Class: IsoDoc::Convert

Inherits:
Common
  • Object
show all
Defined in:
lib/isodoc/css.rb,
lib/isodoc/convert.rb

Constant Summary

Constants included from Function::Utils

Function::Utils::CLAUSE_ANCESTOR, Function::Utils::DOCTYPE_HDR, Function::Utils::NOKOHEAD, Function::Utils::NOTE_CONTAINER_ANCESTOR

Constants included from Function::Table

Function::Table::SW

Constants included from Function::Section

Function::Section::TERM_CLAUSE

Constants included from Function::Lists

Function::Lists::OL_STYLE

Constants included from Function::Cleanup

Function::Cleanup::FIGURE_WITH_FOOTNOTES

Constants included from Function::Blocks

Function::Blocks::EXAMPLE_TBL_ATTR, Function::Blocks::EXAMPLE_TD_ATTR

Instance Attribute Summary collapse

Instance Method Summary collapse

Methods included from ClassUtils

#case_strict, #case_strict1, #case_with_markup, #date_range, #liquid, #ns

Methods included from Function::Utils

#attr_code, #cleanup_entities, #date_range, #emf?, #empty2nil, #external_path, #extract_delims, #from_xhtml, #get_clause_id, #get_note_container_id, #header_strip, #header_strip_elem?, #image_localfile, #insert_tab, #labelled_ancestor, #liquid, #noko, #ns, #populate_template, #save_dataimage, #sentence_join, #to_xhtml, #to_xhtml_fragment, #to_xhtml_prep

Methods included from Function::ToWordHtml

#body_attr, #boilerplate, #define_head, #in_sourcecode, #info, #init_dir, #init_file, #make_body, #make_body1, #make_body2, #make_body3, #middle, #note?, #parse, #rel_tmpimagedir, #set_termdomain, #tmpimagedir

Methods included from Function::Terms

#admitted_term_parse, #definition_parse, #deprecated_term_parse, #modification_parse, #para_then_remainder, #term_parse, #termdef_parse, #termdocsource_parse, #termnote_delim, #termnote_parse, #termref_parse

Methods included from Function::Table

#colgroup, #make_tr_attr, #table_attrs, #table_parse, #table_title_parse, #tbody_parse, #tcaption, #tfoot_parse, #thead_parse, #tr_parse

Methods included from Function::Section

#abstract, #acknowledgements, #annex, #annex_attrs, #annex_name, #clause, #clause_attrs, #clause_name, #clause_parse, #clause_parse_subtitle, #clause_parse_title, #clause_parse_title1, #clause_title_depth, #clausedelim, #clausedelimspace, #copyright_parse, #feedback_parse, #foreword, #inline_header_title, #introduction, #is_clause?, #legal_parse, #license_parse, #preface, #preface_block, #scope, #symbols_abbrevs, #symbols_parse, #terms_defs, #terms_parse, #variant_title

Methods included from Function::References

#bibitem_ref_code, #biblio_list, #bibliography, #bibliography_parse, #bibliography_xpath, #bracket_if_num, #date_note_process, #docid_l10n, #docid_prefix, #implicit_reference, #iso_bibitem_entry_attrs, #iso_title, #nonstd_bibitem, #norm_ref, #norm_ref_xpath, #omit_docid_prefix, #pref_ref_code, #prefix_bracketed_ref, #ref_entry_code, #reference_format, #render_identifier, #standard?, #std_bibitem_entry, #unbracket

Methods included from Function::Lists

#dl_attrs, #dl_parse, #dt_dd?, #dt_parse, #li_parse, #ol_attrs, #ol_depth, #ol_parse, #ol_style, #ul_attrs, #ul_parse

Methods included from Function::Inline

#add_parse, #bookmark_parse, #br_parse, #callout_parse, #del_parse, #em_parse, #eref_parse, #eref_target, #error_parse, #hr_parse, #image_parse, #image_title_parse, #index_parse, #index_xref_parse, #keyword_parse, #link_parse, #no_locality_parse, #origin_parse, #page_break, #pagebreak_parse, #section_break, #smallcap_parse, #stem_parse, #strike_parse, #strong_parse, #sub_parse, #suffix_url, #sup_parse, #termrefelem_parse, #text_parse, #tt_parse, #underline_parse, #xref_parse

Methods included from Function::Form

#form_parse, #input_parse, #label_parse, #option_parse, #select_parse, #text_input, #textarea_parse

