Module: Bolognese::MetadataUtils

Includes:
AuthorUtils, DataciteUtils, DoiUtils, Readers::BibtexReader, Readers::CiteprocReader, Readers::CodemetaReader, Readers::CrossciteReader, Readers::CrossrefReader, Readers::DataciteJsonReader, Readers::DataciteReader, Readers::NpmReader, Readers::RisReader, Readers::SchemaOrgReader, Utils, Writers::BibtexWriter, Writers::CitationWriter, Writers::CiteprocWriter, Writers::CodemetaWriter, Writers::CrossciteWriter, Writers::CrossrefWriter, Writers::CsvWriter, Writers::DataciteJsonWriter, Writers::DataciteWriter, Writers::JatsWriter, Writers::RdfXmlWriter, Writers::RisWriter, Writers::SchemaOrgWriter, Writers::TurtleWriter
Included in:
Metadata
Defined in:
lib/bolognese/metadata_utils.rb

Constant Summary

Constants included from Readers::SchemaOrgReader

Readers::SchemaOrgReader::SO_TO_DC_RELATION_TYPES, Readers::SchemaOrgReader::SO_TO_DC_REVERSE_RELATION_TYPES

Constants included from Readers::RisReader

Readers::RisReader::RIS_TO_BIB_TRANSLATIONS, Readers::RisReader::RIS_TO_CP_TRANSLATIONS, Readers::RisReader::RIS_TO_SO_TRANSLATIONS

Constants included from Readers::CiteprocReader

Readers::CiteprocReader::CP_TO_RIS_TRANSLATIONS, Readers::CiteprocReader::CP_TO_SO_TRANSLATIONS

Constants included from Readers::BibtexReader

Readers::BibtexReader::BIB_TO_CP_TRANSLATIONS, Readers::BibtexReader::BIB_TO_RIS_TRANSLATIONS, Readers::BibtexReader::BIB_TO_SO_TRANSLATIONS

Constants included from Utils

Utils::BIB_TO_CR_TRANSLATIONS, Utils::BIB_TO_DC_TRANSLATIONS, Utils::CP_TO_DC_TRANSLATIONS, Utils::CR_TO_BIB_TRANSLATIONS, Utils::CR_TO_CP_TRANSLATIONS, Utils::CR_TO_DC_TRANSLATIONS, Utils::CR_TO_JATS_TRANSLATIONS, Utils::CR_TO_RIS_TRANSLATIONS, Utils::CR_TO_SO_TRANSLATIONS, Utils::DC_TO_CP_TRANSLATIONS, Utils::DC_TO_RIS_TRANSLATIONS, Utils::DC_TO_SO_TRANSLATIONS, Utils::NORMALIZED_LICENSES, Utils::RIS_TO_DC_TRANSLATIONS, Utils::SO_TO_BIB_TRANSLATIONS, Utils::SO_TO_CP_TRANSLATIONS, Utils::SO_TO_DC_TRANSLATIONS, Utils::SO_TO_JATS_TRANSLATIONS, Utils::SO_TO_RIS_TRANSLATIONS, Utils::UNKNOWN_INFORMATION

Constants included from AuthorUtils

AuthorUtils::IDENTIFIER_SCHEME_URIS

Instance Attribute Summary collapse

Instance Method Summary collapse

Methods included from Writers::TurtleWriter

#turtle

Methods included from Writers::SchemaOrgWriter

#schema_hsh, #schema_org

Methods included from Writers::RisWriter

#ris

Methods included from Writers::RdfXmlWriter

#rdf_xml

Methods included from Writers::JatsWriter

#date, #insert_authors, #insert_citation, #insert_citation_title, #insert_contributor, #insert_editors, #insert_fpage, #insert_issue, #insert_lpage, #insert_pub_id, #insert_publication_date, #insert_publisher_name, #insert_source, #insert_version, #insert_volume, #is_article?, #is_chapter?, #is_data?, #jats, #publication_type

