diff options
| author | Vsevolod Stakhov <vsevolod@highsecure.ru> | 2015-11-12 17:08:40 +0000 |
|---|---|---|
| committer | Vsevolod Stakhov <vsevolod@highsecure.ru> | 2015-11-12 17:08:40 +0000 |
| commit | 5784f28e0f559f4dbde268276e3bc46ebf3c3e7f (patch) | |
| tree | 2785caa6b187c021d6dc6c0cd99d5ac1e32ef00b | |
| parent | cb7faebd1e4167be3fdb78aa7d77519ec812a00f (diff) | |
| download | rspamd-5784f28e0f559f4dbde268276e3bc46ebf3c3e7f.tar.gz rspamd-5784f28e0f559f4dbde268276e3bc46ebf3c3e7f.zip | |
Do not convert 7 bit encodings to utf8.
| -rw-r--r-- | src/libmime/message.c | 12 |
1 files changed, 10 insertions, 2 deletions
diff --git a/src/libmime/message.c b/src/libmime/message.c index 1551f3d18..6074b8437 100644 --- a/src/libmime/message.c +++ b/src/libmime/message.c @@ -27,6 +27,7 @@ #include "rspamd.h" #include "message.h" #include "cfg_file.h" +#include "libutil/regexp.h" #include "html.h" #include "images.h" #include "utlist.h" @@ -50,6 +51,7 @@ static ac_trie_t *gtube_trie = NULL; static const gchar gtube_pattern[] = "XJS*C4JDBQADN1.NSBN3*2IDNEN*" "GTUBE-STANDARD-ANTI-UBE-TEST-EMAIL*C.34X"; +static rspamd_regexp_t *utf_compatible_re = NULL; static GQuark rspamd_message_quark (void) @@ -751,6 +753,13 @@ convert_text_to_utf (struct rspamd_task *task, return part_content; } + if (utf_compatible_re == NULL) { + utf_compatible_re = rspamd_regexp_new ( + "^(?:utf-?8.*)|(?:us-ascii)|(?:ascii)|(?:us)|(?:ISO-8859-1)|" + "(?:latin.*)|(?:CSASCII)$", + "i", NULL); + } + if ((charset = g_mime_content_type_get_parameter (type, "charset")) == NULL) { SET_PART_RAW (text_part); @@ -764,8 +773,7 @@ convert_text_to_utf (struct rspamd_task *task, return part_content; } - if (g_ascii_strcasecmp (ocharset, - "utf-8") == 0 || g_ascii_strcasecmp (ocharset, "utf8") == 0) { + if (!rspamd_regexp_match (utf_compatible_re, ocharset, strlen (ocharset), TRUE)) { if (g_utf8_validate (part_content->data, part_content->len, NULL)) { SET_PART_UTF (text_part); return part_content; |