Added improved encoding of charsets in mm1

13 years ago · e64d8d322d
parent 17a2cbfc90
commit e64d8d322d
7 changed files with 578 additions and 12 deletions
--- a/mbuni/ChangeLog
+++ b/mbuni/ChangeLog
@ -1,3 +1,5 @@
+2010-11-29 P. A. Bagyenda <bagyenda@dsmagic.com>
+	* Add better handling of character sets in mms fields
 2010-11-26 P. A. Bagyenda <bagyenda@dsmagic.com>
 	* Minor fix for default log level (thanks to Piotr Isajew (pki at ex.com.pl)
 2010-11-23 P. A. Bagyenda <bagyenda@dsmagic.com>
--- a/mbuni/mmlib/charmibenum.def
+++ b/mbuni/mmlib/charmibenum.def
@ -0,0 +1,340 @@
+/*
+ * Mbuni - Open  Source MMS Gateway 
+ * 
+ * Charset MIBenum assignments - pre-processor magic.
+ * 
+ * Copyright (C) 2003 - , Digital Solutions Ltd. - http://www.dsmagic.com
+ *
+ * Paul Bagyenda <bagyenda@dsmagic.com>
+ * 
+ * This program is free software, distributed under the terms of
+ * the GNU General Public License, with a few exceptions granted (see LICENSE)
+ */
+
+#if !defined(CharMIBEnum)
+#error "Required macro CharMIBEnum is missing."
+#endif
+
+CharMIBEnum(3,ASCII)
+CharMIBEnum(3,ANSI_X3.4-1968)
+CharMIBEnum(3,ANSI_X3.4-1986)
+CharMIBEnum(3,CP367)
+CharMIBEnum(3,IBM367)
+CharMIBEnum(3,ISO-IR-6)
+CharMIBEnum(3,ISO646-US)
+CharMIBEnum(3,ISO_646.IRV:1991)
+CharMIBEnum(3,US)
+CharMIBEnum(3,US-ASCII)
+CharMIBEnum(3,CSASCII)
+CharMIBEnum(4,CP819)
+CharMIBEnum(4,IBM819)
+CharMIBEnum(4,ISO-8859-1)
+CharMIBEnum(4,ISO-IR-100)
+CharMIBEnum(4,ISO8859-1)
+CharMIBEnum(4,ISO_8859-1)
+CharMIBEnum(4,ISO_8859-1:1987)
+CharMIBEnum(4,L1)
+CharMIBEnum(4,LATIN1)
+CharMIBEnum(4,CSISOLATIN1)
+CharMIBEnum(5,ISO-8859-2)
+CharMIBEnum(5,ISO-IR-101)
+CharMIBEnum(5,ISO8859-2)
+CharMIBEnum(5,ISO_8859-2)
+CharMIBEnum(5,ISO_8859-2:1987)
+CharMIBEnum(5,L2)
+CharMIBEnum(5,LATIN2)
+CharMIBEnum(5,CSISOLATIN2)
+CharMIBEnum(6,ISO-8859-3)
+CharMIBEnum(6,ISO-IR-109)
+CharMIBEnum(6,ISO8859-3)
+CharMIBEnum(6,ISO_8859-3)
+CharMIBEnum(6,ISO_8859-3:1988)
+CharMIBEnum(6,L3)
+CharMIBEnum(6,LATIN3)
+CharMIBEnum(6,CSISOLATIN3)
+CharMIBEnum(7,ISO-8859-4)
+CharMIBEnum(7,ISO-IR-110)
+CharMIBEnum(7,ISO8859-4)
+CharMIBEnum(7,ISO_8859-4)
+CharMIBEnum(7,ISO_8859-4:1988)
+CharMIBEnum(7,L4)
+CharMIBEnum(7,LATIN4)
+CharMIBEnum(7,CSISOLATIN4)
+CharMIBEnum(8,CYRILLIC)
