Fix utf8-handling.

[citadel.git] / citadel / modules / xmpp / serv_xmpp.c

diff --git a/citadel/modules/xmpp/serv_xmpp.c b/citadel/modules/xmpp/serv_xmpp.c
index e23ede887bd2bddf2190356afa4d37c41f584d0a..64dc8b7aa2f4e723a1d8a75be85e536506a546b1 100644 (file)
--- a/citadel/modules/xmpp/serv_xmpp.c
+++ b/citadel/modules/xmpp/serv_xmpp.c
@@ -87,13 +87,32 @@ static void xmpp_entity_declaration(void *userData, const XML_Char *entityName,
  * Given a source string and a target buffer, returns the string
  * properly escaped for insertion into an XML stream.  Returns a
  * pointer to the target buffer for convenience.
- *
- * BUG: this does not properly handle UTF-8
  */
+static inline int Ctdl_GetUtf8SequenceLength(const char *CharS, const char *CharE)
+{
+       int n = 0;
+        unsigned char test = (1<<7);
+
+       if ((*CharS & 0xC0) != 0xC0) 
+               return 1;
+
+       while ((n < 8) && 
+              ((test & ((unsigned char)*CharS)) != 0)) 
+       {
+               test = test >> 1;
+               n ++;
+       }
+       if ((n > 6) || ((CharE - CharS) < n))
+               n = 0;
+       return n;
+}
+
 char *xmlesc(char *buf, char *str, int bufsiz)
 {
-       char *ptr;
+       int IsUtf8Sequence;
+       char *ptr, *pche;
        unsigned char ch;
+       int inlen;
        int len = 0;
 
        if (!buf) return(NULL);
@@ -102,6 +121,8 @@ char *xmlesc(char *buf, char *str, int bufsiz)
        if (!str) {
                return(buf);
        }
+       inlen = strlen(str);
+       pche = str + inlen;
 
        for (ptr=str; *ptr; ptr++) {
                ch = *ptr;
@@ -127,10 +148,24 @@ char *xmlesc(char *buf, char *str, int bufsiz)
                        buf[len] = 0;
                }
                else {
-                       char oct[10];
-                       sprintf(oct, "&#%o;", ch);
-                       strcpy(&buf[len], oct);
-                       len += strlen(oct);
+                       IsUtf8Sequence =  Ctdl_GetUtf8SequenceLength(ptr, pche);
+                       if (IsUtf8Sequence)
+                       {
+                               while ((IsUtf8Sequence > 0) && 
+                                      (ptr < pche))
+                               {
+                                       buf[len] = *ptr;
+                                       ptr ++;
+                                       --IsUtf8Sequence;
+                               }
+                       }
+                       else
+                       {
+                               char oct[10];
+                               sprintf(oct, "&#%o;", ch);
+                               strcpy(&buf[len], oct);
+                               len += strlen(oct);
+                       }
                }
                if ((len + 6) > bufsiz) {
                        return(buf);