void linkLayer::loadFromDoc(PoDoFo::PdfMemDocument* doc) {
qDebug() << "linkLayer: Loading named destinations ...";
try {
PoDoFo::PdfNamesTree* pNames = doc->GetNamesTree( PoDoFo::ePdfDontCreateObject );
if( ! pNames ) return;
PoDoFo::PdfDictionary destsDict;
pNames->ToDictionary( PoDoFo::PdfName("Dests"), destsDict );
PoDoFo::TKeyMap keyMap = destsDict.GetKeys();
QString tName;
PoDoFo::PdfDestination *dest;
PoDoFo::PdfObject *obj;
for(PoDoFo::TKeyMap::const_iterator it = keyMap.begin(); it != keyMap.end(); ++it ) {
try {
tName = QString::fromUtf8( it->first.GetName().c_str() );
qDebug() << "Processing "<< tName;
obj = pdfUtil::resolveRefs( doc, it->second );
if ( obj->IsArray() ) dest = new PoDoFo::PdfDestination( obj );
else if ( obj->IsDictionary() ) {
obj->GetDictionary().GetKey("D")->SetOwner( &doc->GetObjects() );
dest = new PoDoFo::PdfDestination( obj->GetDictionary().GetKey("D") );
}
else {
qDebug() << "Element is neither an array, nor a dictionary:"<< obj->GetDataTypeString();
continue;
}
addTarget( tName, dest );
} catch ( PoDoFo::PdfError e ) {
qDebug() << "linkLayer: Error adding named destination ("<<tName<<"):"<<e.what();
}
}
} catch ( PoDoFo::PdfError e ) {
qDebug() << "linkLayer: Error processing names tree:" << e.what();
};
qDebug() << "linkLayer: Done loading named destinations.";
}
bool parsepdf(const char* filename) {
PoDoFo::PdfVecObjects objects;
PoDoFo::PdfParser parser(&objects, filename);
PoDoFo::TIVecObjects it = objects.begin();
bool result = false;
do {
PoDoFo::PdfObject *obj = (*it);
#if (PODOFO_VERSION_MAJOR > 0) || (PODOFO_VERSION_MINOR > 8) || (PODOFO_VERSION_PATCH >= 3)
if (obj->HasStream() && (obj->GetObjectLength(PoDoFo::ePdfWriteMode_Compact) > 10000)) {
#else
if (obj->HasStream() && (obj->GetObjectLength() > 10000)) {
#endif
PoDoFo::PdfStream *stream = obj->GetStream();
char *buffer;
PoDoFo::pdf_long bufferLen;
stream->GetFilteredCopy(&buffer, &bufferLen);
//std::cerr << "Buffer length : " << bufferLen << std::endl;
if (bufferLen > 1000)
result = parseStream(buffer, bufferLen);
free(buffer);
}
it++;
} while (it != objects.end());
return result;
}
int main(int argc, char** argv) {
if (argc != 2) {
std::cerr << "ERROR: wrong number of argument" << std::endl;
return -1;
} else {
parsepdf(argv[1]);
}
}
int main (int argc, char *argv[])
{
using namespace PoDoFo;
PoDoFo::PdfMemDocument *doc = NULL;
int result = 0;
try {
PoDoFo::PdfError::EnableDebug(false);
if (argc != 2 && argc != 4)
{
cout << "Syntax" << endl;
cout << " " << argv[0] << " <pdf file> - display the XMP in a file (use \"-\" to specify stdin)" << endl;
cout << "or" << endl;
cout << " " << argv[0] << " <src pdf file> <xmp file> <new pdf file> - create a new PDF with the XMP in" << endl;
return EXIT_FAILURE;
}
if ( string("-") == argv[1] )
{
cin >> std::noskipws;
#ifdef _MSC_VER
_setmode(_fileno(stdin), _O_BINARY); // @TODO: MSVC specific binary setmode -- not sure if other platforms need it
cin.sync_with_stdio();
#endif
istream_iterator<char> it(std::cin);
istream_iterator<char> end;
string buffer(it, end);
doc = new PoDoFo::PdfMemDocument();
doc->Load( buffer.c_str(), (long)buffer.size() );
}
else
{
doc = new PoDoFo::PdfMemDocument(argv[1]);
}
if (argc == 2)
{
PoDoFo::PdfObject *metadata;
if ((metadata = doc->GetMetadata()) == NULL)
cout << "No metadata" << endl;
else
{
PoDoFo::PdfStream *str = metadata->GetStream();
if (str != NULL)
{
char *buf;
PoDoFo::pdf_long len;
str->GetFilteredCopy(&buf, &len);
for (PoDoFo::pdf_long i = 0; i < len; ++i)
printf("%c", buf[i]);
printf("\n");
fflush(stdout);
free(buf);
}
}
}
if (argc == 4)
{
char *xmpBuf;
FILE *fp;
if ((fp = fopen(argv[2], "rb")) == NULL)
cout << "Cannot open " << argv[2] << endl;
else
{
if( fseek( fp, 0, SEEK_END ) == -1 )
{
fclose( fp );
PODOFO_RAISE_ERROR_INFO( ePdfError_InvalidDeviceOperation, "Failed to seek to the end of the file" );