+CharMIBEnum(8,ISO-8859-5)
+CharMIBEnum(8,ISO-IR-144)
+CharMIBEnum(8,ISO8859-5)
+CharMIBEnum(8,ISO_8859-5)
+CharMIBEnum(8,ISO_8859-5:1988)
+CharMIBEnum(8,CSISOLATINCYRILLIC)
+CharMIBEnum(9,ARABIC)
+CharMIBEnum(9,ASMO-708)
+CharMIBEnum(9,ECMA-114)
+CharMIBEnum(9,ISO-8859-6)
+CharMIBEnum(9,ISO-IR-127)
+CharMIBEnum(9,ISO8859-6)
+CharMIBEnum(9,ISO_8859-6)
+CharMIBEnum(9,ISO_8859-6:1987)
+CharMIBEnum(9,CSISOLATINARABIC)
+CharMIBEnum(10,ECMA-118)
+CharMIBEnum(10,ELOT_928)
+CharMIBEnum(10,GREEK)
+CharMIBEnum(10,GREEK8)
+CharMIBEnum(10,ISO-8859-7)
+CharMIBEnum(10,ISO-IR-126)
+CharMIBEnum(10,ISO8859-7)
+CharMIBEnum(10,ISO_8859-7)
+CharMIBEnum(10,ISO_8859-7:1987)
+CharMIBEnum(10,ISO_8859-7:2003)
+CharMIBEnum(10,CSISOLATINGREEK)
+CharMIBEnum(11,HEBREW)
+CharMIBEnum(11,ISO-8859-8)
+CharMIBEnum(11,ISO-IR-138)
+CharMIBEnum(11,ISO8859-8)
+CharMIBEnum(11,ISO_8859-8)
+CharMIBEnum(11,ISO_8859-8:1988)
+CharMIBEnum(11,CSISOLATINHEBREW)
+CharMIBEnum(12,ISO-8859-9)
+CharMIBEnum(12,ISO-IR-148)
+CharMIBEnum(12,ISO8859-9)
+CharMIBEnum(12,ISO_8859-9)
+CharMIBEnum(12,ISO_8859-9:1989)
+CharMIBEnum(12,L5)
+CharMIBEnum(12,LATIN5)
+CharMIBEnum(12,CSISOLATIN5)
+CharMIBEnum(13,ISO-8859-10)
+CharMIBEnum(13,ISO-IR-157)
+CharMIBEnum(13,ISO8859-10)
+CharMIBEnum(13,ISO_8859-10)
+CharMIBEnum(13,ISO_8859-10:1992)
+CharMIBEnum(13,L6)
+CharMIBEnum(13,LATIN6)
+CharMIBEnum(13,CSISOLATIN6)
+CharMIBEnum(15,JISX0201-1976)
+CharMIBEnum(15,JIS_X0201)
+CharMIBEnum(15,X0201)
+CharMIBEnum(15,CSHALFWIDTHKATAKANA)
+CharMIBEnum(17,MS_KANJI)
+CharMIBEnum(17,SHIFT-JIS)
+CharMIBEnum(17,SHIFT_JIS)
+CharMIBEnum(17,SJIS)
+CharMIBEnum(17,CSSHIFTJIS)
+CharMIBEnum(17,SHIFT_JIS-2004)
+CharMIBEnum(17,SHIFT_JISX0213)
+CharMIBEnum(18,EUC-JP)
+CharMIBEnum(18,EUCJP)
+CharMIBEnum(18,EXTENDED_UNIX_CODE_PACKED_FORMAT_FOR_JAPANESE)
+CharMIBEnum(18,CSEUCPKDFMTJAPANESE)
+CharMIBEnum(36,ISO-IR-149)
+CharMIBEnum(36,KOREAN)
+CharMIBEnum(36,KSC_5601)
+CharMIBEnum(36,KS_C_5601-1987)
+CharMIBEnum(36,KS_C_5601-1989)
+CharMIBEnum(36,CSKSC56011987)
+CharMIBEnum(37,ISO-2022-KR)
+CharMIBEnum(37,CSISO2022KR)
+CharMIBEnum(38,EUC-KR)
+CharMIBEnum(38,EUCKR)
+CharMIBEnum(38,CSEUCKR)
+CharMIBEnum(39,ISO-2022-JP)
+CharMIBEnum(39,CSISO2022JP)
+CharMIBEnum(39,ISO-2022-JP-1)
+CharMIBEnum(39,ISO-2022-JP-2)
