Merge branch 'master' into issue#5148

2 files changed, 214 insertions, 0 deletions

diff --git a/lualib/lua_mime.lua b/lualib/lua_mime.lua
index ea6bf5125..795a803e5 100644
--- a/lualib/lua_mime.lua
+++ b/lualib/lua_mime.lua
@@ -760,4 +760,141 @@ exports.message_to_ucl_schema = function()
   }
 end
 
+--[[[
+-- @function lua_mime.remove_attachments(task, settings)
+-- Removes all attachments from a message, keeping only text parts
+-- @param {task} task Rspamd task object
+-- @param {table} settings Table with the following fields:
+--   * keep_images: boolean, whether to keep inline images (default: false)
+--   * min_text_size: number, minimum text part size to keep (default: 0)
+--   * max_text_size: number, maximum text part size to keep (default: inf)
+-- @return {table} modified message state similar to other modification functions:
+-- * out: new content (body only)
+--]]
+exports.remove_attachments = function(task, settings)
+  local newline_s = newline(task)
+  local state = {
+    newline_s = newline_s
+  }
+  local out = {}
+
+  settings = settings or {}
+  local keep_images = settings.keep_images or false
+  local min_text_size = settings.min_text_size or 0
+  local max_text_size = settings.max_text_size or math.huge
+
+  -- Process message structure
+  local boundaries = {}
+  local cur_boundary
+  local has_attachments = false
+  local parts_to_keep = {}
+  local parts_indexes_to_keep = {}
+
+  -- First pass: identify parts to keep
+  for i, part in ipairs(task:get_parts()) do
+    local keep_part = false
+
+    if part:is_text() and not part:is_attachment() then
+      local length = part:get_length()
+      if length >= min_text_size and length <= max_text_size then
+        keep_part = true
+      end
+    elseif keep_images and part:is_image() then
+      local cd = part:get_header('Content-Disposition')
+      if cd and cd:lower():match('inline') then
+        keep_part = true
+      end
+    end
+
+    if keep_part then
+      table.insert(parts_to_keep, part)
+      parts_indexes_to_keep[i] = true
+    else
+      has_attachments = true
+    end
+  end
+
+  -- If no attachments found, return false to indicate that no alterations are required
+  if not has_attachments then
+    return false
+  end
+
+  -- Second pass: reconstruct message
+  for i, part in ipairs(task:get_parts()) do
+    local boundary = part:get_boundary()
+    if part:is_multipart() then
+      if cur_boundary then
+        out[#out + 1] = { string.format('--%s',
+            boundaries[#boundaries]), true }
+      end
+
+      boundaries[#boundaries + 1] = boundary or '--XXX'
+      cur_boundary = boundary
+
+      local rh = part:get_raw_headers()
+      if #rh > 0 then
+        out[#out + 1] = { rh, true }
+      end
+    elseif part:is_message() then
+      if boundary then
+        if cur_boundary and boundary ~= cur_boundary then
+          -- Need to close boundary
+          out[#out + 1] = { string.format('--%s--',
+              boundaries[#boundaries]), true }
+          table.remove(boundaries)
+          cur_boundary = nil
+        end
+        out[#out + 1] = { string.format('--%s',
+            boundary), true }
+      end
+
+      out[#out + 1] = { part:get_raw_headers(), true }
+    else
+      if parts_indexes_to_keep[i] then
+        if boundary then
+          if cur_boundary and boundary ~= cur_boundary then
+            -- Need to close previous boundary
+            out[#out + 1] = { string.format('--%s--',
+                boundaries[#boundaries]), true }
+            table.remove(boundaries)
+            cur_boundary = boundary
+          end
+          out[#out + 1] = { string.format('--%s',
+              boundary), true }
+        end
+        -- Add part headers
+        local headers = part:get_raw_headers()
+
+        if headers then
+          out[#out + 1] = {
+            headers,
+            true
+          }
+        end
+
+        -- Add content
+        out[#out + 1] = {
+          part:get_raw_content(),
+          false
+        }
+      end
+
+    end
+  end
+
+  -- Close remaining boundaries
+  local b = table.remove(boundaries)
+  while b do
+    out[#out + 1] = { string.format('--%s--', b), true }
+    if #boundaries > 0 then
+      out[#out + 1] = { '', true }
+    end
+    b = table.remove(boundaries)
+  end
+
+  state.out = out
+
+  return state
+end
+
 return exports
diff --git a/lualib/rspamadm/mime.lua b/lualib/rspamadm/mime.lua
index 6a589d66a..7750c5a78 100644
--- a/lualib/rspamadm/mime.lua
+++ b/lualib/rspamadm/mime.lua
@@ -160,6 +160,25 @@ modify:option "-H --html-footer"
       :description "Adds footer to text/html parts from a specific file"
       :argname "<file>"
 
+local strip = parser:command "strip"
+                    :description "Strip attachments from a message"
+strip:argument "file"
+     :description "File to process"
+     :argname "<file>"
+     :args "+"
+strip:flag "-i --keep-images"
+     :description "Keep images"
+strip:option "--min-text-size"
+     :description "Minimal text size to keep"
+     :argname "<size>"
+     :convert(tonumber)
+     :default(0)
+strip:option "--max-text-size"
+     :description "Max text size to keep"
+     :argname "<size>"
+     :convert(tonumber)
+     :default(math.huge)
+
 local sign = parser:command "sign"
                    :description "Performs DKIM signing"
 sign:argument "file"
@@ -893,6 +912,62 @@ local function sign_handler(opts)
   end
 end
 
+local function strip_handler(opts)
+  load_config(opts)
+  rspamd_url.init(rspamd_config:get_tld_path())
+
+  for _, fname in ipairs(opts.file) do
+    local task = load_task(opts, fname)
+    local newline_s = newline(task)
+
+    local rewrite = lua_mime.remove_attachments(task, {
+      keep_images = opts.keep_images,
+      min_text_size = opts.min_text_size,
+      max_text_size = opts.max_text_size
+    }) or {}
+    local out = {} -- Start with headers
+
+    local function process_headers_cb(name, hdr)
+      out[#out + 1] = hdr.raw:gsub('\r?\n?$', '')
+    end
+
+    task:headers_foreach(process_headers_cb, { full = true })
+    -- End of headers
+    out[#out + 1] = ''
+
+    if rewrite.out then
+      for _, o in ipairs(rewrite.out) do
+        out[#out + 1] = o
+      end
+    else
+      out[#out + 1] = { task:get_rawbody(), false }
+    end
+
+    for _, o in ipairs(out) do
+      if type(o) == 'string' then
+        io.write(o)
+        io.write(newline_s)
+      elseif type(o) == 'table' then
+        io.flush()
+        if type(o[1]) == 'string' then
+          io.write(o[1])
+        else
+          o[1]:save_in_file(1)
+        end
+
+        if o[2] then
+          io.write(newline_s)
+        end
+      else
+        o:save_in_file(1)
+        io.write(newline_s)
+      end
+    end
+
+    task:destroy() -- No automatic dtor
+  end
+end
+
 -- Strips directories and .extensions (if present) from a filepath
 local function filename_only(filepath)
   local filename = filepath:match(".*%/([^%.]+)")
@@ -995,6 +1070,8 @@ local function handler(args)
     urls_handler(opts)
   elseif command == 'modify' then
     modify_handler(opts)
+  elseif command == 'strip' then
+    strip_handler(opts)
   elseif command == 'sign' then
     sign_handler(opts)
   elseif command == 'dump' then