}
long xmpLen = ftell(fp);
if( xmpLen == -1 )
{
fclose( fp );
PODOFO_RAISE_ERROR_INFO( ePdfError_InvalidDeviceOperation, "Failed to read size of the file" );
}
xmpBuf = new char[xmpLen];
if( !xmpBuf )
{
fclose( fp );
PODOFO_RAISE_ERROR( ePdfError_OutOfMemory );
}
if( fseek( fp, 0, SEEK_SET ) == -1 )
{
delete [] xmpBuf;
fclose( fp );
PODOFO_RAISE_ERROR_INFO( ePdfError_InvalidDeviceOperation, "Failed to seek to the beginning of the file" );
}
if( static_cast<long>( fread( xmpBuf, 1, xmpLen, fp ) ) != xmpLen )
{
delete [] xmpBuf;
fclose( fp );
PODOFO_RAISE_ERROR_INFO( ePdfError_InvalidDeviceOperation, "Failed to read whole file into the memory" );
}
PoDoFo::PdfObject *metadata;
if ((metadata = doc->GetMetadata()) != NULL)
metadata->GetStream()->Set(xmpBuf, xmpLen, PoDoFo::TVecFilters());
else
{
metadata = doc->GetObjects().CreateObject("Metadata");
metadata->GetDictionary().AddKey(PoDoFo::PdfName("Subtype"), PoDoFo::PdfName("XML"));
metadata->GetStream()->Set(xmpBuf, xmpLen, PoDoFo::TVecFilters());
doc->GetCatalog()->GetDictionary().AddKey(PoDoFo::PdfName("Metadata"), metadata->Reference());
}
delete[] xmpBuf;
doc->Write(argv[3]);
}
}
} catch( PdfError & e ) {
std::list<FormImage> extract(const std::string& filename, Form& form)
{
std::list<FormImage> images;
ColorSpace colorspace;
PoDoFo::pdf_int64 componentbits;
PoDoFo::PdfObject* obj = nullptr;
PoDoFo::PdfObject* color = nullptr;
PoDoFo::PdfObject* component = nullptr;
PoDoFo::PdfMemDocument document(filename.c_str());
PoDoFo::TCIVecObjects it = document.GetObjects().begin();
while (it != document.GetObjects().end())
{
if ((*it)->IsDictionary())
{
PoDoFo::PdfObject* objType = (*it)->GetDictionary().GetKey(PoDoFo::PdfName::KeyType);
PoDoFo::PdfObject* objSubType = (*it)->GetDictionary().GetKey(PoDoFo::PdfName::KeySubtype);
if ((objType && objType->IsName() && objType->GetName().GetName() == "XObject") ||
(objSubType && objSubType->IsName() && objSubType->GetName().GetName() == "Image" ))
{
// Colorspace
color = (*it)->GetDictionary().GetKey(PoDoFo::PdfName("ColorSpace"));
colorspace = ColorSpace::Unknown;
if (color && color->IsReference())
color = document.GetObjects().GetObject(color->GetReference());
// Follow ICCBased reference to the Alternate colorspace
if (color && color->IsArray() && color->GetArray().GetSize() == 2 &&
// First item is ICCBased
color->GetArray()[0].IsName() &&
color->GetArray()[0].GetName().GetName() == "ICCBased" &&
// Second item is reference to color space
color->GetArray()[1].IsReference())
{
color = document.GetObjects().GetObject(color->GetArray()[1].GetReference());
if (color)
color = color->GetDictionary().GetKey(PoDoFo::PdfName("Alternate"));
}
// Check if either RGB or Grayscale (either the specified
// colorspace or the alternate if using an ICCBased colorspace)
if (color && color->IsName())
{
std::string col = color->GetName().GetName();
if (col == "DeviceRGB")
colorspace = ColorSpace::RGB;
else if (col == "DeviceGray")
colorspace = ColorSpace::Gray;
}
// Bits per component
component = (*it)->GetDictionary().GetKey(PoDoFo::PdfName("BitsPerComponent"));
componentbits = 8;
if (component && component->IsNumber())
componentbits = component->GetNumber();
// Stream
obj = (*it)->GetDictionary().GetKey(PoDoFo::PdfName::KeyFilter);
// JPEG and Flate are in another array
if (obj && obj->IsArray() && obj->GetArray().GetSize() == 1 &&
((obj->GetArray()[0].IsName() && obj->GetArray()[0].GetName().GetName() == "DCTDecode") ||
(obj->GetArray()[0].IsName() && obj->GetArray()[0].GetName().GetName() == "FlateDecode")))
obj = &obj->GetArray()[0];
Pixels pixels;
if (obj && obj->IsName())
{
std::string name = obj->GetName().GetName();
if (name == "DCTDecode")
pixels = readPDFImage(*it, PixelType::JPG, colorspace, componentbits, filename, form);
