua/external_services.lua
@@ -274,7 +274,7 @@ if opts and type(opts) == 'table' then
             for _, p in ipairs(m['patterns']) do
               if type(p) == 'table' then
                 for sym in pairs(p) do
-                  rspamd_logger.debugm(N, rspamd_config, 'registering: %1', {
+                  lua_util.debugm(N, rspamd_config, 'registering: %1', {
                     type = 'virtual',
                     name = sym,
                     parent = m['symbol'],
@@ -307,7 +307,7 @@ if opts and type(opts) == 'table' then
             for _, p in ipairs(m['patterns_fail']) do
               if type(p) == 'table' then
                 for sym in pairs(p) do
-                  rspamd_logger.debugm(N, rspamd_config, 'registering: %1', {
+                  lua_util.debugm(N, rspamd_config, 'registering: %1', {
                     type = 'virtual',
                     name = sym,
                     parent = m['symbol'],
diff --git a/src/plugins/lua/neural.lua b/src/plugins/lua/neural.lua
index f3b26f11a..ea40fc4f7 100644
--- a/src/plugins/lua/neural.lua
+++ b/src/plugins/lua/neural.lua
@@ -687,31 +687,31 @@ local function maybe_train_existing_ann(worker, ev_base, rule, set, profiles)
                 return l >= rule.train.max_trains * (1.0 - rule.train.classes_bias)
               end
               if max_len >= rule.train.max_trains and fun.all(len_bias_check_pred, lens) then
-                rspamd_logger.debugm(N, rspamd_config,
+                lua_util.debugm(N, rspamd_config,
                     'can start ANN %s learn as it has %s learn vectors; %s required, after checking %s vectors',
                     ann_key, lens, rule.train.max_trains, what)
                 cont_cb()
               else
-                rspamd_logger.debugm(N, rspamd_config,
+                lua_util.debugm(N, rspamd_config,
                     'cannot learn ANN %s now: there are not enough %s learn vectors (has %s vectors; %s required)',
                     ann_key, what, lens, rule.train.max_trains)
               end
             else
               -- Probabilistic mode, just ensure that at least one vector is okay
               if min_len > 0 and max_len >= rule.train.max_trains then
-                rspamd_logger.debugm(N, rspamd_config,
+                lua_util.debugm(N, rspamd_config,
                     'can start ANN %s learn as it has %s learn vectors; %s required, after checking %s vectors',
                     ann_key, lens, rule.train.max_trains, what)
                 cont_cb()
               else
-                rspamd_logger.debugm(N, rspamd_config,
+                lua_util.debugm(N, rspamd_config,
                     'cannot learn ANN %s now: there are not enough %s learn vectors (has %s vectors; %s required)',
                     ann_key, what, lens, rule.train.max_trains)
               end
             end
 
           else
-            rspamd_logger.debugm(N, rspamd_config,
+            lua_util.debugm(N, rspamd_config,
                 'checked %s vectors in ANN %s: %s vectors; %s required, need to check other class vectors',
                 what, ann_key, ntrains, rule.train.max_trains)
             cont_cb()
diff --git a/src/plugins/lua/p0f.lua b/src/plugins/lua/p0f.lua
index 97757c23a..727e6d1bb 100644
--- a/src/plugins/lua/p0f.lua
+++ b/src/plugins/lua/p0f.lua
@@ -105,7 +105,7 @@ if rule then
   end
 
   for sym in pairs(rule.patterns) do
-    rspamd_logger.debugm(N, rspamd_config, 'registering: %1', {
+    lua_util.debugm(N, rspamd_config, 'registering: %1', {
       type = 'virtual',
       name = sym,
       parent = id,
diff --git a/src/plugins/lua/phishing.lua b/src/plugins/lua/phishing.lua
index 05e08c0f4..3f5c9e634 100644
--- a/src/plugins/lua/phishing.lua
+++ b/src/plugins/lua/phishing.lua
@@ -71,7 +71,7 @@ end
 local function phishing_cb(task)
   local function check_phishing_map(table)
     local phishing_data = {}
-    for k,v in pairs(table) do
+    for k, v in pairs(table) do
       phishing_data[k] = v
     end
     local url = phishing_data.url
@@ -89,6 +89,8 @@ local function phishing_cb(task)
       local data = nil
 
       if elt then
+        lua_util.debugm(N, task, 'found host element: %s',
+            host)
         local path = url:get_path()
         local query = url:get_query()
 
