disentangle 16bit pascal-strings and zero terminated pstrings, etc.

one method to read pascal strings, zero-terminated pascal strings
and non-zero terminated non-pascal strings is a disaster when
length of 0 toggle into pascal-string mode, shudder!

6 files changed, 35 insertions, 47 deletions

diff --git a/sw/source/filter/ww8/ww8graf2.cxx b/sw/source/filter/ww8/ww8graf2.cxx
index cb40a02d9b94..a382f4923d2a 100644
--- a/sw/source/filter/ww8/ww8graf2.cxx
+++ b/sw/source/filter/ww8/ww8graf2.cxx
@@ -269,7 +269,7 @@ bool SwWW8ImplReader::ReadGrafFile(String& rFileName, Graphic*& rpGraphic,
         case 99: // TIFF-File ( nicht embeddet )
             pSt->Seek(nPosFc);
             // Name als P-String einlesen
-            rFileName = WW8ReadPascalString(*pSt, eStructCharSet);
+            rFileName = read_uInt8_PascalString(*pSt, eStructCharSet);
             if (rFileName.Len())
                 rFileName = URIHelper::SmartRel2Abs(
                     INetURLObject(sBaseURL), rFileName,
diff --git a/sw/source/filter/ww8/ww8par.cxx b/sw/source/filter/ww8/ww8par.cxx
index 703e1db3bbb9..c708a2ce50ae 100644
--- a/sw/source/filter/ww8/ww8par.cxx
+++ b/sw/source/filter/ww8/ww8par.cxx
@@ -5337,13 +5337,13 @@ const String* SwWW8ImplReader::GetAnnotationAuthor(sal_uInt16 nIdx)
         {
             if( bVer67 )
             {
-                mpAtnNames->push_back(WW8ReadPascalString(rStrm,
+                mpAtnNames->push_back(read_uInt8_PascalString(rStrm,
                     RTL_TEXTENCODING_MS_1252));
                 nRead += mpAtnNames->rbegin()->Len() + 1;   // Laenge + sal_uInt8 Count
             }
             else
             {
-                mpAtnNames->push_back(WW8Read_xstz(rStrm, 0, false));
+                mpAtnNames->push_back(read_LEuInt16_PascalString(rStrm));
                 // UNICode: doppelte Laenge + sal_uInt16 Count
                 nRead += mpAtnNames->rbegin()->Len() * 2 + 2;
             }
diff --git a/sw/source/filter/ww8/ww8par3.cxx b/sw/source/filter/ww8/ww8par3.cxx
index 4094d00a433f..41df2b0da259 100644
--- a/sw/source/filter/ww8/ww8par3.cxx
+++ b/sw/source/filter/ww8/ww8par3.cxx
@@ -702,7 +702,7 @@ bool WW8ListManager::ReadLVL(SwNumFmt& rNumFmt, SfxItemSet*& rpItemSet,
     //
     // 4. den Nummerierungsstring einlesen: ergibt Prefix und Postfix
     //
-    String sNumString(WW8Read_xstz(rSt, 0, false));
+    String sNumString(read_LEuInt16_PascalString(rSt));
 
     //
     // 5. gelesene Werte in Writer Syntax umwandeln
@@ -2225,7 +2225,7 @@ void WW8FormulaControl::FormulaRead(SwWw8ControlType nWhich,
         maListEntries.reserve(nNoStrings);
         for (sal_uInt32 nI = 0; nI < nNoStrings; ++nI)
         {
-            String sEntry =  WW8Read_xstz(*pDataStream, 0, false);
+            String sEntry =  read_LEuInt16_PascalString(*pDataStream);
             maListEntries.push_back(sEntry);
         }
     }
diff --git a/sw/source/filter/ww8/ww8scan.cxx b/sw/source/filter/ww8/ww8scan.cxx
index 5f4f714ddd0e..babb3319c364 100644
--- a/sw/source/filter/ww8/ww8scan.cxx
+++ b/sw/source/filter/ww8/ww8scan.cxx
@@ -1900,35 +1900,32 @@ Err:
     return false;
 }
 
