Class: Relaton3gpp::DataFetcher

Inherits:
Object
  • Object
show all
Defined in:
lib/relaton_3gpp/data_fetcher.rb

Constant Summary collapse

CURRENT = 
"current.yaml".freeze

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(output, format) ⇒ DataFetcher

Data fetcher initializer

Parameters:

  • output (String) —

    directory to save files

  • format (String) —

    format of output files (xml, yaml, bibxml)



10
11
12
13
14
15
16
17
18
19
 
# File 'lib/relaton_3gpp/data_fetcher.rb', line 10

def initialize(output, format)
  require "fileutils"
  require "net/ftp"
  require "csv"

  @output = output
  @format = format
  @ext = format.sub(/^bib/, "")
  @files = []
end

Class Method Details

.fetch(source, output: "data", format: "yaml") ⇒ Object

Initialize fetcher and run fetch

Parameters:

  • source (Strin) —

    source name

  • output (Strin) (defaults to: "data") —

    directory to save files, default: “data”

  • format (Strin) (defaults to: "yaml") —

    format of output files (xml, yaml, bibxml), default: yaml



32
33
34
35
36
37
38
39
40
 
# File 'lib/relaton_3gpp/data_fetcher.rb', line 32

def self.fetch(source, output: "data", format: "yaml")
  t1 = Time.now
  puts "Started at: #{t1}"
  FileUtils.mkdir_p output
  new(output, format).fetch(source == "status-smg-3GPP-force")
  t2 = Time.now
  puts "Stopped at: #{t2}"
  puts "Done in: #{(t2 - t1).round} sec."
end

Instance Method Details

#add_affiliation(contrib, affiliation) ⇒ Object 



250
251
252
253
254
255
256
257
258
259
260
261
 
# File 'lib/relaton_3gpp/data_fetcher.rb', line 250

def add_affiliation(contrib, affiliation)
  changed = false

  affiliation.each do |a|
    unless contrib.entity.affiliation.include? a
      contrib.entity.affiliation << a
      changed = true
    end
  end

  changed
end

#add_contributor(bib1, bib2) ⇒ Object

rubocop:disable Metrics/MethodLength,Metrics/AbcSize



231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
 
# File 'lib/relaton_3gpp/data_fetcher.rb', line 231

def add_contributor(bib1, bib2) # rubocop:disable Metrics/MethodLength,Metrics/AbcSize
  changed = false

  bib2.contributor.each do |bc|
    next if bc.entity.is_a? RelatonBib::Organization

    existed = bib1.contributor.find { |ic| ic.entity.name == bc.entity.name }
    if existed
      chng = add_affiliation existed, bc.entity.affiliation
      changed ||= chng
    else
      bib1.contributor << bc
      changed = true
    end
  end

  changed
end

#add_transposed_relation(bib1, bib2) ⇒ Relaton3gpp::BibliographicItem

Add transposed relation

Parameters:

Returns:



223
224
225
226
227
228
229
 
# File 'lib/relaton_3gpp/data_fetcher.rb', line 223

def add_transposed_relation(bib1, bib2)
  bib2.relation.each { |r| bib1.relation << r }
  bib2.instance_variable_set :@relation, RelatonBib::DocRelationCollection.new([])
  dec = RelatonBib::FormattedString.new content: "equivalent"
  rel = RelatonBib::DocumentRelation.new(type: "adoptedAs", bibitem: bib2, description: dec)
  bib1.relation << rel
end

#check_transposed_date(bib, existed) ⇒ Array<Relaton3gpp::BibliographicItem, Boolean>

Check if date of one bibliographic item is transposed to another

Parameters:

Returns:



204
205
206
207
208
209
210
211
212
213
 
# File 'lib/relaton_3gpp/data_fetcher.rb', line 204

def check_transposed_date(bib, existed)
  if bib.date[0].on < existed.date[0].on
    add_transposed_relation bib, existed
    [bib, existed, true]
  elsif bib.date[0].on > existed.date[0].on
    add_transposed_relation existed, bib
    [existed, bib, true]
  else [bib, existed, false]
  end
end

#fetch(renewal) ⇒ Object

Parse documents

Parameters:

  • renewal (Boolean) —

    force to update all documents



47
48
49
50
51
52
53
54
55
56
57
58
59
60
 
# File 'lib/relaton_3gpp/data_fetcher.rb', line 47

def fetch(renewal) # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
  file = get_file renewal
  return unless file && File.exist?(file) && File.size(file) > 20_000_000

  if renewal
    FileUtils.rm_f File.join(@output, "/*") # if renewal && dbs["2001-04-25_schedule"].any?
    index.remove_all # if renewal
  end
  CSV.open(file, "r:bom|utf-8", headers: true).each do |row|
    save_doc Parser.parse(row)
  end
  File.write CURRENT, @current.to_yaml, encoding: "UTF-8"
  index.save
end

#file_name(bib) ⇒ String

Generate file name

Parameters:

  • bib (RelatonW3c::W3cBibliographicItem) —

    bibliographic item

Returns:

  • (String) —

    file name



278
279
280
281
 
# File 'lib/relaton_3gpp/data_fetcher.rb', line 278

def file_name(bib)
  name = bib.docnumber.gsub(/[\s,:\/]/, "_").squeeze("_").upcase
  File.join @output, "#{name}.#{@ext}"
end

#get_file(renewal) ⇒ String?

