Module: Metanorma::Standoc::Bibitem

Includes:
Attachment
Included in:
Cleanup
Defined in:
lib/metanorma/cleanup/bibitem.rb

Constant Summary collapse

BIBITEM_ELEM_ORDER =

canonical element order of BibliographicItem per biblio.rnc

%w(fetched formattedref title uri docidentifier docnumber date
 contributor edition version note language locale script abstract
 status copyright relation series medium place price extent size
 accesslocation license classification keyword validity depiction)
.freeze

Instance Method Summary collapse

Methods included from Attachment

#attachment_cleanup, #attachment_location, #attachment_uri, #datauri_attachment, #init_attachments, #save_attachment, #valid_attachment?

Instance Method Details

#bibitem_cleanup(xmldoc) ⇒ Object 



245
246
247
248
249
250
251
252
253
254
255
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 245

def bibitem_cleanup(xmldoc)
  bibitem_nested_id(xmldoc) # feeds remove_dup_bibtem_id
  ref_dl_cleanup(xmldoc)
  formattedref_spans(xmldoc)
  fetch_local_bibitem(xmldoc)
  remove_empty_docid(xmldoc)
  empty_docid_to_title(xmldoc)
  remove_dup_bibtem_id(xmldoc)
  bibitem_i18n(xmldoc)
  attachment_cleanup(xmldoc)
end

#bibitem_i18n(xmldoc) ⇒ Object 



226
227
228
229
230
231
232
233
234
235
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 226

def bibitem_i18n(xmldoc)
  xmldoc.xpath("//references/bibitem").each do |b|
    s = b.at("./script")
    l = b.at("./language")
    s && l and next
    ins = bibitem_i18n_insert(b)
    s or ins.next = "<script>#{@script}</script>"
    l or ins.next = "<language>#{@lang}</language>"
  end
end

#bibitem_i18n_insert(bib) ⇒ Object 



237
238
239
240
241
242
243
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 237

def bibitem_i18n_insert(bib)
  bib.at("./note[last()]") || bib.at("./version[last()]") ||
    bib.at("./edition[last()]") || bib.at("./contributor[last()]") ||
    bib.at("./date[last()]") || bib.at("./docnumber[last()]") ||
    bib.at("./docidentifier[last()]") || bib.at("./source[last()]") ||
    bib.at("./title[last()]") || bib.children.first
end

#bibitem_id_docid_hash(xmldoc) ⇒ Object 



189
190
191
192
193
194
195
196
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 189

def bibitem_id_docid_hash(xmldoc)
  xmldoc.xpath("//bibitem[@anchor]").each_with_object({}) do |b, m|
    m[b["anchor"]] ||= {}
    docid = b.at("./docidentifier")&.text || "NO ID"
    m[b["anchor"]][docid] ||= []
    m[b["anchor"]][docid] << b
  end
end

#bibitem_nested_id(xmldoc) ⇒ Object 



168
169
170
171
172
173
174
175
176
177
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 168

def bibitem_nested_id(xmldoc)
  xmldoc.xpath("//bibitem//bibitem").each do |b|
    b.delete("id")
    b.delete("anchor")
  end
  xmldoc.xpath("//bibdata//bibitem").each do |b|
    b.delete("id")
    b.delete("anchor")
  end
end

#bibitem_reorder_elems(bib) ⇒ Object

the spans-derived fragment is appended after any elements already on the bibitem (e.g. formattedref, docnumber, date), which violates the ordered relaton content model; restore grammar order, stably



51
52
53
54
55
56
57
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 51

def bibitem_reorder_elems(bib)
  bib.element_children
    .sort_by
    .with_index do |e, i|
      [BIBITEM_ELEM_ORDER.index(e.name) || BIBITEM_ELEM_ORDER.size, i]
    end.each { |e| bib << e }
end

#bibitem_title_to_id(bibitem) ⇒ Object 



216
217
218
219
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 216

def bibitem_title_to_id(bibitem)
  t = bibitem.at("./title") || bibitem.at("./formattedref") or return
  t.text
end

#biblio_hidden_inherit(xmldoc) ⇒ Object 



83
84
85
86
87
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 83

def biblio_hidden_inherit(xmldoc)
  xmldoc.xpath("//references[@hidden = 'true']").each do |r|
    r.xpath("./bibitem").each { |b| b["hidden"] = true }
  end
end

