[Feature] Allow rspamadm mime to process multiple files

diff --git a/lualib/rspamadm/mime.lua b/lualib/rspamadm/mime.lua
index 37a7d66deed43c53a9c72a9a457be71e85d224e9..dc46a2a06358b578932fdbfbef9784d3c4e7e7dc 100644 (file)
--- a/lualib/rspamadm/mime.lua
+++ b/lualib/rspamadm/mime.lua
@@ -15,6 +15,7 @@ limitations under the License.
 ]]--
 
 local argparse = require "argparse"
+local ansicolors = require "ansicolors"
 --local rspamd_util = require "rspamd_util"
 local rspamd_task = require "rspamd_task"
 local rspamd_logger = require "rspamd_logger"
@@ -42,7 +43,7 @@ local extract = parser:command "extract ex e"
 extract:argument "file"
        :description "File to process"
        :argname "<file>"
-       :args "1"
+       :args "+"
 
 extract:mutex(
     extract:flag "-t --text"
@@ -61,13 +62,15 @@ extract:option "-o --output"
           decoded_utf = "raw_utf"
        }
        :default "content"
+extract:flag "--no-file"
+    :description "Do not print filename"
 
 local stat = parser:command "stat st s"
                     :description "Extracts statistical data from MIME messages"
 stat:argument "file"
     :description "File to process"
     :argname "<file>"
-    :args "1"
+    :args "+"
 stat:mutex(
     stat:flag "-m --meta"
         :description "Lua metatokens",
@@ -76,13 +79,15 @@ stat:mutex(
     stat:flag "-F --fuzzy"
         :description "Fuzzy hashes"
 )
+stat:flag "--no-file"
+    :description "Do not print filename"
 
 local urls = parser:command "urls url u"
                    :description "Extracts URLs from MIME messages"
 urls:argument "file"
     :description "File to process"
     :argname "<file>"
-    :args "1"
+    :args "+"
 urls:mutex(
     urls:flag "-t --tld"
         :description "Get TLDs only",
@@ -100,6 +105,8 @@ urls:flag "-c --count"
     :description "Print count of each printed element"
 urls:flag "-r --reverse"
     :description "Reverse sort order"
+urls:flag "--no-file"
+    :description "Do not print filename"
 
 
 local function load_config(opts)
@@ -117,182 +124,222 @@ local function load_config(opts)
   end
 end
 
-local function load_task(opts)
-  if not opts.file then
+local function load_task(opts, fname)
+  if not fname then
     parser:error('no file specified')
   end
 
-  local res,task = rspamd_task.load_from_file(opts.file)
+  local res,task = rspamd_task.load_from_file(fname)
 
   if not res then
-    parser:error(string.format('cannot read message from %s: %s', opts.file,
+    parser:error(string.format('cannot read message from %s: %s', fname,
         task))
   end
 
   if not task:process_message() then
-    parser:error(string.format('cannot read message from %s: %s', opts.file,
+    parser:error(string.format('cannot read message from %s: %s', fname,
         'failed to parse'))
   end
 
   return task
 end
 
+local function highlight(fmt, ...)
+  return ansicolors.white .. string.format(fmt, ...) .. ansicolors.reset
+end
+
+local function maybe_print_fname(opts, fname)
+  if not opts.json and not opts['no-file'] then
+    rspamd_logger.messagex(highlight('File: %s', fname))
+  end
+end
+
 local function extract_handler(opts)
-  local task = load_task(opts)
-
-  if opts.text or opts.html then
-    local tp = task:get_text_parts() or {}
-
-    for _,part in ipairs(tp) do
-      local how = opts.output
-      if opts.text and not part:is_html() then
-        part:get_content(how):write()
-        io.write('\n')
-      elseif opts.html and part:is_html() then
-        part:get_content(how):write()
-        io.write('\n')
+  for i,fname in ipairs(opts.file) do
+    local task = load_task(opts, fname)
+
+    maybe_print_fname(opts, fname)
+
+    if opts.text or opts.html then
+      local tp = task:get_text_parts() or {}
+
+      for _,part in ipairs(tp) do
+        local how = opts.output
+        if opts.text and not part:is_html() then
+          part:get_content(how):write()
+          io.write('\n')
+        elseif opts.html and part:is_html() then
+          part:get_content(how):write()
+          io.write('\n')
+        end
       end
     end
-  end
 
-  task:destroy() -- No automatic dtor
+    task:destroy() -- No automatic dtor
+  end
 end
 
 local function stat_handler(opts)
   load_config(opts)
   rspamd_url.init(rspamd_config:get_tld_path())
   rspamd_config:init_subsystem('langdet,stat') -- Needed to gen stat tokens
-  local task = load_task(opts)
 
-  if opts.meta then
-    local mt = lua_meta.gen_metatokens_table(task)
-    for k,v in pairs(mt) do
-      rspamd_logger.messagex('%s = %s', k, v)
+  for i,fname in ipairs(opts.file) do
+    local task = load_task(opts, fname)
+
+    maybe_print_fname(opts, fname)
+    if opts.meta then
+      local mt = lua_meta.gen_metatokens_table(task)
+      for k,v in pairs(mt) do
+        rspamd_logger.messagex('%s = %s', k, v)
+      end
     end
-  end
 
-  task:destroy() -- No automatic dtor
+    task:destroy() -- No automatic dtor
+
+    if i > 1 then
+      rspamd_logger.messagex('')
+    end
+  end
 end
 
 local function urls_handler(opts)
   load_config(opts)
   rspamd_url.init(rspamd_config:get_tld_path())
-  local task = load_task(opts)
-  local elts = {}
-
-  local function process_url(u)
-    local s
-    if opts.tld then
-      s = u:get_tld()
-    elseif opts.host then
-      s = u:get_host()
-    elseif opts.json then
-      s = u:get_text()
-    else
-      s = u:get_text()
-    end
 
