Changeset 453 for branches/thune/thread_safe/make.c

Timestamp:

08/25/07 22:49:09 (15 months ago)

Author:

krobillard

Message:

UTF-8 strings can now be read.
ur_makeString() now takes UCell argument.

Files:

• branches/thune/thread_safe/make.c (modified) (9 diffs)

branches/thune/thread_safe/make.c

@@ -1 +1 @@
 /*============================================================================
     Thune Interpreter
-    Copyright (C) 2005-2006  Karl Robillard
+    Copyright (C) 2005-2007  Karl Robillard
 
     This library is free software; you can redistribute it and/or
@@ -335 +335 @@
 
 
-#ifndef UR_CONFIG_UDS
 /*
    Create context from initializer block.
@@ -458 +457 @@
     }
 }
-#endif
 
 
@@ -614 +612 @@
 
 
+/*
+   Convert UTF-8 to UCS-2
+*/
+static void _makeString16( UBinary* str, const uint8_t* cp, int len )
+{
+    uint16_t  ch;
+    uint16_t* out;
+    const uint8_t* end;
+
+    ur_arrayReserve( str, 1, len * sizeof(uint16_t) );
+    out = str->ptr.u16;
+
+    end = cp + len;
+    while( cp != end )
+    {
+        ch = *cp++;
+
+        if( ch <= 0x7f )
+        {
+            *out++ = ch;
+        }
+        else if( ch >= 0xc2 && ch <= 0xdf )
+        {
+            if( cp != end )
+            {
+                *out++ = ((ch & 0x1f) << 6) | (*cp & 0x3f);
+                ++cp;
+            }
+        }
+        else if( ch >= 0xe0 && ch <= 0xef )
+        {
+            if( (end - cp) < 2 )
+                break;
+            *out++ = ((ch    & 0x0f) << 12) |
+                     ((cp[0] & 0x3f) <<  6) |
+                      (cp[1] & 0x3f);
+            cp += 2;
+        }
+        else if( ch >= 0xf0 && ch <= 0xf3 )
+        {
+            if( (end - cp) < 3 )
+                break;
+            *out++ = 0;     // Only handle UCS-2
+            cp += 3;
+        }
+    }
+
+    str->avail /= 2;
+    str->used = out - str->ptr.u16;
+}
+
+
 /**
   If len is less than zero then the length is automatically determined.
   Special characters are translated.
 */
-UIndex ur_makeStringT( UThread* ut, const char* txt, int len )
+UIndex ur_makeStringT( UThread* ut, UCell* res, const char* txt, int len )
 {
     UIndex strN;
@@ -625 +675 @@
     const char* end;
     char* out;
+    int ch;
 
     if( len == 0 )
     {
-        return ur_makeBinary( 0, 0 );
+        strN = ur_makeBinary( 0, 0 );
+        goto init;
     }
 
@@ -653 +705 @@
         else
         {
-            *out++ = *cp++;
+            ch = *cp++;
+            if( ((unsigned int) ch) > 0x7f )
+            {
+                _makeString16( str, (const uint8_t*) txt, len );
+                ur_initString( res, strN, 0 );
+                ur_setEncoding( res, UR_ENC_UTF16 );
+                return strN;
+            }
+            *out++ = ch;
         }
     }
 
     str->used = out - str->ptr.c;
-
     str->ptr.c[ str->used ] = '\0';
+
+init:
+
+    ur_initString( res, strN, 0 );
     return strN;
 }
@@ -1326 +1389 @@
 
 
-#ifndef UR_CONFIG_UDS
 /*
   (datatype! proto -- value)
@@ -1720 +1782 @@
             {
                 if( spA )
-                    binN = ur_makeString( spA, spB - spA );
+                {
+                    binN = ur_makeString( res, spA, spB - spA );
+                }
                 else
+                {
                     binN = ur_makeBinary( 0, 0 );
 init_str:
-                ur_initString( res, binN, 0 );
+                    ur_initString( res, binN, 0 );
+                }
             }
         }
@@ -1871 +1937 @@
     ur_throwErr( UR_ERR_DATATYPE, "Invalid make values" );
 }
-#endif