decode stream names from UTF-8

This is the encoding that libgsf accepts.

2 files changed, 65 insertions, 17 deletions

diff --git a/libmsi/msipriv.h b/libmsi/msipriv.h
index d1c2290..14a3453 100644
--- a/libmsi/msipriv.h
+++ b/libmsi/msipriv.h
@@ -372,7 +372,7 @@ extern bool MSI_RecordsAreFieldsEqual(LibmsiRecord *a, LibmsiRecord *b, unsigned
 /* stream internals */
 extern void enum_stream_names( IStorage *stg );
 extern WCHAR *encode_streamname(bool bTable, const WCHAR *in);
-extern bool decode_streamname(const WCHAR *in, WCHAR *out);
+extern void decode_streamname(const WCHAR *in, WCHAR *out);
 
 /* database internals */
 extern unsigned msi_get_raw_stream( LibmsiDatabase *, const WCHAR *, IStream **);
@@ -563,6 +563,32 @@ static inline char *strcpynA( char *dst, const char *src, unsigned count )
     return dst;
 }
 
+static inline char *strdupWtoUTF8( const WCHAR *str )
+{
+    char *ret = NULL;
+    unsigned len;
+
+    if (!str) return ret;
+    len = WideCharToMultiByte( CP_UTF8, 0, str, -1, NULL, 0, NULL, NULL);
+    ret = msi_alloc( len );
+    if (ret)
+        WideCharToMultiByte( CP_UTF8, 0, str, -1, ret, len, NULL, NULL );
+    return ret;
+}
+
+static inline WCHAR *strdupUTF8toW( const char *str )
+{
+    WCHAR *ret = NULL;
+    unsigned len;
+
+    if (!str) return ret;
+    len = MultiByteToWideChar( CP_UTF8, 0, str, -1, NULL, 0 );
+    ret = msi_alloc( len * sizeof(WCHAR) );
+    if (ret)
+        MultiByteToWideChar( CP_UTF8, 0, str, -1, ret, len );
+    return ret;
+}
+
 static inline char *strdupWtoA( const WCHAR *str )
 {
     char *ret = NULL;
diff --git a/libmsi/table.c b/libmsi/table.c
index 9bc61e8..3b89a9a 100644
--- a/libmsi/table.c
+++ b/libmsi/table.c
@@ -185,30 +185,52 @@ static int mime2utf(int x)
     return '_';
 }
 
-bool decode_streamname(const WCHAR *in, WCHAR *out)
+void decode_streamname(const WCHAR *in, WCHAR *out)
 {
-    WCHAR ch;
     unsigned count = 0;
+    uint8_t *enc_utf8 = strdupWtoUTF8(in);
+    uint8_t *p = enc_utf8;
 
-    while ( (ch = *in++) )
+    uint8_t *dec_utf8 = strdup(enc_utf8);
+    uint8_t *q = dec_utf8;
+
+    while ( *p )
     {
-        if( (ch >= 0x3800 ) && (ch < 0x4840 ) )
+        uint8_t ch = *p;
+        if( (ch == 0xe3 && p[1] >= 0xa0) || (ch == 0xe4 && p[1] < 0xa0) )
         {
-            if( ch >= 0x4800 )
-                ch = mime2utf(ch-0x4800);
-            else
-            {
-                ch -= 0x3800;
-                *out++ = mime2utf(ch&0x3f);
-                count++;
-                ch = mime2utf((ch>>6)&0x3f);
-            }
+            /* UTF-8 encoding of 0x3800..0x47ff.  */
+            *q++ = mime2utf(p[2]&0x7f);
+            *q++ = mime2utf(p[1]^0xa0);
+            p += 3;
+            count += 2;
+            continue;
+        }
+        if( ch == 0xe4 && p[1] == 0xa0 ) {
+            /* UTF-8 encoding of 0x4800..0x483f.  */
+            *q++ = mime2utf(p[2]&0x7f);
+            p += 3;
+            count++;
+            continue;
+        }
+        *q++ = *p++;
+        if( ch >= 0xc1) {
+            *q++ = *p++;
+        }
+        if( ch >= 0xe0) {
+            *q++ = *p++;
+        }
+        if( ch >= 0xf0) {
+            *q++ = *p++;
         }
-        *out++ = ch;
         count++;
     }
-    *out = 0;
-    return count;
+    *q = 0;
+    msi_free(enc_utf8);
+
+    /* convert UTF8 to WCHAR */
+    MultiByteToWideChar( CP_UTF8, 0, dec_utf8, -1, out, count + 1 );
+    msi_free(dec_utf8);
 }
 
 void enum_stream_names( IStorage *stg )