Configuration examples

Research & Education Space production configuration

[crawler]
detach=yes
verbose=no
threads=1

[cluster]
name=anansi
registry=pgsql://USER:PASS@HOST/DBNAME
environment=live

[processor]
name=lod

[lod:licenses]
predicate="http://purl.org/dc/terms/rights"
predicate="http://purl.org/dc/terms/license"
predicate="http://purl.org/dc/terms/accessRights"
predicate="http://creativecommons.org/ns#license"
predicate="http://www.w3.org/1999/xhtml/vocab#license"
predicate="http://www.w3.org/1999/xhtml/vocab#copyright"
whitelist="http://creativecommons.org/publicdomain/zero/1.0/"
whitelist="http://id.loc.gov/about/"
whitelist="http://creativecommons.org/licenses/by/4.0/"
whitelist="http://creativecommons.org/licenses/by-sa/4.0/"
whitelist="http://reference.data.gov.uk/id/open-government-licence"
whitelist="http://bbcarchdev.github.io/licences/dps/1.0#id"
whitelist="http://creativecommons.org/licenses/by/1.0/"
whitelist="https://creativecommons.org/licenses/by/1.0/"
whitelist="http://creativecommons.org/licenses/by-sa/1.0/"
whitelist="https://creativecommons.org/licenses/by-sa/1.0/"
whitelist="http://creativecommons.org/licenses/by/2.5/"
whitelist="https://creativecommons.org/licenses/by/2.5/"
whitelist="http://creativecommons.org/licenses/by-sa/2.5/"
whitelist="https://creativecommons.org/licenses/by-sa/2.5/"
whitelist="http://creativecommons.org/licenses/by/3.0/"
whitelist="https://creativecommons.org/licenses/by/3.0/"
whitelist="http://creativecommons.org/licenses/by-sa/3.0/"
whitelist="https://creativecommons.org/licenses/by-sa/3.0/"
whitelist="http://creativecommons.org/licenses/by/3.0/us/"
whitelist="https://creativecommons.org/licenses/by/3.0/us/"
whitelist="http://creativecommons.org/licenses/by-sa/3.0/us/"
whitelist="https://creativecommons.org/licenses/by-sa/3.0/us/"
whitelist="https://www.nationalarchives.gov.uk/doc/open-government-licence/version/3/"
whitelist="http://www.nationalarchives.gov.uk/doc/open-government-licence/version/3/"
whitelist="https://www.nationalarchives.gov.uk/doc/open-government-licence/version/2/"
whitelist="http://www.nationalarchives.gov.uk/doc/open-government-licence/version/2/"
whitelist="https://www.nationalarchives.gov.uk/doc/open-government-licence/version/1/"
whitelist="http://www.nationalarchives.gov.uk/doc/open-government-licence/version/1/"
whitelist="http://www.ordnancesurvey.co.uk/business-and-government/licensing/using-creating-data-with-os-products/os-opendata.html"
whitelist="http://www.ordnancesurvey.co.uk/oswebsite/opendata/licence/docs/licence.pdf"
whitelist="http://opendatacommons.org/licenses/pddl/1.0/"
whitelist="http://opendatacommons.org/licenses/by/1.0/"
whitelist="http://opendatacommons.org/licenses/odbl/1.0/"
whitelist="http://collection.britishmuseum.org/licensing.html"

[cache]
uri=s3://BUCKETNAME/
; endpoint=ALTERNATIVE-S3-HOST
username=ACCESS-KEY
password=SECRET-KEY

[queue]
name=db
uri=pgsql://USER:PASS@HOST/DBNAME
debug-queries=no
debug-errors=yes

[policy:schemes]
whitelist=http,https,ftp

[log]
<% if @environment == 'dev' %>
level=debug
<% else %>
level=info
<% end %>
facility=daemon