Class: Metanorma::Collection::FileLookup

Inherits:
Object
  • Object
show all
Defined in:
lib/metanorma/collection/filelookup/filelookup.rb,
lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(path, parent) ⇒ FileLookup

hash for each document in collection of document identifier to: document reference (fileref or id), type of document reference, and bibdata entry for that file

Parameters:

  • path (String) —

    path to collection



15
16
17
18
19
20
21
22
23
24
25
26
27
28
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 15

def initialize(path, parent)
  @c = HTMLEntities.new
  @files = {}
  @parent = parent
  @xml = parent.xml
  @isodoc = parent.isodoc
  @path = path
  @compile = parent.compile
  @documents = parent.documents
  @files_to_delete = []
  @disambig = Util::DisambigFiles.new
  @manifest = parent.manifest
  read_files(@manifest.entry)
end

Instance Attribute Details

#files_to_delete ⇒ Object

Returns the value of attribute files_to_delete.



9
10
11
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 9

def files_to_delete
  @files_to_delete
end

#parent ⇒ Object

Returns the value of attribute parent.



9
10
11
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 9

def parent
  @parent
end

Instance Method Details

#add_document_suffix(identifier, doc) ⇒ Object 



111
112
113
114
115
116
117
118
119
120
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 111

def add_document_suffix(identifier, doc)
  document_suffix = Metanorma::Utils::to_ncname(identifier)
  Metanorma::Utils::anchor_attributes.each do |(tag_name, attribute_name)|
    Util::add_suffix_to_attributes(doc, document_suffix, tag_name,
                                   attribute_name, @isodoc)
  end
  url_in_css_styles(doc, document_suffix)
  doc.root["document_suffix"] ||= ""
  doc.root["document_suffix"] += document_suffix
end

#add_section_split ⇒ Object 



6
7
8
9
10
11
12
13
14
15
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 6

def add_section_split
  ret = @files.keys.each_with_object({}) do |k, m|
    if @files[k][:sectionsplit] && !@files[k][:attachment]
      process_section_split_instance(k, m)
      cleanup_section_split_instance(k, m)
    end
    m[k] = @files[k]
  end
  @files = ret
end

#add_section_split_cover(manifest, ident) ⇒ Object 



32
33
34
35
36
37
38
39
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 32

def add_section_split_cover(manifest, ident)
  cover = @sectionsplit
    .section_split_cover(manifest, @parent.dir_name_cleanse(ident),
                         one_doc_collection?)
  @files[ident][:out_path] = cover
  { attachment: true, index: false, out_path: cover,
    ref: File.join(File.dirname(manifest.file), cover) }
end

#add_section_split_instance(file, manifest, key, idx) ⇒ Object 



52
53
54
55
56
57
58
59
60
61
62
63
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 52

def add_section_split_instance(file, manifest, key, idx)
  presfile, newkey, xml =
    add_section_split_instance_prep(file, key)
  manifest[newkey] =
    { parentid: key, presentationxml: true, type: "fileref",
      rel_path: file[:url], out_path: File.basename(file[:url]),
      anchors: read_anchors(xml), ids: read_ids(xml),
      sectionsplit_output: true,
      bibdata: @files[key][:bibdata], ref: presfile }
  @files_to_delete << file[:url]
  manifest[newkey][:bare] = true unless idx.zero?
end

#add_section_split_instance_prep(file, key) ⇒ Object 



65
66
67
68
69
70
71
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 65

def add_section_split_instance_prep(file, key)
  presfile = File.join(File.dirname(@files[key][:ref]),
                       File.basename(file[:url]))
  newkey = key("#{key.strip} #{file[:title]}")
  xml = Nokogiri::XML(File.read(presfile), &:huge)
  [presfile, newkey, xml]
end

#bibdata_process(entry, ident) ⇒ Object 



57
58
59
60
61
62
63
64
65
66
67
68
69
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 57

def bibdata_process(entry, ident)
  if entry[:attachment]
    entry[:bibdata] =
      Metanorma::Collection::Document.attachment_bibitem(ident).root
  else
    file, _filename = targetfile(entry, read: true)
    xml = Nokogiri::XML(file, &:huge)
    add_document_suffix(ident, xml)
    entry.merge!(anchors: read_anchors(xml), ids: read_ids(xml),
                 bibdata: xml.at(ns("//bibdata")),
                 document_suffix: xml.root["document_suffix"])
  end
end

#bibitem_process(entry) ⇒ Object 



71
72
73
74
75
76
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 71

def bibitem_process(entry)
  entry[:bibitem] = entry[:bibdata].dup
  entry[:bibitem].name = "bibitem"
  entry[:bibitem]["hidden"] = "true"
  entry[:bibitem].at("./*[local-name() = 'ext']")&.remove
end