+CharMIBEnum(39,CSISO2022JP2)
+CharMIBEnum(39,ISO-2022-JP-2004)
+CharMIBEnum(39,ISO-2022-JP-3)
+CharMIBEnum(40,ISO-2022-JP-2)
+CharMIBEnum(40,CSISO2022JP2)
+CharMIBEnum(40,ISO-2022-JP-2004)
+CharMIBEnum(40,ISO-2022-JP-3)
+CharMIBEnum(42,ISO-IR-14)
+CharMIBEnum(42,ISO646-JP)
+CharMIBEnum(42,JIS_C6220-1969-RO)
+CharMIBEnum(42,JP)
+CharMIBEnum(42,CSISO14JISC6220RO)
+CharMIBEnum(56,CN)
+CharMIBEnum(56,GB_1988-80)
+CharMIBEnum(56,ISO-IR-57)
+CharMIBEnum(56,ISO646-CN)
+CharMIBEnum(56,CSISO57GB1988)
+CharMIBEnum(57,CHINESE)
+CharMIBEnum(57,GB_2312-80)
+CharMIBEnum(57,ISO-IR-58)
+CharMIBEnum(57,CSISO58GB231280)
+CharMIBEnum(63,ISO-IR-87)
+CharMIBEnum(63,JIS0208)
+CharMIBEnum(63,JIS_C6226-1983)
+CharMIBEnum(63,JIS_X0208)
+CharMIBEnum(63,JIS_X0208-1983)
+CharMIBEnum(63,JIS_X0208-1990)
+CharMIBEnum(63,X0208)
+CharMIBEnum(63,CSISO87JISX0208)
+CharMIBEnum(98,ISO-IR-159)
+CharMIBEnum(98,JIS_X0212)
+CharMIBEnum(98,JIS_X0212-1990)
+CharMIBEnum(98,JIS_X0212.1990-0)
+CharMIBEnum(98,X0212)
+CharMIBEnum(98,CSISO159JISX02121990)
+CharMIBEnum(103,UNICODE-1-1-UTF-7)
+CharMIBEnum(103,UTF-7)
+CharMIBEnum(103,CSUNICODE11UTF7)
+CharMIBEnum(104,ISO-2022-CN)
+CharMIBEnum(104,CSISO2022CN)
+CharMIBEnum(104,ISO-2022-CN-EXT)
+CharMIBEnum(105,ISO-2022-CN-EXT)
+CharMIBEnum(106,UTF-8)
+CharMIBEnum(106,UTF-8-MAC)
+CharMIBEnum(106,UTF8-MAC)
+CharMIBEnum(109,ISO-8859-13)
+CharMIBEnum(109,ISO-IR-179)
+CharMIBEnum(109,ISO8859-13)
+CharMIBEnum(109,ISO_8859-13)
+CharMIBEnum(109,L7)
+CharMIBEnum(109,LATIN7)
+CharMIBEnum(110,ISO-8859-14)
+CharMIBEnum(110,ISO-CELTIC)
+CharMIBEnum(110,ISO-IR-199)
+CharMIBEnum(110,ISO8859-14)
+CharMIBEnum(110,ISO_8859-14)
+CharMIBEnum(110,ISO_8859-14:1998)
+CharMIBEnum(110,L8)
+CharMIBEnum(110,LATIN8)
+CharMIBEnum(111,ISO-8859-15)
+CharMIBEnum(111,ISO-IR-203)
+CharMIBEnum(111,ISO8859-15)
+CharMIBEnum(111,ISO_8859-15)
+CharMIBEnum(111,ISO_8859-15:1998)
+CharMIBEnum(111,LATIN-9)
+CharMIBEnum(112,ISO-8859-16)
+CharMIBEnum(112,ISO-IR-226)
+CharMIBEnum(112,ISO8859-16)
+CharMIBEnum(112,ISO_8859-16)
+CharMIBEnum(112,ISO_8859-16:2001)
+CharMIBEnum(112,L10)
+CharMIBEnum(112,LATIN10)
+CharMIBEnum(113,GBK)
+CharMIBEnum(114,GB18030)
+CharMIBEnum(1000,ISO-10646-UCS-2)
+CharMIBEnum(1000,UCS-2)
+CharMIBEnum(1000,CSUNICODE)
+CharMIBEnum(1001,ISO-10646-UCS-4)
