relaton-un 1.0.1 → 1.4.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -119,6 +119,9 @@
119
119
  <optional>
120
120
  <ref name="session"/>
121
121
  </optional>
122
+ <optional>
123
+ <ref name="job_number"/>
124
+ </optional>
122
125
  </define>
123
126
  <define name="preface">
124
127
  <element name="preface">
@@ -138,6 +141,109 @@
138
141
  <ref name="Basic-Section"/>
139
142
  </element>
140
143
  </define>
144
+ <define name="Clause-Section">
145
+ <optional>
146
+ <attribute name="id">
147
+ <data type="ID"/>
148
+ </attribute>
149
+ </optional>
150
+ <optional>
151
+ <attribute name="language"/>
152
+ </optional>
153
+ <optional>
154
+ <attribute name="script"/>
155
+ </optional>
156
+ <optional>
157
+ <attribute name="inline-header">
158
+ <data type="boolean"/>
159
+ </attribute>
160
+ </optional>
161
+ <optional>
162
+ <attribute name="obligation">
163
+ <choice>
164
+ <value>normative</value>
165
+ <value>informative</value>
166
+ </choice>
167
+ </attribute>
168
+ </optional>
169
+ <optional>
170
+ <attribute name="unnumbered">
171
+ <data type="boolean"/>
172
+ </attribute>
173
+ </optional>
174
+ <optional>
175
+ <ref name="section-title"/>
176
+ </optional>
177
+ <group>
178
+ <group>
179
+ <zeroOrMore>
180
+ <ref name="BasicBlock"/>
181
+ </zeroOrMore>
182
+ <zeroOrMore>
183
+ <ref name="note"/>
184
+ </zeroOrMore>
185
+ </group>
186
+ <zeroOrMore>
187
+ <choice>
188
+ <ref name="clause-subsection"/>
189
+ <ref name="terms"/>
190
+ <ref name="definitions"/>
191
+ </choice>
192
+ </zeroOrMore>
193
+ </group>
194
+ </define>
195
+ <define name="Annex-Section">
196
+ <optional>
197
+ <attribute name="id">
198
+ <data type="ID"/>
199
+ </attribute>
200
+ </optional>
201
+ <optional>
202
+ <attribute name="language"/>
203
+ </optional>
204
+ <optional>
205
+ <attribute name="script"/>
206
+ </optional>
207
+ <optional>
208
+ <attribute name="inline-header">
209
+ <data type="boolean"/>
210
+ </attribute>
211
+ </optional>
212
+ <optional>
213
+ <attribute name="obligation">
214
+ <choice>
215
+ <value>normative</value>
216
+ <value>informative</value>
217
+ </choice>
218
+ </attribute>
219
+ </optional>
220
+ <optional>
221
+ <attribute name="unnumbered">
222
+ <data type="boolean"/>
223
+ </attribute>
224
+ </optional>
225
+ <optional>
226
+ <ref name="section-title"/>
227
+ </optional>
228
+ <group>
229
+ <group>
230
+ <zeroOrMore>
231
+ <ref name="BasicBlock"/>
232
+ </zeroOrMore>
233
+ <zeroOrMore>
234
+ <ref name="note"/>
235
+ </zeroOrMore>
236
+ </group>
237
+ <zeroOrMore>
238
+ <choice>
239
+ <ref name="annex-subsection"/>
240
+ <ref name="terms"/>
241
+ <ref name="definitions"/>
242
+ <ref name="references"/>
243
+ </choice>
244
+ </zeroOrMore>
245
+ </group>
246
+ </define>
141
247
  </include>
142
248
  <define name="session">
143
249
  <element name="session">
@@ -227,6 +333,11 @@
227
333
  </choice>
228
334
  </element>
229
335
  </define>
336
+ <define name="job_number">
337
+ <element name="job_number">
338
+ <text/>
339
+ </element>
340
+ </define>
230
341
  <define name="un-standard">
231
342
  <element name="un-standard">
232
343
  <ref name="bibdata"/>
@@ -1,4 +1,4 @@
1
- require "relaton_iso_bib"
1
+ require "relaton_bib"
2
2
  require "relaton_un/version"
3
3
  require "relaton_un/un_bibliographic_item"
4
4
  require "relaton_un/un_bibliography"
@@ -10,6 +10,11 @@ module RelatonUn
10
10
  @committee = committee
11
11
  end
12
12
 