#cleanup_section_split_instance(key, manifest) ⇒ Object 



26
27
28
29
30
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 26

def cleanup_section_split_instance(key, manifest)
  @files_to_delete << manifest["#{key}:index.html"][:ref]
  # @files[key].delete(:ids).delete(:anchors)
  @files[key][:indirect_key] = @sectionsplit.key
end

#each ⇒ Object 



237
238
239
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 237

def each
  @files.each
end

#each_with_index ⇒ Object 



241
242
243
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 241

def each_with_index
  @files.each_with_index
end

#file_entry(ref, identifier) ⇒ Object

ref is the absolute source file address rel_path is the relative source file address, relative to the YAML locaton out_path is the destination file address, with any references outside the working directory (../../…) truncated, and based on relative path identifier is the id with only spaces, no nbsp



83
84
85
86
87
88
89
90
91
92
93
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 83

def file_entry(ref, identifier)
  ref.file or return
  abs = @documents[Util::key identifier].file
  ret = if ref.file
          { type: "fileref", ref: abs, rel_path: ref.file, url: ref.url,
            out_path: output_file_path(ref) }
        else { type: "id", ref: ref.id }
        end
  file_entry_copy(ref, ret)
  ret.compact
end

#file_entry_copy(ref, ret) ⇒ Object 



103
104
105
106
107
108
109
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 103

def file_entry_copy(ref, ret)
  %w(attachment sectionsplit index presentation-xml url
     bare-after-first).each do |s|
    ref.respond_to?(s.to_sym) and
      ret[s.gsub("-", "").to_sym] = ref.send(s)
  end
end

#get(ident, attr = nil) ⇒ Object 



227
228
229
230
231
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 227

def get(ident, attr = nil)
  if attr then @files[key(ident)][attr]
  else @files[key(ident)]
  end
end

#key(ident) ⇒ Object 



218
219
220
221
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 218

def key(ident)
  @c.decode(ident).gsub(/(\p{Zs})+/, " ").sub(/^metanorma-collection /,
                                              "")
end

#keys ⇒ Object 



223
224
225
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 223

def keys
  @files.keys
end

#ns(xpath) ⇒ Object 



245
246
247
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 245

def ns(xpath)
  @isodoc.ns(xpath)
end

#one_doc_collection? ⇒ Boolean

Returns:

  • (Boolean) 


41
42
43
44
45
46
47
48
49
50
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 41

def one_doc_collection?
  return false
  docs = 0
  @files.each_value do |v|
    v[:attachment] and next
    v[:presentationxml] and next
    docs += 1
  end
  docs > 1
end

#output_file_path(ref) ⇒ Object

TODO make the output file location reflect source location universally, not just for attachments: no File.basename



97
98
99
100
101
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 97

def output_file_path(ref)
  f = File.basename(ref.file)
  ref.attachment and f = ref.file
  @disambig.source2dest_filename(f)
end

#process_section_split_instance(key, manifest) ⇒ Object 



17
18
19
20
21
22
23
24
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 17

def process_section_split_instance(key, manifest)
  s, sectionsplit_manifest = sectionsplit(key)
  s.each_with_index do |f1, i|
    add_section_split_instance(f1, manifest, key, i)
  end
  manifest["#{key}:index.html"] =
    add_section_split_cover(sectionsplit_manifest, key)
end

#read_anchors(xml) ⇒ Object

map locality type and label (e.g. “clause” “1”) to id = anchor for a document Note: will only key clauses, which have unambiguous reference label in locality. Notes, examples etc with containers are just plunked against UUIDs, so that their IDs can at least be registered to be tracked as existing.



187
188
189
190
191
192
193
194
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 187

def read_anchors(xml)
  xrefs = @isodoc.xref_init(@lang, @script, @isodoc, @isodoc.i18n,
                            { locale: @locale })
  xrefs.parse xml
  xrefs.get.each_with_object({}) do |(k, v), ret|
    read_anchors1(k, v, ret)
  end
end

#read_anchors1(key, val, ret) ⇒ Object 



196
197
198
199
200
201
202
203
204
205
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 196

def read_anchors1(key, val, ret)
  val[:type] ||= "clause"
  ret[val[:type]] ||= {}
  index = if val[:container] || val[:label].nil? || val[:label].empty?
            UUIDTools::UUID.random_create.to_s
          else val[:label]
          end
  ret[val[:type]][index] = key
  ret[val[:type]][val[:value]] = key if val[:value]
end

#read_file(manifest) ⇒ Object 



37
38
39
40
41
42
43
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 37

def read_file(manifest)
  i, k = read_file_idents(manifest)
  entry = file_entry(manifest, k) or return
  bibdata_process(entry, i)
  bibitem_process(entry)
  @files[key(i)] = entry
end