Methods included from Writers::DataciteJsonWriter

#datacite_json

Methods included from Writers::DataciteWriter

#datacite

Methods included from Writers::CsvWriter

#csv

Methods included from Writers::CrossrefWriter

#crossref

Methods included from Writers::CrossciteWriter

#crosscite

Methods included from Writers::CodemetaWriter

#codemeta

Methods included from Writers::CiteprocWriter

#citeproc

Methods included from Writers::CitationWriter

#citation

Methods included from Writers::BibtexWriter

#bibtex

Methods included from Readers::SchemaOrgReader

#get_schema_org, #read_schema_org, #schema_org_has_part, #schema_org_is_identical_to, #schema_org_is_new_version_of, #schema_org_is_part_of, #schema_org_is_previous_version_of, #schema_org_is_referenced_by, #schema_org_is_supplement_to, #schema_org_is_supplemented_by, #schema_org_references, #schema_org_related_identifier, #schema_org_reverse_related_identifier

Methods included from Readers::RisReader

#read_ris, #ris_meta

Methods included from Readers::NpmReader

#get_npm, #read_npm

Methods included from Readers::DataciteJsonReader

#read_datacite_json

Methods included from Readers::DataciteReader

#get_datacite, #get_titles, #read_datacite, #set_container

Methods included from Readers::CrossrefReader

#crossref_alternate_identifiers, #crossref_date_published, #crossref_description, #crossref_funding_reference, #crossref_is_part_of, #crossref_license, #crossref_people, #crossref_references, #get_crossref, #read_crossref

Methods included from Readers::CrossciteReader

#read_crosscite

Methods included from Readers::CodemetaReader

#get_codemeta, #read_codemeta

Methods included from Readers::CiteprocReader

#read_citeproc

Methods included from Readers::BibtexReader

#read_bibtex

Methods included from Utils

#find_from_format, #find_from_format_by_ext, #find_from_format_by_filename, #find_from_format_by_id, #find_from_format_by_string, #from_citeproc, #from_datacite_json, #from_schema_org, #from_schema_org_contributors, #from_schema_org_creators, #get_contributor, #get_date, #get_date_from_date_parts, #get_date_from_parts, #get_date_parts, #get_date_parts_from_parts, #get_datetime_from_iso8601, #get_datetime_from_time, #get_identifier, #get_identifier_type, #get_iso8601_date, #get_series_information, #get_year_month, #get_year_month_day, #github_as_codemeta_url, #github_as_owner_url, #github_as_release_url, #github_as_repo_url, #github_from_url, #github_owner_from_url, #github_release_from_url, #github_repo_from_url, #hsh_to_fos, #hsh_to_spdx, #jsonlint, #map_hash_keys, #name_to_fos, #name_to_spdx, #normalize_cc_url, #normalize_id, #normalize_ids, #normalize_issn, #normalize_licenses, #normalize_orcid, #normalize_url, #orcid_as_url, #orcid_from_url, #parse_attributes, #sanitize, #to_citeproc, #to_datacite_json, #to_identifier, #to_ris, #to_schema_org, #to_schema_org_container, #to_schema_org_contributors, #to_schema_org_creators, #to_schema_org_funder, #to_schema_org_identifiers, #to_schema_org_relation, #to_schema_org_spatial_coverage, #validate_orcid, #validate_orcid_scheme, #validate_url

Methods included from DataciteUtils

#datacite_errors, #datacite_xml, #insert_alternate_identifiers, #insert_contributors, #insert_creators, #insert_dates, #insert_descriptions, #insert_formats, #insert_funding_references, #insert_geo_locations, #insert_identifier, #insert_language, #insert_person, #insert_publication_year, #insert_publisher, #insert_related_identifiers, #insert_related_items, #insert_resource_type, #insert_rights_list, #insert_sizes, #insert_subjects, #insert_titles, #insert_version, #insert_work, #root_attributes

Methods included from AuthorUtils

#authors_as_string, #cleanup_author, #get_affiliations, #get_authors, #get_one_author, #is_personal_name?, #name_exists?

Methods included from DoiUtils

#doi_api_url, #doi_as_url, #doi_from_url, #doi_resolver, #get_doi_ra, #normalize_doi, #validate_doi, #validate_funder_doi, #validate_prefix

Instance Attribute Details

#name_detector ⇒ Object (readonly)

recognize given name. Can be loaded once as ::NameDetector, e.g. in a Rails initializer



106
107
108
 
# File 'lib/bolognese/metadata_utils.rb', line 106

def name_detector
  @name_detector
end

#reverse ⇒ Object (readonly)

Returns the value of attribute reverse.



