21598438235

Committed 02 Feb 2026 04:29PM UTC coverage: 68.413% (-0.1%) from 68.529%

Build # 21598438235

Build Type

push

github

Committed by

web-flow

Commit Message

Merge pull request #4896 from benwbrum/4880-latex-export-refactor

4880 - Latex export refactor

Run Details

2240 of 3800 branches covered (58.95%)

Branch coverage included in aggregate %.

169 of 183 new or added lines in 4 files covered. (92.35%)

186 existing lines in 4 files now uncovered.

9280 of 13039 relevant lines covered (71.17%)

150.57 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

90.79

/app/models/xml_source_processor.rb

module XmlSourceProcessor
  def validate_source
    if self.source_text.blank?
      return
    end
    # Skip subject linking validation for field-based collections
    # and collections with subjects disabled
    if self.collection&.field_based || self.collection&.subjects_disabled
      return
    end
    validate_links(self.source_text)
  end

  def validate_source_translation
    if self.source_translation.blank?
      return
    end
    # Skip subject linking validation for field-based collections
    # and collections with subjects disabled
    if self.collection&.field_based || self.collection&.subjects_disabled
      return
    end
    validate_links(self.source_translation)
  end

  # check the text for problems or typos with the subject links
  def validate_links(text)
    error_scope = [:activerecord, :errors, :models, :xml_source_processor]
    # split on all begin-braces
    tags = text.split('[[')
    # remove the initial string which occurs before the first tag
    debug("validate_source: tags to process are #{tags.inspect}")
    tags = tags - [tags[0]]
    debug("validate_source: massaged tags to process are #{tags.inspect}")
    for tag in tags
      debug(tag)

      if tag.include?(']]]')
        errors.add(:base, I18n.t('subject_linking_error', scope: error_scope) + I18n.t('tags_should_not_use_3_brackets', scope: error_scope))
        return
      end
      unless tag.include?(']]')
        tag = tag.strip
        errors.add(:base, I18n.t('subject_linking_error', scope: error_scope) + I18n.t('wrong_number_of_closing_braces', tag: '"[['+tag+'"', scope: error_scope))
      end

      # just pull the pieces between the braces
      inner_tag = tag.split(']]')[0]
      if inner_tag =~ /^\s*$/
        errors.add(:base, I18n.t('subject_linking_error', scope: error_scope) + I18n.t('blank_tag_in', tag: '"[['+tag+'"', scope: error_scope))
      end

      # check for unclosed single bracket
      if inner_tag.include?('[')
        unless inner_tag.include?(']')
          errors.add(:base, I18n.t('subject_linking_error', scope: error_scope) + I18n.t('unclosed_bracket_within', tag: '"'+inner_tag+'"', scope: error_scope))
        end
      end
      # check for blank title or display name with pipes
      if inner_tag.include?('|')
        tag_parts = inner_tag.split('|')
        debug("validate_source: inner tag parts are #{tag_parts.inspect}")
        if tag_parts[0] =~ /^\s*$/
          errors.add(:base, I18n.t('subject_linking_error', scope: error_scope) + I18n.t('blank_subject_in', tag: '"[['+inner_tag+']]"', scope: error_scope))
        end
        if tag_parts[1] =~ /^\s*$/
          errors.add(:base, I18n.t('subject_linking_error', scope: error_scope) + I18n.t('blank_text_in', tag: '"[['+inner_tag+']]"', scope: error_scope))
        end
      end
    end
    #    return errors.size > 0
  end

  def source_text=(text)
    self.source_text_will_change!
    super
  end

  def source_translation=(text)
    self.source_translation_will_change!
    super
  end

  ##############################################
  # All code to convert transcriptions from source
  # format to canonical xml format belongs here.
  ##############################################
  def process_source
    if source_text_changed?
      self.xml_text = wiki_to_xml(self, Page::TEXT_TYPE::TRANSCRIPTION)
    end

    if self.respond_to?(:source_translation) && source_translation_changed?
      self.xml_translation = wiki_to_xml(self, Page::TEXT_TYPE::TRANSLATION)
    end
  end

  def wiki_to_xml(page, text_type, preview_mode = false)
    subjects_disabled = page.collection.subjects_disabled

    source_text = case text_type
    when Page::TEXT_TYPE::TRANSCRIPTION
                    page.source_text
    when Page::TEXT_TYPE::TRANSLATION
                    page.source_translation
    else
                    ''
    end