#read_file_idents(manifest) ⇒ Object 



45
46
47
48
49
50
51
52
53
54
55
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 45

def read_file_idents(manifest)
  id = manifest.identifier
  sanitised_id = key(@isodoc.docid_prefix("", manifest.identifier.dup))
  #       if manifest.bibdata and # NO, DO NOT FISH FOR THE GENUINE IDENTIFIER IN BIBDATA
  #         d = manifest.bibdata.docidentifier.detect { |x| x.primary } ||
  #           manifest.bibdata.docidentifier.first
  #         k = d.id
  #         i = key(@isodoc.docid_prefix(d.type, d.id.dup))
  #       end
  [id, sanitised_id]
end

#read_files(entries) ⇒ Object 



30
31
32
33
34
35
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 30

def read_files(entries)
  Array(entries).each do |e|
    e.file and read_file(e)
    read_files(e.entry)
  end
end

#read_ids(xml) ⇒ Object

Also parse all ids in doc (including ones which won’t be xref targets)



208
209
210
211
212
213
214
215
216
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 208

def read_ids(xml)
  ret = {}
  xml.traverse do |x|
    x.text? and next
    /^semantic__/.match?(x.name) and next
    x["id"] and ret[x["id"]] = true
  end
  ret
end

#ref_file(ref, out, read, doc) ⇒ Object 



168
169
170
171
172
173
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 168

def ref_file(ref, out, read, doc)
  file = File.read(ref, encoding: "utf-8") if read
  filename = out.dup
  filename.sub!(/\.xml$/, ".html") if doc
  [file, filename]
end

#sectionsplit(ident) ⇒ Object 



73
74
75
76
77
78
79
80
81
82
83
84
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 73

def sectionsplit(ident)
  file = @files[ident][:ref]
  @sectionsplit = ::Metanorma::Collection::Sectionsplit
    .new(input: file, base: @files[ident][:out_path], dir: File.dirname(file),
         output: @files[ident][:out_path], compile_opts: @parent.compile_options,
         fileslookup: self, ident: ident, isodoc: @isodoc)
  coll = @sectionsplit.sectionsplit.sort_by { |f| f[:order] }
  xml = Nokogiri::XML(File.read(file, encoding: "UTF-8"), &:huge)
  [coll, @sectionsplit
    .collection_manifest(File.basename(file), coll, xml, nil,
                         File.dirname(file))]
end

#set(ident, attr, value) ⇒ Object 



233
234
235
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 233

def set(ident, attr, value)
  @files[key(ident)][attr] = value
end

#targetfile(data, options) ⇒ Array<String, nil>

return file contents + output filename for each file in the collection, given a docref entry so my URL should end with html or pdf or whatever formed relative to YAML file, not input path, relative to calling function

Parameters:

  • data (Hash) —

    docref entry

  • read (Boolean) —

    read the file in and return it

  • doc (Boolean) —

    I am a Metanorma document,

  • relative (Boolean) —

    Return output path,

Returns:



154
155
156
157
158
159
160
161
162
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 154

def targetfile(data, options)
  options = { read: false, doc: true, relative: false }.merge(options)
  path = options[:relative] ? data[:rel_path] : data[:ref]
  if data[:type] == "fileref"
    ref_file path, data[:out_path], options[:read], options[:doc]
  else
    xml_file data[:id], options[:read]
  end
end

#targetfile_id(ident, options) ⇒ Object 



164
165
166
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 164

def targetfile_id(ident, options)
  targetfile(get(ident), options)
end

#url(ident, options) ⇒ Object

return citation url for file so my URL should end with html or pdf or whatever

Parameters:

  • doc (Boolean) —

    I am a Metanorma document,



133
134
135
136
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 133

def url(ident, options)
  data = get(ident)
  data[:url] || targetfile(data, options)[1]
end

#url?(ident) ⇒ Boolean

are references to the file to be linked to a file in the collection, or externally? Determines whether file suffix anchors are to be used

Returns:

  • (Boolean) 


140
141
142
143
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 140

def url?(ident)
  data = get(ident) or return false
  data[:url]
end

#url_in_css_styles(doc, document_suffix) ⇒ Object

update relative URLs, url(#…), in CSS in @style attrs (including SVG)



123
124
125
126
127
128
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 123

def url_in_css_styles(doc, document_suffix)
  doc.xpath("//*[@style]").each do |s|
    s["style"] = s["style"]
      .gsub(%r{url\(#([^)]+)\)}, "url(#\\1_#{document_suffix})")
  end
end

#xml_file(id, read) ⇒ Object 



175
176
177
178
179
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 175

def xml_file(id, read)
  file = @xml.at(ns("//doc-container[@id = '#{id}']")).to_xml if read
  filename = "#{id}.html"
  [file, filename]
end