#!/usr/bin/ruby

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

$stderr.puts("Getting our names...")
names = Name.connection.select_rows %(
  SELECT MIN(id), text_name
  FROM names
  WHERE correct_spelling_id IS NULL
  GROUP BY text_name
)

# names = [
#   [1, 'Agaricus'],
#   [2, 'Agarica'],
#   [3, 'Physcina'],
#   [4, 'Physcia'],
#   [5, 'Phycsia'],
#   [6, 'Bogus'],
#   [7, 'Bogis'],
#   [8, 'Bogas'],
# ]

$stderr.puts("Building lookup for our names...")
lookup = Hash.new
used = Hash.new
for id, name in names
  lookup[id] = name
  used[name] = [id]
end

$stderr.puts("Getting index fungorum's names...")
index = Hash.new
# Expect: "Genus species\tblah blah blah"
File.open('index_fungorum.txt') do |fh|
  fh.each_line do |line|
    name = line.split("\t", 2).first
    index[name] = true
  end
end

$stderr.puts("Getting Esslinger's names...")
# Expect: "Genus species"
File.open('esslinger.txt') do |fh|
  fh.each_line do |line|
    name = line.chomp
    index[name] = true
  end
end

# index = {
#   'Agaricus' => true,
#   'Physcia'  => true,
#   'Physcina' => true,
# }

$stderr.puts("Identifying potential misspellings...")
used2 = used.dup
for id, name in names
  for i in 1..name.length

    # Do one letter at a time.
    name2 = name.dup
    name2[i-1,1] = ''
    used2[name2] ||= Array.new
    used2[name2] << id

    # Do two letters at a time.
    # name2 = name.dup
    # name2[i-1,2] = ''
    # used2[name2] ||= Array.new
    # used2[name2] << id

    # Do one letter in two different places (slow!)
    # for j in 1..name2.length
    #   name3 = name2.dup
    #   name3[j-1,1] = ''
    #   used2[name3] ||= Array.new
    #   used2[name3] << id
    # end
  end
end

# puts "names=#{names.inspect}"
# puts "lookup=#{lookup.inspect}"
# puts "used=#{used.inspect}"
# puts "used2=#{used2.inspect}"

syns = @syns = Hash.new
@syn_id = 0

def synonymize(ids)
  sids = ids.map {|id| @syns[id]}.reject {|x| x.nil?}.uniq
  sid = sids.first
  if !sid
    sid = (@syn_id += 1)
  end
  for sid2 in sids - [sid]
    for key, val in @syns
      if val == sid2
        @syns[key] = sid
      end
    end
  end
  for id in ids
    @syns[id] = sid
  end
end

$stderr.puts("Grouping misspellings...")
for name, ids in used2
  ids = ids.uniq
  if ids.length > 1
    synonymize(ids)
  end
end

$stderr.puts("Sorting into good and bad...")
goods = []
bads = []
for sid in syns.values.uniq.sort
  list = []
  num_legit = 0
  for key, val in syns
    if val == sid
      name = lookup[key]
      legit = index[name]
      list << [key, name, legit]
      num_legit += 1 if legit
    end
  end
  if num_legit == 1
    goods << list
  elsif num_legit < list.length
    bads << list
  end
end

$stderr.puts("Dumping good ones...")
for list in goods
  for id, name, legit in list.sort(&lambda do |a,b|
    x = (a[2]?0:1) <=> (b[2]?0:1)
    x = a[1] <=> b[1] if x == 0
    x = a[0] <=> b[0] if x == 0
    x
  end)
    star = legit ? '*' : ''
    puts "#{star}#{id}\t#{name}"
  end
  puts
end

puts "############################################################"
puts

$stderr.puts("Dumping bad ones...")
for list in bads
  for id, name, legit in list.sort(&lambda do |a,b|
    x = (a[2]?0:1) <=> (b[2]?0:1)
    x = a[1] <=> b[1] if x == 0
    x = a[0] <=> b[0] if x == 0
    x
  end)
    star = legit ? '*' : ''
    puts "#{star}#{id}\t#{name}"
  end
  puts
end