Sha256: e752f496d74b8e9454b16703750a5a96fbcbe0ea3e38af27d189ee896d0ee252
Contents?: true
Size: 1.31 KB
Versions: 14
Compression:
Stored size: 1.31 KB
Contents
module Sufia
module GenericFile
module FullTextIndexing
extend ActiveSupport::Concern
included do
has_file_datastream 'full_text', versionable: false
end
def append_metadata
super
extract_content
end
private
def extract_content
url = Blacklight.solr_config[:url] ? Blacklight.solr_config[:url] : Blacklight.solr_config["url"] ? Blacklight.solr_config["url"] : Blacklight.solr_config[:fulltext] ? Blacklight.solr_config[:fulltext]["url"] : Blacklight.solr_config[:default]["url"]
uri = URI("#{url}/update/extract?extractOnly=true&wt=json&extractFormat=text")
req = Net::HTTP.new(uri.host, uri.port)
resp = req.post(uri.to_s, self.content.content, {
'Content-type' => "#{self.mime_type};charset=utf-8",
'Content-Length' => self.content.content.size.to_s
})
raise "URL '#{uri}' returned code #{resp.code}" unless resp.code == "200"
self.content.content.rewind if self.content.content.respond_to?(:rewind)
extracted_text = JSON.parse(resp.body)[''].rstrip
full_text.content = extracted_text if extracted_text.present?
rescue => e
logger.error("Error extracting content from #{self.pid}: #{e.inspect}")
end
end
end
end
Version data entries
14 entries across 14 versions & 2 rubygems