Module: RelatonBib::BibXMLParser
Constant Summary collapse
- SERIESINFONAMES =
SeriesInfo what should be saved as docidentifiers in the Relaton model.
["DOI"].freeze
- RFCPREFIXES =
%w[RFC BCP FYI STD].freeze
- FLAVOR =
nil
- ORGNAMES =
{ "IEEE" => "Institute of Electrical and Electronics Engineers", "W3C" => "World Wide Web Consortium", "3GPP" => "3rd Generation Partnership Project", }.freeze
Instance Method Summary collapse
- #abstracts(ref) ⇒ Array<RelatonBib::FormattedString>
- #add_contact(conts, type, value) ⇒ Object
- #address(postal) ⇒ Object
- #affiliation(author) ⇒ Array<RelatonBib::Affiliation>
- #bib_item(**attrs) ⇒ RelatonBib::BibliographicItem
- #committee(wgr) ⇒ RelatonBib::TechnicalCommittee
- #contacts(addr) ⇒ Array<RelatonBib::Address, RelatonBib::Phone>
- #contributor_role(author) ⇒ Hash
- #contributors(reference) ⇒ Array<Hash>
-
#create_docid(id, ver) ⇒ Object
rubocop:disable Metrics/MethodLength.
-
#dates(reference) ⇒ Array<RelatonBib::BibliographicDate>
Extract date from reference.
-
#docids(reference, ver) ⇒ Array<RelatonBib::DocumentIdentifier>
Extract document identifiers from reference.
- #docnumber(reference) ⇒ Object
- #doctype(anchor) ⇒ String
- #editorialgroup(reference) ⇒ RelatonBib::EditorialGroup?
- #fetch_rfc(reference, is_relation: false, url: nil, ver: nil) ⇒ RelatonBib::BibliographicItem
-
#forename(initials, lang = nil, script = nil) ⇒ Array<RelatonBib::Forename>
Create forenames with initials.
- #formattedref(reference) ⇒ RelatonBib::FormattedRef?
- #full_name(fname, sname, inits, lang) ⇒ RelatonBib::FullName
- #id_to_pref_num(id) ⇒ Object
- #language(reference) ⇒ String
- #link(reference, url, ver) ⇒ Array<Hash>
- #localized_string(content, lang, script = nil) ⇒ RelatonBib::LocalizedString?
- #month(mon) ⇒ Object
- #new_org(name, abbr = nil) ⇒ RelatonBib::Organization
- #organization(contrib) ⇒ RelatonBib::Organization?
-
#parse(bibxml, url: nil, is_relation: false, ver: nil) ⇒ <Type>
Parse BibXML content.
- #person(author, lang) ⇒ RelatonBib::Person?
-
#pubid_type(id) ⇒ String
Extract document identifier type from identifier.
- #relations(reference) ⇒ Hash
-
#series(reference) ⇒ Array<RelatonBib::Series>
Extract series form reference.
-
#status(reference) ⇒ RelatonBib::DocumentStatus
extract status.
- #titles(reference) ⇒ Array<Hash>
Instance Method Details
#abstracts(ref) ⇒ Array<RelatonBib::FormattedString>
205 206 207 208 209 210 211 212 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 205 def abstracts(ref) ref.xpath("./front/abstract").map do |a| c = a.inner_html.gsub(/\s*(<\/?)t(>)\s*/, '\1p\2') .gsub(/[\t\n]/, " ").squeeze " " FormattedString.new(content: c, language: language(ref), script: "Latn", format: "text/html") end end |
#add_contact(conts, type, value) ⇒ Object
332 333 334 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 332 def add_contact(conts, type, value) conts << Contact.new(type: type, value: value.text) if value end |
#address(postal) ⇒ Object
318 319 320 321 322 323 324 325 326 327 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 318 def address(postal) # rubocop:disable Metrics/CyclomaticComplexity street = [postal.at("./postalLine | ./street")&.text].compact Address.new( street: street, city: postal.at("./city")&.text, postcode: postal.at("./code")&.text, country: postal.at("./country")&.text, state: postal.at("./region")&.text, ) end |
#affiliation(author) ⇒ Array<RelatonBib::Affiliation>
279 280 281 282 283 284 285 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 279 def affiliation() o = .at("./organization") return [] if o.nil? || o.text.empty? org = new_org o.text, o[:abbrev] [Affiliation.new(organization: org)] end |
#bib_item(**attrs) ⇒ RelatonBib::BibliographicItem
68 69 70 71 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 68 def bib_item(**attrs) # attrs[:place] = ["Fremont, CA"] BibliographicItem.new(**attrs) end |
#committee(wgr) ⇒ RelatonBib::TechnicalCommittee
379 380 381 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 379 def committee(wgr) TechnicalCommittee.new wgr end |
#contacts(addr) ⇒ Array<RelatonBib::Address, RelatonBib::Phone>
304 305 306 307 308 309 310 311 312 313 314 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 304 def contacts(addr) conts = [] return conts unless addr postal = addr.at("./postal") conts << address(postal) if postal&.at("./city") && postal&.at("./country") add_contact(conts, "phone", addr.at("./phone")) add_contact(conts, "email", addr.at("./email")) add_contact(conts, "uri", addr.at("./uri")) conts end |
#contributor_role(author) ⇒ Hash
338 339 340 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 338 def contributor_role() { type: [:role] || "author" } end |
#contributors(reference) ⇒ Array<Hash>
216 217 218 219 220 221 222 223 224 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 216 def contributors(reference) lang = language reference reference.xpath("./front/author").map do |contrib| entity = person(contrib, lang) || organization(contrib) next unless entity { entity: entity, role: [contributor_role(contrib)] } end.compact end |
#create_docid(id, ver) ⇒ Object
rubocop:disable Metrics/MethodLength
122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 122 def create_docid(id, ver) # rubocop:disable Metrics/MethodLength pref, num = id_to_pref_num(id) if RFCPREFIXES.include?(pref) pid = "#{pref} #{num.sub(/^-?0+/, '')}" type = pubid_type id elsif %w[I-D draft].include?(pref) pid = "draft-#{num}" pid.sub!(/(?<=-)\d{2}$/, ver) if ver type = "Internet-Draft" else pid = pref ? "#{pref} #{num}" : id type = pubid_type id end DocumentIdentifier.new(type: type, id: pid, primary: true) end |
#dates(reference) ⇒ Array<RelatonBib::BibliographicDate>
Extract date from reference.