+CharMIBEnum(1001,UCS-4)
+CharMIBEnum(1001,CSUCS4)
+CharMIBEnum(1010,UCS-2BE)
+CharMIBEnum(1010,UNICODE-1-1)
+CharMIBEnum(1010,UNICODEBIG)
+CharMIBEnum(1010,CSUNICODE11)
+CharMIBEnum(1010,UNICODE-1-1-UTF-7)
+CharMIBEnum(1010,UTF-7)
+CharMIBEnum(1010,CSUNICODE11UTF7)
+CharMIBEnum(1012,UNICODE-1-1-UTF-7)
+CharMIBEnum(1012,UTF-7)
+CharMIBEnum(1012,CSUNICODE11UTF7)
+CharMIBEnum(1013,UTF-16BE)
+CharMIBEnum(1014,UTF-16LE)
+CharMIBEnum(1015,UTF-16)
+CharMIBEnum(1015,UTF-16BE)
+CharMIBEnum(1015,UTF-16LE)
+CharMIBEnum(1017,UTF-32)
+CharMIBEnum(1017,UTF-32BE)
+CharMIBEnum(1017,UTF-32LE)
+CharMIBEnum(1018,UTF-32BE)
+CharMIBEnum(1019,UTF-32LE)
+CharMIBEnum(2004,HP-ROMAN8)
+CharMIBEnum(2004,R8)
+CharMIBEnum(2004,ROMAN8)
+CharMIBEnum(2004,CSHPROMAN8)
+CharMIBEnum(2009,850)
+CharMIBEnum(2009,CP850)
+CharMIBEnum(2009,IBM850)
+CharMIBEnum(2009,CSPC850MULTILINGUAL)
+CharMIBEnum(2013,862)
+CharMIBEnum(2013,CP862)
+CharMIBEnum(2013,IBM862)
+CharMIBEnum(2013,CSPC862LATINHEBREW)
+CharMIBEnum(2025,CHINESE)
+CharMIBEnum(2025,GB_2312-80)
+CharMIBEnum(2025,ISO-IR-58)
+CharMIBEnum(2025,CSISO58GB231280)
+CharMIBEnum(2025,CN-GB)
+CharMIBEnum(2025,EUC-CN)
+CharMIBEnum(2025,EUCCN)
+CharMIBEnum(2025,GB2312)
+CharMIBEnum(2025,CSGB2312)
+CharMIBEnum(2026,BIG-5)
+CharMIBEnum(2026,BIG-FIVE)
+CharMIBEnum(2026,BIG5)
+CharMIBEnum(2026,BIGFIVE)
+CharMIBEnum(2026,CN-BIG5)
+CharMIBEnum(2026,CSBIG5)
+CharMIBEnum(2026,BIG5-HKSCS:1999)
+CharMIBEnum(2026,BIG5-HKSCS:2001)
+CharMIBEnum(2026,BIG5-HKSCS)
+CharMIBEnum(2026,BIG5-HKSCS:2004)
+CharMIBEnum(2026,BIG5HKSCS)
+CharMIBEnum(2026,BIG5-2003)
+CharMIBEnum(2027,MAC)
+CharMIBEnum(2027,MACINTOSH)
+CharMIBEnum(2027,MACROMAN)
+CharMIBEnum(2027,CSMACINTOSH)
+CharMIBEnum(2082,VISCII)
+CharMIBEnum(2082,VISCII1.1-1)
+CharMIBEnum(2082,CSVISCII)
+CharMIBEnum(2084,KOI8-R)
+CharMIBEnum(2084,CSKOI8R)
+CharMIBEnum(2084,KOI8-RU)
+CharMIBEnum(2085,HZ)
+CharMIBEnum(2085,HZ-GB-2312)
+CharMIBEnum(2086,866)
+CharMIBEnum(2086,CP866)
+CharMIBEnum(2086,IBM866)
+CharMIBEnum(2086,CSIBM866)
+CharMIBEnum(2088,KOI8-U)
+CharMIBEnum(2101,BIG5-HKSCS:1999)
+CharMIBEnum(2101,BIG5-HKSCS:2001)
+CharMIBEnum(2101,BIG5-HKSCS)
+CharMIBEnum(2101,BIG5-HKSCS:2004)
+CharMIBEnum(2101,BIG5HKSCS)
+CharMIBEnum(2103,CP154)