Get file from FTP. If file does not exist or changed, return nil

Parameters:

  • renewal (Boolean) —

    force to update all documents

Returns:

  • (String, nil) —

    file name



69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
 
# File 'lib/relaton_3gpp/data_fetcher.rb', line 69

def get_file(renewal) # rubocop:disable Metrics/MethodLength, Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
  @current = YAML.load_file CURRENT if File.exist? CURRENT
  @current ||= {}
  n = 0
  begin
    ftp = Net::FTP.new("www.3gpp.org")
    ftp.resume = true
    ftp.
    ftp.chdir "/Information/Databases/"
    file_path = ftp.list("*.csv").first
    return unless file_path

    d, t, _, file = file_path.split
    dt = DateTime.strptime("#{d} #{t}", "%m-%d-%y %I:%M%p")
    if !renewal && file == @current["file"] && !@current["date"].empty? && dt == DateTime.parse(@current["date"])
      return
    end

    tmp_file = File.join Dir.tmpdir, "3gpp.csv"
    ftp.get(file, tmp_file)
  rescue Net::ReadTimeout => e
    n += 1
    retry if n < 5
    raise e
  end
  @current["file"] = file
  @current["date"] = dt.to_s
  tmp_file
end

#index ⇒ Object 



21
22
23
 
# File 'lib/relaton_3gpp/data_fetcher.rb', line 21

def index
  @index ||= Relaton::Index.find_or_create "3gpp", file: "index-v1.yaml"
end

#merge_duplication(bib, file) ⇒ Relaton3gpp::BibliographicItem?

Merge duplication

Parameters:

Returns:



148
149
150
151
152
153
154
155
156
157
 
# File 'lib/relaton_3gpp/data_fetcher.rb', line 148

def merge_duplication(bib, file)
  hash = YAML.load_file file
  existed = BibliographicItem.from_hash hash
  changed = update_link bib, existed
  bib1, bib2, chng = transposed_relation bib, existed
  changed ||= chng
  chng = add_contributor(bib1, bib2)
  changed ||= chng
  bib1 if changed
end

#save_doc(bib) ⇒ Object

Save document to file

Parameters:

  • bib (RelatonW3c::W3cBibliographicItem, nil) —

    bibliographic item



125
126
127
128
129
130
131
132
133
134
135
136
137
138
 
# File 'lib/relaton_3gpp/data_fetcher.rb', line 125

def save_doc(bib) # rubocop:disable Metrics/MethodLength
  return unless bib

  bib1 = bib
  file = file_name(bib1)
  if @files.include? file
    bib1 = merge_duplication bib1, file
    Util.warn "File #{file} already exists. Document: #{bib.docnumber}" if bib1.nil?
  else
    @files << file
    index.add_or_update bib1.docnumber, file
  end
  File.write file, serialise(bib1), encoding: "UTF-8" unless bib1.nil?
end

#serialise(bib) ⇒ Object 



263
264
265
266
267
268
269
 
# File 'lib/relaton_3gpp/data_fetcher.rb', line 263

def serialise(bib)
  case @format
  when "xml" then bib.to_xml(bibdata: true)
  when "yaml" then bib.to_hash.to_yaml
  else bib.send("to_#{@format}")
  end
end

#transposed_relation(bib, existed) ⇒ Array<Relaton3gpp::BibliographicItem, Boolean>

If one of bibliographic items has date gereater than anotherm=, make it relation

Parameters:

Returns:



187
188
189
190
191
192
193
 
# File 'lib/relaton_3gpp/data_fetcher.rb', line 187

def transposed_relation(bib, existed) # rubocop:disable Metrics/CyclomaticComplexity
  return [bib, existed, false] if bib.date.none? && existed.date.none? ||
    bib.date.any? && existed.date.none?
  return [existed, bib, true] if bib.date.none? && existed.date.any?

  check_transposed_date bib, existed
end

#update_link(bib1, bib2) ⇒ Boolean

Update link in case one of bibliographic items has no link

Parameters:

Returns:

  • (Boolean) —

    true if link has been updated



167
168
169
170
171
172
173
174
175
176
 
# File 'lib/relaton_3gpp/data_fetcher.rb', line 167

def update_link(bib1, bib2)
  if bib1.link.any? && bib2.link.empty?
    bib2.instance_variable_set(:@link, bib1.link)
    true
  elsif bib1.link.empty? && bib2.link.any?
    bib1.instance_variable_set(:@link, bib2.link)
    true
  else false
  end
end