Some more work on VSD5 parser

7 files changed, 105 insertions, 25 deletions

diff --git a/src/lib/VSD11Parser.h b/src/lib/VSD11Parser.h
index 3557af9..bf9155a 100644
--- a/src/lib/VSD11Parser.h
+++ b/src/lib/VSD11Parser.h
@@ -47,7 +47,7 @@ public:
   explicit VSD11Parser(WPXInputStream *input, libwpg::WPGPaintInterface *painter);
   ~VSD11Parser();
 private:
-  bool getChunkHeader(WPXInputStream *input);
+  virtual bool getChunkHeader(WPXInputStream *input);
   void readText(WPXInputStream *input);
   void readCharIX(WPXInputStream *input);
   void readParaIX(WPXInputStream *input);
diff --git a/src/lib/VSD5Parser.cpp b/src/lib/VSD5Parser.cpp
index a29834a..1662e85 100644
--- a/src/lib/VSD5Parser.cpp
+++ b/src/lib/VSD5Parser.cpp
@@ -55,5 +55,69 @@ void libvisio::VSD5Parser::readPointer(WPXInputStream *input, Pointer &ptr)
   ptr.Length = readU32(input);
 }
 
+void libvisio::VSD5Parser::readPointerInfo(WPXInputStream *input, unsigned ptrType, unsigned shift, unsigned &listSize, unsigned &pointerCount)
+{
+  VSD_DEBUG_MSG(("VSD5Parser::readPointerInfo\n"));
+  input->seek(shift+0x6, WPX_SEEK_SET);
+  listSize = readU16(input);
+  switch (ptrType)
+  {
+  case VSD_TRAILER_STREAM:
+    input->seek(shift+0x82, WPX_SEEK_SET);
+    break;
+  case VSD_STENCILS:
+    input->seek(shift+0x1e, WPX_SEEK_SET);
+    break;
+  case VSD_STENCIL_PAGE:
+    input->seek(shift+0x36, WPX_SEEK_SET);
+    break;
+  default:
+    input->seek(shift+0xa, WPX_SEEK_SET);
+    break;
+  }
+  pointerCount = readU16(input);
+  VSD_DEBUG_MSG(("VSD5Parser::readPointerInfo ptrType %u shift %u listSize 0x%x pointerCount 0x%x\n", ptrType, shift, listSize, pointerCount));
+}
+
+bool libvisio::VSD5Parser::getChunkHeader(WPXInputStream *input)
+{
+  unsigned char tmpChar = 0;
+  while (!input->atEOS() && !tmpChar)
+    tmpChar = readU8(input);
+
+  if (input->atEOS())
+    return false;
+  else
+    input->seek(-1, WPX_SEEK_CUR);
+
+  m_header.chunkType = readU16(input);
+  m_header.id = readU16(input);
+  m_header.level = readU8(input);
+  m_header.unknown = readU8(input);
+
+  m_header.list = readU16(input);
+
+  // Certain chunk types seem to always have a trailer
+  m_header.trailer = 0;
+  if (m_header.list != 0 || m_header.chunkType == 0x76 || m_header.chunkType == 0x73 ||
+      m_header.chunkType == 0x72 || m_header.chunkType == 0x71 || m_header.chunkType == 0x70 ||
+      m_header.chunkType == 0x6f || m_header.chunkType == 0x6e || m_header.chunkType == 0x6d ||
+      m_header.chunkType == 0x6c || m_header.chunkType == 0x6b || m_header.chunkType == 0x6a ||
+      m_header.chunkType == 0x69 || m_header.chunkType == 0x68 || m_header.chunkType == 0x67 ||
+      m_header.chunkType == 0x66 || m_header.chunkType == 0x65 || m_header.chunkType == 0x64 ||
+      m_header.chunkType == 0x2c || m_header.chunkType == 0xd)
+    m_header.trailer += 8; // 8 byte trailer
+
+  m_header.dataLength = readU32(input);
+
+  // 0x1f (OLE data) and 0xc9 (Name ID) never have trailer
+  if (m_header.chunkType == 0x1f || m_header.chunkType == 0xc9)
+  {
+    m_header.trailer = 0;
+  }
+  return true;
+}
+
+
 