Methods included from Function::Cleanup

#admonition_cleanup, #break_up_long_strings, #break_up_long_strings1, #cleanup, #example_cleanup, #figure_aside_process, #figure_cleanup, #figure_get_or_make_dl, #footnote_cleanup, #footnote_reference_format, #inline_header_cleanup, #merge_fnref_into_fn_text, #new_fullcolspan_row, #passthrough_cleanup, #remove_bottom_border, #symbols_cleanup, #table_cleanup, #table_footnote_cleanup, #table_footnote_cleanup_propagate, #table_footnote_reference_format, #table_get_or_make_tfoot, #table_long_strings_cleanup, #table_note_cleanup, #textcleanup

Methods included from Function::Blocks

#admonition_attrs, #admonition_class, #admonition_name, #admonition_name_parse, #admonition_parse, #annotation_parse, #example_div_attr, #example_div_parse, #example_label, #example_parse, #example_table_attr, #example_table_parse, #figure_attrs, #figure_key, #figure_name_parse, #figure_parse, #figure_parse1, #formula_attrs, #formula_parse, #formula_parse1, #formula_where, #keep_style, #middle_admonitions, #middle_title, #note_attrs, #note_delim, #note_p_parse, #note_parse, #note_parse1, #para_attrs, #para_class, #para_parse, #passthrough_parse, #permission_parse, #pre_parse, #pseudocode_attrs, #pseudocode_parse, #quote_attribution, #quote_parse, #recommendation_attr_keyvalue, #recommendation_attr_parse, #recommendation_attributes, #recommendation_attributes1, #recommendation_labels, #recommendation_name, #recommendation_parse, #reqt_attrs, #reqt_component_attrs, #reqt_metadata_node, #requirement_component_parse, #requirement_parse, #requirement_skip_parse, #sourcecode_attrs, #sourcecode_name_parse, #sourcecode_parse, #svg_parse, #toc_parse

Constructor Details

#initialize(options) ⇒ Convert

htmlstylesheet: Generic stylesheet for HTML htmlstylesheet_override: Override stylesheet for HTML wordstylesheet: Generic stylesheet for Word wordstylesheet_override: Override stylesheet for Word standardsheet: Stylesheet specific to Standard header: Header file for Word htmlcoverpage: Cover page for HTML wordcoverpage: Cover page for Word htmlintropage: Introductory page for HTML wordintropage: Introductory page for Word normalfontsize: Font size for body text smallerfontsize: Font size for smaller than body text monospacefontsize: Font size for monospace font footnotefontsize: Font size for footnotes i18nyaml: YAML file for internationalisation of text ulstyle: list style in Word CSS for unordered lists olstyle: list style in Word CSS for ordered lists bodyfont: font to use for body text headerfont: font to use for header text monospace: font to use for monospace text suppressheadingnumbers: suppress heading numbers for clauses scripts: Scripts file for HTML scripts_override: Override scripts file for HTML scripts_pdf: Scripts file for PDF (not used in XSLT PDF) datauriimage: Encode images in HTML output as data URIs break_up_urls_in_tables: whether to insert spaces in URLs in tables

every 40-odd chars

sectionsplit: split up HTML output on sections bare: do not insert any prefatory material (coverpage, boilerplate)



