PdfContentsTokenizer::PdfContentsTokenizer( PdfCanvas* pCanvas )
: PdfTokenizer()
{
PdfObject* pContents = pCanvas->GetContents();
if( pContents && pContents->IsArray() )
{
PdfArray& a = pContents->GetArray();
for ( PdfArray::iterator it = a.begin(); it != a.end() ; ++it )
{
if ( !(*it).IsReference() )
{
PODOFO_RAISE_ERROR_INFO( ePdfError_InvalidDataType, "/Contents array contained non-references" );
}
m_lstContents.push_back( pContents->GetOwner()->GetObject( (*it).GetReference() ) );
}
}
else if ( pContents && pContents->HasStream() )
{
m_lstContents.push_back( pContents );
}
else
{
PODOFO_RAISE_ERROR_INFO( ePdfError_InvalidDataType, "Page /Contents not stream or array of streams" );
}
if( m_lstContents.size() )
{
SetCurrentContentsStream( m_lstContents.front() );
m_lstContents.pop_front();
}
}
PDFFont PDFAnalyzer::getFontInfo(PdfObject* fontObj)
{
PDFFont currFont;
PdfObject* subtype = fontObj->GetIndirectKey("Subtype");
if (subtype && subtype->IsName())
{
PdfObject* fontDesc = fontObj->GetIndirectKey("FontDescriptor");
if (subtype->GetName() == "Type1")
currFont.fontType = F_Type1;
else if (subtype->GetName() == "MMType1")
currFont.fontType = F_MMType1;
else if (subtype->GetName() == "TrueType")
currFont.fontType = F_TrueType;
else if (subtype->GetName() == "Type3")
{
currFont.fontType = F_Type3;
currFont.isEmbedded = true;
fontDesc = NULL;
}
else if (subtype->GetName() == "Type0")
{
PdfObject* descendantFonts = fontObj->GetIndirectKey("DescendantFonts");
if (descendantFonts && descendantFonts->IsArray())
{
PdfObject descendantFont = descendantFonts->GetArray()[0];
descendantFont.SetOwner(descendantFonts->GetOwner());
PdfObject* subtypeDescFont = descendantFont.GetIndirectKey("Subtype");
fontDesc = descendantFont.MustGetIndirectKey("FontDescriptor");
if (subtypeDescFont && subtypeDescFont->IsName())
{
if (subtypeDescFont->GetName() == "CIDFontType0")
currFont.fontType = F_CIDFontType0;
else if (subtypeDescFont->GetName() == "CIDFontType2")
currFont.fontType = F_CIDFontType2;
}
}
}
if (fontDesc)
{
PdfObject* fontFile = fontDesc->GetIndirectKey("FontFile");
PdfObject* fontFile2 = fontDesc->GetIndirectKey("FontFile2");
PdfObject* fontFile3 = fontDesc->GetIndirectKey("FontFile3");
if (fontFile && fontFile->HasStream())
currFont.isEmbedded = true;
if (fontFile2 && fontFile2->HasStream())
currFont.isEmbedded = true;
if (fontFile3 && fontFile3->HasStream())
{
currFont.isEmbedded = true;
PdfObject* ff3Subtype = fontFile3->GetIndirectKey("Subtype");
if (ff3Subtype && ff3Subtype->IsName() && ff3Subtype->GetName() == "OpenType")
currFont.isOpenType = true;
}
}
}
return currFont;
}
PdfContentsTokenizer::PdfContentsTokenizer( PdfCanvas* pCanvas )
: PdfTokenizer(), m_readingInlineImgData(false)
{
if( !pCanvas )
{
PODOFO_RAISE_ERROR( ePdfError_InvalidHandle );
}
PdfObject* pContents = pCanvas->GetContents();
if( pContents && pContents->IsArray() )
{
PdfArray& a = pContents->GetArray();
for ( PdfArray::iterator it = a.begin(); it != a.end() ; ++it )
{
if ( !(*it).IsReference() )
{
PODOFO_RAISE_ERROR_INFO( ePdfError_InvalidDataType, "/Contents array contained non-references" );
}
if ( !pContents->GetOwner()->GetObject( (*it).GetReference() ) )
{
// some damaged PDFs may have dangling references
PODOFO_RAISE_ERROR_INFO( ePdfError_InvalidDataType, "/Contents array NULL reference" );
}
m_lstContents.push_back( pContents->GetOwner()->GetObject( (*it).GetReference() ) );
}
}
else if ( pContents && pContents->HasStream() )
{
m_lstContents.push_back( pContents );
}
else if ( pContents && pContents->IsDictionary() )
{
m_lstContents.push_back( pContents );
PdfError::LogMessage(eLogSeverity_Information,
"PdfContentsTokenizer: found canvas-dictionary without stream => empty page");
// OC 18.09.2010 BugFix: Found an empty page in a PDF document:
// 103 0 obj
// <<
// /Type /Page
// /MediaBox [ 0 0 595 842 ]
// /Parent 3 0 R
// /Resources <<
// /ProcSet [ /PDF ]
// >>
// /Rotate 0
// >>
// endobj
}
else
{
PODOFO_RAISE_ERROR_INFO( ePdfError_InvalidDataType, "Page /Contents not stream or array of streams" );
}
if( m_lstContents.size() )
{
SetCurrentContentsStream( m_lstContents.front() );
m_lstContents.pop_front();
}
}