 /* vim:set shiftwidth=2 softtabstop=2 expandtab: */
diff --git a/src/lib/VSD5Parser.h b/src/lib/VSD5Parser.h
index e3caeb2..2eff54d 100644
--- a/src/lib/VSD5Parser.h
+++ b/src/lib/VSD5Parser.h
@@ -49,6 +49,8 @@ public:
 
 protected:
   virtual void readPointer(WPXInputStream *input, Pointer &ptr);
+  virtual bool getChunkHeader(WPXInputStream *input);
+  virtual void readPointerInfo(WPXInputStream *input, unsigned ptrType, unsigned shift, unsigned &listSize, unsigned &pointerCount);
 
 private:
   VSD5Parser();
diff --git a/src/lib/VSD6Parser.h b/src/lib/VSD6Parser.h
index 6c403b5..2615abe 100644
--- a/src/lib/VSD6Parser.h
+++ b/src/lib/VSD6Parser.h
@@ -46,8 +46,9 @@ class VSD6Parser : public VSDParser
 public:
   explicit VSD6Parser(WPXInputStream *input, libwpg::WPGPaintInterface *painter);
   ~VSD6Parser();
+protected:
+  virtual bool getChunkHeader(WPXInputStream *input);
 private:
-  bool getChunkHeader(WPXInputStream *input);
   void readText(WPXInputStream *input);
   void readCharIX(WPXInputStream *input);
   void readParaIX(WPXInputStream *input);
diff --git a/src/lib/VSDDocumentStructure.h b/src/lib/VSDDocumentStructure.h
index 774fe29..77eba00 100644
--- a/src/lib/VSDDocumentStructure.h
+++ b/src/lib/VSDDocumentStructure.h
@@ -35,6 +35,7 @@
 #define VSD_OLE_LIST 0x0d
 #define VSD_TEXT 0x0e
 
+#define VSD_TRAILER_STREAM 0x14
 #define VSD_PAGE 0x15
 #define VSD_COLORS 0x16
 #define VSD_FONT_LIST 0x18
diff --git a/src/lib/VSDParser.cpp b/src/lib/VSDParser.cpp
index 99f8860..971777a 100644
--- a/src/lib/VSDParser.cpp
+++ b/src/lib/VSDParser.cpp
@@ -62,14 +62,15 @@ bool libvisio::VSDParser::parseMain()
   // Seek to trailer stream pointer
   m_input->seek(0x24, WPX_SEEK_SET);
 
-  m_input->seek(8, WPX_SEEK_CUR);
-  unsigned offset = readU32(m_input);
-  unsigned length = readU32(m_input);
-  unsigned short format = readU16(m_input);
-  bool compressed = ((format & 2) == 2);
+  Pointer trailerPointer;
+  readPointer(m_input, trailerPointer);
+  bool compressed = ((trailerPointer.Format & 2) == 2);
+  unsigned shift = 0;
+  if (compressed)
+    shift = 4;
 
-  m_input->seek(offset, WPX_SEEK_SET);
-  VSDInternalStream trailerStream(m_input, length, compressed);
+  m_input->seek(trailerPointer.Offset, WPX_SEEK_SET);
+  VSDInternalStream trailerStream(m_input, trailerPointer.Length, compressed);
 
   std::vector<std::map<unsigned, XForm> > groupXFormsSequence;
   std::vector<std::map<unsigned, unsigned> > groupMembershipsSequence;
@@ -78,7 +79,7 @@ bool libvisio::VSDParser::parseMain()
   VSDStylesCollector stylesCollector(groupXFormsSequence, groupMembershipsSequence, documentPageShapeOrders);
   m_collector = &stylesCollector;
   VSD_DEBUG_MSG(("VSDParser::parseMain 1st pass\n"));
-  if (!parseDocument(&trailerStream))
+  if (!parseDocument(&trailerStream, shift))
     return false;
 
   _handleLevelChange(0);
@@ -88,17 +89,17 @@ bool libvisio::VSDParser::parseMain()
   VSDContentCollector contentCollector(m_painter, groupXFormsSequence, groupMembershipsSequence, documentPageShapeOrders, styles, m_stencils);
   m_collector = &contentCollector;
   VSD_DEBUG_MSG(("VSDParser::parseMain 2nd pass\n"));
-  if (!parseDocument(&trailerStream))
+  if (!parseDocument(&trailerStream, shift))
     return false;
 
   return true;
 }
 
-bool libvisio::VSDParser::parseDocument(WPXInputStream *input)
+bool libvisio::VSDParser::parseDocument(WPXInputStream *input, unsigned shift)
 {
   try
   {
-    handleStreams(input, 4, 0);
+    handleStreams(input, VSD_TRAILER_STREAM, shift, 0);
     return true;
   }
   catch (...)
@@ -122,8 +123,20 @@ void libvisio::VSDParser::readPointer(WPXInputStream *input, Pointer &ptr)
   ptr.Format = readU16(input);
 }
 