#biblio_no_ext(xmldoc) ⇒ Object 



89
90
91
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 89

def biblio_no_ext(xmldoc)
  xmldoc.xpath("//bibitem/ext").each(&:remove)
end

#empty_docid_to_title(xmldoc) ⇒ Object 



203
204
205
206
207
208
209
210
211
212
213
214
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 203

def empty_docid_to_title(xmldoc)
  xmldoc.xpath("//references/bibitem").each do |b|
    b.at("./docidentifier[not(@type = 'metanorma' or @type = 'DOI' or " \
     "@type = 'metanorma-ordinal')]") and next
    empty_docid_to_title?(b) or next
    ins = b.at("./title[last()]") || b.at("./formattedref")
    id = bibitem_title_to_id(b) or return
    ins.next = <<~XML
      <docidentifier type='title' primary='true'>#{id}</docidentifier>
    XML
  end
end

#empty_docid_to_title?(bibitem) ⇒ Boolean

normative references only, biblio uses ordinal code instead

Returns:

  • (Boolean) 


222
223
224
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 222

def empty_docid_to_title?(bibitem)
  bibitem.parent["normative"] == "true"
end

#extract_notes_from_biblio(refs) ⇒ Object 



74
75
76
77
78
79
80
81
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 74

def extract_notes_from_biblio(refs)
  refs.xpath("./bibitem").each do |r|
    r.xpath("./note[@appended]").reverse_each do |n|
      n.delete("appended")
      r.next = n
    end
  end
end

#fetch_local_bibitem(xmldoc) ⇒ Object

if citation uri points to local file, get bibitem from it



157
158
159
160
161
162
163
164
165
166
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 157

def fetch_local_bibitem(xmldoc)
  xmldoc.xpath("//bibitem[formattedref][uri[@type = 'citation']]")
    .each do |b|
      uri = b&.at("./uri[@type = 'citation']")&.text
      bibitem = read_local_bibitem(uri) or next
      bibitem["id"] = b["id"]
      bibitem["anchor"] = b["anchor"]
      b.replace(bibitem)
  end
end

#fetch_termbase(_termbase, _id) ⇒ Object 



132
133
134
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 132

def fetch_termbase(_termbase, _id)
  ""
end

#fold_notes_into_biblio(refs) ⇒ Object 



65
66
67
68
69
70
71
72
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 65

def fold_notes_into_biblio(refs)
  refs.xpath("./bibitem").each do |r|
    while r&.next_element&.name == "note"
      r.next_element["appended"] = true
      r << r.next_element.remove
    end
  end
end

#format_ref(ref, type) ⇒ Object 



93
94
95
96
97
98
99
100
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 93

def format_ref(ref, type)
  ret = Nokogiri::XML.fragment(ref)
  ret.traverse { |x| x.remove if x.name == "fn" }
  ref = to_xml(ret)
  type != "metanorma" and return @isodoc.docid_prefix(type, ref)
  /^\d+$/.match(ref) && !/^\[.*\]$/.match(ref) and return "[#{ref}]"
  ref
end

#formattedref_spans(xmldoc) ⇒ Object 



8
9
10
11
12
13
14
15
16
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 8

def formattedref_spans(xmldoc)
  xmldoc.xpath("//bibitem[formattedref//span]").each do |b|
    ret = new_bibitem_from_formattedref_spans(b)
    merge_bibitem_from_formattedref_spans(b, ret)
  end
  xmldoc.xpath("//bibitem[@amend]").each do |b|
    b.delete("amend")
  end
end

#idtype2cit(ref) ⇒ Object 



113
114
115
116
117
118
119
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 113

def idtype2cit(ref)
  ref.xpath("./docidentifier/@type").each_with_object({}) do |t, m|
    m[t.text] and next
    docid = select_docid(ref, t.text) or next
    m[t.text] = format_ref(docid.children.to_xml, docid["type"])
  end
end

#merge_bibitem_from_formattedref_span_attrs(bib, new) ⇒ Object 



59
60
61
62
63
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 59

def merge_bibitem_from_formattedref_span_attrs(bib, new)
  %w(type).each do |k|
    new[k] and bib[k] = new[k]
  end
end

