<?php
/**
 * analyze a text field data before switching a to country field
 * uses dev-php/pecl-syck
 *
 * grabs needed languages into countries-i18n/*.yml;
for i in $(drush ev "print implode(' ', array_keys(language_list('language')))); do 
 wget "https://github.com/onomojo/i18n-country-translations/raw/master/rails/locale/$i.yml"
done

define("TEXT_COUNTRY_FIELD", "field_country");
$data = array();
$iterator = new DirectoryIterator(__DIR__ . "/countries-i18n");
foreach ($iterator as $fileinfo) {
  if ($fileinfo->isFile()) {
    @$a = array_pop(array_pop(syck_load(file_get_contents(__DIR__ . "/countries-i18n/" . $fileinfo->getFilename() ))));
    foreach($a as $k => $word) $data[preg_replace('/\..*$/', '', $fileinfo->getFilename()) . '-' . $k] = $word;
  }
}

$res = db_query("SELECT DISTINCT " . TEXT_COUNTRY_FIELD . "_value AS pays FROM field_data_" . TEXT_COUNTRY_FIELD . " ORDER BY pays")->fetchCol();
$res_leven = array();
foreach($res as $v) {
  $levs = NULL;
  foreach($data as $k => $name) {
    $lev = levenshtein($v, $name);
    if(is_null($levs) || $lev < $levs) {
      $levs = $lev;
      $res_leven['$v'] = $k;
    }
  }
  if($levs == 0 ) // or < 2
    printf("$v levenshtein 0 found: {$res_leven['$v']}\n");
  else
    printf("nearest match for $v is $levs ({$res_leven['$v']})\n");
}