Pixels readPDFImage(PoDoFo::PdfObject* object, const PixelType type,
const ColorSpace colorspace, const PoDoFo::pdf_int64 componentbits,
const std::string& filename, Form& form)
{
Pixels pixels;
if (!object->GetDictionary().HasKey(PoDoFo::PdfName("Width")) ||
!object->GetDictionary().HasKey(PoDoFo::PdfName("Height")))
return pixels;
const unsigned int width = object->GetDictionary().GetKey(PoDoFo::PdfName("Width"))->GetNumber();
const unsigned int height = object->GetDictionary().GetKey(PoDoFo::PdfName("Height"))->GetNumber();
if (type == PixelType::JPG)
{
PoDoFo::PdfMemStream* stream = dynamic_cast<PoDoFo::PdfMemStream*>(object->GetStream());
pixels = Pixels(IL_JPG, stream->Get(), stream->GetLength(), filename);
}
else if (type == PixelType::TIF)
{
// Monochrome, otherwise wouldn't have used CCITT
const unsigned int bits = 1;
const unsigned int samples = 1;
if (bits != componentbits)
form.log("BitsPerComponent is not 1 for CCITTFaxDecode image in PDF", LogType::Warning);
std::ostringstream os;
TIFF* tif = TIFFStreamOpen("Input", &os);
TIFFSetField(tif, TIFFTAG_IMAGEWIDTH, width);
TIFFSetField(tif, TIFFTAG_IMAGELENGTH, height);
TIFFSetField(tif, TIFFTAG_BITSPERSAMPLE, bits);
TIFFSetField(tif, TIFFTAG_SAMPLESPERPIXEL, samples);
TIFFSetField(tif, TIFFTAG_FILLORDER, FILLORDER_MSB2LSB);
TIFFSetField(tif, TIFFTAG_PLANARCONFIG, PLANARCONFIG_CONTIG);
TIFFSetField(tif, TIFFTAG_PHOTOMETRIC, PHOTOMETRIC_MINISWHITE);
TIFFSetField(tif, TIFFTAG_XRESOLUTION, 204.0); // From fax2tiff
TIFFSetField(tif, TIFFTAG_YRESOLUTION, 196.0); // ditto
TIFFSetField(tif, TIFFTAG_RESOLUTIONUNIT, RESUNIT_INCH);
TIFFSetField(tif, TIFFTAG_COMPRESSION, COMPRESSION_CCITTFAX4);
TIFFSetField(tif, TIFFTAG_ORIENTATION, ORIENTATION_TOPLEFT);
TIFFSetField(tif, TIFFTAG_FAXMODE, FAXMODE_CLASSF);
TIFFSetField(tif, TIFFTAG_ROWSPERSTRIP, (uint32)-1L);
// stream->Get returns read-only, so copy it
PoDoFo::PdfMemStream* stream = dynamic_cast<PoDoFo::PdfMemStream*>(object->GetStream());
const char* readonly = stream->Get();
PoDoFo::pdf_long len = stream->GetLength();
char* buffer = new char[len];
std::memcpy(buffer, readonly, len);
TIFFWriteRawStrip(tif, 0, buffer, len);
TIFFWriteDirectory(tif);
TIFFClose(tif);
delete[] buffer;
pixels = Pixels(IL_TIF, os.str().c_str(), os.tellp(), filename);
}
else
{
std::ostringstream os;
// P4 is bitmap (1 bit), P5 is graymap (8 bit), P6 is pixmap (24 bit or
// 8-bit/channel). See:
// http://netpbm.sourceforge.net/doc/pbm.html
// http://en.wikipedia.org/wiki/Portable_anymap#File_format_description
if (colorspace == ColorSpace::Gray)
os << ((componentbits == 1)?"P4\n":"P5\n");
else
os << "P6\n";
os << width << " "
<< height << "\n";
if (componentbits != 1)
os << "255\n";
std::string s = os.str();
char* buffer;
PoDoFo::pdf_long len;
object->GetStream()->GetFilteredCopy(&buffer, &len);
// Warn if unknown colorspace
if (colorspace == ColorSpace::Unknown)
form.log("unknown color space on PDF image", LogType::Warning);
// If the buffer isn't the correct size for the image data, don't try
// reading the image from this invalid data
if (len != correctLength(width, height, colorspace, componentbits))
{
std::ostringstream ss;
ss << "wrong buffer size for PDF image of size "
<< width << "x" << height
<< " (" << colorspace << "): " << len;
form.log(ss.str(), LogType::Warning);
std::free(buffer);
return pixels;
}
std::unique_ptr<char> stream(new char[len+s.size()]);
std::memcpy(stream.get(), s.c_str(), s.size());
std::memcpy(stream.get()+s.size(), buffer, len);
std::free(buffer);
pixels = Pixels(IL_PNM, stream.get(), len+s.size(), filename);
}
return pixels;
}
