dina_fetch_duplicate_images.rb

#!/usr/bin/env ruby
# encoding: utf-8

require 'csv'
require 'dina'

Dina.config = {
    authorization_url: 'https://dina.biodiversity.agr.gc.ca/auth',
    endpoint_url: 'https://dina.biodiversity.agr.gc.ca/api',
    server_name: 'dina-prod',
    realm: 'dina',
    client_id: 'dina-public',
    user: 'USERNAME',
    password: 'PASSWORD',
    token_store_file: File.join(Dir.pwd, "token.json")
}

def images_by_tag(tag:)
    payload = {
      query: {
        bool: {
          must: [
            { term: { "data.attributes.acTags.keyword": "#{tag}" }},
            { term: { "data.attributes.group":"dao" } }
          ]
        }
      },
      size: 10_000
    }
    Dina::Search.execute(index: "object_store", payload: payload)
end

CSV.open(File.join(Dir.pwd, "biomob_duplicate_barcodes.csv"), "w") do |csv|
    csv << ["barcode", "folder_barcode", "original_directory_name", "dina_uuid", "dina_url"]
    images_by_tag(tag: "duplicated barcode").each do |obj|
        attributes = obj.attributes["managed_attributes"]
        csv << [
            attributes["barcode"],
            attributes["folder_barcode"],
            attributes["original_directory_name"],
            obj.id,
            "https://dina.biodiversity.agr.gc.ca/object-store/object/view?id=#{obj.id}"
        ]
    end
end