-String WW8ReadPascalString(SvStream& rStrm, rtl_TextEncoding eEnc)
+String read_uInt8_PascalString(SvStream& rStrm, rtl_TextEncoding eEnc)
 {
-    sal_uInt8 b(0);
-    rStrm >> b;
-    return rtl::OStringToOUString(read_uInt8s_AsOString(rStrm, b), eEnc);
+    sal_uInt8 nLen(0);
+    rStrm >> nLen;
+    return rtl::OStringToOUString(read_uInt8s_AsOString(rStrm, nLen), eEnc);
 }
 
-String WW8ReadBeltAndBracesString(SvStream& rStrm, rtl_TextEncoding eEnc)
+String read_LEuInt16_PascalString(SvStream& rStrm)
 {
-    String aRet = WW8ReadPascalString(rStrm, eEnc);
-    rStrm.SeekRel( 1 ); // skip null-byte at end
-    return aRet;
+    sal_uInt16 nLen(0);
+    rStrm >> nLen;
+    return read_LEuInt16s_AsOUString(rStrm, nLen);
 }
 
-String WW8Read_xstz(SvStream& rStrm, sal_uInt16 nChars, bool bAtEndSeekRel1)
+String read_uInt8_BeltAndBracesString(SvStream& rStrm, rtl_TextEncoding eEnc)
 {
-    sal_uInt16 b(0);
-
-    if( nChars )
-        b = nChars;
-    else
-        rStrm >> b;
-
-    rtl::OUString sRet = read_LEuInt16s_AsOUString(rStrm, nChars);
-
-    if( bAtEndSeekRel1 )
-        rStrm.SeekRel( 2 ); // ueberspringe das Null-Character am Ende.
+    String aRet = read_uInt8_PascalString(rStrm, eEnc);
+    rStrm.SeekRel(sizeof(sal_uInt8)); // skip null-byte at end
+    return aRet;
+}
 
-    return sRet;
+String read_LEuInt16_BeltAndBracesString(SvStream& rStrm)
+{
+    String aRet = read_LEuInt16_PascalString(rStrm);
+    rStrm.SeekRel(sizeof(sal_Unicode)); // skip null-byte at end
+    return aRet;
 }
 
 xub_StrLen WW8ScannerBase::WW8ReadString( SvStream& rStrm, String& rStr,
@@ -1961,7 +1958,7 @@ xub_StrLen WW8ScannerBase::WW8ReadString( SvStream& rStrm, String& rStr,
             nLen = USHRT_MAX - 1;
 
         if( bIsUnicode )
-            rStr.Append(WW8Read_xstz(rStrm, (sal_uInt16)nLen, false));
+            rStr.Append(String(read_LEuInt16s_AsOUString(rStrm, nLen)));
         else
         {
             // Alloc method automatically sets Zero at the end
@@ -3864,7 +3861,7 @@ void WW8ReadSTTBF(bool bVer8, SvStream& rStrm, sal_uInt32 nStart, sal_Int32 nLen
             for (sal_uInt16 i=0; i < nStrings; ++i)
             {
                 if (bUnicode)
-                    rArray.push_back(WW8Read_xstz(rStrm, 0, false));
+                    rArray.push_back(read_LEuInt16_PascalString(rStrm));
                 else
                 {
                     sal_uInt8 nBChar(0);
@@ -3897,7 +3894,7 @@ void WW8ReadSTTBF(bool bVer8, SvStream& rStrm, sal_uInt32 nStart, sal_Int32 nLen
                 for (sal_uInt16 i=0; i < nStrings; ++i)
                 {
                     if( bUnicode )
-                        pValueArray->push_back(WW8Read_xstz(rStrm, 0, false));
+                        pValueArray->push_back(read_LEuInt16_PascalString(rStrm));
                     else
                     {
                         sal_uInt8 nBChar(0);
@@ -6142,7 +6139,7 @@ WW8_STD* WW8Style::Read1Style( short& rSkip, String* pString, short* pcbStd )
                 case 6:
                 case 7:
                     // lies Pascal-String
-                    *pString = WW8ReadBeltAndBracesString(rSt, RTL_TEXTENCODING_MS_1252);
+                    *pString = read_uInt8_BeltAndBracesString(rSt, RTL_TEXTENCODING_MS_1252);
                     // leading len and trailing zero --> 2
                     rSkip -= 2+ pString->Len();
                     break;
@@ -6151,7 +6148,7 @@ WW8_STD* WW8Style::Read1Style( short& rSkip, String* pString, short* pcbStd )
                     // trailing zero
                     if (ww8String::TestBeltAndBraces(rSt))
                     {
-                        *pString = WW8Read_xstz(rSt, 0, true);
+                        *pString = read_LEuInt16_BeltAndBracesString(rSt);
                         rSkip -= (pString->Len() + 2) * 2;
                     }
                     else
@@ -6166,7 +6163,7 @@ WW8_STD* WW8Style::Read1Style( short& rSkip, String* pString, short* pcbStd )
                         they are not corrupt. If they are then we try them as
                         8bit ones
                         */
-                        *pString = WW8ReadBeltAndBracesString(rSt,RTL_TEXTENCODING_MS_1252);
+                        *pString = read_uInt8_BeltAndBracesString(rSt,RTL_TEXTENCODING_MS_1252);
                         // leading len and trailing zero --> 2
                         rSkip -= 2+ pString->Len();
                     }
diff --git a/sw/source/filter/ww8/ww8scan.hxx b/sw/source/filter/ww8/ww8scan.hxx
index 5118eca44a24..7d732c81b4e7 100644
--- a/sw/source/filter/ww8/ww8scan.hxx
+++ b/sw/source/filter/ww8/ww8scan.hxx
@@ -152,12 +152,14 @@ public:
 