    xml_string = String.new(source_text)
    start_time=Time.now
    xml_string = process_latex_snippets(xml_string)
    logger.info("wiki_to_xml: process_latex_snippets finished at #{Time.now - start_time} seconds")
    xml_string = clean_bad_braces(xml_string)
    logger.info("wiki_to_xml: clean_bad_braces finished at #{Time.now - start_time} seconds")
    xml_string = clean_script_tags(xml_string)
    logger.info("wiki_to_xml: clean_script_tags finished at #{Time.now - start_time} seconds")
    xml_string = process_square_braces(xml_string) unless subjects_disabled
    logger.info("wiki_to_xml: process_square_braces finished at #{Time.now - start_time} seconds")
    xml_string = process_initial_whitespace(xml_string)
    logger.info("wiki_to_xml: process_initial_whitespace finished at #{Time.now - start_time} seconds")
    xml_string = process_linewise_markup(xml_string)
    logger.info("wiki_to_xml: process_linewise_markup finished at #{Time.now - start_time} seconds")
    xml_string = process_line_breaks(xml_string, is_field_based: page.collection.field_based?)
    logger.info("wiki_to_xml: process_line_breaks finished at #{Time.now - start_time} seconds")
    xml_string = valid_xml_from_source(xml_string)
    logger.info("wiki_to_xml: valid_xml_from_source finished at #{Time.now - start_time} seconds")
    xml_string = update_links_and_xml(xml_string, preview_mode, text_type)
    logger.info("wiki_to_xml: update_links_and_xml finished at #{Time.now - start_time} seconds")
    xml_string = postprocess_xml_markup(xml_string)
    logger.info("wiki_to_xml: postprocess_xml_markup finished at #{Time.now - start_time} seconds")
    postprocess_sections
    logger.info("wiki_to_xml: postprocess_sections finished at #{Time.now - start_time} seconds")
    xml_string
  end

  # remove script tags from HTML to prevent javascript injection
  def clean_script_tags(text)
    # text.gsub(/<script.*?<\/script>/m, '')
    text.gsub(/<\/?script.*?>/m, '')
  end

  BAD_SHIFT_REGEX = /\[\[([[[:alpha:]][[:blank:]]|,\(\)\-[[:digit:]]]+)\}\}/
  def clean_bad_braces(text)
    text.gsub BAD_SHIFT_REGEX, '[[\\1]]'
  end

  BRACE_REGEX = /\[\[.*?\]\]/m
  def process_square_braces(text)
    # find all the links
    wikilinks = text.scan(BRACE_REGEX)
    wikilinks.each do |wikilink_contents|
      # strip braces
      munged = wikilink_contents.sub('[[', '')
      munged = munged.sub(']]', '')

      # extract the title and display
      if munged.include? '|'
        parts = munged.split '|'
        title = parts[0]
        verbatim = parts[1]
      else
        title = munged
        verbatim = munged
      end

      title = canonicalize_title(title)

      replacement = "<link target_title=\"#{title}\">#{verbatim}</link>"
      text.sub!(wikilink_contents, replacement)
    end

    text
  end

  def remove_square_braces(text)
    new_text = text.scan(BRACE_REGEX)
    new_text.each do |results|
      changed = results
      # remove title
      if results.include?('|')
        changed = results.sub(/\[\[.*?\|/, '')
      end
      changed = changed.sub('[[', '')
      changed = changed.sub(']]', '')

      text.sub!(results, changed)
    end
    text
  end

  LATEX_SNIPPET = /(\{\{tex:?(.*?):?tex\}\})/m
  def process_latex_snippets(text)
    return text unless self.respond_to? :tex_figures
    replacements = {}
    figures = self.tex_figures.to_a

    text.scan(LATEX_SNIPPET).each_with_index do |pair, i|
      with_tags = pair[0]
      contents = pair[1]

      replacements[with_tags] = "<texFigure position=\"#{i+1}\"/>" # position attribute in acts as list starts with 1

      figure = figures[i] || TexFigure.new
      figure.source = contents unless figure.source == contents
      figures[i] = figure
    end

    self.tex_figures = figures
    replacements.each_pair do |s, r|
      text.sub!(s, r)
    end

    text
  end

  HEADER = /\s\|\s/
  SEPARATOR = /---.*\|/
  ROW = HEADER

  def process_linewise_markup(text)
    @tables = []
    @sections = []
    new_lines = []
    current_table = nil
    text.lines.each do |line|
      # first deal with any sections
      line = process_any_sections(line)
      # look for a header
      if !current_table
        if line.match(HEADER)
          line = line.chomp
          current_table = { header: [], rows: [], section: @sections.last }
          # fill the header
          cells = line.split(/\s*\|\s*/)
          cells.shift if line.match(/^\|/) # remove leading pipe