+CharMIBEnum(2103,CYRILLIC-ASIAN)
+CharMIBEnum(2103,PT154)
+CharMIBEnum(2103,PTCP154)
+CharMIBEnum(2103,CSPTCP154)
+CharMIBEnum(2109,CP874)
+CharMIBEnum(2109,WINDOWS-874)
+CharMIBEnum(2250,CP1250)
+CharMIBEnum(2250,MS-EE)
+CharMIBEnum(2250,WINDOWS-1250)
+CharMIBEnum(2251,CP1251)
+CharMIBEnum(2251,MS-CYRL)
+CharMIBEnum(2251,WINDOWS-1251)
+CharMIBEnum(2252,CP1252)
+CharMIBEnum(2252,MS-ANSI)
+CharMIBEnum(2252,WINDOWS-1252)
+CharMIBEnum(2253,CP1253)
+CharMIBEnum(2253,MS-GREEK)
+CharMIBEnum(2253,WINDOWS-1253)
+CharMIBEnum(2254,CP1254)
+CharMIBEnum(2254,MS-TURK)
+CharMIBEnum(2254,WINDOWS-1254)
+CharMIBEnum(2255,CP1255)
+CharMIBEnum(2255,MS-HEBR)
+CharMIBEnum(2255,WINDOWS-1255)
+CharMIBEnum(2256,CP1256)
+CharMIBEnum(2256,MS-ARAB)
+CharMIBEnum(2256,WINDOWS-1256)
+CharMIBEnum(2257,CP1257)
+CharMIBEnum(2257,WINBALTRIM)
+CharMIBEnum(2257,WINDOWS-1257)
+CharMIBEnum(2258,CP1258)
+CharMIBEnum(2258,WINDOWS-1258)
+CharMIBEnum(2259,ISO-IR-166)
+CharMIBEnum(2259,TIS-620)
+CharMIBEnum(2259,TIS620)
+CharMIBEnum(2259,TIS620-0)
+CharMIBEnum(2259,TIS620.2529-1)
+CharMIBEnum(2259,TIS620.2533-0)
+CharMIBEnum(2259,TIS620.2533-1)
+
+#undef CharMIBEnum
+
--- a/mbuni/mmlib/mms_mm7soap.c
+++ b/mbuni/mmlib/mms_mm7soap.c
@ -370,14 +370,16 @@ static int append_address(Octstr *p, Octstr *addr_spec, char *prefix, int add_ty
 {
     Octstr *v = addr_spec, *z;
     char *y, *typ = "";
+     
     int j, ch = octstr_get_char(v, 0);
     if (ch == '-')
 	  y = " displayOnly=\"true\"";
     else 
 	  y = "";
     j = octstr_case_search(v, octstr_imm("/TYPE=PLMN"),0);
-     if (j >= 0 && octstr_search_char(v, '@', 0) < 0) { /* NO '@' */
-	  z = octstr_copy(v, 2, j-2); /* skip the initial char that is only for info purposes. */
+     if ((j >= 0 || isphonenum_ex(v, 2)) && 
+	 octstr_search_char(v, '@', 0) < 0) { /* NO '@' */
+	  z = j>=0 ? octstr_copy(v, 2, j-2) : octstr_copy(v, 2, octstr_len(v)); /* skip the initial char that is only for info purposes. */
 	  typ = octstr_len(z) <= 6 ? "ShortCode" : "Number";
     } else {
 	  z  = octstr_copy(v, 2, octstr_len(v));
@ -814,7 +816,7 @@ int mm7_get_envelope(MSoapMsg_t *m,
     if ((s = http_header_value(m->envelope, octstr_imm("SenderAddress"))) == NULL)
 	  s = http_header_value(m->envelope, octstr_imm("Sender"));

