Map grabber :: Yandex.Maps + jQuery + PHP/DOMDocument/XPathby Proger_XP :: http://proger.meOriginal article (Russian): http://habrahabr.ru/post/184334/

magrabx.php

<?php
  // Map grabber  :: Yandex.Maps + jQuery + PHP/DOMDocument/XPath
  // by Proger_XP :: http://proger.me
  // Original article (Russian): http://habrahabr.ru/post/184334/
  // In public domain. I appreciate backlinks.

  function dl($url) {
    require_once 'sqobot/lib/downwind.php';

    return new Downwind($url, array(
      'User-Agent'            => 'Mozilla/5.0 (rv:29.0) Gecko/20100101 Firefox/29.0',
      'Accept-Language'       => 'ru-ru,ru;q=0.8,en-us;q=0.5,en;q=0.3',
      'Accept-Encoding'       => 'gzip, deflate',
      'Accept-Charset'        => 'windows-1251 utf-8 koi8-u big-5 euc-jp shift-jis iso-8859-1',
      'Accept'                => 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
    ));
  }

  function parse($html) {
    $html = '<?xml encoding="UTF-8">'.iconv('cp1251', 'utf-8', $html);

    $doc = new DOMDocument('1.0', 'utf-8');
    @$doc->loadHTML($html) or die('loadHTML: '.$html);
    //die($doc->saveHTML());

    $xpath = new DOMXPath($doc);
    $nodes = $xpath->query('//table[@class="results"]/tr[th[@class="head_kvart"] or td[@width or @class="tooltip"]]');

    $results = array();
    $roomCount = 1;

    foreach ($nodes as $row) {
      $cells = array();
      $cell = $row->firstChild;

      while ($cell) {
        $cell->nodeType == XML_ELEMENT_NODE and $cells[] = trim($cell->nodeValue);
        $cell = $cell->nextSibling;
      }

      if (count($cells) == 1) {
        $roomCount = (int) reset($cells);
      } else {
        $cells[0] = $roomCount;
        $html = $row->ownerDocument->saveXML($row);

        if (count($cells) == 10) {
          array_splice($cells, 6, 1, array(0, '', $cells[6], ''));
        }

        if (preg_match('~<a href="([^"]+)~u', $html, $match)) {
          array_unshift($cells, 'http://www.bn.ru'.$match[1]);
        } else {
          array_unshift($cells, '');
        }

        $keys = array('url', 'rooms', 'address', 'floors', 'houseType', 'area', 'areaLiving', 'areaKitchen', 'toilet', 'price', 'conditions', 'seller', 'phone', 'notes');

        $result[] = array_combine($keys, $cells);
      }
    }

    return $result;
  }

  $regions = array(
    1 => 'Адмиралтейский',
    6 => 'Василеостровский',
    7 => 'Выборгский',
    39 => 'Калининский',
    8 => 'Кировский',
    271 => 'Колпинский',
    40 => 'Красногвардейский',
    41 => 'Красносельский',
    272 => 'Кронштадтский',
    273 => 'Курортный',
    42 => ' Московский',
    43 => 'Невский',
    100 => 'Область',
    44 => 'Петроградский',
    275 => 'Петродворцовый',
    45 => 'Приморский',
    276 => 'Пушкинский',
    46 => 'Фрунзенский',
    47 => 'Центральный',
  );

  set_time_limit(180);

  if (!empty($_REQUEST['find'])) {
    $all = $counts = array();
    $baseURL = 'http://www.bn.ru/zap_fl.phtml?print=printall&';
    $empty = iconv('utf-8', 'cp1251', 'Количество найденных вариантов 0');

    foreach ((array) $_REQUEST['region'] as $region) {
      $url = $baseURL."region$region=$region&";

      foreach (range(0, 10000, 1000) as $price0) {
        if ($price0 + 1000 >= $_REQUEST['price0'] and
            $price0 - 1000 <= $_REQUEST['price1']) {
          $reqURL = $url."price1=$price0&price2=".($price0 + 999);
          $data = dl($reqURL)->fetchData();
          $counts[$price0] = 0;

          if (!strpos($data, $empty)) {
            $offers = parse($data);
            $counts[$price0] = count($offers);
            $all = array_merge($all, $offers);
          }

