Class: ODDB::Import::PharmNet::PiParser

Inherits:

TermedRtf

• Object
• Rtf
• TermedRtf
• ODDB::Import::PharmNet::PiParser

Defined in:

lib/oddb/import/pharmnet.rb

Constant Summary

Constants inherited from Rtf

Rtf::SYMBOLS

Instance Method Summary

• #_sanitize_text(value) ⇒ Object
• #identify_chapter(buffer) ⇒ Object

Methods inherited from TermedRtf

#initialize

Methods inherited from Rtf

#_import_text, #_next_paragraph, #align, #current_chapter, #current_group, #current_table, #ignore?, #import, #import_control, #import_group, #import_text, #import_token, #init, #next_paragraph, #parent_group, #set_font_size

Constructor Details

This class inherits a constructor from ODDB::Import::PharmNet::TermedRtf

Instance Method Details

#_sanitize_text(value) ⇒ Object

# File 'lib/oddb/import/pharmnet.rb', line 228

def _sanitize_text(value)
  ## some rtfs have unusable information prior to the actual PI
  case value
  when /^PCX\b/
    init
  when /Gebrauchsinformation/
    init if /Recyclinglogo/.match(current_chapter.to_s)
  end
  if @buffer.empty? && @buffer.is_a?(Text::Paragraph)
    value.gsub! /^([P][A-Z0-9]{1,2})?\b/, ''
    value.lstrip!
  end
end

#identify_chapter(buffer) ⇒ Object

# File 'lib/oddb/import/pharmnet.rb', line 155

def identify_chapter buffer
  name = nil
  if(/\b#@term\b/i.match buffer)
    name = case buffer 
           when /wof(ü|Ü|ue)r\s+(wird|werden)\s+(es|sie)\s+(angewendet|eingenommen)/i,
                /wird\s+angewendet$/i
             'indications'
           when /^(3\.?\s*)?Wie\s+(ist|sind).+?(anzuwenden|einzunehmen)\?/i
             'application'
           when /vor\s+der\s+(Anwendung|Einnahme)\s+von/i 
             'precautions'
           when /^([56]\.?\s*)?Wie\s+(ist|sind).+?aufzubewahren/i
             'storage'
           when /^Bitte\s.+für\s+Kinder\s+nicht\s+erreichbar/i
             'personal'
           when /^([45]\.?\s*)?Welche\s+Nebenwirkungen/i, /^Nebenwirkungen:?$/i
             'unwanted_effects'
           when /Behandlungserfolg/i
             nil ## prevent composition if this is a dodgy match
           else
             'composition'
           end
  else
    name = case buffer
           when /^([45]\.?\s*)?Welche\s+Nebenwirkungen/i, /^Nebenwirkungen:?$/i
             'unwanted_effects'
           when /^(4\.?\s*)?Verhalten\s+im\s+Notfall/i
             'emergency'
           when /^(6\.?\s*)?(Weitere\s+)?(Informationen|Angaben)/i,
                /^(6\.?\s*)?Gebrauchsanleitung/i,
                /^Zusätzliche\s+Informationen/i
             'additional_information'
           when /^Anwendungsgebiete/i
             'indications'
           when /^Vorsichtsma(ss|ß)nahmen/i 
             'precautions'
           when /^Dosierung\s*($|und)/i, /^Dosierungsanleitung/
             'application'
           when /Angaben\s+zur\s+Haltbarkeit/i 
             'storage'
           when /^Gegenanzeigen/i
             'counterindications'
           when /^Darreichungsform/i
             'packaging'
           when /^(Hersteller.+)?Pharmazeutischer\s+Unternehmer/i,
                /^Pharmazeutischer\s+Hersteller/i
             'company'
           when /^\s*Stand\b/, /wurde\s+zuletzt\s+überarbeitet/i
             'date'
           when /^(Sehr\s+geehrte|Liebe)r?\s+Patient/i, 
                /^Bitte\s.+für\s+Kinder\s+nicht\s+erreichbar/i
             'personal'
           end
  end
  composition = @document.chapter('composition')
  if(name && (name == 'composition' || composition))
    chapter = @document.chapter(name)
    if(chapter.nil?)
      @document.add_chapter Text::Chapter.new(name)
    else
      pars = chapter.paragraphs.select do |par| !par.to_s.strip.empty? end
      if(pars.size == 1 && /^\d+/.match(pars.first))
        ## some PI insert a document-overview after the composition, in which
        #  case we have an erroneous chapter, identified by only consisting of
        #  a heading. In that case:
        composition.append chapter
        @document.remove_chapter chapter
        @document.add_chapter Text::Chapter.new(name)
      end
    end
  end
  super
end