else if (name == "CCITTFaxDecode")
pixels = readPDFImage(*it, PixelType::TIF, colorspace, componentbits, filename, form);
// PNM is the default
//else if (name == "FlateDecode")
// pixels = readPDFImage(*it, PixelType::PNM, colorspace, componentbits, filename, form);
else
pixels = readPDFImage(*it, PixelType::PNM, colorspace, componentbits, filename, form);
}
else
{
pixels = readPDFImage(*it, PixelType::PNM, colorspace, componentbits, filename, form);
}
document.FreeObjectMemory(*it);
if (pixels.isLoaded())
images.push_back(FormImage(form, std::move(pixels)));
}
}
++it;
}
return images;
}
int main (int argc, char *argv[])
{
PoDoFo::PdfError::EnableDebug(false);
if (argc != 2 && argc != 4)
{
cout << "Syntax" << endl;
cout << " " << argv[0] << " <pdf file> - display the XMP in a file" << endl;
cout << "or" << endl;
cout << " " << argv[0] << " <src pdf file> <xmp file> <new pdf file> - create a new PDF with the XMP in" << endl;
return EXIT_FAILURE;
}
PoDoFo::PdfMemDocument *doc = new PoDoFo::PdfMemDocument(argv[1]);
if (argc == 2)
{
PoDoFo::PdfObject *metadata;
if ((metadata = doc->GetMetadata()) == NULL)
cout << "No metadata" << endl;
else
{
PoDoFo::PdfStream *str = metadata->GetStream();
if (str != NULL)
{
char *buf;
PoDoFo::pdf_long len;
str->GetFilteredCopy(&buf, &len);
for (PoDoFo::pdf_long i = 0; i < len; ++i)
printf("%c", buf[i]);
printf("\n");
fflush(stdout);
free(buf);
}
}
}
if (argc == 4)
{
char *xmpBuf;
FILE *fp;
if ((fp = fopen(argv[2], "rb")) == NULL)
cout << "Cannot open " << argv[2] << endl;
else
{
fseek(fp, 0, SEEK_END);
long xmpLen = ftell(fp);
xmpBuf = new char[xmpLen];
fseek(fp, 0, SEEK_SET);
fread(xmpBuf, 1, xmpLen, fp);
fclose(fp);
PoDoFo::PdfObject *metadata;
if ((metadata = doc->GetMetadata()) != NULL)
metadata->GetStream()->Set(xmpBuf, xmpLen, PoDoFo::TVecFilters());
else
{
metadata = doc->GetObjects().CreateObject("Metadata");
metadata->GetDictionary().AddKey(PoDoFo::PdfName("Subtype"), PoDoFo::PdfName("XML"));
metadata->GetStream()->Set(xmpBuf, xmpLen, PoDoFo::TVecFilters());
doc->GetCatalog()->GetDictionary().AddKey(PoDoFo::PdfName("Metadata"), metadata->Reference());
}
delete[] xmpBuf;
doc->Write(argv[3]);
}
}
delete doc;
return EXIT_SUCCESS;
}
std::ostream& ossimGeoPdfInfo::print(std::ostream& out) const
{
static const char MODULE[] = "ossimGeoPdfInfo::print";
int count = m_PdfMemDocument->GetPageCount();
PoDoFo::PdfString author = m_PdfMemDocument->GetInfo()->GetAuthor();
PoDoFo::PdfString creator = m_PdfMemDocument->GetInfo()->GetCreator();
PoDoFo::PdfString title = m_PdfMemDocument->GetInfo()->GetTitle();
PoDoFo::PdfString subject = m_PdfMemDocument->GetInfo()->GetSubject();
PoDoFo::PdfString keywords = m_PdfMemDocument->GetInfo()->GetKeywords();
PoDoFo::PdfString producer = m_PdfMemDocument->GetInfo()->GetProducer();
ossimString createDate;
ossimString modifyDate;
PoDoFo::PdfObject* obj = m_PdfMemDocument->GetInfo()->GetObject();
if (obj->IsDictionary())
{
PoDoFo::PdfDictionary pdfDictionary = obj->GetDictionary();
PoDoFo::TKeyMap keyMap = pdfDictionary.GetKeys();
PoDoFo::TKeyMap::iterator it = keyMap.begin();
while (it != keyMap.end())
{
ossimString refName = ossimString(it->first.GetName());
PoDoFo::PdfObject* refObj = it->second;
std::string objStr;
refObj->ToString(objStr);
if (refName == "CreationDate")
{
createDate = ossimString(objStr);
createDate = createDate.substitute("(", "", true).trim();
createDate = createDate.substitute(")", "", true).trim();
createDate = createDate.substitute("D:", "", true).trim();
}
else if (refName == "ModDate")
{
modifyDate = ossimString(objStr);
modifyDate = modifyDate.substitute("(", "", true).trim();
modifyDate = modifyDate.substitute(")", "", true).trim();
modifyDate = modifyDate.substitute("D:", "", true).trim();
}
it++;
}
}
try
{
m_PdfMemDocument->FreeObjectMemory(obj);
}
catch (...)
