Bugzilla 52784 - replace ISO control characters with question marks in SXSSF to be consistent with XSSF

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1294657 13f79535-47bb-0310-9956-ffa450edef68

1 files changed, 9 insertions, 8 deletions

diff --git a/src/ooxml/java/org/apache/poi/xssf/streaming/SheetDataWriter.java b/src/ooxml/java/org/apache/poi/xssf/streaming/SheetDataWriter.java
index d575e05324..c9a34a4630 100644
--- a/src/ooxml/java/org/apache/poi/xssf/streaming/SheetDataWriter.java
+++ b/src/ooxml/java/org/apache/poi/xssf/streaming/SheetDataWriter.java
@@ -250,6 +250,7 @@ public class SheetDataWriter {
                     break;

                 // Special characters

                 case '\n':

+                case '\r':

                     if (counter > last) {

                         _out.write(chars, last, counter - last);

                     }

@@ -263,13 +264,6 @@ public class SheetDataWriter {
                     _out.write("	");

                     last = counter + 1;

                     break;

-                case '\r':

-                    if (counter > last) {

-                        _out.write(chars, last, counter - last);

-                    }

-                    _out.write("
");

-                    last = counter + 1;

-                    break;

                 case 0xa0:

                     if (counter > last) {

                         _out.write(chars, last, counter - last);

@@ -278,7 +272,14 @@ public class SheetDataWriter {
                     last = counter + 1;

                     break;

                 default:

-                    if (c < ' ' || c > 127) {

+                    // YK: XmlBeans silently replaces all ISO control characters ( < 32) with question marks.

+                    // the same rule applies to unicode surrogates and "not a character" symbols.

+                    if( c < ' ' || Character.isLowSurrogate(c) || Character.isHighSurrogate(c) ||

+                            ('\uFFFE' <= c && c <= '\uFFFF')) {

+                        _out.write('?');

+                        last = counter + 1;

+                    }

+                    else if (c > 127) {

                         if (counter > last) {

                             _out.write(chars, last, counter - last);

                         }