Add layers management feature

[enzo-billis]

Describe your idea

It could be great to add an easy layer API working with OCProperties.
Something like:
const pdfDoc = await PDFDocument.load(...); const layers = pdfDoc.ocProperties.layers.get(); const updatedLayers = layers.filter(e => e.name === "SOMETHING"); pdfDoc.ocProperties.layers.set(updatedLayers); const pdfBytes = await pdfDoc.saveAsBase64();

How could this be implemented?

1. Adding an object ocProperties to PDFDocument class, copying the firstPage.doc.catalog.get(PDFName.of("OCProperties")); data
2. Allow editing with accessor set (IDK how to edit catalog data actually)

What problem are you trying to solve?

Would like to offer the feature of removing layers when displaying PDF

Why does this matter to you?

This is the key feature of my project

Would others find this helpful?

OC

Are you interested in implementing your proposal?

No

Why are you submitting a proposal?

First, would like to know if it feasible (should be, but i'm not experienced enought with this lib)

Additional Notes

No response

[Sharcoux]

We don't use this, so this will be left to the community. Feel free to open a PR if you want. I'll follow and merge it.

[enzo-billis]

Thanks for you answer. I will try to do this feature. Also, could you confirm me there isn't a way to manipulate properties returned by firstPage.doc.catalog.get(PDFName.of("OCProperties")); ? It seems to be an array of number.
This is the only pain point for this feature. The rest seems only to be a faciliting way to do acces and edit.

[Sharcoux]

Honestly, I didn't directly work on manipulating properties since we took over the maintenance. But you can ask Cursor AI or ChatGPT. They are of generally of great help and appears to know a bit of the project code (which is probably a part of their training base)

[enzo-billis]

Yes, indeed ChatGPT seem's to be experienced with the lib. He helped me changing the type of PDFObject to PDFDict.
Before (PDFObject):
const ocProperties = firstPage.doc.catalog.get(PDFName.of("OCProperties"))
Now (Cast as PDFDict):
const ocProperties = firstPage.doc.catalog.get(PDFName.of("OCProperties")) as PDFDict;

It gives me more method, like the get to acces to D, ON, OFF and OCGs properties.

Thanks to that I achieve to read all layers (with name), and remove them by updating the ON and OFF properties.
I share my code for those who needs help:

const getLayers = async () => {
      const pdfDoc = await PDFDocument.load(base64);
      const pages = pdfDoc.getPages();
      const firstPage = pages[0];
      const context = firstPage.doc.context;
      const ocProperties = firstPage.doc.catalog.get(
        PDFName.of("OCProperties")
      ) as PDFDict;

      const ocgs = ocProperties.get(PDFName.of("OCGs")) as PDFArray;
      const dDict = ocProperties.get(PDFName.of("D")) as PDFDict;

      const onArray = dDict?.get(PDFName.of("ON")) as PDFArray;
      const visibleRefs = onArray
        ? new Set(onArray.asArray().map((ref) => ref?.toString()))
        : new Set();

      const ocgInfo = ocgs.asArray().map((ref) => {
        const ocgDict = context.lookup(ref, PDFDict) as PDFDict;
        const nameObj = ocgDict.get(PDFName.of("Name"));
        const name = nameObj?.toString() ?? "[Sans nom]";
        const isVisible = visibleRefs.has(ref?.toString());
        return { ref, name, visible: isVisible };
      });

      setLayers(ocgInfo);
};

const applyLayers = async () => {
    const pdfDoc = await PDFDocument.load(base64);
    const pages = pdfDoc.getPages();
    const firstPage = pages[0];
    const context = firstPage.doc.context;

    const ocProperties = firstPage.doc.catalog.get(
      PDFName.of("OCProperties")
    ) as PDFDict;

    const ocgs = ocProperties.get(PDFName.of("OCGs")) as PDFArray;

    const allOCGRefs = ocgs.asArray(); // Toutes les refs connues
    const visibleRefs = layers.filter((l) => l.visible).map((l) => l.ref);
    const hiddenRefs = layers.filter((l) => !l.visible).map((l) => l.ref);

    // Create /D if doesn't exists
    let dDict = ocProperties.get(PDFName.of("D")) as PDFDict;
    if (!dDict) {
      dDict = PDFDict.withContext(context);
      ocProperties.set(PDFName.of("D"), dDict);
    }

    // Update /ON with visible layers
    const onArray = PDFArray.withContext(context);
    visibleRefs.forEach((ref) => onArray.push(ref));
    dDict.set(PDFName.of("ON"), onArray);

    // Update /OF with hidden layers
    const offArray = PDFArray.withContext(context);
    hiddenRefs.forEach((ref) => offArray.push(ref));
    dDict.set(PDFName.of("OFF"), offArray);

    firstPage.doc.catalog.set(PDFName.of("OCProperties"), ocProperties);

    const pdfBytes = await pdfDoc.saveAsBase64();
    onChangePDF(pdfBytes);
  };

As this proof of concept works, if I got time I would prepare a PR for this !

[jorgedanisc]

Hey! What would be the state of this feature? After doing some research I found there is no PDF creation js library that supports OCG, so this is actually unique and useful!