          usleep(200000);
        }
      }
    }
  }

false and
  $all = array( 0 => array ( 'url' => 'http://www.bn.ru/detail/flats/17469.html?from=search', 'rooms' => 1, 'address' => '7 Советская ул. 32/25', 'floors' => '1\\5', 'houseType' => 'СФ', 'area' => '30', 'areaLiving' => '18.3', 'areaKitchen' => '6', 'toilet' => ' ', 'price' => '3100', 'conditions' => ' ', 'seller' => 'АмпиР Недвижимость', 'phone' => '(965) 0165440', 'notes' => 'ПП свободна ХС торг', ), 1 => array ( 'url' => 'http://www.bn.ru/detail/pflats/168605.html?from=psearch', 'rooms' => 1, 'address' => 'Ломоносова ул. 14', 'floors' => '1/6', 'houseType' => 'К', 'area' => '38', 'areaLiving' => '36', 'areaKitchen' => 0, 'toilet' => '', 'price' => '3300 тыс.руб.', 'conditions' => '', 'seller' => 'частное', 'phone' => '(961) 8114432', 'notes' => 'Две 1-комн. Квартиры продаются в...', ), 2 => array ( 'url' => 'http://www.bn.ru/detail/flats/230375.html?from=search', 'rooms' => 1, 'address' => 'Лиговский пр. 259', 'floors' => '1\\5', 'houseType' => 'К', 'area' => '32', 'areaLiving' => '18', 'areaKitchen' => '6', 'toilet' => 'С', 'price' => '3300', 'conditions' => ' ', 'seller' => 'АН Альтаир', 'phone' => '9626023', 'notes' => '(921) 348-4194 комерц расп', ),);
?>

<!DOCTYPE html>
<html>
  <head>
    <meta charset="utf-8">
    <title>Map grabber</title>

    <script src="//api-maps.yandex.ru/2.0/?load=package.standard&lang=ru-RU" type="text/javascript"></script>
    <script src="//cdnjs.cloudflare.com/ajax/libs/jquery/1.9.0/jquery.min.js"></script>

    <style>
      html, body, .regions, #map { margin: 0; padding: 0; }
      form { margin: 1em; }
      .regions { -moz-column-width: 11em; list-style-position: inside; }
      p[data-marker] { display: inline-block; margin: 3px; }
      p[data-marker] img { vertical-align: bottom; height: 24px; }
      p input[type=submit]:first-child { font-weight: bold; }
      #map { width: 100%; height: 1000px; }

      .ymaps-b-balloon__content-body table {
        border-collapse: collapse;
      }

      .ymaps-b-balloon__content-body tr > * {
        padding: 0 6px;
        text-align: center;
      }
    </style>
  </head>
  <body>
    <form action="magrabx.php" method="get">
      <ul class="regions">
        <?foreach ($regions as $id => $title) {?>
          <li><label>
            <input type="checkbox" name="region[]" value="<?=$id?>"
                   <?=in_array($id, (array) @$_REQUEST['region']) ? 'checked="checked"' : ''?>>
            <?=$title?>
          </label></li>
        <?}?>
      </ul>

      <p>
        <b>Цена:</b>
        от <input name="price0" value="<?=@$_REQUEST['price0'] ?: 1000?>">
        до <input name="price1" value="<?=@$_REQUEST['price1'] ?: 4600?>">
      </p>

      <p>
        <input type="submit" name="find" value="Найти">
        <button type="button" onclick="postJS()">Обновить маркеры</button>

        <?if (isset($all)) {?>
          <b><?=count($all)?> квартир</b> (без фильтра по цене)
          / <?=count($counts)?> запросов
        <?}?>
      </p>

      <?foreach (array('twirl#blueDotIcon', 'twirl#orangeDotIcon', 'twirl#darkblueDotIcon', 'twirl#pinkDotIcon', 'twirl#whiteDotIcon', 'twirl#greyDotIcon', 'twirl#darkgreenDotIcon', 'twirl#redDotIcon', 'twirl#darkorangeDotIcon', 'twirl#violetDotIcon', 'twirl#greenDotIcon', 'twirl#yellowDotIcon', 'twirl#lightblueDotIcon', 'twirl#brownDotIcon', 'twirl#nightDotIcon', 'twirl#blackDotIcon') as $marker) {?>
        <p data-marker="<?=$marker?>">
          <img src="http://api.yandex.ru/maps/doc/jsapi/2.x/ref/images/styles/dot<?=substr($marker, 6, strpos($marker, 'Dot') - 6)?>.png">

