Fetching PURLs doesn't follow redirects

[thatbudakguy]

discovered when working with @cbeer that fetching a particular PURL via HTTP doesn't seem to follow redirection on our aliased domains. the output of the following two (example) commands is different:

echo "zt944np1693" | traject -c config.rb -s purl.url=https://purl-stage.stanford.edu/ --stdin
echo "zt944np1693" | traject -c config.rb -s purl.url=https://sul-purl-stage.stanford.edu/ --stdin

because purl-stage is an alias for sul-purl-stage, the former command fails somewhat silently — the empty response body of the redirect results in the fetched record having no unique identifier, so it is skipped for indexing. if sul-purl-stage is used instead, the real publicXML can be fetched.

this behavior may be ruby implementation dependent — a first place to look might be:

searchworks_traject_indexer/lib/sdr_stuff.rb

Lines 27 to 35 in b2175d2

	def self.fetch(url)
	if defined?(JRUBY_VERSION)
	response = Manticore.get(url)
	response.body if response.code == 200
	else
	response = HTTP.get(url)
	response.body if response.status.ok?
	end
	end