#!/usr/bin/ruby

require 'rexml/document'

class App

  def initialize
    @ns = {
      "x"=>"urn:schemas-microsoft-com:office:excel", 
      "ss"=>"urn:schemas-microsoft-com:office:spreadsheet"
    }
  end

  def run args
    for file in args
      convert file
    end
  end

  def parsehash doc, sheet
    faxsheet = REXML::XPath.first(doc, <<-XPATH, @ns)
    /ss:Workbook/ss:Worksheet[@ss:Name="#{sheet}"]
    XPATH
    raise unless faxsheet
    nodes = REXML::XPath.match(faxsheet, <<-XPATH, @ns)
    ss:Table/ss:Row[1]/ss:Cell/ss:Data//text()
    XPATH
    cols = nodes.map {|cell| REXML::Text.unnormalize(cell.to_s) }
    nodes = nil
    nodes = REXML::XPath.match(faxsheet, "ss:Table/ss:Row[position()>1]", @ns)
    nodes.each {|rownode|
      row = {}
      irow = 0
      nodes2 = REXML::XPath.match(rownode, "ss:Cell", @ns)
      nodes2.each {|cellnode|
        if a = REXML::XPath.first(cellnode, '@ss:Index', @ns)
	then irow = a.value.to_i
	else irow += 1
	end
	text = REXML::XPath.first(cellnode, 'ss:Data//text()', @ns)
        text = REXML::Text.unnormalize(text.to_s)
	row[cols[irow - 1]] = text
      }
      nodes2 = nil
      yield row
    }
    nodes = nil
  end

  def serialize md
    doc = REXML::Document.new
    doc << REXML::XMLDecl.new('1.0', 'UTF-8')
    root = doc.add_element('metadata')
    root.add_namespace("http://www.gisc.kishou.go.jp/xsd/jmd0.1")
    for key, val in md
      for sval in Array(val)
        node = root.add_element(key)
	node.add_text(sval)
	case key
	when /bc$/ then
          node.add_attribute('units', 'deg')
	when /^(avail|maxftime)$/ then
          node.add_attribute('units', 'h')
	end
      end
    end
    doc
  end

  def convert_row row
    throw(:ng, "STAMP undefined") if row['STAMP'].to_s.strip.empty?
    stamp = row['STAMP'].strip.gsub(/\s/, ' and ')
    row['*fid'] = fid = row['STAMP'].strip.split(/\s+/).sort.join('-')
    for key in %w(URL Theme Level TFcst Area.Desc North)
      throw(:ng, "#{fid}: #{key} undefined") if row[key].to_s.strip.empty?
    end
    title = "Fax chart #{stamp}"
    xtheme = row['Theme'].split(/\s+/).map{|s| s.gsub(/_/, ' ')}.join('/')
    title += " - #{xtheme}" unless xtheme.empty?
    xlev = row['Level'].split(/\s+/).map{|s| s.gsub(/_/, ' ')}.join('/')
    title += " on #{xlev}" unless xlev.empty?
    xft = row['TFcst'].split(/\s+/).map{|s| s.gsub(/_/, ' ')}.join('/')
    title += " at #{xft}" unless xft.empty?
    title += " for #{row['Area.Desc']}" unless row['Area.Desc'].to_s.empty?
    fmt = row['URL'].strip.split(/\s+/).map{|u|
      u.sub(/.*\./, '').sub(/html/, 'png').upcase
    }.uniq.sort
    res = "Automated black-and-white chart"
    res.sub!(/Automated/, row['Manual']) unless row['Manual'].to_s.empty?
    md = {
      'mdfid' => "urn:x-wmo:md:jp.go.jma.wis.dcpc-geogr::FAX-#{fid}",
      'wisorg' => "DCPC/RSMC-Geographical Tokyo, Japan Meteorological Agency",
      'wiscont' => 'mailto:wis-jma@met.kishou.go.jp',
      'mddate' => Time.now.utc.strftime('%Y-%m-%d'),
      'subjkey' => 'synopticMeteorology',
      'topicat' => 'climatologyMeteorologyAtmosphere',
      'title' => title,
      'abstract' => (<<ABST).gsub(/  +/, ' ').gsub(/^[^:]+:\s+;\n/, ''),
Resource: #{res};
Originator: DCPC/RSMC-Geographical Tokyo, Japan Meteorological Agency;
FAX Stamp:  #{stamp};
Contents: #{row['Theme'].gsub(/\s+/, ', ').gsub(/_/, ' ')};
Vertical Level: #{row['Level'].gsub(/\s+/, ', ').gsub(/_/, ' ')};
Area: #{row['Area.AA']} (#{row['Area.Desc']});
Forecast time: #{row['TFcst'].strip.gsub(/_/, ' ').split(/\s+/).join(', ')};
GTS TTAAii: #{row['TTAAii']};
Radio transmission: #{row['Radio']};
Format on web: #{fmt.join(', ')};
ABST
      'themekey' => row['Theme'].strip.split(/\s+/),
      'southbc' => '%3.1f' % row['South'].to_f,
      'northbc' => '%3.1f' % row['North'].to_f,
      'westbc' => '%3.1f' % row['West'].to_f,
      'eastbc' => '%3.1f' % row['East'].to_f,
      'stratkey' => row['Level'].strip.split(/\s+/),
      'avail' => '24',
      'maxftime' => row['TFcst'].strip.split(/\s+/).map{|s|
	s.sub(/(Analysis|Observation)/, '0').sub(/_h/, '').to_i
      }.max.to_s,
      'updcycle' => 'daily',
      'placekey' => row['Area.Desc'].strip.gsub(/\s/, '_'),
      'format' => fmt,
      'url' => row['URL'].strip.split(/\s+/),
    }
  end

  def convert file
    doc = REXML::Document.new(File.open(file, 'r'))
    areatab = {}
    parsehash(doc, 'area') {|row|
      areatab[row['Area']] = row
    }
    parsehash(doc, 'FAX') {|row|
      area = row['Area']
      unless areatab[area] 
        puts "area #{area} undefined"
        next
      end
      row.update(areatab[area])
      msg = catch(:ng) {
	md = convert_row(row)
	fid = row['*fid']
	puts fid
	jmd = serialize(md)
	File.open("md#{fid}.jmd", 'w') {|ofp| jmd.write(ofp, 1, true) }
	nil
      }
      puts msg if msg
    }
  end

end

App.new.run(ARGV)