When we have a PDF Asset with work_source_pdf role, we want to make a pdf derivative, with a smaller file size and scaled down images. Some of our "born digital" Distillations PDFs have 300 dpi images for printing, and are quite big.
We define a derivative on AssetUploader, in the standard way -- it will be stored in the derivatives hash in the shrine file!
we mark it default_create: false, so it won't be automatically created on ingest, it will be triggered manually. (on ingest, role isn't set yet telling us this is a work_source_pdf we want to create the deriv for!)
The setup_work_from_pdf_source action -- which was already extracting page images as Assets -- will now also see if the derivative is already on the work_source_pdf, and if not launch a CreateScaledDownPdfDerivativeJob to create one. (It can take 120+ seconds to create)
The actual work of creating the PDF is done by ScaleDownPdf class -- which shells out to gs (ghostscript) command line
Then the recently added WorkDownloadOptionsCreator class was enhanced to notice the presence of the derivative (called scaled_down_pdf, with key kept in a constant), and add it to the whole-work options shown on Work page and in Downloads popup menu.
On top of #2747
When we have a PDF Asset with
work_source_pdfrole, we want to make a pdf derivative, with a smaller file size and scaled down images. Some of our "born digital" Distillations PDFs have 300 dpi images for printing, and are quite big.We define a derivative on
AssetUploader, in the standard way -- it will be stored in the derivatives hash in the shrine file!default_create: false, so it won't be automatically created on ingest, it will be triggered manually. (on ingest, role isn't set yet telling us this is awork_source_pdfwe want to create the deriv for!)setup_work_from_pdf_sourceaction -- which was already extracting page images as Assets -- will now also see if the derivative is already on thework_source_pdf, and if not launch aCreateScaledDownPdfDerivativeJobto create one. (It can take 120+ seconds to create)The actual work of creating the PDF is done by
ScaleDownPdfclass -- which shells out togs(ghostscript) command lineGhost script has a "pseudo-device" pre-set for
ebook, that downsamples images to 150dpi, among other things. We're using that. There was ascreenpreset for 72dpi, but that made images look noticeably bad. 150 dpi was fine on devices I tested, still with significant file size reduction for our sample files. https://ghostscript.readthedocs.io/en/gs10.0.0/VectorDevices.html#the-family-of-pdf-and-postscript-output-devicesAdded check for
gsto system_env_specgswas already present on my Mac (prob a brew dependency of vips?) and on herokuBut it was removed from heroku-24. Could be re-added with a custom buildpack, but it also looks like they may be re-adding it back. https://www.reddit.com/r/Heroku/comments/1fj3cr4/ghostscript_on_heroku24/.
Then the recently added
WorkDownloadOptionsCreatorclass was enhanced to notice the presence of the derivative (calledscaled_down_pdf, with key kept in a constant), and add it to the whole-work options shown on Work page and in Downloads popup menu.