#merge_bibitem_from_formattedref_spans(bib, new) ⇒ Object 



27
28
29
30
31
32
33
34
35
36
37
38
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 27

def merge_bibitem_from_formattedref_spans(bib, new)
  merge_bibitem_from_formattedref_span_attrs(bib, new)
  if bib.at("./title") && bib["amend"]
    # there already is a fetched record here: merge
    bib.children = Metanorma::Standoc::Cleanup::MergeBibitems
      .new(bib.to_xml, new.to_xml).merge.to_noko.children
  elsif bib.at("./title") # replace record
    bib.children = new.children.to_xml
  else bib << new.children.to_xml
  end
  bibitem_reorder_elems(bib)
end

#new_bibitem_from_formattedref_spans(bib) ⇒ Object 



18
19
20
21
22
23
24
25
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 18

def new_bibitem_from_formattedref_spans(bib)
  ret = Metanorma::Standoc::Cleanup::SpansToBibitem.new(bib).convert
  ret.err.each do |e|
    @log.add(e[:fatal] ? "STANDOC_52" : "STANDOC_53",
             bib, params: [e[:msg]])
  end
  ret.out
end

#read_local_bibitem(uri) ⇒ Object 



136
137
138
139
140
141
142
143
144
145
146
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 136

def read_local_bibitem(uri)
  xml = read_local_bibitem_file(uri) or return nil
  ret = xml.at("//*[local-name() = 'bibdata']") or return nil
  ret = Nokogiri::XML(ret.to_xml
    .sub(%r{(<bibdata[^>]*?) xmlns=("[^"]+"|'[^']+')}, "\\1")).root
  ret.name = "bibitem"
  ins = ret.at("./*[local-name() = 'docidentifier']") or return nil
  ins.previous = %{<uri type="citation">#{uri}</uri>}
  ret&.at("./*[local-name() = 'ext']")&.remove
  ret
end

#read_local_bibitem_file(uri) ⇒ Object 



148
149
150
151
152
153
154
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 148

def read_local_bibitem_file(uri)
  %r{^https?://}.match?(uri) and return nil
  file = "#{@localdir}#{uri}.rxl"
  File.file?(file) or file = "#{@localdir}#{uri}.xml"
  File.file?(file) or return nil
  Nokogiri::XML(File.read(file, encoding: "utf-8"))
end

#reference_names(xmldoc) ⇒ Object 



102
103
104
105
106
107
108
109
110
111
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 102

def reference_names(xmldoc)
  xmldoc.xpath("//bibitem[not(ancestor::bibitem)]").each do |ref|
    @anchors[ref["anchor"]] =
      if docid = select_docid(ref)
        reference = format_ref(docid.children.to_xml, docid["type"])
        { xref: reference, id: idtype2cit(ref) }
      else { xref: ref["anchor"], id: { "" => ref["anchor"] } }
      end
  end
end

#remove_dup_bibtem_id(xmldoc) ⇒ Object

remove dupes if both same ID and same docid, in case dupes introduced through termbases



181
182
183
184
185
186
187
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 181

def remove_dup_bibtem_id(xmldoc)
  bibitem_id_docid_hash(xmldoc).each_value do |v|
    v.each_value do |v1|
      v1[1..].each(&:remove)
    end
  end
end

#remove_empty_docid(xmldoc) ⇒ Object 



198
199
200
201
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 198

def remove_empty_docid(xmldoc)
  xmldoc.xpath("//bibitem/docidentifier[normalize-space(.)='']")
    .each(&:remove)
end

#select_docid(ref, type = nil) ⇒ Object 



121
122
123
124
125
126
127
128
129
130
 
# File 'lib/metanorma/cleanup/bibitem.rb', line 121

def select_docid(ref, type = nil)
  type and t = "[@type = '#{type}']"
  ref.at("./docidentifier[@type = 'metanorma']#{t}") ||
    ref.at("./docidentifier[@primary = 'true'][@language = '#{@lang}']#{t}") ||
    ref.at("./docidentifier[@primary = 'true'][not(@language)]#{t}") ||
    ref.at("./docidentifier[@primary = 'true']#{t}") ||
    ref.at("./docidentifier[not(@type = 'DOI')][@language = '#{@lang}']#{t}") ||
    ref.at("./docidentifier[not(@type = 'DOI')][not(@language)]#{t}") ||
    ref.at("./docidentifier[not(@type = 'DOI')]#{t}")
end