-    if opts.unique then
-      if elts[s] then
-        elts[s].count = elts[s].count + 1
+  if opts.json then rspamd_logger.messagex('[') end
+
+  for i,fname in ipairs(opts.file) do
+    local task = load_task(opts, fname)
+
+    maybe_print_fname(opts, fname)
+    if opts.json then rspamd_logger.messagex('{"file":"%s",', fname) end
+    local task = load_task(opts)
+    local elts = {}
+
+    local function process_url(u)
+      local s
+      if opts.tld then
+        s = u:get_tld()
+      elseif opts.host then
+        s = u:get_host()
+      elseif opts.json then
+        s = u:get_text()
       else
-        elts[s] = {
-          count = 1,
-          url = u
-        }
+        s = u:get_text()
       end
-    else
-      if opts.json then
-        table.insert(elts, u)
+
+      if opts.unique then
+        if elts[s] then
+          elts[s].count = elts[s].count + 1
+        else
+          elts[s] = {
+            count = 1,
+            url = u
+          }
+        end
       else
-        table.insert(elts, s)
+        if opts.json then
+          table.insert(elts, u)
+        else
+          table.insert(elts, s)
+        end
       end
     end
-  end
 
-  for _,u in ipairs(task:get_urls(true)) do
-    process_url(u)
-  end
+    for _,u in ipairs(task:get_urls(true)) do
+      process_url(u)
+    end
 
-  local json_elts = {}
+    local json_elts = {}
 
-  local function process_elt(s, u)
-    if opts.unique then
-      -- s is string, u is {url = url, count = count }
-      if not opts.json then
-        if opts.count then
-          rspamd_logger.messagex('%s : %s', s, u.count)
+    local function process_elt(s, u)
+      if opts.unique then
+        -- s is string, u is {url = url, count = count }
+        if not opts.json then
+          if opts.count then
+            rspamd_logger.messagex('%s : %s', s, u.count)
+          else
+            rspamd_logger.messagex('%s', s)
+          end
         else
-          rspamd_logger.messagex('%s', s)
+          local tb = u.url:to_table()
+          tb.count = u.count
+          table.insert(json_elts, tb)
         end
       else
-        local tb = u.url:to_table()
-        tb.count = u.count
-        table.insert(json_elts, tb)
-      end
-    else
-      -- s is index, u is url or string
-      if opts.json then
-        local tb = u:to_table()
-        table.insert(json_elts, tb)
-      else
-        rspamd_logger.messagex('%s', u)
+        -- s is index, u is url or string
+        if opts.json then
+          local tb = u:to_table()
+          table.insert(json_elts, tb)
+        else
+          rspamd_logger.messagex('%s', u)
+        end
       end
     end
-  end
 
-  if opts.sort then
-    local sfunc
-    if opts.unique then
-      sfunc = function(t, a, b)
-        if t[a].count ~= t[b].count then
-          if opts.reverse then
-            return t[a].count > t[b].count
+    if opts.sort then
+      local sfunc
+      if opts.unique then
+        sfunc = function(t, a, b)
+          if t[a].count ~= t[b].count then
+            if opts.reverse then
+              return t[a].count > t[b].count
+            else
+              return t[a].count < t[b].count
+            end
           else
-            return t[a].count < t[b].count
+            -- Sort lexicography
+            if opts.reverse then
+              return a > b
+            else
+              return a < b
+            end
+          end
+        end
+      else
+        sfunc = function(t, a, b)
+          local va, vb
+          if opts.json then
+            va = t[a]:get_text()
+            vb = t[b]:get_text()
+          else
+            va = t[a]
+            vb = t[b]
           end
-        else
-          -- Sort lexicography
           if opts.reverse then
-            return a > b
+            return va > vb
           else
-            return a < b
+            return va < vb
           end
         end
       end
+
+
+      for s,u in lua_util.spairs(elts, sfunc) do
+        process_elt(s, u)
+      end
     else
-      sfunc = function(t, a, b)
-        local va, vb
-        if opts.json then
-          va = t[a]:get_text()
-          vb = t[b]:get_text()
-        else
-          va = t[a]
-          vb = t[b]
-        end
-        if opts.reverse then
-          return va > vb
-        else
-          return va < vb
-        end
+      for s,u in pairs(elts) do
+        process_elt(s, u)
       end
     end
 
-
-    for s,u in lua_util.spairs(elts, sfunc) do
-      process_elt(s, u)
+    if opts.json then
+      rspamd_logger.messagex('"urls": %s', ucl.to_format(json_elts, 'json'))
     end
-  else
-    for s,u in pairs(elts) do
-      process_elt(s, u)
+
+    if opts.json then
+      if i == #opts.file then
+        rspamd_logger.messagex('}')
+      else
+        rspamd_logger.messagex('},')
+      end
     end
-  end
 
-  if opts.json then
-    rspamd_logger.messagex('%s', ucl.to_format(json_elts, 'json'))
+    task:destroy() -- No automatic dtor
   end
-
-  task:destroy() -- No automatic dtor
+  if opts.json then rspamd_logger.messagex(']') end
 end
 
 local function handler(args)
@@ -300,6 +347,12 @@ local function handler(args)
 
   local command = opts.command
 
+  if type(opts.file) == 'string' then
+    opts.file = {opts.file}
+  elseif type(opts.file) == 'none' then
+    opts.file = {}
+  end
+
   if command == 'extract' then
     extract_handler(opts)
   elseif command == 'stat' then