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;
}
