Skip to content

Commit

Permalink
optionally change number of facets in DOI REST API. #709
Browse files Browse the repository at this point in the history
  • Loading branch information
Martin Fenner committed Jan 27, 2021
1 parent 3794f90 commit 1397676
Show file tree
Hide file tree
Showing 4 changed files with 200 additions and 162 deletions.
140 changes: 55 additions & 85 deletions app/controllers/datacite_dois_controller.rb
View file
Original file line number Diff line number Diff line change
Expand Up @@ -68,6 +68,9 @@ def index
params[:state] = "findable"
end

# default number of facets
facet_count = (params[:facet_count] || 10).to_i

if params[:id].present?
response = DataciteDoi.find_by_id(params[:id])
elsif params[:ids].present?
Expand Down Expand Up @@ -122,6 +125,7 @@ def index
sample_size: params[:sample],
source: params[:source],
scroll_id: params[:scroll_id],
facet_count: facet_count,
page: page,
sort: sort,
random: params[:random],
Expand Down Expand Up @@ -206,91 +210,56 @@ def index
status: :ok
end
else
states =
if total.positive?
facet_by_key(response.aggregations.states.buckets)
end
resource_types =
if total.positive?
facet_by_combined_key(response.aggregations.resource_types.buckets)
end
published =
if total.positive?
facet_by_range(response.aggregations.published.buckets)
end
created =
if total.positive?
facet_by_key_as_string(response.aggregations.created.buckets)
end
registered =
if total.positive?
facet_by_key_as_string(response.aggregations.registered.buckets)
end
providers =
if total.positive?
facet_by_combined_key(response.aggregations.providers.buckets)
end
clients =
if total.positive?
facet_by_combined_key(response.aggregations.clients.buckets)
end
prefixes =
if total.positive?
facet_by_key(response.aggregations.prefixes.buckets)
end
schema_versions =
if total.positive?
facet_by_schema(response.aggregations.schema_versions.buckets)
end

affiliations =
if total.positive?
facet_by_combined_key(response.aggregations.affiliations.buckets)
end
# sources = total.positive? ? facet_by_key(response.aggregations.sources.buckets) : nil
subjects =
if total.positive?
facet_by_key(response.aggregations.subjects.buckets)
end
fields_of_science =
if total.positive?
facet_by_fos(
response.aggregations.fields_of_science.subject.buckets,
)
end
certificates =
if total.positive?
facet_by_key(response.aggregations.certificates.buckets)
end
licenses =
if total.positive?
facet_by_license(response.aggregations.licenses.buckets)
end

link_checks_status =
if total.positive?
facet_by_cumulative_year(
response.aggregations.link_checks_status.buckets,
)
end
# links_with_schema_org = total.positive? ? facet_by_cumulative_year(response.aggregations.link_checks_has_schema_org.buckets) : nil
# link_checks_schema_org_id = total.positive? ? response.aggregations.link_checks_schema_org_id.value : nil
# link_checks_dc_identifier = total.positive? ? response.aggregations.link_checks_dc_identifier.value : nil
# link_checks_citation_doi = total.positive? ? response.aggregations.link_checks_citation_doi.value : nil
# links_checked = total.positive? ? response.aggregations.links_checked.value : nil

citations =
if total.positive?
metric_facet_by_year(response.aggregations.citations.buckets)
end
views =
if total.positive?
metric_facet_by_year(response.aggregations.views.buckets)
end
downloads =
if total.positive?
metric_facet_by_year(response.aggregations.downloads.buckets)
end
if total.positive? && facet_count.positive?
states = facet_by_key(response.aggregations.states.buckets)
resource_types = facet_by_combined_key(response.aggregations.resource_types.buckets)
published = facet_by_range(response.aggregations.published.buckets)
created = facet_by_key_as_string(response.aggregations.created.buckets)
registered = facet_by_key_as_string(response.aggregations.registered.buckets)
providers = facet_by_combined_key(response.aggregations.providers.buckets)
clients = facet_by_combined_key(response.aggregations.clients.buckets)
prefixes = facet_by_key(response.aggregations.prefixes.buckets)
schema_versions = facet_by_schema(response.aggregations.schema_versions.buckets)
affiliations = facet_by_combined_key(response.aggregations.affiliations.buckets)
# sources = total.positive? ? facet_by_key(response.aggregations.sources.buckets) : nil
subjects = facet_by_key(response.aggregations.subjects.buckets)
fields_of_science = facet_by_fos(
response.aggregations.fields_of_science.subject.buckets,
)
certificates = facet_by_key(response.aggregations.certificates.buckets)
licenses = facet_by_license(response.aggregations.licenses.buckets)
link_checks_status = facet_by_cumulative_year(
response.aggregations.link_checks_status.buckets,
)
# links_with_schema_org = total.positive? ? facet_by_cumulative_year(response.aggregations.link_checks_has_schema_org.buckets) : nil
# link_checks_schema_org_id = total.positive? ? response.aggregations.link_checks_schema_org_id.value : nil
# link_checks_dc_identifier = total.positive? ? response.aggregations.link_checks_dc_identifier.value : nil
# link_checks_citation_doi = total.positive? ? response.aggregations.link_checks_citation_doi.value : nil
# links_checked = total.positive? ? response.aggregations.links_checked.value : nil