43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
 
# File 'lib/isodoc/convert.rb', line 43

def initialize(options)
  @libdir ||= File.dirname(__FILE__) # rubocop:disable Lint/DisjunctiveAssignmentInConstructor
  options.merge!(default_fonts(options)) do |_, old, new|
    old || new
  end.merge!(default_file_locations(options)) do |_, old, new|
    old || new
  end
  @options = options
  @files_to_delete = []
  @tempfile_cache = []
  @htmlstylesheet_name = options[:htmlstylesheet]
  @wordstylesheet_name = options[:wordstylesheet]
  @htmlstylesheet_override_name = options[:htmlstylesheet_override]
  @wordstylesheet_override_name = options[:wordstylesheet_override]
  @standardstylesheet_name = options[:standardstylesheet]
  @sourcefilename = options[:sourcefilename]
  @header = options[:header]
  @htmlcoverpage = options[:htmlcoverpage]
  @wordcoverpage = options[:wordcoverpage]
  @htmlintropage = options[:htmlintropage]
  @wordintropage = options[:wordintropage]
  @normalfontsize = options[:normalfontsize]
  @smallerfontsize = options[:smallerfontsize]
  @monospacefontsize = options[:monospacefontsize]
  @footnotefontsize = options[:footnotefontsize]
  @scripts = options[:scripts] ||
    File.join(File.dirname(__FILE__), "base_style", "scripts.html")
  @scripts_pdf = options[:scripts_pdf]
  @scripts_override = options[:scripts_override]
  @i18nyaml = options[:i18nyaml]
  @ulstyle = options[:ulstyle]
  @olstyle = options[:olstyle]
  @datauriimage = options[:datauriimage]
  @suppressheadingnumbers = options[:suppressheadingnumbers]
  @break_up_urls_in_tables = options[:break_up_urls_in_tables] == "true"
  @sectionsplit = options[:sectionsplit] == "true"
  @suppressasciimathdup = options[:suppressasciimathdup] == "true"
  @bare = options[:bare]
  @termdomain = ""
  @termexample = false
  @note = false
  @sourcecode = false
  @footnotes = []
  @comments = []
  @in_footnote = false
  @in_comment = false
  @in_table = false
  @in_figure = false
  @seen_footnote = Set.new
  @c = HTMLEntities.new
  @openmathdelim = "`"
  @closemathdelim = "`"
  @lang = options[:language] || "en"
  @script = options[:script] || "Latn"
  @maxwidth = 1200
  @maxheight = 800
  @wordToClevels = options[:doctoclevels].to_i
  @wordToClevels = 2 if @wordToClevels.zero?
  @htmlToClevels = options[:htmltoclevels].to_i
  @htmlToClevels = 2 if @htmlToClevels.zero?
  @bookmarks_allocated = { "X" => true }
  @fn_bookmarks = {}
  @baseassetpath = options[:baseassetpath]
  @aligncrosselements = options[:aligncrosselements]
  @tmpimagedir_suffix = tmpimagedir_suffix
  @tmpfilesdir_suffix = tmpfilesdir_suffix
end

Instance Attribute Details

#i18n ⇒ Object

Returns the value of attribute i18n.



12
13
14
 
# File 'lib/isodoc/convert.rb', line 12

def i18n
  @i18n
end

#meta ⇒ Object

Returns the value of attribute meta.



12
13
14
 
# File 'lib/isodoc/convert.rb', line 12

def meta
  @meta
end

#options ⇒ Object

Returns the value of attribute options.



12
13
14
 
# File 'lib/isodoc/convert.rb', line 12

def options
  @options
end

#xrefs ⇒ Object

Returns the value of attribute xrefs.



12
13
14
 
# File 'lib/isodoc/convert.rb', line 12

def xrefs
  @xrefs
end

Instance Method Details

#bibitem_lookup(docxml) ⇒ Object 



141
142
143
144
145
146
 
# File 'lib/isodoc/convert.rb', line 141

def bibitem_lookup(docxml)
  @bibitems = docxml.xpath(ns("//references/bibitem"))
    .each_with_object({}) do |b, m|
    m[b["id"]] = b
  end
end

#convert(input_filename, file = nil, debug = false, output_filename = nil) ⇒ Object 



177
178
179
180
181
182
183
184
185
186
187
188
 
# File 'lib/isodoc/convert.rb', line 177

def convert(input_filename, file = nil, debug = false,
            output_filename = nil)
  file = File.read(input_filename, encoding: "utf-8") if file.nil?
  @openmathdelim, @closemathdelim = extract_delims(file)
  docxml, filename, dir = convert_init(file, input_filename, debug)
  result = convert1(docxml, filename, dir)
  return result if debug

  output_filename ||= "#{filename}.#{@suffix}"
  postprocess(result, output_filename, dir)
  FileUtils.rm_rf dir
end

#convert1(docxml, filename, dir) ⇒ Object 



127
128
129
130
131
132
133
134
135
136
137
138
139
 
# File 'lib/isodoc/convert.rb', line 127

def convert1(docxml, filename, dir)
  @xrefs.parse docxml
  bibitem_lookup(docxml)
  noko do |xml|
    xml.html **{ lang: @lang.to_s } do |html|
      html.parent.add_namespace("epub", "http://www.idpf.org/2007/ops")
      info docxml, nil
      populate_css
      html.head { |head| define_head head, filename, dir }
      make_body(html, docxml)
    end
  end.join("\n")
end