-     if (s && octstr_get_char(s, 0) == '+')
+     if (s && (octstr_get_char(s, 0) == '+'))
 	  octstr_delete(s, 0, 2);
     else if (s) {
 	  octstr_destroy(s);
@ -954,7 +956,7 @@ MmsMsg *mm7_soap_to_mmsmsg(MSoapMsg_t *m, Octstr *from)
 	  }

 	  if ((f = http_header_value(m->envelope, octstr_imm("From"))) != NULL && 
-	      octstr_get_char(f, 0) == '+')
+	      (octstr_get_char(f, 0) == '+'))
 	       octstr_delete(f, 0, 2);
 	  else if (f) {
 	       octstr_destroy(f);
--- a/mbuni/mmlib/mms_msg.c
+++ b/mbuni/mmlib/mms_msg.c
@ -264,14 +264,17 @@ static Octstr *decode_encoded_string_value(int ret, ParseContext *context, unsig
     
     if (ret2 == WSP_FIELD_VALUE_DATA) { /* expect charset text. */
 	  long charset; /* Get it and ignore it. */
+	  Octstr *xs;
 	  wsp_secondary_field_value(context, &charset);
-	  res = parse_get_nul_string(context); /* XXX Currently we ignore charset */
+	  xs = parse_get_nul_string(context);

 	  if (ret < 0) {
 	       parse_skip_to_limit(context);
 	       parse_pop_limit(context);
 	  }

+	  res = pack_rfc2047_text(xs, charset);
+	  octstr_destroy(xs);
     } else if (ret2 != WSP_FIELD_VALUE_NUL_STRING) {
 	  mms_warning(0, "mms_msg", NULL, "Faulty header value for %s! [ret=%d,ret2=%d]\n", hname,ret,ret2);
 	  res = octstr_imm("");
@ -694,8 +697,19 @@ static void mms_pack_well_known_field(Octstr *os, int field_type, Octstr *value)
     case MMS_HEADER_RETRIEVE_TEXT:

     case MMS_HEADER_STORE_STATUS_TEXT:
-	  
-	  wsp_pack_text(os, value); /* XXX need to deal with charset issues. */
+       {
+	    int charset;
+	    Octstr *xvalue = parse_rfc2047_text(value, &charset);
+	    
+	    if (charset == US_ASCII_MIB_VAL)
+		 wsp_pack_text(os, xvalue);
+	    else { /* Need to use the other encoding */
+		 wsp_pack_integer_value(encoded, charset);
+		 wsp_pack_text(encoded,xvalue);
+		 wsp_pack_value(os, encoded);
+	    }
+	    octstr_destroy(xvalue);
+       }
 	  break;	       
 	  
     case MMS_HEADER_RESPONSE_TEXT: /* make sure response status does not begin with digit!! Has special meaning*/
--- a/mbuni/mmlib/mms_queue.c
+++ b/mbuni/mmlib/mms_queue.c
@ -383,9 +383,17 @@ static MmsEnvelope *mms_queue_readenvelope(char *qf, char *mms_queuedir, int sho

     /* We should properly validate the queue file here. */
     if (!okfile) {
+	  struct stat st;
+	  int rx;
+
+	  if ((rx = stat(xqf, &st)) < 0 ||
+	      st.st_size == 0) { /* Attempt removal */
+	       unlink(xqf);
+	       mms_error(0, "mms_queue", NULL, "Corrupt queue control file [%s], removed!",  xqf);	       
+	  } else 
+	       mms_error(0, "mms_queue", NULL, "Corrupt queue control file: %s",  xqf);
 	  free_envelope(e,0);
-	  e = NULL;
-	  mms_error(0, "mms_queue", NULL, "Corrupt queue control file: %s",  xqf);
+	  e = NULL;	  
     }
     return e;     
 }
--- a/mbuni/mmlib/mms_util.c
+++ b/mbuni/mmlib/mms_util.c
@ -1365,13 +1365,13 @@ unsigned long _mshash(char *s)
     return h;
 }

-int isphonenum(Octstr *s)
+int isphonenum_ex(Octstr *s, int offset)
 {
-     int i = 0, n = octstr_len(s);
+     int i = offset, n = octstr_len(s);
     char *cs;

     if (s && octstr_len(s) >= 1 && 
-	 octstr_get_cstr(s)[0] == '+')
+	 octstr_get_char(s, i) == '+')
 	  i++;
     for ( cs = octstr_get_cstr(s); i<n; i++)
 	  if (!gw_isdigit(cs[i]))
@ -1379,6 +1379,10 @@ int isphonenum(Octstr *s)
     return 1;
 }

