diff --git a/bin/export-items b/bin/export-items index a6dbaddb..909247f4 100755 --- a/bin/export-items +++ b/bin/export-items @@ -7,6 +7,17 @@ COLLECTION_DRUIDS_LIST = 'collection_druids.txt' GRAVEYARD_APO = 'druid:kg712km1576' +# This list from: https://docs.google.com/spreadsheets/d/1Gu0TIUpNByNgNtFDn5MJARUvtsgcNXJAHiKf5s_NsUc/edit#gid=0 +# NOTE: we don't need to worry about the "hydrus object with DOI", because these are not returned in the query +# for Hydrus::Items as they have been converted to Dor::Item. +ITEMS_TO_EXCLUDE = %w[ + druid:ty334nd6571 + druid:bx749bs2681 + druid:zc000fq4044 + druid:pr213sh5046 + druid:jr671fk0644 +] + require_relative '../config/environment' # Monkey-patch AF to allow using HTTP POST (for querying items by their collection) @@ -128,6 +139,9 @@ list.each do |item| elsif item.catkey.present? warn "Skipping #{item.pid} because it has a catkey" next + elsif ITEMS_TO_EXCLUDE.include? item.pid + warn "Skipping #{item.pid} because it is on the list of items to exclude" + next end attributes = serialize(item) puts attributes.compact.to_json