#convert_init(file, input_filename, debug) ⇒ Object 



165
166
167
168
169
170
171
172
173
174
175
 
# File 'lib/isodoc/convert.rb', line 165

def convert_init(file, input_filename, debug)
  docxml = Nokogiri::XML(file) { |config| config.huge }
  filename, dir = init_file(input_filename, debug)
  docxml.root.default_namespace = ""
  lang = docxml&.at(ns("//bibdata/language"))&.text and @lang = lang
  script = docxml&.at(ns("//bibdata/script"))&.text and @script = script
  i18n_init(@lang, @script)
  metadata_init(@lang, @script, @i18n)
  xref_init(@lang, @script, self, @i18n, {})
  [docxml, filename, dir]
end

#convert_scss(filename, stylesheet, stripwordcss) ⇒ Object 



77
78
79
80
81
82
83
84
85
86
87
88
89
 
# File 'lib/isodoc/css.rb', line 77

def convert_scss(filename, stylesheet, stripwordcss)
  require "sassc"
  require "isodoc/sassc_importer"

  [File.join(Gem.loaded_specs["isodoc"].full_gem_path,
             "lib", "isodoc"),
             File.dirname(filename)].each do |name|
               SassC.load_paths << name
             end
             SassC::Engine.new(scss_fontheader(stripwordcss) + stylesheet,
                               syntax: :scss, importer: SasscImporter)
               .render
end

#default_file_locations(_options) ⇒ Object

none for this parent gem, but will be populated in child gems which have access to stylesheets &c



48
49
50
 
# File 'lib/isodoc/css.rb', line 48

def default_file_locations(_options)
  {}
end

#default_fonts(_options) ⇒ Object 



38
39
40
41
42
43
44
 
# File 'lib/isodoc/css.rb', line 38

def default_fonts(_options)
  {
    bodyfont: "Arial",
    headerfont: "Arial",
    monospacefont: "Courier New",
  }
end

#fonts_options ⇒ Object 



52
53
54
55
56
57
58
59
60
61
62
 
# File 'lib/isodoc/css.rb', line 52

def fonts_options
  {
    "bodyfont" => options[:bodyfont] || "Arial",
    "headerfont" => options[:headerfont] || "Arial",
    "monospacefont" => options[:monospacefont] || "Courier New",
    "normalfontsize" => options[:normalfontsize],
    "monospacefontsize" => options[:monospacefontsize],
    "smallerfontsize" => options[:smallerfontsize],
    "footnotefontsize" => options[:footnotefontsize],
  }
end

#generate_css(filename, stripwordcss) ⇒ Object

stripwordcss if HTML stylesheet, !stripwordcss if DOC stylesheet



92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
 
# File 'lib/isodoc/css.rb', line 92

def generate_css(filename, stripwordcss)
  return nil if filename.nil?

  filename = precompiled_style_or_original(filename)
  stylesheet = File.read(filename, encoding: "UTF-8")
  stylesheet = populate_template(stylesheet, :word)
  stylesheet.gsub!(/(\s|\{)mso-[^:]+:[^;]+;/m, "\\1") if stripwordcss
  if File.extname(filename) == ".scss"
    stylesheet = convert_scss(filename, stylesheet, stripwordcss)
  end
  Tempfile.open([File.basename(filename, ".*"), "css"],
                encoding: "utf-8") do |f|
    f.write(stylesheet)
    f
  end
end

#html_doc_path(*file) ⇒ Object 



119
120
121
122
123
124
125
 
# File 'lib/isodoc/convert.rb', line 119

def html_doc_path(*file)
  file.each do |f|
    ret = File.join(@libdir, File.join("html", f))
    File.exist?(ret) and return ret
  end
  nil
end

#i18n_init(lang, script, i18nyaml = nil) ⇒ Object 



157
158
159
 
# File 'lib/isodoc/convert.rb', line 157

def i18n_init(lang, script, i18nyaml = nil)
  @i18n = I18n.new(lang, script, i18nyaml || @i18nyaml)
end

#l10n(expr, lang = @lang, script = @script) ⇒ Object 



161
162
163
 
# File 'lib/isodoc/convert.rb', line 161

def l10n(expr, lang = @lang, script = @script)
  @i18n.l10n(expr, lang, script)
end