+int isphonenum(Octstr *s)
+{
+     return isphonenum_ex(s, 0);
+}
 void mms_normalize_phonenum(Octstr **num, char *unified_prefix, List *strip_prefixes)
 {
     int i, n;
@ -2347,3 +2351,187 @@ MIMEEntity *multipart_from_urls(List *url_list)
     
     return m;
 }
+
+const char *mibenum_to_charset(int val)
+{
+#define CharMIBEnum(c,cset) else if (val == (c)) return #cset;
+     
+     if (0)
+	  return "";
+#include "charmibenum.def"
+     
+     return "ASCII"; /* Assume default ASCII ? */
+}
+
+int charset_to_mibenum(char *charset)
+{
+     
+#define CharMIBEnum(c,cset) else if (strcasecmp(charset,(#cset)) == 0) return (c);
+     
+     if (0)
+	  return US_ASCII_MIB_VAL;
+#include "charmibenum.def"
+     
+     return US_ASCII_MIB_VAL; /* Assume default ASCII ? */
+     
+     
+}
+
+/* XXX right now we do not honour the 75 char length limit */
+static Octstr *parse_rfc2047_atom(Octstr *in, int offset, int *mibenum, int *end_marker)
+{
+     Octstr *xs;
+     int enc, n, i, ch, ch2;
+     
+     *end_marker = offset;
+
+#if 0
+     if (in == NULL || 
+	 octstr_get_char(in, offset) != '=' ||
+	 octstr_get_char(in, offset + 1) != '?') {
+	  *mibenum = US_ASCII_MIB_VAL;
+	  return in ? octstr_duplicate(in) : NULL;
+     }
+#endif
+
+     offset += 2;
+     xs = octstr_copy(in, offset, octstr_len(in));
+     
+     /* Get charset */
+     if ((n = octstr_search_char(xs, '?', 0)) >= 0) {
+	  Octstr *charset = octstr_copy(xs, 0, n);
+
+	  octstr_delete(xs, 0, 1+octstr_len(charset));
+	  *mibenum = charset_to_mibenum(octstr_get_cstr(charset));
+	  octstr_destroy(charset);
+
+	  offset += n+1;
+     } else 
+	  *mibenum =  US_ASCII_MIB_VAL;
+
+     /* Get encoding */
+     if ((n = octstr_search_char(xs, '?', 0)) >= 0) {
+	  enc = octstr_get_char(xs, n-1);
+	  octstr_delete(xs, 0, 2); /* Delete the encoding, and the following '?' */
+	  offset += n+2;	  
+     } else 
+	  enc = 'q';
+
+
+     /* Look for end marker */
+     if ((n = octstr_search(xs, octstr_imm("?="), 0)) >= 0) {
+	  octstr_delete(xs, n, octstr_len(xs)); /* Remove everything after that */
+	  offset += n+2;
+     } else 
+	  offset += octstr_len(xs);
+	 
+     if (toupper(enc) == 'B') 
+	  octstr_base64_to_binary(xs);
+     else /* Assume plain text, deal with =20 and _ issues */
+	  for (i = 0; i < octstr_len(xs); i++)
+	       if (octstr_get_char(xs, i) == '=' && 
+		   (ch = octstr_get_char(xs,i+1)) > 0 && 
+		   (ch2 = octstr_get_char(xs,i+2)) > 0 && 
+		   isxdigit(ch) && isxdigit(ch2)) { 
+		    sscanf(octstr_get_cstr(xs) + i + 1, "%2x", &ch);
+		    
+		    octstr_delete(xs, i, 3); 
+		    octstr_insert_char(xs, i,ch);
+	       } else if (octstr_get_char(xs, i) == '_')  {
+		    octstr_delete(xs, i, 1); 
+		    octstr_insert_char(xs, i,0x20);
+	       }
+	 
+     *end_marker = offset;
+     return xs;
+}
+
+/* Output charset will be forced to UTF8 if multiple atoms exist */
+Octstr *parse_rfc2047_text(Octstr *in, int *mibenum)
+{
+     int n1, mib, end, last_mib = -1, offset = 0;
+     Octstr *x = NULL, *xs = NULL;
+     
+     *mibenum = US_ASCII_MIB_VAL; /* Default */
+     if (in == NULL)
+	  goto done;
+     x = octstr_create("");
+     while ((n1 = octstr_search(in, octstr_imm("=?"), offset)) >= 0 && 
