#!/usr/bin/env ruby
#
#  USAGE::
#
#    script/refresh_name_lister_cache
#
#  DESCRIPTION::
#
#    This is a nightly cronjob that updates the name_lister data cache.
#    It just writes the file NAME_LISTER_CACHE_FILE:
#
#      RAILS_ROOT/public/javascripts/name_list_data.js
#
################################################################################

require File.expand_path('../../config/boot.rb', __FILE__)
require File.expand_path('../../config/environment.rb', __FILE__)

# Tired of fighting with ActionView to get it to let me use the one in
# helpers/javascript_helper.rb, so I just copied it verbatim.
def escape(str)
  str.to_s.gsub('\\','\0\0').gsub('</','<\/').gsub(/\r\n|\n|\r/, "\\n").gsub(/["']/) { |m| "\\#{m}" }
end

# Instead of doing all of this every time someone requests name_lister, let's
# cache as a separate JS file and serve it up directly (bypassing Rails
# altogether if all goes well!)  This should plug one of our major memory
# leaks.  (fingers crossed) 

genera = Name.connection.select_all %(
  SELECT text_name as n, deprecated as d
  FROM names
  WHERE rank = 'Genus' AND correct_spelling_id IS NULL
  ORDER BY text_name
)

species = Name.connection.select_all %(
  SELECT text_name as n, author as a, deprecated as d, synonym_id as s
  FROM names
  WHERE (rank = 'Species' OR rank = 'Subspecies' OR rank = 'Variety' OR rank = 'Form')
        AND correct_spelling_id IS NULL
  ORDER BY text_name
)

# Place "*" after all accepted genera.
seen = {}
genera = genera.map do |rec|
  n, d = rec.values_at('n', 'd')
  val = d.to_i == 1 ? n : n + '*'
  seen[val] = true
  val
end.uniq
genera.reject! do |n|
  seen[n + '*']
end

# How many times is each name used?
occurs = {}
for rec in species
  n = rec['n']
  occurs[n] ||= 0
  occurs[n] += 1
end

# Build map from synonym_id to list of valid names.
valid = {}
for rec in species
  n, a, d, s = rec.values_at('n', 'a', 'd', 's')
  need_author = occurs[n] > 1
  n += '|' + a if !a.blank? && need_author
  if s.to_i > 0 && d.to_i != 1
    l = valid[s] ||= []
    l.push(n) if !l.include?(n)
  end
end

# Now insert valid synonyms after each deprecated name.  Stick a "*" after
# all accepted names (including, of course, the accepted synonyms).
# Include author after names, using a "|" to help make it easy for
# javascript to parse it correctly.
species = species.map do |rec|
  n, a, d, s = rec.values_at('n', 'a', 'd', 's')
  need_author = occurs[n] > 1
  n += '|' + a if !a.blank? && need_author
  n += '*' if d.to_i != 1
  d.to_i == 1 && valid[s] ? ([n] + valid[s].map {|x| "= #{x}"}) : n
end.flatten

# Write cache.
File.open(NAME_LISTER_CACHE_FILE, 'w') do |fh|
  fh.write %(
    var NL_GENERA  = [#{genera.map  {|n| "'#{escape(n)}'"}.join(",")}];
    var NL_SPECIES = [#{species.map {|n| "'#{escape(n)}'"}.join(",")}];
    var NL_NAMES   = [];
  )
end

exit 0