13
+ # @return [true]
14
+ def presence?
15
+ true
16
+ end
17
+
13
18
  # @param builder [Nokogiri::XML::Builder]
14
19
  def to_xml(builder)
15
20
  builder.editorialgroup do |b|
@@ -21,5 +26,13 @@ module RelatonUn
21
26
  def to_hash
22
27
  single_element_array(committee.map { |c| { "committee" => c } })
23
28
  end
29
+
30
+ # @param prefix [String]
31
+ # @return [String]
32
+ def to_asciibib(prefix)
33
+ pref = prefix.empty? ? prefix : prefix + "."
34
+ pref += "editorialgroup"
35
+ committee.map { |c| "#{pref}.committee:: #{c}\n" }.join
36
+ end
24
37
  end
25
38
  end
@@ -1,5 +1,5 @@
1
1
  module RelatonUn
2
- class HashConverter < RelatonIsoBib::HashConverter
2
+ class HashConverter < RelatonBib::HashConverter
3
3
  class << self
4
4
  # @override RelatonIsoBib::HashConverter.hash_to_bib
5
5
  # @param args [Hash]
@@ -9,6 +9,7 @@ module RelatonUn
9
9
  ret = super
10
10
  return if ret.nil?
11
11
 
12
+ ret[:submissionlanguage] = array ret[:submissionlanguage]
12
13
  session_hash_to_bib ret
13
14
  ret
14
15
  end
@@ -59,7 +59,7 @@ module RelatonUn
59
59
  # rubocop:disable Metrics/MethodLength
60
60
 
61
61
  # @return [RelatonUn::UnBibliographicItem]
62
- def un_bib_item
62
+ def un_bib_item # rubocop:disable Metrics/AbcSize
63
63
  UnBibliographicItem.new(
64
64
  type: "standard",
65
65
  fetched: Date.today.to_s,
@@ -75,6 +75,7 @@ module RelatonUn
75
75
  distribution: fetch_distribution,
76
76
  editorialgroup: fetch_editorialgroup,
77
77
  classification: fetch_classification,
78
+ job_number: hit[:job_number]
78
79
  )
79
80
  end
80
81
  # rubocop:enable Metrics/MethodLength
@@ -88,11 +89,7 @@ module RelatonUn
88
89
 
89
90
  # @return [Array<RelatonBib::TypedTitleString>]
90
91
  def fetch_title
91
- # fs = RelatonBib::FormattedString.new(
92
- # content: hit[:title], language: "en", script: "Latn",
93
- # )
94
- # [RelatonBib::TypedTitleString.new(type: "main", title: fs)]
95
- [{ title_main: hit[:title], language: "en", script: "Latn" }]
92
+ RelatonBib::TypedTitleString.from_string hit[:title], "en", "Latn"
96
93
  end
97
94
 
98
95
  # @return [Array<RelatonBib::BibliographicDate>]
@@ -134,7 +131,7 @@ module RelatonUn
134
131
  def fetch_editorialgroup
135
132
  tc = hit[:ref].match(/^[\S]+/).to_s.split(/\/|-/).reduce([]) do |m, v|
136
133
  if BODY[v] then m << BODY[v]
137
- elsif v =~ /(AC|C|CN|CONF|GC|SC|Sub|WG).\d+|PC/ then m << v
134
+ elsif v.match? /(AC|C|CN|CONF|GC|SC|Sub|WG).\d+|PC/ then m << v
138
135
  else m
139
136
  end
140
137
  end.uniq
@@ -6,18 +6,23 @@ require "http-cookie"
6
6
  module RelatonUn
7
7
  # Page of hit collection.
8
8
  class HitCollection < RelatonBib::HitCollection
9
- AGENT = "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_3) "\
10
- "AppleWebKit/537.36 (KHTML, like Gecko) Chrome/79.0.3945.130 Safari/537.36"
9
+ AGENT = "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_5) "\
10
+ "AppleWebKit/537.36 (KHTML, like Gecko) Chrome/85.0.4183.102 Safari/537.36"
11
11
  DOMAIN = "https://documents.un.org"
12
- BOUNDARY = "----WebKitFormBoundary6hkaBvITDck8dHCn"
12
+ BOUNDARY = "----WebKitFormBoundaryVarT9Z7AFUzw2lma"
13
13
 
