Fixes #2635, crash during gpx export

Added utf-16 codepoint surrogate handling to xml serializer

2 files changed, 15 insertions, 5 deletions

diff --git a/main/src/cgeo/geocaching/export/GpxExport.java b/main/src/cgeo/geocaching/export/GpxExport.java
index 8303137..c2a58b7 100644
--- a/main/src/cgeo/geocaching/export/GpxExport.java
+++ b/main/src/cgeo/geocaching/export/GpxExport.java
@@ -14,6 +14,7 @@ import cgeo.geocaching.geopoint.Geopoint;
 import cgeo.geocaching.utils.BaseUtils;
 import cgeo.geocaching.utils.Log;
 import cgeo.geocaching.utils.XmlUtils;
+import cgeo.org.kxml2.io.KXmlSerializer;
 
 import org.apache.commons.lang3.StringUtils;
 import org.xmlpull.v1.XmlSerializer;
@@ -27,7 +28,6 @@ import android.content.Intent;
 import android.net.Uri;
 import android.os.AsyncTask;
 import android.os.Environment;
-import android.util.Xml;
 import android.view.ContextThemeWrapper;
 import android.view.View;
 import android.widget.CheckBox;
@@ -138,7 +138,7 @@ class GpxExport extends AbstractExport {
                 final File exportLocation = new File(Settings.getGpxExportDir());
                 exportLocation.mkdirs();
 
-                final XmlSerializer gpx = Xml.newSerializer();
+                final XmlSerializer gpx = new KXmlSerializer();
                 writer = new FileWriter(exportFile);
                 gpx.setOutput(writer);
 
diff --git a/main/src/cgeo/org/kxml2/io/KXmlSerializer.java b/main/src/cgeo/org/kxml2/io/KXmlSerializer.java
index a9dde8c..027ff53 100644
--- a/main/src/cgeo/org/kxml2/io/KXmlSerializer.java
+++ b/main/src/cgeo/org/kxml2/io/KXmlSerializer.java
@@ -130,11 +130,21 @@ public class KXmlSerializer implements XmlSerializer {
                     }
                     // BEGIN android-changed: refuse to output invalid characters
                     // See http://www.w3.org/TR/REC-xml/#charsets for definition.
-                    // No other Java XML writer we know of does this, but no Java
-                    // XML reader we know of is able to parse the bad output we'd
-                    // otherwise generate.
+                    // Corrected for c:geo to handle utf-16 codepoint surrogates correctly
+                    // See http://en.wikipedia.org/wiki/UTF-16#Code_points_U.2B10000_to_U.2B10FFFF
                     // Note: tab, newline, and carriage return have already been
                     // handled above.
+                    // Check for lead surrogate
+                    if (c >= 0xd800 && c <= 0xdbff) {
+
+                        if (i + 1 < s.length()) {
+                            writer.write(s.substring(i, i + 1));
+                            i++;
+                            break;
+                        }
+                        // if the lead surrogate is at the string end, it's not valid utf-16
+                        reportInvalidCharacter(c);
+                    }
                     boolean valid = (c >= 0x20 && c <= 0xd7ff) || (c >= 0xe000 && c <= 0xfffd);
                     if (!valid) {
                         reportInvalidCharacter(c);