Apply Felix Meschberger's patch to fix the high unicode issue

kobjects · Aug 18, 2017 · 49758b4 · 49758b4
1 parent f9aba22
commit 49758b4
Show file tree

Hide file tree

Showing 2 changed files with 18 additions and 4 deletions.
diff --git a/src/org/kxml2/io/KXmlParser.java b/src/org/kxml2/io/KXmlParser.java
@@ -595,7 +595,14 @@ private final void push(int c) {
             txtBuf = bigger;
         }
 
-        txtBuf[txtPos++] = (char) c;
+	if (c > 0xffff) {
+            // write high Unicode value as surrogate pair
+            int offset = c - 0x010000;
+            txtBuf[txtPos++] = (char)((offset >>> 10) + 0xd800);
+            txtBuf[txtPos++] = (char)((offset & 0x3ff) + 0xdc00);
+        } else {
+            txtBuf[txtPos++] = (char) c;
+        }
     }
 
     /** Sets name and attributes */

diff --git a/src/org/kxml2/io/KXmlSerializer.java b/src/org/kxml2/io/KXmlSerializer.java
@@ -119,11 +119,18 @@ private final void writeEscaped(String s, int quot)
                 	//if(c < ' ')
 					//	throw new IllegalArgumentException("Illegal control code:"+((int) c));
 
-                    if (c >= ' ' && c !='@' && (c < 127 || unicode))
+                    if (c >= 0xd800 && c <= 0xdfff && i < s.length() - 1) {
+                        // write surrogate pair as single code value
+                        i++;
+                        int h = c;
+                        int l = s.charAt(i);
+                        int n = ((h - 0xd800) << 10) + (l - 0xdc00) + 0x010000;
+                        writer.write("&#" + n + ";");
+                    } if (c >= ' ' && c !='@' && (c < 127 || unicode)) {
                         writer.write(c);
-                    else
+                    } else {
                         writer.write("&#" + ((int) c) + ";");
-
+                    }
             }
         }
     }