+	    (xs = parse_rfc2047_atom(in, n1, &mib, &end)) != NULL) {
+	  Octstr *x1 = octstr_copy(in, offset, n1 - offset); /* Copy all the stuff up to this point. */
+	  
+	  octstr_append(x, x1);
+
+	  if (last_mib >= 0 && mib != last_mib) { 
+	       /* Force all to be utf-8 encoded */
+	       const char *charset1 = mibenum_to_charset(mib);
+	       Octstr *xcharset1 = octstr_imm(charset1);
+
+	       const char *charset2 = mibenum_to_charset(last_mib);
+	       Octstr *xcharset2 = octstr_imm(charset2);
+	       Octstr *xout = NULL;
+	       
+	       /* Convert old stuff to utf8 */
+	       if (last_mib != UTF8_MIB_VAL) {
+		    if (charset_to_utf8(x, &xout, xcharset2) > 0 && xout) {
+			 octstr_destroy(x);
+			 x  = xout;
+		    } else 
+			 octstr_destroy(xout);
+		    xout = NULL;
+	       }
+	       
+	       /* Convert new string and append it */
+	       if (mib != UTF8_MIB_VAL) {
+		    if (charset_to_utf8(xs, &xout, xcharset1) > 0 && xout) {
+			 octstr_destroy(xs);
+			 xs  = xout;
+		    } else 
+			 octstr_destroy(xout);
+		    xout = NULL;
+	       }
+	       
+	       mib = UTF8_MIB_VAL; /* Force all subsequent to utf8 */	       
+	  } 
+	  
+	  octstr_append(x, xs); /* Put it in */
+
+	  last_mib = *mibenum = mib;
+	  
+	  offset = end;
+
+	  octstr_destroy(x1);
+	  octstr_destroy(xs);
+	  xs = NULL;
+     }
+
+     octstr_destroy(xs); /* Because it might have escaped */
+     if (offset < octstr_len(in)) {
+	  Octstr *x1 = octstr_copy(in, offset, octstr_len(in)); /* Copy all the stuff up to this point. Ignore charset*/
+	  octstr_append(x, x1);
+	  octstr_destroy(x1);
+     }
+
+done:
+     return x;
+}
+
+Octstr *pack_rfc2047_text(Octstr *in, int charset_mib_enum)
+{
+     int i;
+     Octstr *xs;
+     if (charset_mib_enum == US_ASCII_MIB_VAL)
+	  return octstr_duplicate(in);
+     
+     xs = octstr_format("=?%s?q?", mibenum_to_charset(charset_mib_enum));
+
+     for (i = 0; i<octstr_len(in); i++) {
+	  int ch = octstr_get_char(in, i);
+	  if (isspace(ch) || ch > 127) 
+	       octstr_format_append(xs,"=%02X", ch);
+	  else 
+	       octstr_append_char(xs, ch);
+     }
+     octstr_append_cstr(xs, "?=");
+     return xs;
+}
--- a/mbuni/mmlib/mms_util.h
+++ b/mbuni/mmlib/mms_util.h
@ -173,6 +173,8 @@ unsigned long _mshash(char *s);

 /* Tell us whether address is a phone number. */
 int isphonenum(Octstr *s);
+int isphonenum_ex(Octstr *s, int offset);
+
 /* Fixup an address: Normalize number (if prefix given), Add type (if keep_suffix is set), etc. */
 void _mms_fixup_address(Octstr **address, char *unified_prefix, List *strip_prefixes, int keep_suffix);

@ -249,6 +251,16 @@ void *_mms_load_module(mCfg *cfg, mCfgGrp *grp, char *config_key, char *symbolna
 void fixup_address_type(List *headers, char *hdr, 
 			char *unified_prefix, 
 			List *strip_prefixes);
+
+/* Convert charset mib enum to charset and vice versa */
+extern const char *mibenum_to_charset(int val);
+extern int charset_to_mibenum(char *charset);
+Octstr *pack_rfc2047_text(Octstr *in, int charset_mib_enum);
+Octstr *parse_rfc2047_text(Octstr *in, int *mibenum);
+
+#define US_ASCII_MIB_VAL 3
+#define UTF8_MIB_VAL 106
+
 #define MAXQTRIES 100
 #define BACKOFF_FACTOR 5*60       /* In seconds */
 #define QUEUERUN_INTERVAL 1*60   /* 1 minutes. */