<?php


/**
 * Implements hook_menu().
 */
function mymodule_menu() {
  $items = array();

  $items['mymodule'] = array(
    'title' => 'mymodule',
    'page callback' => 'pull_data',
    'page arguments' => array(),
    'access arguments' => array('access content'),
  );
  return $items;
}


function insert_term_if_not_exists($term_name, $vocabulary_name) {
  $vocabulary = taxonomy_vocabulary_machine_name_load($vocabulary_name);
  $matched_terms = taxonomy_get_term_by_name($term_name, $vocabulary->machine_name);
  if(empty($matched_terms)) {
    $term = new stdClass();
    $term->vid = $vocabulary->vid;
    $term->name = $term_name;
    taxonomy_term_save($term);
  } else {
    $term = array_shift($matched_terms);
  }
  return $term;
}


function pull_data() {
   $str = file_get_contents('http://www.remotesite.tld/news.php?S=left');
   $re = "/\?st=\K\S+(?=\s*target)/";
   preg_match_all($re, $str, $matches);
   $content ='<h3>The following remote URLs have been processed:</h3>';

foreach ($matches[0] as $match) {
      $last_time = db_query('SELECT created FROM {node} WHERE created = :created', array(':created' => $match))->fetchField();

   if ($last_time<>$match) {

      $article = iconv('windows-1251', 'utf-8', file_get_contents('http://www.remotesite.tld/news2.php?st='.$match));

      $reg_country = "/(?<=target=left class=datn>).*?(?=<\/a> &nbsp;)/";
      preg_match_all($reg_country, $article, $country);
      $country = $country[0][0];
      if ($country == "Europe") {$country="Planet";}

      $reg_title = "/(?<=\<title>).*?(?= \| Europe)/";
      preg_match_all($reg_title, $article, $title);
      $title = $title[0][0];
      $title = str_replace('(фото)', '', $title);
      $title = str_replace('(фрейм)', '', $title);

      $reg_body = "/(?<=\<p align=justify>).*?(?=<p>)/";
      preg_match_all($reg_body, $article, $body);
      $body = $body[0][0];

      $reg_source = "/(?<=target=_blank class=datn>).*?(?=<\/a><br>Permalink)/";
      preg_match_all($reg_source, $article, $source);
      $source = $source[0][0];

      $node = new stdClass();
      $node->uid      = '1';
      $node->type     = 'article';
      node_object_prepare($node);
      $node->created = $match;
      $node->changed = $match;
      $node->title    = $title;
      $node->status   = 0;
      $node->promote  = 0;
      $node->sticky   = 0;

      if ($source == "Europe") { $node->flash=1; } else { $node->flash=0; }

      $node->language = 'en';
      $node->body['und'][0]['value']   = $body;
     $node->body['und'][0]['format']  = 'filtered_html';

      insert_term_if_not_exists($country, 'tags');
      $term_tags = taxonomy_get_term_by_name($country);
      $tid_tags = key($term_tags);
      $lang_tags = field_language('node', $node, 'field_tags');
      $node->field_tags[$lang_tags][0]['tid'] = $tid_tags;

      insert_term_if_not_exists($source, 'source');
      $term_source = taxonomy_get_term_by_name($source);
      $tid_source = key($term_source);
      $lang_source = field_language('node', $node, 'field_source');
      $node->field_source[$lang_source][0]['tid'] = $tid_source;

      $reg_image = "/(?<=\<td><IMG src=).*?(?=border=1)/";
      preg_match_all($reg_image, $article, $image);
      if (isset($image[0][0])) {
        $image_url = 'http://www.remotesite.tld/'.$image[0][0];
        $file = file_save_data(file_get_contents($image_url), file_default_scheme().'://field/image/'.basename($image_url));
        $file->status = 1;
        $node->field_image['und'][0] = (array)$file;
      }
      node_save($node);
      $content .= $title.'<br>';
      } else { $content .='http://www.remotesite.tld/news2.php?st='.$match.' had been previously added.<br>'; }

} // for each end

return $content;

}