14
14
  # @param text [String] reference to search
15
15
  def initialize(text)
16
16
  super
17
17
  @uri = URI.parse DOMAIN
18
18
  @jar = HTTP::CookieJar.new
19
- @http = Net::HTTP.new @uri.host, @uri.port
19
+ @http = Net::HTTP.new @uri.host, @uri.port # , "localhost", "8000"
20
20
  @http.use_ssl = true
21
+ # @http.verify_mode = OpenSSL::SSL::VERIFY_PEER
22
+ # @http.cert_store = OpenSSL::X509::Store.new
23
+ # ca_file = "/Users/andrej/Library/Preferences/httptoolkit/ca.pem"
24
+ # @http.cert_store.set_default_paths
25
+ # @http.cert_store.add_file ca_file
21
26
  @http.read_timeout = 120
22
27
  if (form_resp = get_page)
23
28
  doc = Nokogiri::HTML page_resp(form_resp, text).body
@@ -48,19 +53,21 @@ module RelatonUn
48
53
  # @param form [Nokogiri::HTML::Document]
49
54
  # @param text [String]
50
55
  # @return [Array<String>]
51
- def form_data(form, text)
56
+ def form_data(form, text) # rubocop:disable Metrics/CyclomaticComplexity
52
57
  fd = form.xpath(
53
- "//input[@type!='radio']",
54
- "//input[@type='radio'][@checked]",
55
- "//select[@name!='view:_id1:_id2:cbLang']",
56
- "//textarea",
58
+ "//input[@type!='radio']|"\
59
+ "//input[@type='radio'][@checked]|"\
60
+ "//select[@name!='view:_id1:_id2:cbLang']|"\
61
+ "//textarea"
57
62
  ).reduce([]) do |m, i|
58
63
  v = case i[:name]
59
64
  when "view:_id1:_id2:txtSymbol" then text
65
+ when "view:_id1:_id2:rgTrunc" then "R"
60
66
  when "view:_id1:_id2:cbType" then "FP"
61
67
  when "view:_id1:_id2:cbSort" then "R"
62
68
  when "$$xspsubmitid" then "view:_id1:_id2:_id130"
63
- when "$$xspsubmitscroll" then "0|167"
69
+ when "$$xspsubmitscroll" then "0|102"
70
+ when "view:_id1" then "view:_id1"
64
71
  else i[:value]
65
72
  end
66
73
  m << %{--#{BOUNDARY}}
@@ -77,7 +84,7 @@ module RelatonUn
77
84
  form = Nokogiri::HTML form_resp.body
78
85
  req = Net::HTTP::Post.new form.at("//form")[:action]
79
86
  set_headers req
80
- req["Content-Type"] = "multipart/form-data, boundary=#{BOUNDARY}"
87
+ req["Content-Type"] = "multipart/form-data; boundary=#{BOUNDARY}"
81
88
  req.body = form_data(form, text).join("\r\n")
82
89
  resp = @http.request req
83
90
  get_page URI.parse(resp["location"]).request_uri
@@ -91,7 +98,7 @@ module RelatonUn
91
98
 
92
99
  # @param item [Nokogiri::XML::Element]
93
100
  # @return [Hash]
94
- def hit_data(item)
101
+ def hit_data(item) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
95
102
  en = item.at("//span[.='ENGLISH']/../..")
96
103
  {
97
104
  ref: item.at("div/div/a")&.text&.sub("\u00A0", ""),
@@ -103,7 +110,8 @@ module RelatonUn
103
110
  link: link(en),
104
111
  session: session(item),
105
112
  agenda: agenda(item),
106
- distribution: distribution(item)
113
+ distribution: distribution(item),
114
+ job_number: job_number(item),
107
115
  }
108
116
  end
109
117
 
@@ -155,18 +163,25 @@ module RelatonUn
155
163
  item.at("//label[.='Distribution:']/following-sibling::span")&.text
156
164
  end
157
165
 
166
+ def job_number(item)
167
+ item.at("//span[contains(@id, 'cfJobNumE')]")&.text
168
+ end
169
+
158
170
  # rubocop:disable Metrics/MethodLength
159
171
 
