public function sCreateCookieJarFile($bUse = true)
{
$oInstance = isset($this) ? $this : self::$oDefault;
if (!file_exists(self::$sCookieJarFolder)) {
mkdir(self::$sCookieJarFolder, 0777, true);
}
$sFile = self::$sCookieJarFolder . '/' . Utilitu::sMicrotime() . '_' . Utilitu::sRandomString(8) . '.cookie';
file_put_contents($sFile, '');
if ($bUse) {
$oInstance->vSetCookieJarFile($sFile);
}
return $sFile;
}
public static function sMakeIncludeHtml($sLink, $sFileType = null)
{
$sReturn = null;
$sLink = self::sFileToLink($sLink);
if (!$sFileType) {
$sFileType = Utilitu::sPregRead('#\\.([^\\.]+)$#', $sLink);
}
if ($sFileType == 'js') {
$sReturn = '<script type="text/javascript" src="' . htmlspecialchars($sLink) . '"></script>';
}
if ($sFileType == 'css') {
$sReturn = '<link rel="stylesheet" type="text/css" media="all" href="' . htmlspecialchars($sLink) . '" />';
}
return $sReturn;
}
public static function oGetCoords($sAddress)
{
$sAddress = trim($sAddress);
$sAddressHash = Utilitu::sConditionalHash($sAddress);
$oAddress = DirectDB::oSelectOne(self::$sTable, array('address_hash' => $sAddressHash));
if (!$oAddress) {
$sResponse = Curl::sGet('http://maps.google.com/maps/api/geocode/json?address=' . urlencode($sAddress) . '&sensor=false');
$oResponse = json_decode($sResponse);
$oFirstResult = $oResponse->results[0];
$oFRC = $oFirstResult->geometry->location;
DirectDB::iInsert(self::$sTable, array('address_hash' => $sAddressHash, 'address_stub' => substr($sAddress, 0, 64), 'fetched' => date('Y-m-d H:i:s'), 'x' => '' . $oFRC->lng . '', 'y' => '' . $oFRC->lat . '', 'response' => $sResponse));
$oAddress = DirectDB::oSelectOne(self::$sTable, array('address_hash' => $sAddressHash));
}
$oCoords = new StdClass();
$oCoords->nX = $oAddress->x;
$oCoords->nY = $oAddress->y;
return $oCoords;
}
public static function oParseHtml($iHtmlID)
{
$oHtml = Ad::oGetHtml($iHtmlID);
$oAd = new Ad();
$oAd->oPage->iHtmlID = $oHtml->id;
$oAd->oPage->sListID = $oHtml->list;
$oAd->oPage->sFetched = $oHtml->fetched;
$oAd->oPage->sDomain = $oHtml->domain_hash;
$oAd->oPage->sUrl = $oHtml->url;
$oDom = HtmlDomParser::str_get_html($oHtml->html);
$oMainInfo = $oDom->find('.panel-body > .row', 0);
$sOrange = $oDom->find('.headline-key-facts', 0)->innertext;
$sSquareMeters = Utilitu::sPregRead('#röße:\\s+([,\\d]+)m#', $sOrange);
$oAd->oPhysical->nSquareMeters = floatval($sSquareMeters);
$aKostenRows = $oMainInfo->find('.col-sm-5 tbody tr');
$aKostenRowsByLabel = array();
foreach ($aKostenRows as $oKostenRow) {
$aCells = $oKostenRow->find('td');
$sLabel = trim($aCells[0]->plaintext, "\t\n :");
$sValue = trim($aCells[1]->plaintext, "\t\n ");
$iPrice = 100 * intval(str_replace(array(',', '€'), array('.', ''), $sValue));
$aKostenRowsByLabel[$sLabel] = array('oDom' => $oKostenRow, 'sLabel' => $sLabel, 'sValue' => $sValue, 'iPrice' => $iPrice);
}
$aKostenMap = array('iCold' => 'Miete', 'iNeben' => 'Nebenkosten', 'iOther' => 'Sonstige Kosten', 'iBail' => 'Kaution', 'iBuy' => 'Abschlagszahlung');