{
}
ossimString authorStr = author.GetString();
ossimString creatorStr = creator.GetString();
ossimString titleStr = title.GetString();
ossimString producerStr = producer.GetString();
ossimString subjectStr = subject.GetString();
ossimString keywordsStr = keywords.GetString();
ossimString prefix = "geopdf.";
out << prefix << "pagecount: "
<< ossimString::toString(count).c_str() << "\n";
if (!authorStr.empty())
{
out << prefix << "author: "
<< authorStr.c_str() << "\n";
}
if (!creatorStr.empty())
{
out << prefix << "creator: "
<< creatorStr.c_str() << "\n";
}
if (!titleStr.empty())
{
out << prefix << "title: "
<< titleStr.c_str() << "\n";
}
if (!producerStr.empty())
{
out << prefix << "producer: "
<< producerStr.c_str() << "\n";
}
if (!subjectStr.empty())
{
out << prefix << "subject: "
<< subjectStr.c_str() << "\n";
}
if (!keywordsStr.empty())
{
out << prefix << "keywords: "
<< keywordsStr.c_str() << "\n";
}
if (!createDate.empty())
{
out << prefix << "creationdate: "
<< createDate.c_str() << "\n";
}
if (!modifyDate.empty())
{
out << prefix << "modificationdate: "
<< modifyDate.c_str() << "\n";
}
if (traceDebug())
{
ossimNotify(ossimNotifyLevel_DEBUG)
<< MODULE << " DEBUG Entered...\n";
}
return out;
}
QStringList FMPDFFontExtractor::list()
{
if(!document)
return mfont.keys();
if(cachedList)
return mfont.keys();
else
cachedList = true;
PoDoFo::TCIVecObjects objIt( document->GetObjects().begin() );
PoDoFo::PdfName pType("Type");
PoDoFo::PdfName pSubtype("Subtype");
PoDoFo::PdfName pFont("Font");
PoDoFo::PdfName pType1("Type1");
PoDoFo::PdfName pTrueType("TrueType");
PoDoFo::PdfName pFontDescriptor( "FontDescriptor" );
PoDoFo::PdfName pFontFile( "FontFile" );
PoDoFo::PdfName pFontFile3( "FontFile3" );
PoDoFo::PdfName pFontName( "FontName" );
while( objIt != document->GetObjects().end() )
{
PoDoFo::PdfObject* obj(*objIt);
if ( obj->IsDictionary() )
{
if(obj->GetIndirectKey(pType))
{
PoDoFo::PdfName type( obj->GetIndirectKey(pType)->GetName() );
if(type == pFont)
{
if(obj->GetIndirectKey( pSubtype ))
{
PoDoFo::PdfName subtype ( obj->GetIndirectKey( pSubtype )->GetName() );
if ((subtype == pType1) || (subtype == pTrueType))
{
PoDoFo::PdfObject * fontDescriptor ( obj->GetIndirectKey ( pFontDescriptor ) );
if (fontDescriptor )
{
bool hasFile(false);
PoDoFo::PdfObject * fontFile ( fontDescriptor->GetIndirectKey ( pFontFile ) );
if ( !fontFile )
{
fontFile = fontDescriptor->GetIndirectKey(pFontFile3) ;
if ( !fontFile )
qWarning ( "Font not embedded not supported yet" );
else
hasFile = true;
}
else
hasFile = true;
if(hasFile)
{
PoDoFo::PdfName fontName(fontDescriptor->GetIndirectKey(pFontName)->GetName());
if(1)
{
QString n(QString::fromStdString( fontName.GetName() ));
mfont[n] = fontFile;
// we know naming it pfb is wrong
mType[n] = (subtype == pType1) ? "pfb" : "ttf";
}
else
qDebug()<<"Error: no /FontName key";
}
}
}
}
}
}
}
objIt++;
}
return mfont.keys();
}