160
172
  # @param req [Net::HTTP::Get, Net::HTTP::Post]
161
- def set_headers(req)
173
+ def set_headers(req) # rubocop:disable Metrics/AbcSize
162
174
  set_cookie req
163
175
  req["Accept"] = "text/html,application/xhtml+xml,application/xml;q=0.9,"\
164
- "image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.9"
176
+ "image/avif,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;"\
177
+ "v=b3;q=0.9"
165
178
  req["Accept-Encoding"] = "gzip, deflate, br"
179
+ req["Accept-Language"] = "en-US;q=0.8,en;q=0.7"
166
180
  req["Cache-Control"] = "max-age=0"
167
181
  req["Connection"] = "keep-alive"
168
182
  req["Origin"] = "https://documents.un.org"
169
183
  req["Referer"] = "https://documents.un.org/prod/ods.nsf/home.xsp"
184
+ req["Sec-Fetch-Dest"] = "document"
170
185
  req["Sec-Fetch-Mode"] = "navigate"
171
186
  req["Sec-Fetch-Site"] = "same-origin"
172
187
  req["Sec-Fetch-User"] = "?1"
@@ -29,7 +29,7 @@ module RelatonUn
29
29
  # @return [RelatonIsoBib::IsoBibliographicItem]
30
30
  def hash_to_bib(hash)
31
31
  item_hash = ::RelatonUn::HashConverter.hash_to_bib(hash)
32
- ::RelatonIsoBib::UnBibliographicItem.new item_hash
32
+ ::RelatonUn::UnBibliographicItem.new item_hash
33
33
  end
34
34
 
35
35
  # Returns hash of XML grammar
@@ -3,7 +3,7 @@ module RelatonUn
3
3
  include RelatonBib
4
4
 
5
5
  # @return [String, NilClass]
6
- attr_reader :session_number, :collaboration, :agenda_id, :item_footnote
6
+ attr_reader :session_number, :collaborator, :agenda_id, :item_footnote
7
7
 
8
8
  # @return [Date, NilClass]
9
9
  attr_reader :session_date
@@ -16,7 +16,7 @@ module RelatonUn
16
16
  # @param item_number [Array<String>]
17
17
  # @pqrqm item_name [Array<String>]
18
18
  # @pqrqm subitem_name [Array<String>]
19
- # @param collaboration [String]
19
+ # @param collaborator [String]
20
20
  # @param agenda_id [String]
21
21
  # @param item_footnote [String]
22
22
  def initialize(**args)
@@ -25,12 +25,12 @@ module RelatonUn
25
25
  @item_number = args.fetch(:item_number, [])
26
26
  @item_name = args.fetch(:item_name, [])
27
27
  @subitem_name = args.fetch(:subitem_name, [])
28
- @collaboration = args[:collaboration]
28
+ @collaborator = args[:collaborator]
29
29
  @agenda_id = args[:agenda_id]
30
30
  @item_footnote = args[:item_footnote]
31
31
  end
32
32
 
33
- # rubocop:disable Metrics/AbcSize
33
+ # rubocop:disable Metrics/CyclomaticComplexity, Metrics/AbcSize
34
34
 
35
35
  # @param [Nokogiri::XML::Builder]
36
36
  def to_xml(builder)
@@ -40,26 +40,48 @@ module RelatonUn
40
40
  item_number.each { |n| b.send "item-number", n }
41
41
  item_name.each { |n| b.send "item-name", n }
42
42
  subitem_name.each { |n| b.send "subitem-name", n }
43
- b.collaboration collaboration if collaboration
43
+ b.collaborator collaborator if collaborator
44
44
  b.send "agenda-id", agenda_id if agenda_id
45
45
  b.send "item-footnote", item_footnote if item_footnote
46
46
  end
47
47
  end
48
48
 