@@ -156,7 +158,7 @@ local function phishing_cb(task)
 
   local function check_phishing_dns(table)
     local phishing_data = {}
-    for k,v in pairs(table) do
+    for k, v in pairs(table) do
       phishing_data[k] = v
     end
     local url = phishing_data.url
@@ -192,7 +194,7 @@ local function phishing_cb(task)
       end
 
       local to_resolve_hp = compose_dns_query({ host, path })
-      rspamd_logger.debugm(N, task, 'try to resolve {%s, %s} -> %s',
+      lua_util.debugm(N, task, 'try to resolve {%s, %s} -> %s',
           host, path, to_resolve_hp)
       r:resolve_txt({
         task = task,
@@ -207,7 +209,7 @@ local function phishing_cb(task)
         end
 
         local to_resolve_hpq = compose_dns_query({ host, path, query })
-        rspamd_logger.debugm(N, task, 'try to resolve {%s, %s, %s} -> %s',
+        lua_util.debugm(N, task, 'try to resolve {%s, %s, %s} -> %s',
             host, path, query, to_resolve_hpq)
         r:resolve_txt({
           task = task,
@@ -256,9 +258,14 @@ local function phishing_cb(task)
       end
 
       if url:is_phished() then
+        local surl = tostring(url)
         local purl
+        lua_util.debugm(N, task, 'found phished url: %s',
+            surl)
 
         if url:is_redirected() then
+          lua_util.debugm(N, task, 'url %s is also been redirected',
+              surl)
           local rspamd_url = require "rspamd_url"
           -- Examine the real redirect target instead of the url
           local redirected_url = url:get_redirected()
@@ -268,6 +275,7 @@ local function phishing_cb(task)
 
           purl = rspamd_url.create(task:get_mempool(), url:get_visible())
           url = redirected_url
+          surl = string.format("redirected(%s)", tostring(url))
         else
           purl = url:get_phished()
         end
@@ -276,6 +284,10 @@ local function phishing_cb(task)
           return
         end
 
+        local spurl = tostring(purl)
+        lua_util.debugm(N, task, 'processing pair %s -> %s',
+            surl, spurl)
+
         local tld = url:get_tld()
         local ptld = purl:get_tld()
 
@@ -308,10 +320,11 @@ local function phishing_cb(task)
         local weight = 1.0
         local spoofed, why = util.is_utf_spoofed(tld, ptld)
         if spoofed then
-          lua_util.debugm(N, task, "confusable: %1 -> %2: %3", tld, ptld, why)
+          lua_util.debugm(N, task, "confusable: %s -> %s: %s", tld, ptld, why)
           weight = 1.0
         else
-          local dist = util.levenshtein_distance(stripped_tld, stripped_ptld, 2)
+          local dist = (stripped_tld == stripped_ptld) and 0
+              or util.levenshtein_distance(stripped_tld, stripped_ptld, 2)
           dist = 2 * dist / (#stripped_tld + #stripped_ptld)
 
           if dist > 0.3 and dist <= 1.0 then
@@ -330,15 +343,18 @@ local function phishing_cb(task)
 
             if a1 ~= a2 then
               weight = 1
-              lua_util.debugm(N, task, "confusable: %1 -> %2: different characters",
-                  tld, ptld, why)
+              lua_util.debugm(N, task, "confusable: %s -> %s: different characters",
+                  tld, ptld)
             else
               -- We have totally different strings in tld, so penalize it somehow
               weight = 0.5
             end
+          elseif dist == 0 then
+            -- Same domains, not phishing!
+            weight = 0.0
           end
 
-          lua_util.debugm(N, task, "distance: %1 -> %2: %3", tld, ptld, dist)
+          lua_util.debugm(N, task, "distance: %s -> %s: %s; weight = %s", tld, ptld, dist, weight)
         end
 
         local function is_url_in_map(map, furl)
@@ -368,15 +384,17 @@ local function phishing_cb(task)
           end
         end
 