68
69
70
 
# File 'lib/bolognese/metadata_utils.rb', line 68

def reverse
  @reverse
end

Instance Method Details

#citeproc_hsh ⇒ Object 



135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
 
# File 'lib/bolognese/metadata_utils.rb', line 135

def citeproc_hsh
  page = container.to_h["firstPage"].present? ? [container["firstPage"], container["lastPage"]].compact.join("-") : nil
  if Array.wrap(creators).size == 1 && Array.wrap(creators).first.fetch("name", nil) == ":(unav)"
    author = nil
  else
    author = to_citeproc(creators)
  end

  if types["resourceTypeGeneral"] == "Software" && version_info.present?
    type = "book"
  else
    type = types["citeproc"]
  end

  {
    "type" => type,
    "id" => normalize_doi(doi),
    "categories" => Array.wrap(subjects).map { |k| parse_attributes(k, content: "subject", first: true) }.presence,
    "language" => language,
    "author" => author,
    "contributor" => to_citeproc(contributors),
    "issued" => get_date(dates, "Issued") ? get_date_parts(get_date(dates, "Issued")) : get_date_parts(publication_year.to_s),
    "submitted" => Array.wrap(dates).find { |d| d["dateType"] == "Submitted" }.to_h.fetch("__content__", nil),
    "abstract" => parse_attributes(descriptions, content: "description", first: true),
    "container-title" => container_title,
    "DOI" => doi,
    "volume" => container.to_h["volume"],
    "issue" => container.to_h["issue"],
    "page" => page,
    "publisher" => publisher,
    "title" => parse_attributes(titles, content: "title", first: true),
    "URL" => url,
    "copyright" => Array.wrap(rights_list).map { |l| l["rights"] }.first,
    "version" => version_info
  }.compact.symbolize_keys
end

#container_title ⇒ Object 



95
96
97
98
99
100
101
102
103
 
# File 'lib/bolognese/metadata_utils.rb', line 95

def container_title
  if container.present?
    container["title"]
  elsif types["citeproc"] == "article-journal"
    publisher
  else
    nil
  end
end

#crosscite_hsh ⇒ Object 



172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
 
# File 'lib/bolognese/metadata_utils.rb', line 172

def crosscite_hsh
  {
    "id" => normalize_doi(doi),
    "doi" => doi,
    "url" => url,
    "types" => types,
    "creators" => creators,
    "titles" => titles,
    "publisher" => publisher,
    "container" => container,
    "subjects" => subjects,
    "contributors" => contributors,
    "dates" => dates,
    "publication_year" => publication_year,
    "language" => language,
    "identifiers" => identifiers,
    "sizes" => sizes,
    "formats" => formats,
    "version" => version_info,
    "rights_list" => rights_list,
    "descriptions" => descriptions,
    "geo_locations" => geo_locations,
    "funding_references" => funding_references,
    "related_identifiers" => related_identifiers,
    "related_items" => related_items,
    "schema_version" => schema_version,
    "provider_id" => provider_id,
    "client_id" => client_id,
    "agency" => agency,
    "state" => state
  }.compact
end

#graph ⇒ Object 



123
124
125
126
127
128
129
130
131
132
133
 
# File 'lib/bolognese/metadata_utils.rb', line 123

def graph
  # preload schema_org context
  JSON::LD::Context.add_preloaded(
    'http://schema.org/',
    JSON::LD::Context.new.parse('resources/schema_org/jsonldcontext.json')
  )

  RDF::Graph.new << JSON::LD::API.toRdf(schema_hsh)
rescue NameError
  nil
end

#locale ⇒ Object 



209
210
211
 
# File 'lib/bolognese/metadata_utils.rb', line 209

def locale
  @locale ||= "en-US"
end

#raw ⇒ Object

replace DOI in XML if provided in options



81
82
83
84
85
86
87
88
89
 
# File 'lib/bolognese/metadata_utils.rb', line 81

def raw
  r = string.present? ? string.strip : nil
  return r unless (from == "datacite" && r.present?)

  doc = Nokogiri::XML(string, nil, 'UTF-8', &:noblanks)
  node = doc.at_css("identifier")
  node.content = doi.to_s.upcase if node.present? && doi.present?
  doc.to_xml.strip
end

#should_passthru ⇒ Object 



91
92
93
 
# File 'lib/bolognese/metadata_utils.rb', line 91

def should_passthru
  (from == "datacite") && regenerate.blank? && raw.present?
end

#style ⇒ Object 



205
206
207
 
# File 'lib/bolognese/metadata_utils.rb', line 205

def style
  @style ||= "apa"
end