49
- # rubocop:disable Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
49
+ # rubocop:disable Metrics/PerceivedComplexity, Metrics/MethodLength
50
50
  # @return [Hash]
51
51
  def to_hash
52
52
  hash = {}
53
53
  hash["session_number"] = session_number if session_number
54
54
  hash["session_date"] = session_date.to_s if session_date
55
- hash["item_number"] = single_element_array(item_number) if item_number.any?
55
+ if item_number.any?
56
+ hash["item_number"] = single_element_array(item_number)
57
+ end
56
58
  hash["item_name"] = single_element_array(item_name) if item_name.any?
57
- hash["subitem_name"] = single_element_array(subitem_name) if subitem_name.any?
58
- hash["collaboration"] = collaboration if collaboration
59
+ if subitem_name.any?
60
+ hash["subitem_name"] = single_element_array(subitem_name)
61
+ end
62
+ hash["collaborator"] = collaborator if collaborator
59
63
  hash["agenda_id"] = agenda_id if agenda_id
60
64
  hash["item_footnote"] = item_footnote if item_footnote
61
65
  hash
62
66
  end
63
- # rubocop:enable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
67
+ # rubocop:enable Metrics/PerceivedComplexity
68
+
69
+ # @param prefix [String]
70
+ # @return [String]
71
+ def to_asciibib(prefix = "")
72
+ pref = prefix.empty? ? prefix : prefix + "."
73
+ pref += "session"
74
+ out = ""
75
+ out += "#{pref}.session_number:: #{session_number}\n" if session_number
76
+ out += "#{pref}.session_date:: #{session_date}\n" if session_date
77
+ item_number.each { |n| out += "#{pref}.item_number:: #{n}\n" }
78
+ item_name.each { |n| out += "#{pref}.item_name:: #{n}\n" }
79
+ subitem_name.each { |n| out += "#{pref}.subitem_name:: #{n}\n" }
80
+ out += "#{pref}.collaborator:: #{collaborator}\n" if collaborator
81
+ out += "#{pref}.agenda_id:: #{agenda_id}\n" if agenda_id
82
+ out += "#{pref}.item_footnote:: #{item_footnote}\n" if item_footnote
83
+ out
84
+ end
85
+ # rubocop:enable Metrics/CyclomaticComplexity, Metrics/AbcSize, Metrics/MethodLength
64
86
  end
65
87
  end
@@ -1,5 +1,7 @@
1
1
  module RelatonUn
2
- class UnBibliographicItem < RelatonIsoBib::IsoBibliographicItem
2
+ class UnBibliographicItem < RelatonBib::BibliographicItem
3
+ include RelatonBib
4
+
3
5
  TYPES = %w[
4
6
  recommendation plenary addendum communication corrigendum reissue agenda
5
7
  budgetary sec-gen-notes expert-report resolution
@@ -12,33 +14,66 @@ module RelatonUn
12
14
  attr_reader :session
13
15
 
14
16
  # @return [String, NilClass]
15
- attr_reader :distribution
17
+ attr_reader :distribution, :job_number
18
+
19
+ # @return [Array<String>]
20
+ attr_reader :submissionlanguage
16
21
 
17
22
  # @param session [RelatonUn::Session, NilClass]
18
- # @param distribution [String]
23
+ # @param distribution [String, nil]
24
+ # @param job_number [String, nil]
19
25
  def initialize(**args)
20
26
  if args[:distribution] && !DISTRIBUTIONS.has_value?(args[:distribution])
21
- warn "[relaton-un] WARNING: invalid distribution: #{args[:distribution]}"
27
+ warn "[relaton-un] WARNING: invalid distribution: "\
28
+ "#{args[:distribution]}"
22
29
  end
30
+ @submissionlanguage = args.delete :submissionlanguage
23
31
  @distribution = args.delete :distribution
24
32
  @session = args.delete :session
33
+ @job_number = args.delete :job_number
25
34
  super **args
26
35
  end
27
36
 
28
37
  # @param builder [Nokogiri::XML::Builder]
29
38
  # @param bibdata [TrueClasss, FalseClass, NilClass]
30
- def to_xml(builder = nil, **opts)
39
+ def to_xml(builder = nil, **opts) # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/MethodLength
31
40
  super(builder, **opts) do |b|
32
- b.distribution distribution if distribution
33
- session&.to_xml b if session
41
+ b.ext do
42
+ b.doctype doctype if doctype
43
+ submissionlanguage&.each { |sl| b.submissionlanguage sl }
44
+ editorialgroup&.to_xml b
45
+ ics&.each { |i| i.to_xml b }
46
+ b.distribution distribution if distribution
47
+ session&.to_xml b if session
48
+ b.job_number job_number if job_number
49
+ end
34
50
  end
35
51
  end
36
52
 
37
53
  # @return [Hash]
38
54
  def to_hash
39
55
  hash = super
56
+ if submissionlanguage&.any?
57
+ hash["submissionlanguage"] = single_element_array submissionlanguage
58
+ end
59
+ hash["distribution"] = distribution if distribution
40
60
  hash["session"] = session.to_hash if session
61
+ hash["job_number"] = job_number if job_number
41
62
  hash
42
63
  end
64
+
65
+ # @param prefix [String]
66
+ # @return [String]
67
+ def to_asciibib(prefix = "") # rubocop:disable Metrics/AbcSize
68
+ pref = prefix.empty? ? prefix : prefix + "."
69
+ out = super
70
+ submissionlanguage.each do |sl|
71
+ out += "#{pref}submissionlanguage:: #{sl}\n"
72
+ end
73
+ out += "#{pref}distribution:: #{distribution}\n" if distribution
74
+ out += session.to_asciibib prefix if session
75
+ out += "#{pref}job_number:: #{job_number}\n" if job_number
76
+ out
77
+ end
43
78
  end
44
79
  end