Contents

#!/usr/bin/env ruby
require 'rubygems'
gem 'dimus-biodiversity'

$LOAD_PATH.unshift(File.expand_path(File.dirname(__FILE__) + "/../lib"))
require 'biodiversity'
require 'yaml'

if ARGV.empty?
  puts "Usage:\n\nnnparse file_with_scientific_names [output_file]\n\ndefault output_file is parsed.yml\n\n"
  exit
end


parser = ScientificNameParser.new

output = ARGV[1] || 'parsed.yml'
o = File.open(output,'w')

# parse a file with names
count = count2 = 0
names = []
IO.foreach(ARGV[0]) do |n|
  name_dict = {}
  puts 'Parsing names' if count2 == 0
  count2 += 1
  p count2  if count2 % 5000 == 0
  n.strip!
  name_dict = {:input => n} 
  parsed = parser.parse n
  unless parsed
    name_dict[:details] = {:parsed => false}
    names << name_dict
    count += 1
  else
    name_dict[:output] = parsed.value
    name_dict[:caononical] = parsed.canonical
    name_dict[:details] = parsed.details
    name_dict[:parsed => true]
    names << name_dict
  end
end
$KCODE = 'UTF8'
puts "Converting results to YAML"
results = YAML.dump(names)
puts "Writing restuls to #{output} file"
o.write(results)
puts "Found #{count2} records, #{count} of them could not be parsed."

Version data entries

2 entries across 2 versions & 1 rubygems

Version	Path
dimus-biodiversity-0.0.5	bin/nnparse
dimus-biodiversity-0.0.6	bin/nnparse