Fetch sentences from the Russian National Corpus.

ruscorpora.rb

#!/usr/bin/env ruby

require 'net/http'
require 'uri'
require 'nokogiri'

Example = Struct.new(:text, :source)

def ruscorpora(word)
  uri = URI('http://search.ruscorpora.ru/download-xml.xml')
  uri.query = URI.encode_www_form({ req: word, text: 'lexform', mode: 'main', doc_tagging: 'manual' })
  doc = Nokogiri::XML(Net::HTTP.get(uri).gsub('́', '')).remove_namespaces!

  doc.xpath('(/Workbook/Worksheet/Table/Row)[position()>1]/Cell[last()]').map do |cell|
    md = cell.text.strip.match(/\A(?<text>.+) +\[(?<source>.+)\] +\[.+\]\z/)
    Example.new(
      md[:text].gsub(/ {2,}/, ''),
      md[:source].tap(&:strip!).gsub(/ *\((\d+(|-\d+))\)\z/, ', \1')
    )
  end
end

# ruscorpora('лопата')