import_trials.rb

# Top-level driver script for getting trials from clinicaltrials.gov into our database.
class ImportTrials

  SEARCH_URL = "http://clinicaltrials.gov/ct2/results?term=%s&recr=Open&studyxml=true"

  INSTITUTIONS = { "Dana-Farber Cancer Institute"         => { :search_url => SEARCH_URL % '%22dana+farber%22',                     :abbrev => 'DFCI' },
                   "Harvard Medical School"               => { :search_url => SEARCH_URL % '%22harvard+medical+school%22',          :abbrev => 'HMS' },
                   "Harvard School of Public Health"      => { :search_url => SEARCH_URL % '%22harvard+school+of+public+health%22', :abbrev => 'HSPH' },
                   "Brigham and Women's Hospital"         => { :search_url => SEARCH_URL % '%22brigham+and+women\'s+hospital%22',   :abbrev => 'BWH' },
                   "Children's Hospital Boston"           => { :search_url => SEARCH_URL % '%22children%27s+hospital+boston%22',    :abbrev => 'CHB' },
                   "Beth-Israel Deaconess Medical Center" => { :search_url => SEARCH_URL % '%22beth+israel+deaconess%22',           :abbrev => 'BIDMC' },
                   "Massachusetts General Hospital"       => { :search_url => SEARCH_URL % '%22massachusetts+general+hospital%22',  :abbrev => 'MGH' } }

  LOG_FILE = File.dirname(__FILE__) + '/../log/import_trials.log'

  # Run the import for all search strings.
  def self.run
    @logger = Logger.new(LOG_FILE)

    time_script do
      INSTITUTIONS.each do |institution_name, details|
        search_string = details[:search_url]
        import(institution_name, search_string)
      end
    end
    
    @logger.close
  end

  # Given a search string, download the results and create/update the database records.
  def self.import(institution_name, search_string)
    log("\n\n\nImport started on #{Time.now}")
    log("Started download from CT.gov for #{institution_name} with search string #{search_string}")
    DownloadTrials.download(search_string)
    
    log("Unzipping results")
    DownloadTrials.unzip_trials

    log("Creating trials from xml files") 
    ImportTrials.create_or_update_trials_from_xml(DownloadTrials::OUTPUT_DIRECTORY, institution_name)
  end

  def self.assign_abbreviations
    INSTITUTIONS.each_key do |institution_name|
      institution = Institution.find_by_name(institution_name)
      institution.abbreviation = INSTITUTIONS[institution_name][:abbrev]
      institution.save!
    end
  end
  
  # Find or create trials for each xml representation in a directory
  def self.create_or_update_trials_from_xml(directory, institution)
    Dir.glob(directory + '/*.xml').each do |file|
      create_or_update_trial(file, institution)
    end
  end

  def self.create_or_update_trial(file, institution_name)
    attributes = Trial.parse_trial_attributes(file)
    trial = Trial.find_by_nct_id(attributes[:nct_id])

    if trial
      trial.update_existing(attributes)
    else
      trial = Trial.new(attributes)
    end

    trial.add_institution(institution_name)

    if trial.new_record?
      trial.save false
    else
      trial.trial_institutions.map(&:save)
    end

    trial
  end

  private

  def self.log(message)
    @logger.info(message) unless Rails.env.test?
  end
    
  # Time how long the script takes
  def self.time_script(&block)
    start = Time.now
    yield
    log "Completed import in #{(Time.now - start) / 60} minutes" unless Rails.env.test?
  end


end