Local API, Seed Uploads without Physical Files

zoul0813

2 months ago

5

payload.create({collection: 'images', data: imageData})

• 

  dribbens

  Payload Team

  2 months ago

  We don't currently have an option that the files wouldn't be written.

  
  
  

  If this is a one time operational task, I would comment out the use of the cloud-storage plugin from your config and enable

  disableLocaleStorage

  on your upload. Then you should be able to call create without actually making any files. If this is something you often do, then you'll want a way to set an env variable or something in your own code that you can conditionally.

  
  
  

  Payload will still need to read the files to populate the database with things like image width and height. Also if you've configured image resizing in your collection, it will make resized files with generated filenames that may be different from your source.

  
  
  

  I hope this helps!

• 

  zoul0813

  2 months ago

  @dribbens thanks - unfortunately, Payload appears to require physical access to the file - even if I disable local storage. I'm unable to call

  payload.create

  and just feed it the raw information ...

  
  
  

  One idea that I'm considering at the moment is having two separate collections, and dynamically returning one or the other based on whether I'm seeding data or launching Payload.

  
  
  

  If I'm seeding, I'll just return a generic collection that isn't upload enabled, but defines all the same fields as an upload collection (filename, filesize, mimetype, width, height) and then I can just seed the data as if it were a regular document ... then when I launch Payload, I return the upload enabled version of the same collection.

  
  
  

  I'm not sure if swapping out collection configs like this dynamically will cause problems ... but so far, it looks like this thought may work.

• 

  dribbens

  Payload Team

  2 months ago

  I don't think it will cause you any issues having the upload feature toggle on/off this way. If you have the data for those extra fields, adding them manually is a good option.

• 

  zoul0813

  2 months ago

  FYI, I've tracked all of this down to

  generateFileData

  which is being passed a "throwOnMissingFile" argument that is set to whether or not the document is a draft ... my images are not version enabled with drafts enabled, so generateFileData throws an exception ...

  
  
  

  What I think could happen here, is generateFileData could look at the data that came in ... and if the file doesn't exist, but the file information is provided ... it can just use that?

  
  

  
  

  @dribbens swapping out the collection with a non-upload with some simple logic seems to have solved the problem ... seems a bit janky, but it works - thanks!

• 

  dribbens

  Payload Team

  2 months ago

  It does seem less than ideal. An alternative approach would be to use

  payload.collections[slug].Model

  directly. That would let you do your DB inserts without relying on Payload's file handling, the risk there is that you may have hooks and other behavior in your collection config that wouldn't be called.