#localpath(path) ⇒ Object 



19
20
21
22
23
24
 
# File 'lib/isodoc/css.rb', line 19

def localpath(path)
  return path if %r{^[A-Z]:|^/|^file:/}.match?(path)
  return path unless (@sourcedir || @localdir) && path

  File.expand_path(File.join((@sourcedir || @localdir), path))
end

#metadata_init(lang, script, i18n) ⇒ Object 



148
149
150
 
# File 'lib/isodoc/convert.rb', line 148

def metadata_init(lang, script, i18n)
  @meta = Metadata.new(lang, script, i18n)
end

#middle_clause(_docxml = nil) ⇒ Object 



190
191
192
193
 
# File 'lib/isodoc/convert.rb', line 190

def middle_clause(_docxml = nil)
  "//clause[parent::sections][not(@type = 'scope')]"\
    "[not(descendant::terms)]"
end

#populate_css ⇒ Object

run this after @meta is populated



27
28
29
30
31
32
33
34
35
36
 
# File 'lib/isodoc/css.rb', line 27

def populate_css
  @htmlstylesheet = generate_css(localpath(@htmlstylesheet_name), true)
  @wordstylesheet = generate_css(localpath(@wordstylesheet_name), false)
  @standardstylesheet =
    generate_css(localpath(@standardstylesheet_name), false)
  @htmlstylesheet_override_name and @htmlstylesheet_override =
    File.open(localpath(@htmlstylesheet_override_name))
  @wordstylesheet_override_name and @wordstylesheet_override =
    File.open(localpath(@wordstylesheet_override_name))
end

#precompiled_style_or_original(stylesheet_path) ⇒ Object

Check if already compiled version(.css) exists,

if not, return original scss file. During release
we compile scss into css files in order to not depend on scss



6
7
8
9
10
11
12
13
14
15
16
17
 
# File 'lib/isodoc/css.rb', line 6

def precompiled_style_or_original(stylesheet_path)
  # Already have compiled stylesheet, use it
  return stylesheet_path if stylesheet_path.nil? ||
    File.extname(stylesheet_path) == ".css"

  basename = File.basename(stylesheet_path, ".*")
  compiled_path = File.join(File.dirname(stylesheet_path),
                            "#{basename}.css")
  return stylesheet_path unless File.file?(compiled_path)

  compiled_path
end

#scss_fontheader(is_html_css) ⇒ Object 



64
65
66
67
68
69
70
71
72
73
74
75
 
# File 'lib/isodoc/css.rb', line 64

def scss_fontheader(is_html_css)
  b = options[:bodyfont] || "Arial"
  h = options[:headerfont] || "Arial"
  m = options[:monospacefont] || "Courier New"
  ns = options[:normalfontsize] || (is_html_css ? "1.0em" : "12.0pt")
  ms = options[:monospacefontsize] || (is_html_css ? "0.8em" : "11.0pt")
  ss = options[:smallerfontsize] || (is_html_css ? "0.9em" : "10.0pt")
  fs = options[:footnotefontsize] || (is_html_css ? "0.9em" : "9.0pt")
  "$bodyfont: #{b};\n$headerfont: #{h};\n$monospacefont: #{m};\n"\
    "$normalfontsize: #{ns};\n$monospacefontsize: #{ms};\n"\
    "$smallerfontsize: #{ss};\n$footnotefontsize: #{fs};\n"
end

#target_pdf(node) ⇒ Object 



195
196
197
198
199
200
201
 
# File 'lib/isodoc/convert.rb', line 195

def target_pdf(node)
  if /#/.match?(node["target"])
    node["target"].sub(/#/, ".pdf#")
  else
    "##{node['target']}"
  end
end

#tmpfilesdir_suffix ⇒ Object 



115
116
117
 
# File 'lib/isodoc/convert.rb', line 115

def tmpfilesdir_suffix
  "_#{SecureRandom.hex(8)}_files"
end

#tmpimagedir_suffix ⇒ Object 



111
112
113
 
# File 'lib/isodoc/convert.rb', line 111

def tmpimagedir_suffix
  "_#{SecureRandom.hex(8)}_images"
end

#xref_init(lang, script, _klass, i18n, options) ⇒ Object 



152
153
154
155
 
# File 'lib/isodoc/convert.rb', line 152

def xref_init(lang, script, _klass, i18n, options)
  html = HtmlConvert.new(language: @lang, script: @script)
  @xrefs = Xref.new(lang, script, html, i18n, options)
end