          # trim whitespace from each header cell
          cells = cells.map(&:strip)

          current_table[:header] = cells.map { |cell_title| cell_title.sub(/^!\s*/, '') }
          heading = cells.map do |cell|
            if cell.match(/^!/)
              "<th class=\"bang\">#{cell.sub(/^!\s*/, '')}</th>"
            else
              "<th>#{cell}</th>"
            end
          end.join(' ')
          new_lines << "<table class=\"tabular\">\n<thead>\n<tr>#{heading}</tr></thead>"
        else
          # no current table, no table contents -- NO-OP
          new_lines << line
        end
      else
        # this is either an end or a separator
        if line.match(SEPARATOR)
          # NO-OP
        elsif line.match(ROW)
          # handle initial blank cells - if line starts with whitespace followed by pipe, preserve empty cell
          line_chomp = line.chomp
          has_initial_empty_cell = line_chomp.match(/^\s+\|/)

          # remove leading and trailing delimiters
          clean_line = line_chomp.sub(/^\s*\|/, '').sub(/\|\s*$/, '')
          # fill the row
          cells = clean_line.split(/\s*\|\s*/, -1) # -1 means "don't prune empty values at the end"

          # trim whitespace from each cell
          cells = cells.map(&:strip)

          # if there was initial whitespace before pipe, add empty cell at beginning
          cells.unshift('') if has_initial_empty_cell
          current_table[:rows] << cells
          rowline = ''
          cells.each_with_index do |cell, _i|
            rowline += "<td>#{cell}</td> "
          end

          if current_table[:rows].size == 1
            new_lines << '<tbody>'
          end
          new_lines << "<tr>#{rowline}</tr>"
        else
          # finished the last row
          unless current_table[:rows].empty? # only process tables with bodies
            @tables << current_table
            new_lines << '</tbody>'
          end
          new_lines << '</table><lb/>'
          current_table = nil
        end
      end
    end

    if current_table
      # unclosed table
      @tables << current_table
      unless current_table[:rows].empty? # only process tables with bodies
        @tables << current_table
        new_lines << '</tbody>'
      end
      new_lines << '</table><lb/>'
    end
    # do something with the table data
    new_lines.join(' ')
  end

  def process_any_sections(line)
    6.downto(2) do |depth|
      line.scan(/(={#{depth}}([^=]+)={#{depth}})/).each do |section_match|
        wiki_title = section_match[1].strip
        if wiki_title.length > 0
          verbatim = XmlSourceProcessor.cell_to_plaintext(wiki_title)
          safe_verbatim = verbatim.gsub(/"/, '&quot;')
          line = line.sub(section_match.first, "<entryHeading title=\"#{safe_verbatim}\" depth=\"#{depth}\" >#{wiki_title}</entryHeading>")
          @sections << Section.new(title: wiki_title, depth: depth)
        end
      end
    end

    line
  end

  def postprocess_sections
    @sections.each do |section|
      doc = XmlSourceProcessor.cell_to_xml(section.title)
      doc.elements.each('//link') do |e|
        title = e.attributes['target_title']
        article = collection.articles.where(title: title).first
        if article
          e.add_attribute('target_id', article.id.to_s)
        end
      end
      section.title = XmlSourceProcessor.xml_to_cell(doc)
    end
  end


  def canonicalize_title(title)
    # kill all tags
    title = title.gsub(/<.*?>/, '')
    # linebreaks -> spaces
    title = title.gsub(/\n/, ' ')
    # multiple spaces -> single spaces
    title = title.gsub(/\s+/, ' ')
    # change double quotes to proper xml
    title = title.gsub(/\"/, '&quot;')
    title
  end

  # Process initial whitespace at the beginning of lines
  # Replaces leading spaces with <indent> elements containing a spaces attribute
  def process_initial_whitespace(text)
    # Beginning of text
    text = text.gsub(/\A( +)(?=\S)/) do
      "<indent spaces=\"#{$1.length}\"/>"
    end