          <input placeholder="javascript..." name="marker[<?=$marker?>]" value="<?=htmlspecialchars(@$_REQUEST['marker'][$marker], ENT_COMPAT, 'utf-8')?>">
        </p>
      <?}?>
    </form>

    <div id="map"></div>

    <?if (isset($all)) {?>
      <script type="text/javascript">
      function MultiGeocoder(options) {
          this._options = options || {};
      }

      MultiGeocoder.prototype.geocode = function (requests, options) {
          var self = this,
              opts = ymaps.util.extend({}, self._options, options),
              size = requests.length,
              promise = new ymaps.util.Promise(),
              result = [],
              geoObjects = new ymaps.GeoObjectArray();

          requests.forEach(function (request, index) {
              ymaps.geocode(request, opts).then(
                  function (response) {
                      var geoObject = response.geoObjects.get(0);

                      geoObject && (result[index] = geoObject);
                      --size || (result.forEach(geoObjects.add, geoObjects), promise.resolve({ geoObjects: geoObjects }));
                  },
                  function (err) {
                      promise.reject(err);
                  }
              );
          });

          return promise;
      };

      var coords = []
      var info = []

      <?foreach ($all as $offer) {?>
        <?if ($offer['price'] >= @$_REQUEST['price0'] and
              $offer['price'] <= @$_REQUEST['price1']) {?>
          coords.push('Санкт-Петербург, <?=$offer['address']?>')
          info.push(<?=json_encode($offer, JSON_UNESCAPED_UNICODE)?>)
        <?}?>
      <?}?>

      function init() {
        var map = new ymaps.Map('map', {
          center: [59.932666, 30.329596],
          zoom: 13,
          behaviors: ['default', 'scrollZoom'],
        })

        window.map = map

        map.controls
          .add('zoomControl', {left: 5, top: 5})
          .add('mapTools', {left: 35, top: 5});

        coder = new MultiGeocoder({boundedBy: map.getBounds()});

        var titles = {price: 'Цена', rooms: 'Комнат', area: 'Площадь',
                      areaLiving: 'Жилая', areaKitchen: 'Кухня',
                      floors: 'Этаж', houseType: 'Дом'}

        coder
          .geocode(coords)
          .then(
            function (res) {
              for (var i = 0; i < res.geoObjects.getLength(); i++) {
                var cells = info[i]
                var text = '<p style="margin-top: 0">' + $('<b>').append($('<a>').attr({href: cells.url, target: '_blank'}).text(cells.address))[0].outerHTML + '</p>'

                var th = []
                var td = []

                for (var key in titles) {
                  th.push('<th>' + titles[key] + '</th>')
                  td.push('<td>' + cells[key] + '</td>')
                }

                text += '<table><tr>' + th.join('') + '</tr>' +
                        '<tr>' + td.join('') + '</tr></table>';

                text += $('<p style="margin-bottom: 0">').text(cells.notes)[0].outerHTML

                var geo = res.geoObjects.get(i)
                info[i].geo = geo
                geo.properties.set('balloonContentBody', text)

                geo.events.add('balloonopen', function () {
                  this.options.set('preset', 'twirl#greyIcon')
                }, geo)
              }

              map.geoObjects.add(res.geoObjects)
              setTimeout(postJS, 100)
            },
            function (err) { alert(err) }
          )
      }

      function postJS() {
        var markers = []

        $('p[data-marker] input').val(function (i, value) {
          var marker = $(this).parent().attr('data-marker')
          value = $.trim(value)

          if (value) {
            markers.push([marker, value])
          }

          return value
        })

        $.each(info, function (i, cells) {
          var colored = false

          for (var i = 0; i < markers.length && !colored; i++) {
            var item = markers[i]
            var code = ''

            $.each(cells, function (key, value) {
              code += 'var ' + key + ' = \'' + (value + '').replace(/([\r\n\/'])/g, '\\$1') + '\';'
            })

            code += 'return ' + item[1]
            var func = new Function('cells', code);

            if (func(cells)) {
              cells.geo.options.set('preset', item[0])
              colored = true
            }
          }

          colored || cells.geo.options.set('preset', 'twirl#blueIcon')
        })
      }

      ymaps.ready(init);
      </script>
    <?}?>
  </body>
</html>

Здравствуйте. Как быть, если количество объектов, которые вернул геокодер не будет совпадать с количеством элементов массива с описанием?