-void libvisio::VSDParser::handleStreams(WPXInputStream *input, unsigned shift, unsigned level)
+void libvisio::VSDParser::readPointerInfo(WPXInputStream *input, unsigned /* ptrType */, unsigned shift, unsigned &listSize, unsigned &pointerCount)
 {
+  VSD_DEBUG_MSG(("VSDParser::readPointerInfo\n"));
+  input->seek(shift, WPX_SEEK_SET);
+  unsigned offset = readU32(input);
+  input->seek(offset+shift-4, WPX_SEEK_SET);
+  listSize = readU32(input);
+  pointerCount = readU32(input);
+  input->seek(4, WPX_SEEK_CUR);
+}
+
+void libvisio::VSDParser::handleStreams(WPXInputStream *input, unsigned ptrType, unsigned shift, unsigned level)
+{
+  VSD_DEBUG_MSG(("VSDParser::HandleStreams\n"));
   std::vector<unsigned> pointerOrder;
   std::map<unsigned, libvisio::Pointer> PtrList;
   std::map<unsigned, libvisio::Pointer> FontFaces;
@@ -132,20 +145,18 @@ void libvisio::VSDParser::handleStreams(WPXInputStream *input, unsigned shift, u
   try
   {
     // Parse out pointers to streams
-    input->seek(shift, WPX_SEEK_SET);
-    unsigned offset = readU32(input);
-    input->seek(offset+shift-4, WPX_SEEK_SET);
-    unsigned listSize = readU32(input);
-    unsigned pointerCount = readU32(input);
-    input->seek(4, WPX_SEEK_CUR);
+    unsigned listSize = 0;
+    unsigned pointerCount = 0;
+    readPointerInfo(input, ptrType, shift, listSize, pointerCount);
     for (i = 0; i < pointerCount; i++)
     {
       Pointer ptr;
-	  readPointer(input, ptr);
+      readPointer(input, ptr);
       if (ptr.Type == VSD_FONTFACES)
         FontFaces[i] = ptr;
       else if (ptr.Type != 0)
         PtrList[i] = ptr;
+      VSD_DEBUG_MSG(("--> Pointer #%u\n", i));
     }
     for (i = 0; i < listSize; ++i)
       pointerOrder.push_back(readU32(input));
@@ -178,9 +189,9 @@ void libvisio::VSDParser::handleStreams(WPXInputStream *input, unsigned shift, u
 
 }
 
-
 void libvisio::VSDParser::handleStream(const Pointer &ptr, unsigned idx, unsigned level)
 {
+  VSD_DEBUG_MSG(("VSDParser::HandleStream\n"));
   m_header.level = level;
   m_header.id = idx;
   m_header.chunkType = ptr.Type;
@@ -248,7 +259,7 @@ void libvisio::VSDParser::handleStream(const Pointer &ptr, unsigned idx, unsigne
     if (ptr.Length > 4)
       handleBlob(&tmpInput, level+1);
     if ((ptr.Format >> 4) == 0x5 && ptr.Type != VSD_COLORS)
-      handleStreams(&tmpInput, shift, level+1);
+      handleStreams(&tmpInput, ptr.Type, shift, level+1);
   }
   else if ((ptr.Format >> 4) == 0xd || (ptr.Format >> 4) == 0x8)
     handleChunks(&tmpInput, level+1);
diff --git a/src/lib/VSDParser.h b/src/lib/VSDParser.h
index f6ddd49..a4594e9 100644
--- a/src/lib/VSDParser.h
+++ b/src/lib/VSDParser.h
@@ -124,16 +124,17 @@ protected:
   void readOLEData(WPXInputStream *input);
 
   // parser of one pass
-  bool parseDocument(WPXInputStream *input);
+  bool parseDocument(WPXInputStream *input, unsigned shift);
 
   // Stream handlers
-  void handleStreams(WPXInputStream *input, unsigned shift, unsigned level);
+  void handleStreams(WPXInputStream *input, unsigned ptrType, unsigned shift, unsigned level);
   void handleStream(const Pointer &ptr, unsigned idx, unsigned level);
   void handleChunks(WPXInputStream *input, unsigned level);
   void handleChunk(WPXInputStream *input);
   void handleBlob(WPXInputStream *input, unsigned level);
 
   virtual void readPointer(WPXInputStream *input, Pointer &ptr);
+  virtual void readPointerInfo(WPXInputStream *input, unsigned ptrType, unsigned shift, unsigned &listSize, unsigned &pointerCount);
   virtual bool getChunkHeader(WPXInputStream *input) = 0;
   void _handleLevelChange(unsigned level);
   Colour _colourFromIndex(unsigned idx);