    # Handle the \r (old Mac) line ending case, which ^ doesn't match
    text = text.gsub(/(\r?\n|\r)( +)(?=\S)/) do
      line_break = $1
      spaces = $2.length
      "#{line_break}<indent spaces=\"#{spaces}\"/>"
    end

    text
  end

  # transformations converting source mode transcription to xml
  def process_line_breaks(text, is_field_based: false)
    if is_field_based
      text = text.gsub(/\n/, '<lb/>')
    else
      text="<p>#{text}</p>"
      text = text.gsub(/\s*\n\s*\n\s*/, '</p><p>')
    end
    text = text.gsub(/([[:word:]]+)-\r\n\s*/, '\1<lb break="no" />')
    text = text.gsub(/\r\n\s*/, '<lb/>')
    text = text.gsub(/([[:word:]]+)-\n\s*/, '\1<lb break="no" />')
    text = text.gsub(/\n\s*/, '<lb/>')
    text = text.gsub(/([[:word:]]+)-\r\s*/, '\1<lb break="no" />')
    text = text.gsub(/\r\s*/, '<lb/>')
    text
  end

  def valid_xml_from_source(source)
    source = source || ''
    safe = source.gsub /\&/, '&amp;'
    safe.gsub! /\&amp;amp;/, '&amp;'
    safe.gsub! /[^\u0009\u000A\u000D\u0020-\uD7FF\uE000-\uFFFD\u10000-\u10FFFF]/, ' '

    string = <<EOF
    <?xml version="1.0" encoding="UTF-8"?>
      <page>
        #{safe}
      </page>
EOF
  end

  def update_links_and_xml(xml_string, preview_mode = false, text_type)
    # first clear out the existing links
    # log the count of articles before and after
    clear_links(text_type) unless preview_mode


    articles_by_title = collection.articles.index_by(&:title) # optimize for common case
    candidate_articles = collection.articles.left_joins(:article_versions) # fall-back
    page_update_timestamp = 1.hour.ago

    processed = ''
    # process it
    doc = REXML::Document.new xml_string
    doc.elements.each('//link') do |element|
      # default the title to the text if it's not specified
      if !(title = element.attributes['target_title'])
        title = element.text
      end
      # display_text = element.text
      display_text = ''
      element.children.each do |e|
        display_text += e.to_s
      end
      debug("link display_text = #{display_text}")
      # change the xml version of quotes back to double quotes for article title
      title = title.gsub('&quot;', '"')

      article = articles_by_title[title]

      if article.nil?
        article = candidate_articles.where('article_versions.title': title)
                                    .where('article_versions.created_on > ?', page_update_timestamp)
                                    .first
        if article.present?
          display_text = article.title
          title = article.title
        end
      end

      # create new blank articles if they don't exist already
      if article.nil?
        article = Article.new
        article.title = title
        article.collection = collection
        article.created_by_id = Current.user.id if Current.user.present?
        article.save! unless preview_mode
        # add the new article to the hash
        articles_by_title[title] = article
      end

      link_id = create_link(article, display_text, text_type) unless preview_mode
      # now update the attribute
      link_element = REXML::Element.new('link')
      element.children.each { |c| link_element.add(c) }
      link_element.add_attribute('target_title', title)
      debug('element='+link_element.inspect)
      debug('article='+article.inspect)
      link_element.add_attribute('target_id', article.id.to_s) unless preview_mode
      link_element.add_attribute('link_id', link_id.to_s) unless preview_mode
      element.replace_with(link_element)
    end
    doc.write(processed)
    processed
  end

  # handle XML-dependent post-processing
  def postprocess_xml_markup(xml_string)
    doc = REXML::Document.new xml_string
    processed = ''
    doc.elements.each('//lb') do |element|
      if element.previous_element && element.previous_sibling.node_type == :element && element.previous_element.name == 'lb'
        pre = doc.to_s
        element.parent.elements.delete(element)
      end
    end
    doc.write(processed)
    processed
  end


  CELL_PREFIX = "<?xml version='1.0' encoding='UTF-8'?><cell>"
  CELL_SUFFIX = '</cell>'

  def self.cell_to_xml(cell)
    REXML::Document.new(CELL_PREFIX + cell.gsub('&', '&amp;') + CELL_SUFFIX)
  end

  def self.xml_to_cell(doc)
    text = ''
    doc.write(text)
    text.sub(CELL_PREFIX, '').sub(CELL_SUFFIX, '')
  end

  def self.cell_to_plaintext(cell)
    doc = cell_to_xml(cell)
    doc.each_element('.//text()') { |e| p e.text }.join
  end

  def self.cell_to_subject(cell)
    doc = cell_to_xml(cell)
    subjects = ''
    doc.elements.each('//link') do |e|
      title = e.attributes['target_title']
      subjects << title
      subjects << "\n"
    end
    subjects
  end

  def self.cell_to_category(cell)
    doc = cell_to_xml(cell)
    categories = ''
    doc.elements.each('//link') do |e|
      id = e.attributes['target_id']
      if id
        article = Article.find(id)
        article.categories.each do |category|
          categories << category.title
          categories << "\n"
        end
      end
    end
    categories
  end