citations = metric_facet_by_year(response.aggregations.citations.buckets)
views = metric_facet_by_year(response.aggregations.views.buckets)
downloads = metric_facet_by_year(response.aggregations.downloads.buckets)
else
states = nil
resource_types = nil
published = nil
created = nil
registered = nil
providers = nil
clients = nil
prefixes = nil
schema_versions = nil
affiliations = nil
subjects = nil
fields_of_science = nil
certificates = nil
licenses = nil
link_checks_status = nil
citations = nil
views = nil
downloads = nil
end

respond_to do |format|
format.json do
Expand Down Expand Up @@ -359,6 +328,7 @@ def index
"has-person" => params[:has_person],
"has-affiliation" => params[:has_affiliation],
"has-funder" => params[:has_funder],
facet_count: params[:facet_count],
detail: params[:detail],
composite: params[:composite],
affiliation: params[:affiliation],
Expand Down
138 changes: 70 additions & 68 deletions app/models/doi.rb
View file
Original file line number Diff line number Diff line change
Expand Up @@ -585,78 +585,80 @@ def self.gql_query_aggregations(facet_count: 10)
end
end

def self.query_aggregations
{
resource_types: { terms: { field: "resource_type_id_and_name", size: 16, min_doc_count: 1 } },
states: { terms: { field: "aasm_state", size: 3, min_doc_count: 1 } },
published: {
date_histogram: {
field: "publication_year",
interval: "year",
format: "year",
order: {
_key: "desc",
def self.query_aggregations(facet_count: 10)
if facet_count.to_i.positive?
{
resource_types: { terms: { field: "resource_type_id_and_name", size: facet_count, min_doc_count: 1 } },
states: { terms: { field: "aasm_state", size: 3, min_doc_count: 1 } },
published: {
date_histogram: {
field: "publication_year",
interval: "year",
format: "year",
order: {
_key: "desc",
},
min_doc_count: 1,
},
min_doc_count: 1,
},
},
registration_agencies: { terms: { field: "agency", size: 10, min_doc_count: 1 } },
created: { date_histogram: { field: "created", interval: "year", format: "year", order: { _key: "desc" }, min_doc_count: 1 },
aggs: { bucket_truncate: { bucket_sort: { size: 10 } } } },
registered: { date_histogram: { field: "registered", interval: "year", format: "year", order: { _key: "desc" }, min_doc_count: 1 },
aggs: { bucket_truncate: { bucket_sort: { size: 10 } } } },
providers: { terms: { field: "provider_id_and_name", size: 10, min_doc_count: 1 } },
clients: { terms: { field: "client_id_and_name", size: 10, min_doc_count: 1 } },
affiliations: { terms: { field: "affiliation_id_and_name", size: 10, min_doc_count: 1 } },
prefixes: { terms: { field: "prefix", size: 10, min_doc_count: 1 } },
schema_versions: { terms: { field: "schema_version", size: 10, min_doc_count: 1 } },
link_checks_status: { terms: { field: "landing_page.status", size: 10, min_doc_count: 1 } },
# link_checks_has_schema_org: { terms: { field: 'landing_page.hasSchemaOrg', size: 2, min_doc_count: 1 } },
# link_checks_schema_org_id: { value_count: { field: "landing_page.schemaOrgId" } },
# link_checks_dc_identifier: { value_count: { field: "landing_page.dcIdentifier" } },
# link_checks_citation_doi: { value_count: { field: "landing_page.citationDoi" } },
# links_checked: { value_count: { field: "landing_page.checked" } },
# sources: { terms: { field: 'source', size: 15, min_doc_count: 1 } },
subjects: { terms: { field: "subjects.subject", size: 10, min_doc_count: 1 } },
pid_entities: {
filter: { term: { "subjects.subjectScheme": "PidEntity" } },
aggs: {
subject: { terms: { field: "subjects.subject", size: 10, min_doc_count: 1,
include: %w(Dataset Publication Software Organization Funder Person Grant Sample Instrument Repository Project) } },
registration_agencies: { terms: { field: "agency", size: facet_count, min_doc_count: 1 } },
created: { date_histogram: { field: "created", interval: "year", format: "year", order: { _key: "desc" }, min_doc_count: 1 },
aggs: { bucket_truncate: { bucket_sort: { size: facet_count } } } },
registered: { date_histogram: { field: "registered", interval: "year", format: "year", order: { _key: "desc" }, min_doc_count: 1 },
aggs: { bucket_truncate: { bucket_sort: { size: facet_count } } } },
providers: { terms: { field: "provider_id_and_name", size: facet_count, min_doc_count: 1 } },
clients: { terms: { field: "client_id_and_name", size: 10, min_doc_count: 1 } },
affiliations: { terms: { field: "affiliation_id_and_name", size: facet_count, min_doc_count: 1 } },
prefixes: { terms: { field: "prefix", size: facet_count, min_doc_count: 1 } },
schema_versions: { terms: { field: "schema_version", size: facet_count, min_doc_count: 1 } },
link_checks_status: { terms: { field: "landing_page.status", size: facet_count, min_doc_count: 1 } },
