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_cover_one_doc_coll(manifest, sectionsplit_manifest, key, entry) ⇒ Object 



60
61
62
63
64
65
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 60

def add_cover_one_doc_coll(manifest, sectionsplit_manifest, key, entry)
  idx = File.join(File.dirname(sectionsplit_manifest.file), "index.html")
  FileUtils.cp entry[:ref], idx
  manifest["#{key}:index1.html"] =
    entry.merge(out_path: "index.html", ref: idx)
end

#add_document_suffix(identifier, doc) ⇒ Object 



123
124
125
126
127
128
129
130
131
132
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 123

def add_document_suffix(identifier, doc)
  document_suffix = Metanorma::Utils::to_ncname(identifier)
  Metanorma::Utils::anchor_attributes.each do |(tag_name, attr_name)|
    Util::add_suffix_to_attrs(doc, document_suffix, tag_name, attr_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 



7
8
9
10
11
12
13
14
15
16
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 7

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_attachments(manifest, ident) ⇒ Object 



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

def add_section_split_attachments(manifest, ident)
  attachments = @sectionsplit
    .section_split_attachments(out: File.dirname(manifest.file))
  attachments or return
  @files[ident][:out_path] = attachments
  { attachment: true, index: false, out_path: attachments,
    ref: File.join(File.dirname(manifest.file), attachments) }
end

#add_section_split_cover(manifest, sectionsplit_manifest, ident) ⇒ Object 



47
48
49
50
51
52
53
54
55
56
57
58
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 47

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

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



86
87
88
89
90
91
92
93
94
95
96
97
98
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 86

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

#add_section_split_instance_prep(file, key) ⇒ Object 



100
101
102
103
104
105
106
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 100

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

#anchors_lookup(anchors) ⇒ Object 



69
70
71
72
73
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 69

def anchors_lookup(anchors)
  anchors.values.each_with_object({}) do |v, m|
    v.each_value { |v1| m[v1] = true }
  end
end

#bibdata_extract(xml) ⇒ Object 



75
76
77
78
79
80
81
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 75

def bibdata_extract(xml)
  anchors = read_anchors(xml)
  { anchors: anchors, anchors_lookup: anchors_lookup(anchors),
    ids: read_ids(xml),
    bibdata: xml.at(ns("//bibdata")),
    document_suffix: xml.root["document_suffix"] }
end

#bibdata_process(entry, ident) ⇒ Object 



57
58
59
60
61
62
63
64
65
66
67
 
# 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!(bibdata_extract(xml))
  end
end

#bibitem_process(entry) ⇒ Object 



83
84
85
86
87
88
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 83

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 



41
42
43
44
45
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 41

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 



249
250
251
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 249

def each
  @files.each
end

#each_with_index ⇒ Object 



253
254
255
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 253

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 location 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



95
96
97
98
99
100
101
102
103
104
105
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 95

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 



115
116
117
118
119
120
121
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 115

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.delete("-").to_sym] = ref.send(s)
  end
end

#get(ident, attr = nil) ⇒ Object 



239
240
241
242
243
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 239

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

#key(ident) ⇒ Object 



230
231
232
233
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 230

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

#keys ⇒ Object 



235
236
237
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 235

def keys
  @files.keys
end

#ns(xpath) ⇒ Object 



257
258
259
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 257

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

#one_doc_collection? ⇒ Boolean

Returns:

  • (Boolean) 


67
68
69
70
71
72
73
74
75
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 67

def one_doc_collection?
  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



109
110
111
112
113
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 109

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 



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

def process_section_split_instance(key, manifest)
  s, sectionsplit_manifest = sectionsplit(key)
  # section_split_instance_threads(s, manifest, key)
  s.each_with_index do |f1, i|
    add_section_split_instance(f1, manifest, key, i)
  end
  a = add_section_split_attachments(sectionsplit_manifest, key) and
    manifest["#{key}:attachments"] = a
  add_section_split_cover(manifest, 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.



199
200
201
202
203
204
205
206
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 199

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 



208
209
210
211
212
213
214
215
216
217
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 208

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)



220
221
222
223
224
225
226
227
228
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 220

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 



180
181
182
183
184
185
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 180

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

#section_split_instance_threads(s, manifest, key) ⇒ Object 



29
30
31
32
33
34
35
36
37
38
39
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 29

def section_split_instance_threads(s, manifest, key)
  @mutex = Mutex.new
  pool = Concurrent::FixedThreadPool.new(4)
  s.each_with_index do |f1, i|
    pool.post do
      add_section_split_instance(f1, manifest, key, i)
    end
  end
  pool.shutdown
  pool.wait_for_termination
end

#sectionsplit(ident) ⇒ Object 



108
109
110
111
112
113
114
115
116
117
118
119
120
121
 
# File 'lib/metanorma/collection/filelookup/filelookup_sectionsplit.rb', line 108

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,
         document_suffix: @files[ident][:document_suffix])
  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 



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

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:



166
167
168
169
170
171
172
173
174
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 166

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 



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

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,



145
146
147
148
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 145

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) 


152
153
154
155
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 152

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)



135
136
137
138
139
140
 
# File 'lib/metanorma/collection/filelookup/filelookup.rb', line 135

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 



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

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