require 'open3'
require 'readahl'

class RegBuilder
  def initialize
    @t2tab = {}
    @t1tab = {}
    @a1tab = {}
    @cctab = {}
    @bucket = []
    @s = nil
    @area = nil
  end
  def learn str
    t1 = str[0]
    t2 = str[1]
    a1 = str[2]
    a2 = str[3]
    ii = str[4,2]
    cc = str[-4..-1]
    @t2tab[t2] = {} unless @t2tab[t2]
    @t2tab[t2][ii] = true
    @t1tab[t1] = {} unless @t1tab[t1]
    @t1tab[t1][a2] = true
    @a1tab[a1] = true
    @cctab[cc] = true
    @bucket.push Regexp.quote(str)
  end
  def sa2re sa
    if sa.size > 1 then
      ra = sa.map{|s| Regexp.quote(s)}
      if (va = ra.grep(/^\d5$/)).size > 2
        ra = ra - va + [ia2re(va.map{|ii| ii[0]}) + '5']
      end
      if (va = ra.grep(/^\d0$/)).size > 2
        ra = ra - va + [ia2re(va.map{|ii| ii[0]}) + '0']
      end
      if (va = ra.grep(/^0\d$/)).size > 2
        ra = ra - va + ['0' + ia2re(va.map{|ii| ii[1]})]
      end
      if (va = ra.grep(/^9\d$/)).size > 2
        ra = ra - va + ['9' + ia2re(va.map{|ii| ii[1]})]
      end
      '(' + ra.join('|') + ')'
    else
      Regexp.quote(sa.join)
    end
  end
  def ia2re ia
    if ia.size > 1 then
      a = []
      a.push ']' if ia.include? ?]
      a.push '-' if ia.include? ?-
      sa = (ia - [?], ?-]).sort
      seq = nil
      while i = sa.shift
        a.push Regexp.quote(i.to_i.chr)
	if sa.first == i.succ
	  iorg = i
	  while sa.first == i.succ
	    i = sa.shift
	  end
	  a.push '-' if (i - iorg) > 1
	  a.push Regexp.quote(i.to_i.chr)
	end
      end
      if a.size > 1 then
	'[' + a.join + ']'
      else
	a.join
      end
    else
      Regexp.quote(ia.first.to_i.chr)
    end
  end
  def compile
    return if @s
    @area = @a1tab.keys.sort.map{|i| i.chr.sub(/[\W]/,'.')}.join
    pin, pout, perr = Open3.popen3("/usr/bin/perl re-assemble.pl")
    @bucket.each {|res| pin.puts res}
    pin.close
    @s = pout.read.chomp
    unless @s.empty?
      @s = @s.sub(/^\(\?-xism:/,'').sub(/\)$/,'').gsub(/\(\?:/, '(').\
	gsub(/\[[A-Z0-9]+\]/){|s| ia2re(s[1..-2].unpack('C*')) }
      return
    end
    # failsafe
    iitab = {}
    for t2 in @t2tab.keys.sort
      iis = sa2re(@t2tab[t2].keys.sort)
      iitab[iis] = [] unless iitab[iis]
      iitab[iis].push t2
    end
    a = []
    for t1 in @t1tab.keys.sort
      b = []
      for iis in iitab.keys.sort
        c = []
	c.push ia2re(iitab[iis].dup)
	c.push ia2re(@a1tab.keys)
	c.push ia2re(@t1tab[t1].keys)
	c.push iis
        b.push c.join
      end
      if b.size > 1
        bs = '(' + b.join('|') + ')'
      else
        bs = b.join
      end
      a.push(t1.chr + bs)
    end
    if a.size > 1
      @s = '(' + a.join('|') + ')'
    else
      @s = a.join
    end
  end
  def to_s
    compile
    @s
  end
  def area
    compile
    @area
  end
end

def areahack str
  a = str.split(/\s+/, 4).map{|s|
    r = Float(s)
    r -= 360.0 if r > 180.0
    r
  }
  s = a[0,2].min
  n = a[0,2].max
  w = a[2,2].min
  e = a[2,2].max
  w, e = e, w if (e - w) > 180 or (e - w) < 10
  vals = [s, n, w, e].join(' ')
  desc = []
  if ((-10)..0) === (e - w)
    desc.push 'all-longitude'
  else
    desc.push [w, e].map{|lo| (lo < 0 ? '%gW' : '%gE') % lo.abs }.join('-')
  end
  case [s, n]
  when [-90, 0] then
    desc[desc.size-1] = 'whole' if desc.last == 'all-longitude'
    desc.push 'SH'
  when [0, 90] then
    desc[desc.size-1] = 'whole' if desc.last == 'all-longitude'
    desc.push 'NH'
  when [-90, 90] then
    if desc.last == 'all-longitude'
      desc.pop; desc.push 'whole globe'
    else
      desc.push '90S-90N'
    end
  else
    desc.push [s, n].map{|la|
      (la < 0 ? '%gS' : la > 0 ? '%gN' : 'EQ') % la.abs }.join('-')
  end
  [vals, desc.join(' ')]
end

ds = {}
as = {}

for line in ARGF
  ttaaii, cccc, model, grid, resolution, area, level, updtime, access \
    = line.chomp.split(/\t/)
  ahl = [ttaaii, cccc].join
  props = [ model, resolution, area ].join("\t")
  ds[props] = {} unless ds.include?(props)
  ds[props][ahl] = true
  unless as.include?(props)
    as[props] = {
      :lev => {}, :upd => {}, :acc => {}, :el => {}, :ft => {}
    }
  end
  ah = as[props]
  ah[:lev][level] = true
  ah[:upd][updtime.sub(/ \[h\]$/, '')] = true
  ah[:acc][access] = true
  ah[:el][AHL.element(ttaaii, cccc)] = true
  ah[:ft][AHL.fcsttime(ttaaii, cccc)] = true
end

for props in ds.keys.sort
  rb = RegBuilder.new
  for ahl in ds[props].keys
    rb.learn ahl
  end
  model, resol, area = props.split(/\t/)
  fres = resol.split(/\s+/).map{|s| Float(s)}.max
  xprops = [ model, fres, areahack(area), resol, area,
      as[props][:lev].keys.sort.join('|'),
      as[props][:upd].keys.sort.join('|'),
      as[props][:acc].keys.sort.join('|'),
      as[props][:el].keys.sort.join('|'),
      as[props][:ft].keys.sort.join('|')
    ].flatten
  name = model.sub(/One-week/, 'Week').gsub(/[- a-z]/, '') \
    + resol.sub(/ .*/, '').sub(/\./, '') + rb.area
  puts [name, ds[props].size, rb.to_s, xprops].join("\t")
end