Module: Metanorma::Standoc::Asciibib

Included in:
Cleanup
Defined in:
lib/metanorma/cleanup/asciibib.rb

Instance Method Summary collapse

Instance Method Details

#add_to_hash(bib, key, val) ⇒ Object 



77
78
79
 
# File 'lib/metanorma/cleanup/asciibib.rb', line 77

def add_to_hash(bib, key, val)
  Metanorma::Utils::set_nested_value(bib, key.split("."), val)
end

#dd_bib_extract(dtd) ⇒ Object 



65
66
67
68
69
70
71
72
73
74
75
 
# File 'lib/metanorma/cleanup/asciibib.rb', line 65

def dd_bib_extract(dtd)
  dtd.children.empty? and return nil
  dtd.at("./dl") and return dl_bib_extract(dtd)
  elems = dtd.remove.elements
  return p_unwrap(dtd) unless elems.size == 1 &&
    %w(ol ul).include?(elems[0].name)

  elems[0].xpath("./li").each_with_object([]) do |li, ret|
    ret << p_unwrap(li)
  end
end

#dl_bib_extract(clause, nested: false) ⇒ Object

definition list, with at most one level of unordered lists



82
83
84
85
86
87
88
89
90
91
92
93
94
95
 
# File 'lib/metanorma/cleanup/asciibib.rb', line 82

def dl_bib_extract(clause, nested: false)
  dl = clause.at("./dl") or return
  key = ""
  bib = dl.xpath("./dt | ./dd").each_with_object({}) do |dtd, m|
    (dtd.name == "dt" and key = dtd.text.sub(/:+$/, "")) and next
    add_to_hash(m, key, dd_bib_extract(dtd))
  end
  clause.xpath("./clause").each do |c1|
    key = c1.at("./title")&.text&.downcase&.strip
    %w(contributor relation series).include?(key) or next
    add_to_hash(bib, key, dl_bib_extract(c1, nested: true))
  end
  dl_bib_extract_title(bib, clause, nested)
end

#dl_bib_extract_title(bib, clause, nested) ⇒ Object 



97
98
99
100
101
102
103
104
105
106
 
# File 'lib/metanorma/cleanup/asciibib.rb', line 97

def dl_bib_extract_title(bib, clause, nested)
  (!nested && clause.at("./title")) or return bib
  title = clause.at("./title").remove.children.to_xml
  bib["title"] = [bib["title"]] if bib["title"].is_a?(Hash) ||
    bib["title"].is_a?(String)
  bib["title"] ||= []
  title.empty? or bib["title"] << { "content" => title, "type" => "main",
                                    "language" => @lang }
  bib
end

#extract_from_p(tag, bib, key) ⇒ Object 



48
49
50
51
52
 
# File 'lib/metanorma/cleanup/asciibib.rb', line 48

def extract_from_p(tag, bib, key)
  return unless bib[tag]

  "<#{key}>#{bib[tag].at('p').children}</#{key}>"
end

#p_unwrap(para) ⇒ Object

if the content is a single paragraph, replace it with its children single links replaced with uri



56
57
58
59
60
61
62
63
 
# File 'lib/metanorma/cleanup/asciibib.rb', line 56

def p_unwrap(para)
  elems = para.elements
  if elems.size == 1 && elems[0].name == "p"
    link_unwrap(elems[0]).children.to_xml.strip
  else
    para.to_xml.strip
  end
end

#ref_dl_cleanup(xmldoc) ⇒ Object 



7
8
9
10
11
12
13
14
15
16
17
18
 
# File 'lib/metanorma/cleanup/asciibib.rb', line 7

def ref_dl_cleanup(xmldoc)
  xmldoc.xpath("//clause[@bibitem = 'true']").each do |c|
    bib = dl_bib_extract(c) or next
    validate_ref_dl(bib, c)
    xml = Relaton::Bib::ItemData.new(
      **Relaton::Bib::HashParserV1.hash_to_bib(bib),
    ).to_xml or next
    bibitem = Nokogiri::XML(xml)
    ref_dl_cleanup_id(bibitem.root, c)
    c.replace(bibitem.root)
  end
end

#ref_dl_cleanup_id(bibitem, clause) ⇒ Object 



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

def ref_dl_cleanup_id(bibitem, clause)
  bibitem["anchor"] = bibitem["id"]
  clause["anchor"] && !/^_/.match(clause["anchor"]) and
    bibitem["anchor"] = clause["anchor"]
  add_id(bibitem)
end

#validate_ref_dl(bib, clause) ⇒ Object

do not accept implicit id



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

def validate_ref_dl(bib, clause)
  id = bib["id"]
  id ||= clause["anchor"] unless /^_/.match?(clause["anchor"])
  unless id
    @log.add("STANDOC_10", clause, params: [clause.to_xml])
    return
  end
  @refids << id
  validate_ref_dl1(bib, id, clause)
end

#validate_ref_dl1(bib, id, clause) ⇒ Object 



39
40
41
42
43
44
45
46
 
# File 'lib/metanorma/cleanup/asciibib.rb', line 39

def validate_ref_dl1(bib, id, clause)
  if !bib["title"]
    @log.add("STANDOC_11", clause, params: [id])
  end
  if !bib["docid"]
    @log.add("STANDOC_12", clause, params: [id])
  end
end