# link_checks_has_schema_org: { terms: { field: 'landing_page.hasSchemaOrg', size: 2, min_doc_count: 1 } },
# link_checks_schema_org_id: { value_count: { field: "landing_page.schemaOrgId" } },
# link_checks_dc_identifier: { value_count: { field: "landing_page.dcIdentifier" } },
# link_checks_citation_doi: { value_count: { field: "landing_page.citationDoi" } },
# links_checked: { value_count: { field: "landing_page.checked" } },
# sources: { terms: { field: 'source', size: 15, min_doc_count: 1 } },
subjects: { terms: { field: "subjects.subject", size: facet_count, min_doc_count: 1 } },
pid_entities: {
filter: { term: { "subjects.subjectScheme": "PidEntity" } },
aggs: {
subject: { terms: { field: "subjects.subject", size: facet_count, min_doc_count: 1,
include: %w(Dataset Publication Software Organization Funder Person Grant Sample Instrument Repository Project) } },
},
},
},
fields_of_science: {
filter: { term: { "subjects.subjectScheme": "Fields of Science and Technology (FOS)" } },
aggs: {
subject: { terms: { field: "subjects.subject", size: 10, min_doc_count: 1,
include: "FOS:.*" } },
fields_of_science: {
filter: { term: { "subjects.subjectScheme": "Fields of Science and Technology (FOS)" } },
aggs: {
subject: { terms: { field: "subjects.subject", size: facet_count, min_doc_count: 1,
include: "FOS:.*" } },
},
},
},
licenses: { terms: { field: "rights_list.rightsIdentifier", size: 10, min_doc_count: 1 } },
languages: { terms: { field: "language", size: 10, min_doc_count: 1 } },
certificates: { terms: { field: "client.certificate", size: 10, min_doc_count: 1 } },
views: {
date_histogram: { field: "publication_year", interval: "year", format: "year", order: { _key: "desc" }, min_doc_count: 1 },
aggs: {
metric_count: { sum: { field: "view_count" } },
bucket_truncate: { bucket_sort: { size: 10 } },
licenses: { terms: { field: "rights_list.rightsIdentifier", size: facet_count, min_doc_count: 1 } },
languages: { terms: { field: "language", size: facet_count, min_doc_count: 1 } },
certificates: { terms: { field: "client.certificate", size: facet_count, min_doc_count: 1 } },
views: {
date_histogram: { field: "publication_year", interval: "year", format: "year", order: { _key: "desc" }, min_doc_count: 1 },
aggs: {
metric_count: { sum: { field: "view_count" } },
bucket_truncate: { bucket_sort: { size: facet_count } },
},
},
},
downloads: {
date_histogram: { field: "publication_year", interval: "year", format: "year", order: { _key: "desc" }, min_doc_count: 1 },
aggs: {
metric_count: { sum: { field: "download_count" } },
bucket_truncate: { bucket_sort: { size: 10 } },
downloads: {
date_histogram: { field: "publication_year", interval: "year", format: "year", order: { _key: "desc" }, min_doc_count: 1 },
aggs: {
metric_count: { sum: { field: "download_count" } },
bucket_truncate: { bucket_sort: { size: facet_count } },
},
},
},
citations: {
date_histogram: { field: "publication_year", interval: "year", format: "year", order: { _key: "desc" }, min_doc_count: 1 },
aggs: {
metric_count: { sum: { field: "citation_count" } },
bucket_truncate: { bucket_sort: { size: 10 } },
citations: {
date_histogram: { field: "publication_year", interval: "year", format: "year", order: { _key: "desc" }, min_doc_count: 1 },
aggs: {
metric_count: { sum: { field: "citation_count" } },
bucket_truncate: { bucket_sort: { size: facet_count } },
},
},
},
}
}
end
end

def self.provider_aggregations
Expand Down Expand Up @@ -720,7 +722,7 @@ def self.find_by_ids(ids, options = {})
must: must,
},
},
aggregations: query_aggregations,
aggregations: query_aggregations(facet_count: options[:facet_count]),
)
end

Expand Down Expand Up @@ -761,7 +763,7 @@ def self.stats_query(options = {})
# query for graphql, removing options that are not needed
def self.gql_query(query, options = {})
options[:page] ||= {}
options[:facet_count] ||= 10
options[:facet_count] = (options[:facet_count] || 10).to_i
aggregations = gql_query_aggregations(facet_count: options[:facet_count])

# cursor nav uses search_after, this should always be an array of values that match the sort.
Expand Down Expand Up @@ -952,7 +954,7 @@ def self.query(query, options = {})
elsif options[:totals_agg] == "prefix"
prefix_aggregations
else
query_aggregations
query_aggregations(facet_count: options[:facet_count])
end

# Cursor nav uses search_after, this should always be an array of values that match the sort.
Expand Down
55 changes: 55 additions & 0 deletions spec/fixtures/vcr_cassettes/DataciteDoisController/downloads/has_downloads.yml
View file

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

Loading

0 comments on commit 1397676

Please sign in to comment.