 //Read a Pascal-style, i.e. single byte string length followed
 //by string contents
-String WW8ReadPascalString(SvStream& rStrm, rtl_TextEncoding eEnc);
+String read_uInt8_PascalString(SvStream& rStrm, rtl_TextEncoding eEnc);
+String read_LEuInt16_PascalString(SvStream& rStrm);
 
 //Belt and Braces strings, i.e. Pascal-style strings followed by
 //null termination, Spolsky calls them "fucked strings" FWIW
 //http://www.joelonsoftware.com/articles/fog0000000319.html
-String WW8ReadBeltAndBracesString(SvStream& rStrm, rtl_TextEncoding eEnc);
+String read_uInt8_BeltAndBracesString(SvStream& rStrm, rtl_TextEncoding eEnc);
+String read_LEuInt16_BeltAndBracesString(SvStream& rStrm);
 
 //--Line abovewhich the code has meaningful comments
 
@@ -168,16 +170,6 @@ struct WW8PLCFxDesc;
 class  WW8PLCFx_PCD;
 
 /**
- Reads a UTF-16 unicode string. If nLen is set then that number of 16bit units
- (not bytes) is read, if its not set, then read a 16bit count from the
- stream and use that for the length.
-
- If bAtEndSeekRel1 is set, exactly ONE extra unit (not byte) is skipped at the
- end of the string.
-*/
-String WW8Read_xstz(SvStream& rStrm, sal_uInt16 nChars, bool bAtEndSeekRel1);
-
-/**
  reads array of strings (see MS documentation: STring TaBle stored in File)
  returns NOT the original pascal strings but an array of converted char*
 
diff --git a/sw/source/filter/ww8/ww8toolbar.cxx b/sw/source/filter/ww8/ww8toolbar.cxx
index 1147e9e6b010..3f2c333a07d7 100644
--- a/sw/source/filter/ww8/ww8toolbar.cxx
+++ b/sw/source/filter/ww8/ww8toolbar.cxx
@@ -29,6 +29,7 @@
  * instead of those above.
  */
 #include "ww8toolbar.hxx"
+#include "ww8scan.hxx"
 #include <rtl/ustrbuf.hxx>
 #include <stdarg.h>
 #include <com/sun/star/ui/XUIConfigurationPersistence.hpp>
@@ -786,9 +787,7 @@ Xst::Read( SvStream& rS )
 {
     OSL_TRACE("Xst::Read() stream pos 0x%x", rS.Tell() );
     nOffSet = rS.Tell();
-    sal_Int16 nChars = 0;
-    rS >> nChars;
-    sString = read_LEuInt16s_AsOUString(rS, nChars);
+    sString = read_LEuInt16_PascalString(rS);
     return true;
 }