356 357 358 359 360 361 362 363 364 365 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 356 def dates(reference) # rubocop:disable Metrics/CyclomaticComplexity, Metrics/AbcSize date = reference.at "./front/date" return [] if date.nil? || date[:year].nil? || date[:year].empty? d = date[:year] d += "-#{month(date[:month])}" if date[:month] && !date[:month].empty? d += "-#{date[:day]}" if date[:day] # date = Time.parse(d).strftime "%Y-%m-%d" [BibliographicDate.new(type: "published", on: d)] end |
#docids(reference, ver) ⇒ Array<RelatonBib::DocumentIdentifier>
Extract document identifiers from reference
87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 87 def docids(reference, ver) # rubocop:disable Metrics/MethodLength,Metrics/CyclomaticComplexity,Metrics/PerceivedComplexity,Metrics/AbcSize ret = [] si = reference.at("./seriesInfo[@name='Internet-Draft']", "./front/seriesInfo[@name='Internet-Draft']") if si id = si[:value] id.sub!(/(?<=-)\d{2}$/, ver) if ver ret << DocumentIdentifier.new(type: "Internet-Draft", id: id, primary: true) else id = reference[:anchor] || reference[:docName] || reference[:number] ret << create_docid(id, ver) if id end %w[anchor docName number].each do |atr| if reference[atr] pref, num = id_to_pref_num reference[atr] atrid = if atr == "anchor" && RFCPREFIXES.include?(pref) "#{pref}#{num.sub(/^-?0+/, '')}" else reference[atr] end type = pubid_type id ret << DocumentIdentifier.new(id: atrid, type: type, scope: atr) end end ret + reference.xpath("./seriesInfo", "./front/seriesInfo").map do |si| next unless SERIESINFONAMES.include? si[:name] id = si[:value] # id.sub!(/(?<=-)\d{2}$/, ver) if ver && si[:name] == "Internet-Draft" DocumentIdentifier.new(id: id, type: si[:name]) end.compact end |
#docnumber(reference) ⇒ Object
62 63 64 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 62 def docnumber(reference) reference[:anchor]&.sub(/^\w+\./, "") end |
#doctype(anchor) ⇒ String
409 410 411 412 413 414 415 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 409 def doctype(anchor) case anchor when /I-D/ then "internet-draft" when /IEEE/ then "ieee" else "rfc" end end |
#editorialgroup(reference) ⇒ RelatonBib::EditorialGroup?
369 370 371 372 373 374 375 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 369 def editorialgroup(reference) tc = reference.xpath("./front/workgroup").map do |ed| wg = WorkGroup.new name: ed.text committee wg end EditorialGroup.new tc if tc.any? end |
#fetch_rfc(reference, is_relation: false, url: nil, ver: nil) ⇒ RelatonBib::BibliographicItem
35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 35 def fetch_rfc(reference, is_relation: false, url: nil, ver: nil) # rubocop:disable Metrics/AbcSize,Metrics/MethodLength return unless reference hash = { is_relation: is_relation, docnumber: docnumber(reference), type: "standard", docid: docids(reference, ver), status: status(reference), language: [language(reference)], script: ["Latn"], link: link(reference, url, ver), title: titles(reference), formattedref: formattedref(reference), abstract: abstracts(reference), contributor: contributors(reference), relation: relations(reference), date: dates(reference), editorialgroup: editorialgroup(reference), series: series(reference), keyword: reference.xpath("front/keyword").map(&:text), doctype: doctype(reference[:anchor]), } # hash[:fetched] = Date.today.to_s unless is_relation bib_item(**hash) end |
#forename(initials, lang = nil, script = nil) ⇒ Array<RelatonBib::Forename>
Create forenames with initials
269 270 271 272 273 274 275 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 269 def forename(initials, lang = nil, script = nil) return [] unless initials initials.split(/\.-?\s?|\s/).map do |i| Forename.new(initial: i, language: lang, script: script) end end |
#formattedref(reference) ⇒ RelatonBib::FormattedRef?