-        found_in_map(strict_domains_maps, purl, 1.0)
-        if not found_in_map(anchor_exceptions_maps) then
-          if not found_in_map(phishing_exceptions_maps, purl, 1.0) then
-            if domains then
-              if is_url_in_map(domains, purl) then
+        if weight > 0 then
+          found_in_map(strict_domains_maps, purl, 1.0)
+          if not found_in_map(anchor_exceptions_maps) then
+            if not found_in_map(phishing_exceptions_maps, purl, 1.0) then
+              if domains then
+                if is_url_in_map(domains, purl) then
+                  task:insert_result(symbol, weight, ptld .. '->' .. tld)
+                end
+              else
                 task:insert_result(symbol, weight, ptld .. '->' .. tld)
               end
-            else
-              task:insert_result(symbol, weight, ptld .. '->' .. tld)
             end
           end
         end
diff --git a/src/plugins/lua/reputation.lua b/src/plugins/lua/reputation.lua
index a3af26c91..bd7d91932 100644
--- a/src/plugins/lua/reputation.lua
+++ b/src/plugins/lua/reputation.lua
@@ -955,7 +955,7 @@ local function reputation_redis_init(rule, cfg, ev_base, worker)
 
   local get_script = lua_util.jinja_template(redis_get_script_tpl,
       { windows = rule.backend.config.buckets })
-  rspamd_logger.debugm(N, rspamd_config, 'added extraction script %s', get_script)
+  lua_util.debugm(N, rspamd_config, 'added extraction script %s', get_script)
   rule.backend.script_get = lua_redis.add_redis_script(get_script, our_redis_params)
 
   -- Redis script to update Redis buckets
@@ -1003,7 +1003,7 @@ local function reputation_redis_init(rule, cfg, ev_base, worker)
 
   local set_script = lua_util.jinja_template(redis_adaptive_emea_script_tpl,
       { windows = rule.backend.config.buckets })
-  rspamd_logger.debugm(N, rspamd_config, 'added emea update script %s', set_script)
+  lua_util.debugm(N, rspamd_config, 'added emea update script %s', set_script)
   rule.backend.script_set = lua_redis.add_redis_script(set_script, our_redis_params)
 
   return true
diff --git a/src/plugins/lua/url_redirector.lua b/src/plugins/lua/url_redirector.lua
index 10b5fb255..c1fa85cae 100644
--- a/src/plugins/lua/url_redirector.lua
+++ b/src/plugins/lua/url_redirector.lua
@@ -182,7 +182,7 @@ local function resolve_cached(task, orig_url, url, key, ntries)
   local function resolve_url()
     if ntries > settings.nested_limit then
       -- We cannot resolve more, stop
-      rspamd_logger.debugm(N, task, 'cannot get more requests to resolve %s, stop on %s after %s attempts',
+      lua_util.debugm(N, task, 'cannot get more requests to resolve %s, stop on %s after %s attempts',
           orig_url, url, ntries)
       cache_url(task, orig_url, url, key, 'nested')
       local str_orig_url = tostring(orig_url)
@@ -223,7 +223,7 @@ local function resolve_cached(task, orig_url, url, key, ntries)
           if loc then
             redir_url = rspamd_url.create(task:get_mempool(), loc)
           end
-          rspamd_logger.debugm(N, task, 'found redirect from %s to %s, err code %s',
+          lua_util.debugm(N, task, 'found redirect from %s to %s, err code %s',
               orig_url, loc, code)
 
           if redir_url then
@@ -239,11 +239,11 @@ local function resolve_cached(task, orig_url, url, key, ntries)
               resolve_cached(task, orig_url, redir_url, key, ntries + 1)
             end
           else
-            rspamd_logger.debugm(N, task, "no location, headers: %s", headers)
+            lua_util.debugm(N, task, "no location, headers: %s", headers)
             cache_url(task, orig_url, url, key)
           end
         else
-          rspamd_logger.debugm(N, task, 'found redirect error from %s to %s, err code: %s',
+          lua_util.debugm(N, task, 'found redirect error from %s to %s, err code: %s',
               orig_url, url, code)
           cache_url(task, orig_url, url, key)
         end
@@ -278,7 +278,7 @@ local function resolve_cached(task, orig_url, url, key, ntries)
       if type(data) == 'string' then
         if data ~= 'processing' then
           -- Got cached result
-          rspamd_logger.debugm(N, task, 'found cached redirect from %s to %s',
+          lua_util.debugm(N, task, 'found cached redirect from %s to %s',
               url, data)
           if data:sub(1, 1) == '^' then
             -- Prefixed url stored