foreach ($aKostenMap as $sTarget => $sSource) {
if (isset($aKostenRowsByLabel[$sSource])) {
$oAd->oPrice->{$sTarget} = $aKostenRowsByLabel[$sSource]['iPrice'];
}
}
$oAd->oPrice->iWarm = $oAd->oPrice->iCold + $oAd->oPrice->iNeben;
$sAddressHtml = $oMainInfo->find('.col-sm-4 > p', 0)->innertext;
$sAddress = trim($sAddressHtml);
$sAddress = str_replace("\n", '', $sAddress);
$sAddress = preg_replace('#<br ?/?>\\s+#', "\n", $sAddress);
$aAddress = explode("\n", $sAddress);
$oAd->oAddress->sCity = 'Aachen';
$oAd->oAddress->sZip = Utilitu::sPregRead('#\\s*(\\d+)#', $aAddress[0]);
$oAd->oAddress->sStreet = trim($aAddress[1]);
$sGeocodeAddress = $oAd->oAddress->sStreet . ', ' . $oAd->oAddress->sZip . ' ' . 'Aachen';
$oCoords = Maps::oGetCoords($sGeocodeAddress);
$oAd->oAddress->oCoords = $oCoords;
$aImageDoms = $oDom->find('img.sp-image');
foreach ($aImageDoms as $oImageDom) {
if (!isset($oImageDom->attr['data-large'])) {
continue;
}
$oImage = new StdClass();
$oImage->sUrl = str_replace('/./', '/', $oImageDom->attr['data-large']);
$sFileType = Utilitu::sPregRead('#\\.([^\\.]+)$#', $oImage->sUrl);
$oImage->sFile = self::$sImagesFolder . md5($oImage->sUrl) . '.' . $sFileType;
if (!file_exists($oImage->sFile)) {
$sImage = Curl::sGet($oImage->sUrl);
if (Curl::iGetLastStatus() == 200) {
file_put_contents($oImage->sFile, $sImage);
}
}
if (file_exists($oImage->sFile)) {
$oAd->oPage->aImages[] = $oImage;
}
}
$aDescription = array();
$aDescriptionBlocks = $oDom->find('#infobox_nachrichtsenden', 0)->parent->find('.freitext');
foreach ($aDescriptionBlocks as $oDescriptionBlock) {
$aDescription[] = $oDescriptionBlock->plaintext;
}
$oAd->oPage->sDescription = implode("\n\n", $aDescription);
$oAd->oPage->sDescription = preg_replace('#\\n\\s+#', "\n", $oAd->oPage->sDescription);
$oAd->oPage->sDescription = str_replace(' ', '', $oAd->oPage->sDescription);
$aPotentialDates = $oDom->find('.col-sm-4 .col-sm-12');
foreach ($aPotentialDates as $oPotentialDate) {
if (preg_match('#^\\s*Angebot vom:\\s*(.+)\\s*$#', $oPotentialDate->plaintext, $aMatch)) {
$sDate = date('Y-m-d H:i:s', strtotime($aMatch[1]));
$oAd->oPage->sCreated = $sDate;
$oAd->oPage->sChanged = $sDate;
}
}
/// TODO: oContact
Ad::iRemoveAdsByUrl($oHtml->url);
$oAd->vSave();
DirectDB::bUpdate('ads_htmls', array('parsed' => true), $oHtml->id);
return $oAd;
}
<?php
require_once '../init.php';
Utilitu::vReturnJson(array('aAds' => Ad::aGet("\n\t\tSELECT a.* FROM ads AS a\n\t\tJOIN ads_htmls AS h ON a.html_id = h.id\n\t\tWHERE h.list != 'wg0' AND h.fetched > '" . date('Y-m-d H:i:s', strtotime('now - 3 days')) . "';\n\t")));
public static function oGetEarliestHtmlForUrl($sUrl)
{
$sExtra = 'ORDER BY fetched ASC LIMIT 1';
$sWhere = array('url_hash' => Utilitu::sConditionalHash($sUrl));
$oReturn = DirectDB::oSelectOne('ads_htmls', $sWhere, '*', $sExtra);
return $oReturn;
}
public static function vInit()
{
self::$sHtdocs = realpath(dirname(__FILE__) . '/..');
}
