VLV clean

[openldap] / libraries / libldap / utf-8.c

diff --git a/libraries/libldap/utf-8.c b/libraries/libldap/utf-8.c
index c2518f41b0f50ab40b5ee1acc0e6032636fe8097..ab42cfbf5e15f8ffb6425f32283a212979e97028 100644 (file)
--- a/libraries/libldap/utf-8.c
+++ b/libraries/libldap/utf-8.c
@@ -70,8 +70,6 @@ int ldap_utf8_offset( const char * p )
 
 /*
  * Returns length indicated by first byte.
- *
- * This function should use a table lookup.
  */
 const char ldap_utf8_lentab[] = {
        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
@@ -88,20 +86,40 @@ int ldap_utf8_charlen( const char * p )
        if (!(*p & 0x80))
                return 1;
 
-       return ldap_utf8_lentab[*(unsigned char *)p ^ 0x80];
+       return ldap_utf8_lentab[*(const unsigned char *)p ^ 0x80];
 }
 
 /*
  * Make sure the UTF-8 char used the shortest possible encoding
  * returns charlen if valid, 0 if not. 
+ *
+ * Here are the valid UTF-8 encodings, taken from RFC 2279 page 4.
+ * The table is slightly modified from that of the RFC.
+ *
+ * UCS-4 range (hex)      UTF-8 sequence (binary)
+ * 0000 0000-0000 007F   0.......
+ * 0000 0080-0000 07FF   110++++. 10......
+ * 0000 0800-0000 FFFF   1110++++ 10+..... 10......
+ * 0001 0000-001F FFFF   11110+++ 10++.... 10...... 10......
+ * 0020 0000-03FF FFFF   111110++ 10+++... 10...... 10...... 10......
+ * 0400 0000-7FFF FFFF   1111110+ 10++++.. 10...... 10...... 10...... 10......
+ *
+ * The '.' bits are "don't cares". When validating a UTF-8 sequence,
+ * at least one of the '+' bits must be set, otherwise the character
+ * should have been encoded in fewer octets. Note that in the two-octet
+ * case, only the first octet needs to be validated, and this is done
+ * in the ldap_utf8_lentab[] above.
  */
 
 /* mask of required bits in second octet */
-const char ldap_utf8_mintab[] = {
-       0x20, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-       0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-       0x30, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-       0x38, 0x80, 0x80, 0x80, 0x3c, 0x80, 0x00, 0x00 };
+#undef c
+#define c const char
+c ldap_utf8_mintab[] = {
+       (c)0x20, (c)0x80, (c)0x80, (c)0x80, (c)0x80, (c)0x80, (c)0x80, (c)0x80,
+       (c)0x80, (c)0x80, (c)0x80, (c)0x80, (c)0x80, (c)0x80, (c)0x80, (c)0x80,
+       (c)0x30, (c)0x80, (c)0x80, (c)0x80, (c)0x80, (c)0x80, (c)0x80, (c)0x80,
+       (c)0x38, (c)0x80, (c)0x80, (c)0x80, (c)0x3c, (c)0x80, (c)0x00, (c)0x00 };
+#undef c
 
 int ldap_utf8_charlen2( const char * p )
 {
@@ -146,12 +164,21 @@ int ldap_x_ucs4_to_utf8( ldap_ucs4_t c, char *buf )
 {
        int len=0;
        unsigned char* p = buf;
-       if(buf == NULL) return 0;
 
-       if ( c < 0 ) {
-               /* not a valid Unicode character */
+       /* not a valid Unicode character */
+       if ( c < 0 ) return 0;
+
+       /* Just return length, don't convert */
+       if(buf == NULL) {
+               if( c < 0x80 ) return 1;
+               else if( c < 0x800 ) return 2;
+               else if( c < 0x10000 ) return 3;
+               else if( c < 0x200000 ) return 4;
+               else if( c < 0x4000000 ) return 5;
+               else return 6;
+       }
 
-       } else if( c < 0x80 ) {
+       if( c < 0x80 ) {
                p[len++] = c;
 
        } else if( c < 0x800 ) {
@@ -185,10 +212,80 @@ int ldap_x_ucs4_to_utf8( ldap_ucs4_t c, char *buf )
                p[len++] = 0x80 | ( c & 0x3f );
        }
 
-       buf[len] = '\0';
        return len;
 }
 
+#define LDAP_UCS_UTF8LEN(c)    \
+       c < 0 ? 0 : (c < 0x80 ? 1 : (c < 0x800 ? 2 : (c < 0x10000 ? 3 : \
+       (c < 0x200000 ? 4 : (c < 0x4000000 ? 5 : 6)))))
+
+/* Convert a string to UTF-8 format. The input string is expected to
+ * have characters of 1, 2, or 4 octets (in network byte order)
+ * corresponding to the ASN.1 T61STRING, BMPSTRING, and UNIVERSALSTRING
+ * types respectively. (Here T61STRING just means that there is one
+ * octet per character and characters may use the high bit of the octet.
+ * The characters are assumed to use ISO mappings, no provision is made
+ * for converting from T.61 coding rules to Unicode.)
+ */
+
+int
+ldap_ucs_to_utf8s( struct berval *ucs, int csize, struct berval *utf8s )
+{
+       unsigned char *in, *end;
+       char *ptr;
+       ldap_ucs4_t u;
+       int i, l = 0;
+
+       utf8s->bv_val = NULL;
+       utf8s->bv_len = 0;
+
+       in = (unsigned char *)ucs->bv_val;
+
+       /* Make sure we stop at an even multiple of csize */
+       end = in + ( ucs->bv_len & ~(csize-1) );
+       
+       for (; in < end; ) {
+               u = *in++;
+               if (csize > 1) {
+                       u <<= 8;
+                       u |= *in++;
+               }
+               if (csize > 2) {
+                       u <<= 8;
+                       u |= *in++;
+                       u <<= 8;
+                       u |= *in++;
+               }
+               i = LDAP_UCS_UTF8LEN(u);
+               if (i == 0)
+                       return LDAP_INVALID_SYNTAX;
+               l += i;
+       }
+
+       utf8s->bv_val = LDAP_MALLOC( l+1 );
+       if (utf8s->bv_val == NULL)
+               return LDAP_NO_MEMORY;
+       utf8s->bv_len = l;
+
+       ptr = utf8s->bv_val;
+       for (in = (unsigned char *)ucs->bv_val; in < end; ) {
+               u = *in++;
+               if (csize > 1) {
+                       u <<= 8;
+                       u |= *in++;
+               }
+               if (csize > 2) {
+                       u <<= 8;
+                       u |= *in++;
+                       u <<= 8;
+                       u |= *in++;
+               }
+               ptr += ldap_x_ucs4_to_utf8(u, ptr);
+       }
+       *ptr = '\0';
+       return LDAP_SUCCESS;
+}
+
 /*
  * Advance to the next UTF-8 character
  *