192 193 194 195 196 197 198 199 200 201 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 192 def formattedref(reference) return if reference.at "./front/title" cont = (reference[:anchor] || reference[:docName] || reference[:number]) if cont FormattedRef.new( content: cont, language: language(reference), script: "Latn", ) end end |
#full_name(fname, sname, inits, lang) ⇒ RelatonBib::FullName
253 254 255 256 257 258 259 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 253 def full_name(fname, sname, inits, lang) initials = localized_string(inits, lang) if inits FullName.new( completename: localized_string(fname, lang), initials: initials, forename: forename(inits, lang), surname: localized_string(sname, lang) ) end |
#id_to_pref_num(id) ⇒ Object
138 139 140 141 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 138 def id_to_pref_num(id) tn = /^(?<pref>I-D|draft|3GPP|W3C|[A-Z]{2,})[._-]?(?<num>.+)/.match id tn && tn.to_a[1..2] end |
#language(reference) ⇒ String
75 76 77 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 75 def language(reference) reference[:lang] || "en" end |
#link(reference, url, ver) ⇒ Array<Hash>
169 170 171 172 173 174 175 176 177 178 179 180 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 169 def link(reference, url, ver) l = [] l << { type: "xml", content: url } if url l << { type: "src", content: reference[:target] } if reference[:target] if /^I-D/.match? reference[:anchor] reference.xpath("format").each do |f| c = ver ? f[:target].sub(/(?<=-)\d{2}(?=\.)/, ver) : f[:target] l << { type: f[:type], content: c } end end l end |
#localized_string(content, lang, script = nil) ⇒ RelatonBib::LocalizedString?
298 299 300 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 298 def localized_string(content, lang, script = nil) LocalizedString.new(content, lang, script) if content end |
#month(mon) ⇒ Object
383 384 385 386 387 388 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 383 def month(mon) # return 1 if !mon || mon.empty? return mon if /^\d+$/.match? mon Date::MONTHNAMES.index { |m| m&.include? mon }.to_s.rjust 2, "0" end |
#new_org(name, abbr = nil) ⇒ RelatonBib::Organization
290 291 292 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 290 def new_org(name, abbr = nil) Organization.new name: name, abbreviation: abbr end |
#organization(contrib) ⇒ RelatonBib::Organization?
239 240 241 242 243 244 245 246 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 239 def organization(contrib) org = contrib.at("./organization") orgname = org.text.strip return if orgname.empty? name = ORGNAMES[orgname] || orgname new_org name, org[:abbrev] end |
#parse(bibxml, url: nil, is_relation: false, ver: nil) ⇒ <Type>
Parse BibXML content
25 26 27 28 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 25 def parse(bibxml, url: nil, is_relation: false, ver: nil) doc = Nokogiri::XML bibxml fetch_rfc doc.at("/referencegroup", "/reference"), url: url, is_relation: is_relation, ver: ver end |
#person(author, lang) ⇒ RelatonBib::Person?
229 230 231 232 233 234 235 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 229 def person(, lang) return unless [:fullname] || [:surname] name = full_name([:fullname], [:surname], [:initials], lang) Person.new(name: name, affiliation: affiliation(), contact: contacts(.at("./address"))) end |
#pubid_type(id) ⇒ String
Extract document identifier type from identifier
150 151 152 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 150 def pubid_type(id) id_to_pref_num(id)&.first end |
#relations(reference) ⇒ Hash
344 345 346 347 348 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 344 def relations(reference) reference.xpath("reference").map do |ref| { type: "includes", bibitem: fetch_rfc(ref, is_relation: true) } end end |
#series(reference) ⇒ Array<RelatonBib::Series>
Extract series form reference
396 397 398 399 400 401 402 403 404 405 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 396 def series(reference) reference.xpath("./seriesInfo", "./front/seriesInfo").map do |si| next if SERIESINFONAMES.include?(si[:name]) || si[:stream] || si[:status] t = TypedTitleString.new( content: si[:name], language: language(reference), script: "Latn", ) Series.new(title: t, number: si[:value], type: "main") end.compact end |
#status(reference) ⇒ RelatonBib::DocumentStatus
extract status
160 161 162 163 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 160 def status(reference) st = reference.at("./seriesinfo[@status]") DocumentStatus.new(stage: st[:status]) if st end |
#titles(reference) ⇒ Array<Hash>
184 185 186 187 188 |
# File 'lib/relaton_bib/bibxml_parser.rb', line 184 def titles(reference) reference.xpath("./front/title").map do |title| { content: title.text, language: language(reference), script: "Latn" } end end |