  ##############################################
  # Code to rename links within the text.
  # This assumes that the name change has already
  # taken place within the article table in the DB
  ##############################################
  def rename_article_links(old_title, new_title)
    title_regex =
      Regexp.escape(old_title)
        .gsub('\\ ', ' ') # Regexp.escape converts ' ' to '\\ ' for some reason -- undo this
        .gsub(/\s+/, '\s+') # convert multiple whitespaces into 1+n space characters

    self.source_text = rename_link_in_text(source_text, title_regex, new_title)

    # Articles don't have translations, but we still need to update pages.source_translation
    if has_attribute?(:source_translation) && !source_translation.nil?
      self.source_translation = rename_link_in_text(source_translation, title_regex, new_title)
    end
  end

  def rename_link_in_text(text, title_regex, new_title)
    if new_title == ''
      # Link deleted, remove [[ ]] but keep the original title text

      # Handle links of the form [[Old Title|Display Text]] => Display Text
      text = text.gsub(/\[\[#{title_regex}\|([^\]]+)\]\]/i, '\1')
      # Handle links of the form [[Old Title]] => Old Title
      text = text.gsub(/\[\[(#{title_regex})\]\]/i, '\1')
    else
      # Replace the title part in [[Old Title|Display Text]]
      text = text.gsub(/\[\[#{title_regex}\|/i, "[[#{new_title}|")
      # Replace [[Old Title]] with [[New Title|Old Title]]
      text = text.gsub(/\[\[(#{title_regex})\]\]/i, "[[#{new_title}|\\1]]")
    end

    text
  end


  def pipe_tables_formatting(text)
    # since Pandoc Pipe Tables extension requires pipe characters at the beginning and end of each line we must add them
    # to the beginning and end of each line
    text.split("\n").map { |line| "|#{line}|" }.join("\n")
  end

  def xml_table_to_markdown_table(table_element, pandoc_format = false, plaintext_export = false)
    text_table = ''

    # clean up in-cell line-breaks
    table_element.xpath('//lb').each { |n| n.replace(' ') }

    # Sanitize single quotes with backticks
    # table_element.xpath('//*').each { |n| n.content.gsub("'", '`') }

    # calculate the widths of each column based on max(header, cell[0...end])
    column_count = ([table_element.xpath('//th').count] + table_element.xpath('//tr').map { |e| e.xpath('td').count }).max
    column_widths = {}
    1.upto(column_count) do |column_index|
      longest_cell = (table_element.xpath("//tr/td[position()=#{column_index}]").map { |e| e.text().length }.max || 0)
      corresponding_heading = heading_length = table_element.xpath("//th[position()=#{column_index}]").first
      heading_length = corresponding_heading.nil? ? 0 : corresponding_heading.text().length
      column_widths[column_index] = [longest_cell, heading_length].max
    end

    # print the header as markdown
    cell_strings = []
    table_element.xpath('//th').each_with_index do |e, i|
      cell_strings << e.text.rjust(column_widths[i+1], ' ')
    end
    text_table << cell_strings.join(' | ') << "\n"

    # print the separator
    text_table << column_count.times.map { |i| ''.rjust(column_widths[i+1], '-') }.join(' | ') << "\n"

    # print each row as markdown
    table_element.xpath('//tr').each do |row_element|
      text_table << row_element.xpath('td').map do |e|
        width = 80 # default for hand-coded tables
        index = e.path.match(/.*td\[(\d+)\]/)
        if index
          width = column_widths[index[1].to_i] || 80
        else
          width = column_widths.values.first
        end

        if plaintext_export
          e.text.rjust(width, ' ')
        else
          inner_html = xml_to_pandoc_md(e.to_s.gsub("'", '&#39;'), false, false, nil, false).gsub("\n", '')
          inner_html.rjust(width, ' ')
        end
      end.join(' | ') << "\n"
    end
    if pandoc_format
      text_table = pipe_tables_formatting(text_table)
    end

    "#{text_table}\n\n"
  end



  def debug(msg)
    logger.debug("DEBUG: #{msg}")
  end
end

benwbrum / fromthepage / 21598438235

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous