Fix tdf#93438 Impress generate abnormal style when copy-pasting

from Writer with Chinese UI. When user configure UI language as
Chinese, style names are also translated. However when Copy-pasting
from Writer to Impress ( select RTF fromat with paste-special),
Chinese (unicode) style names are borken into multiparts, where
some of them are empty string. Neither deleting the abnormal
style nor saving the file can be achieved. The only remedy is to
remove style with emtpy name with macro. With this patch:
1) Catch NoSuchElementException and ignores it, so
   user still has chance saving file.
2) Make sure style has valid number before inserting it.
3) Prevent text breaking into multiple tokens by handling  ucN in
 ScanText().

Change-Id: I417f70b81c23ac63c175cc13c548068873d13a38
Reviewed-on: https://gerrit.libreoffice.org/18148
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>
Reviewed-on: https://gerrit.libreoffice.org/19685

1 files changed, 19 insertions, 2 deletions

diff --git a/svtools/source/svrtf/parrtf.cxx b/svtools/source/svrtf/parrtf.cxx
index fc98b8781e60..e7be520573c0 100644
--- a/svtools/source/svrtf/parrtf.cxx
+++ b/svtools/source/svrtf/parrtf.cxx
@@ -431,6 +431,24 @@ void SvRTFParser::ScanText( const sal_Unicode cBreak )
                             aToken = sSave;
                             bRTF_InTextRead = false;
                         }
+                        else if ( 'c' == nNextCh )
+                        {
+                            // Prevent text breaking into multiple tokens.
+                            rInput.SeekRel( 2 );
+                            nNextCh = GetNextChar();
+                            if (RTF_ISDIGIT( nNextCh ))
+                            {
+                                sal_uInt8 nNewOverread = 0 ;
+                                do {
+                                    nNewOverread *= 10;
+                                    nNewOverread += nNextCh - '0';
+                                    nNextCh = GetNextChar();
+                                } while ( RTF_ISDIGIT( nNextCh ) );
+                                nUCharOverread = nNewOverread;
+                                aParserStates.top().nUCharOverread = nNewOverread;
+                            }
+                            bNextCh = 0x20 == nNextCh;
+                        }
                         else
                         {
                             nNextCh = '\\';
@@ -448,8 +466,7 @@ void SvRTFParser::ScanText( const sal_Unicode cBreak )
             }
             break;
 
-        case sal_Unicode(EOF):
-                eState = SVPAR_ERROR;
+        case sal_Unicode(EOF): eState = SVPAR_ERROR;
                 // continue
         case '{':
         case '}':