Maps::vWipeDatabase();
Ad::vWipeDatabase();
}
if (isset($_REQUEST['ad_html'])) {
$oAds = DirectDB::oSelectOne('ads', array('id' => intval($_REQUEST['ad_html'])));
$oHtml = DirectDB::oSelectOne('ads_htmls', array('id' => intval($oAds->html_id)));
$sHtml = $oHtml->html;
ODT::vDump(WgGesuchtReader::oParseHtml(null, $sHtml));
exit($sHtml);
}
$bFetch = isset($_REQUEST['fetch']) || isset($_REQUEST['fetch_and_parse']);
$bParse = isset($_REQUEST['parse']) || isset($_REQUEST['fetch_and_parse']);
if (isset($_REQUEST['search_html'])) {
if (isset($_REQUEST['url'])) {
$aWhere = array('url' => array('%like%' => $_REQUEST['url']));
$oHtml = DirectDB::oSelectOne('ads_htmls', $aWhere, 'id , url', 'ORDER BY fetched DESC LIMIT 1');
ODT::vDump($oHtml);
}
}
if ($bFetch) {
Ad::vDeleteDuplicateUrlAds();
WgGesuchtReader::vFetch();
}
if ($bParse) {
$iParse = intval($_REQUEST['parse']);
if ($iParse) {
$oAd = WgGesuchtReader::oParseHtml($iParse);
ODT::vDump($oAd);
$oHtml = Ad::oGetHtml($iParse);
ODT::vDump($oHtml);
} else {
public static function vSetDefault($aAccessData)
{
self::$oDefault = new self($aAccessData);
}
foreach ($aDirFiles as $sDirFile) {
$aFiles[] = $sDirFile;
}
} else {
$aFiles[] = $sFile;
}
}
return $aFiles;
}
require 'lib/foreign/include_all.php';
spl_autoload_register(function ($sClass) {
require_once 'lib/' . $sClass . '.php';
if (method_exists($sClass, 'vInit')) {
$sClass::vInit();
}
});
DirectDB::vSetDefault(array('sUser' => 'wohnungssuche', 'sPass' => 'c54effb3740c91a40760eda1e5c15319', 'sDaba' => 'wohnungssuche'));
#$aInitClasses = array();
#$aLibFiles = aListFiles('lib');
#foreach ($aLibFiles as $sLibFile) {
# if (preg_match('#\.php$#', $sLibFile)) {
# $sClass = preg_replace('#^.+/#', '', preg_replace('#\.php$#', '', $sLibFile));
# require_once('lib/' . $sClass . '.php');
# if (method_exists($sClass, 'vInit')) {
# $aInitClasses []= $sClass;
# }
# }
#}
#foreach ($aInitClasses as $sInitClass) {
# $sInitClass::vInit();
#}
public static function oParseHtml($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 $oAd;
}
public static function vInit()
{
DirectDB::mQuery("\n\t\t\t\tCREATE TABLE IF NOT EXISTS `" . self::$sTable . "` (\n\t\t\t\t\taddress_hash varchar(32) NOT NULL,\n\t\t\t\t\taddress_stub varchar(64) NOT NULL,\n\t\t\t\t\tfetched DATETIME,\n\t\t\t\t\tx varchar(32),\n\t\t\t\t\ty varchar(32),\n\t\t\t\t\tresponse mediumtext,\n\t\t\t\t\tPRIMARY KEY (`address_hash`)\n\t\t\t\t) ENGINE=MyISAM AUTO_INCREMENT=1 DEFAULT CHARSET=utf8;\n\t\t\t");
}
public static function vInit()
{
DirectDB::mQuery("\n\t\t\t\tCREATE TABLE IF NOT EXISTS ads_htmls (\n\t\t\t\t\tid int(9) NOT NULL AUTO_INCREMENT,\n\t\t\t\t\tfetched datetime,\n\t\t\t\t\tparsed tinyint(1),\n\t\t\t\t\tlist varchar(32),\n\t\t\t\t\tdomain_hash varchar(32),\n\t\t\t\t\turl_hash varchar(32),\n\t\t\t\t\thtml_hash varchar(32),\n\t\t\t\t\thtml longtext,\n\t\t\t\t\turl text,\n\t\t\t\t\tINDEX (url_hash),\n\t\t\t\t\tPRIMARY KEY (id)\n\t\t\t\t) ENGINE=MyISAM AUTO_INCREMENT=1 DEFAULT CHARSET=utf8;\n\t\t\t");
DirectDB::mQuery("\n\t\t\t\tCREATE TABLE IF NOT EXISTS ads (\n\t\t\t\t\tid int(9) NOT NULL AUTO_INCREMENT,\n\t\t\t\t\thtml_id int(9),\n\t\t\t\t\tcreated datetime,\n\t\t\t\t\tchanged datetime,\n\t\t\t\t\tjson_data longtext,\n\t\t\t\t\tPRIMARY KEY (id)\n\t\t\t\t) ENGINE=MyISAM AUTO_INCREMENT=1 DEFAULT CHARSET=utf8;\n\t\t\t");
}