commit d8c76f9: [Minor] Improve lua_magic HTML content type check

[Anton Yuzhaninov]

Author: Anton Yuzhaninov
Date: 2020-08-19 17:39:59 +0100
URL: https://github.com/rspamd/rspamd/commit/d8c76f9fcb23fc62db8fc539db4736dfaedb11f8 (refs/pull/3477/head)

[Minor] Improve lua_magic HTML content type check
Types which are used for html (xhtml) attachments:
text/html
application/html
application/xhtml+xml

Type which cannot be found in the wild:
text/htm

---
 lualib/lua_magic/heuristics.lua | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/lualib/lua_magic/heuristics.lua b/lualib/lua_magic/heuristics.lua
index 8c63cce72..d36b5633f 100644
--- a/lualib/lua_magic/heuristics.lua
+++ b/lualib/lua_magic/heuristics.lua
@@ -434,7 +434,7 @@ exports.text_part_heuristic = function(part, log_obj, _)
         end
       end
 
-      if mtype == 'text' and (msubtype == 'html' or msubtype == 'htm') then
+      if (mtype == 'text' or mtype == 'application') and (msubtype == 'html' or msubtype == 'xhtml+xml') then
         return 'html',21
       end
 
@@ -458,4 +458,4 @@ exports.pdf_format_heuristic = function(input, log_obj, pos, part)
   return 'pdf',weight
 end
 
-return exports
\ No newline at end of file
+return exports