PhoenixKit.Modules.Storage.TesseraAdapter (phoenix_kit v1.7.108)

Copy Markdown View Source

Tessera.Storage adapter that lands tile / manifest writes inside PhoenixKit's storage pipeline:

  • writes the binary into every configured bucket via PhoenixKit.Modules.Storage.Manager.store_file/2 under the _tiles/ prefix (multi-bucket redundancy applies)
  • creates a system-managed File row + one "original" FileInstance row for the chunk via Storage.store_system_file/3, so the chunk is tracked in the DB and cascades away when the source File is deleted

Required storage_opts

Callers must pass the source image's UUID so the chunk knows its parent:

Tessera.generate_tile(
  original_path, {level, col, row}, "<uuid>/<uuid>",
  image_width: w,
  image_height: h,
  storage: PhoenixKit.Modules.Storage.TesseraAdapter,
  storage_opts: [
    parent_file_uuid: source_file_uuid,
    mime_type: "image/jpeg"
  ]
)

Reads, existence checks, and deletes happen from the consumer side (see PhoenixKitWeb.FileController.serve_tile/2 and the cascade FK on parent_file_uuid in V112).

Summary

Functions

destination_for(key)

Returns the storage-side destination path for a tile/manifest key (prepends the _tiles/ prefix this adapter uses for writes).

put(content_path, key, opts)

Persist content_path under _tiles/<key> in every configured bucket and create a system-managed File + FileInstance pair in the DB.

Functions

destination_for(key)

Returns the storage-side destination path for a tile/manifest key (prepends the _tiles/ prefix this adapter uses for writes).

put(content_path, key, opts)

Persist content_path under _tiles/<key> in every configured bucket and create a system-managed File + FileInstance pair in the DB.

Returns :ok or {:error, reason}. If the bucket write succeeded but DB row creation failed, the bucket data is left in place (orphaned) — the controller will treat the next request as a cache miss and try again.