#!/usr/bin/ruby

class ParseVolc1

  def initialize input
    @fp = File.open(input)
    case input
    when /\.xml$/i then @each = :each_xml
    when /\.txt$/i then @each = :each_csv
    else raise "unknown suffix #{input}"
    end
    @block = nil
  end

  def close
    @fp.close
  end

  def each_xml
    require 'rubygems'
    require 'libxml'
    xparser = LibXML::XML::Reader.io(@fp)
    tagstack = []
    record = nil
    loop do
      xparser.read or break
      case xparser.node_type
      when LibXML::XML::Reader::TYPE_ELEMENT then
        tagstack.push xparser.name
	if tagstack.size == 2 then
	  record = {}
	end
      when LibXML::XML::Reader::TYPE_END_ELEMENT then
	if tagstack.size == 2 then
	  @block.call record
	  record = nil
	end
        tagstack.pop
      when LibXML::XML::Reader::TYPE_TEXT then
        if tagstack.size == 3 then
	  record[tagstack.last.downcase] = xparser.value
	end
      end
    end
  end

  def each_csv
    heading = nil
    row = []
    token = nil
    @fp.each_line do |line|
      line.sub!(/\r?\n$/, '')
      for word in line.split(/,/, -1)
        if token.nil?
	  case word
	  when /^"([^"]*)"$/ then # quoted finalized cell
	    row.push $1
	  when /^"(([^"]|"")*)"$/ then # quoted finalized cell
	    row.push $1.gsub(/""/, '"') 
	  when /^(([^"]|"")*)$/ then # unquoted finalized cell
	    row.push $1.gsub(/""/, '"') 
	  when /^"(([^"]|"")*)$/ then # quoted continued cell
	    token = $1.gsub(/""/, '"') 
	  else
	    raise [token, word, row].inspect
	  end
	else
	  case word
	  when /^([^"]*)$/ then # unclosed continued cell
	    token = [token, $1].join(',')
	  when /^(([^"]|"")*)$/ then # unclosed continued cell
	    token = [token, $1.gsub(/""/, '"')].join(',')
	  when /^(([^"]|"")*)"$/ then # closed continued cell
	    row.push [token, $1.gsub(/""/, '"')].join(',')
	    token = nil
	  else
	    raise [token, word, row].inspect
	  end
	end
      end
      if token.nil? then
        if heading.nil? then
	  heading = row.collect { |key| key.downcase }
	else
	  @block.call Hash[*heading.zip(row).flatten]
	end
	row = []
      end
    end
  end

  def each &block
    @block = block
    self.send(@each)
    @block = nil
  end

  def self::read input, &block
    par = self.new(input)
    par.each &block
    par.close
  end

end

if __FILE__ == $0
  for arg in ARGV
    ParseVolc1.read(arg) do |row|
      p row
    end
  end
end