config.yml

# The list of repositories to be harvested. The name will be used in logging/tracing and to name the resulting file.
# The url denotes the so-called sitemap of the repository (at best it's really a sitemap).
# The sitemap specifies the parser for the sitemap url above. Currently two are available:
#   - xml (for sitemaps conforming to the xml standard defined a https://sitemaps.org)
#   - openagrar (the sitemap conists of a list of datasets from which the ID needs to be extracted and appended to
#               the base url for downloading datasets)
# metadata denotes the metadata parser. Currently two are available:
#   - embedded_jsonld (for schema.org JSON-LD metadata embedded into HTML)
#   - jsonld (for standalone schema.org JSON-LD files)
sitemaps:
  - name: bonares
    url: https://maps.bonares.de/finder/resources/googleds/sitemap.xml
    sitemap: xml
    metadata: embedded_jsonld
  - name: edal
    url: https://doi.ipk-gatersleben.de/sitemap.xml
    sitemap: xml
    metadata: embedded_jsonld
  - name: openagrar_native
    url: "https://www.openagrar.de/servlets/solr/select?core=main&q=category.top%3A%22mir_genres%3Aresearch_data%22+AND+objectType%3Amods+AND+category.top%3A%22state%3Apublished%22&rows=1000&fl=id%2Cmods.identifier&wt=json&XSL.Style=xml"
    sitemap: openagrar
    metadata: embedded_jsonld
  - name: publisso_native
    url: "https://frl.publisso.de/find?q=contentType:researchData&from=0&until=200&format=json"
    sitemap: publisso
  # - name: thunen_atlas
  #   url: "https://atlas.thuenen.de/api/v2/resources?page_size=200&format=json"
  #   sitemap: thunen_atlas

# Configures the HTTP client uses to download sitemaps and datasets
http_client:
  connection_limit: 20
  receive_timeout: 120
  connect_timeout: 60

# Configures the git client
# Note on paths: relative paths refer to the current working directory.
# You can omit the ssh_key_path if you have a local git setup with your personal key.
# If branch is omitted, the main branch is used
git:
  repo_url: https://github.com/fairagro/middleware_repo.git
  branch: main
  local_path: output
  user_name: "FAIRagro middleware"
  user_email: "middleware@fairagro.net"

# If you operate an opentelemetry endpoint you can configure it here
opentelemetry: {}
#  endpoint:
#  insecure: true  # set to true if you do not use TLS