feat(images): support multi-page jpg reps for tif files

[ikerr]

No description provided.

[CLAassistant]

All committers have signed the CLA.

[jstoffan]

Can we avoid duplicating the logic from AssetLoader? Can we call super.determineViewer and then further filter the list as needed?

[ikerr]

We could probably do something like that if determineViewer() returned a set of viewers rather than a single viewer. Then we could call super.determineViewer() and then further filter the results. Happy to make that change if you'd like, but it'll increase the scope of this change.

[ikerr]

Alternatively, I could probably create a new method, AssetLoader.determineViewerSet() that returns a set of viewers and then do this:

// AssetLoader
determineViewer(file, disabledViewers = []) {
  const viewers = determineViewerSet(file, disabledViewers);
  if (viewers.length > 0) {
    return viewers[0];
  }

  return undefined;
}

and then in ImageLoader:

// ImageLoader
determineViewer(file, disabledViewers = []) {
  const viewers = super.determineViewerSet(file, disabledViewers);
  // perform further filtering...
}

[jstoffan]

I'm finding the logic here hard to follow. Can we have non-paged TIFF files? What case is this guarding against?

[ikerr]

@jstoffan I was seeing the following issue... Suppose we have the following reps available:

representations: {                                              
  entries: [                                                  
    {                                                       
      representation: 'jpg',                              
      properties: {                                       
        dimensions: '1024x1024',                        
        paged: 'false',                                 
      },                                                  
    },                                                      
    {                                                       
      representation: 'png',                              
      properties: {                                       
        dimensions: '2048x2048',                        
        paged: 'true',
      },                                                  
    },                                                      
  ],                                                          
}

When AssetLoader.determineViewer(file) is called for a TIFF file, it will see that a "jpg" representation is available (the first entry), and select the following viewer:

{                                                                           
  NAME: 'MultiImage',                                                     
  CONSTRUCTOR: MultiImageViewer,                                          
  REP: 'jpg',                                                             
  EXT: ['tif', 'tiff'],                                                   
  ASSET: '{page}.jpg',                                                    
}

However, the 1024x1024 jpg rep that is available is not a paged rep. So, when we then call ImageLoader.determineRepresentation(), we get back null/undefined because it specifically filters out these reps:

determineRepresentation(file, viewer) {                                     
  return file.representations.entries.find(entry => {                     
    // Do not use the dimensions=1024x1024&paged=false rep that is for document preloading
    if (entry.properties && entry.properties.paged === 'false') {       
      return false;                                                   
    }                                                                   
                                                                                
    return viewer.REP === entry.representation;                         
   });                                                                     
}

This most recent change aims to only use the MultiImage viewer (which is currently only used for TIFF files) when a paged representation exists.

[jstoffan]

Thank you!