public function process_listing($a_sListingURL, $a_aDOM, $a_sListingDOM = 0, $a_bSave = true)
{
//$this->m_bIsTestRun = true;
$this->m_oDOM->clear();
//$this->m_oDOM->load_file($a_sListingURL);
$l_sHtmlStr = '';
$l_sListingNoHref = '';
if ($this->load_content($a_sListingURL, $l_sHtmlStr)) {
$l_sListingNoHref = str_replace($a_sListingDOM['add_href'], '', $a_sListingURL);
$this->m_oDOM->load($l_sHtmlStr);
$this->m_bPageLoadOK = true;
} else {
$this->m_bPageLoadOK = false;
$this->m_oTheLogger->write_log("**WARNING: cURL cant load page " . $a_sListingURL);
return false;
}
$l_aPropResults = array();
foreach ($a_aDOM as $l_aProperty) {
$l_aDetail = array();
$l_iPropertyID = $l_aProperty['id'];
$l_sContainerDOM = $l_aProperty['container_dom'];
$l_sKeyDomET = $l_aProperty['property_key_dom']['extract_type'];
$l_sValDomET = $l_aProperty['property_dom']['extract_type'];
$l_oExtractor = new CExtractor();
$l_oExtractor->reset_values($l_sKeyDomET, '');
$l_aProperty['property_key_dom']['extract_type'] = $l_oExtractor->get_extract_type();
$l_oExtractor->reset_values($l_sValDomET, '');
$l_aProperty['property_dom']['extract_type'] = $l_oExtractor->get_extract_type();
//print "\n Property $l_iPropertyID = ";
if ($l_aProperty['source'] == 'parent') {
// IF THE SOURCE IS NOT A CHILD
# $this->m_oTheLogger->write_log("\n * PROPERTY id = ".$l_iPropertyID);
if ($l_aProperty['property_key_dom']['main_dom'] == _PROP_VAL_STATIC) {
$l_aDetail[0] = $l_aProperty['property_key_dom']['element_id'];
} else {
//first get all the links from the site
// reset the href since we will check if its the right one
$l_sTmpContainer = $this->extract_container($this->m_sListingHTML, $a_sListingDOM['container'], $l_sListingNoHref);
$this->m_oDOM->clear();
$this->m_oDOM->load($l_sTmpContainer);
$l_aDetail = $this->extract_detail_dom($l_aProperty['property_dom'], $l_sContainerDOM, $l_aProperty['property_key_dom']);
$this->m_oDOM->clear();
$this->m_oDOM->load($l_sHtmlStr);
unset($l_sHtmlStr);
if ($l_aProperty['group']) {
if (strlen(trim($l_aDetail[0])) >= 1) {
$l_aDetail[0] = htmlspecialchars_decode(trim($l_aDetail[0]));
$l_oExtractor->reset_values($l_sValDomET, $l_aDetail[0]);
$l_aPropResults["{$l_iPropertyID}"] = $l_aPropResults["{$l_iPropertyID}"] . $l_oExtractor->get_result() . ', ';
}
} else {
if (strlen(trim($l_aDetail[0])) >= 1) {
$l_aDetail[0] = htmlspecialchars_decode(trim($l_aDetail[0]));
$l_oExtractor->reset_values($l_sValDomET, $l_aDetail[0]);
$l_aPropResults["{$l_iPropertyID}"] = $l_oExtractor->get_result();
}
}
# $this->m_oTheLogger->write_log("\n * RESULT = ".$l_aPropResults["$l_iPropertyID"]);
if ($a_bIsPreview) {
return htmlspecialchars_decode($l_aDetail[0]);
}
}
} else {
# $this->m_oTheLogger->write_log("\n * PROPERTY id = ".$l_iPropertyID);
if ($l_aProperty['group']) {
if ($l_aProperty['property_key_dom']['main_dom'] == _PROP_VAL_STATIC) {
$l_aDetail[0] = $l_aProperty['property_key_dom']['element_id'];
} else {
$l_aDetail = $this->extract_detail_dom($l_aProperty['property_dom'], $l_sContainerDOM, $l_aProperty['property_key_dom']);
}
//print "\n result : ".$l_aDetail[0];
if (strlen(trim($l_aDetail[0])) >= 1) {
$l_aDetail[0] = htmlspecialchars_decode(trim($l_aDetail[0]));
$l_oExtractor->reset_values($l_sValDomET, $l_aDetail[0]);
$l_aPropResults["{$l_iPropertyID}"] = $l_aPropResults["{$l_iPropertyID}"] . $l_oExtractor->get_result() . ', ';
}
} else {
if ($l_aProperty['property_key_dom']['main_dom'] == _PROP_VAL_STATIC) {
$l_aDetail[0] = $l_aProperty['property_key_dom']['element_id'];
} else {
$l_aDetail = $this->extract_detail_dom($l_aProperty['property_dom'], $l_sContainerDOM, $l_aProperty['property_key_dom']);
}
if (strlen(trim($l_aDetail[0])) >= 1) {
$l_aDetail[0] = htmlspecialchars_decode(trim($l_aDetail[0]));
$l_oExtractor->reset_values($l_sValDomET, $l_aDetail[0]);
$l_aPropResults["{$l_iPropertyID}"] = $l_oExtractor->get_result();
}
}
# $this->m_oTheLogger->write_log("\n * RESULT = ".$l_aPropResults["$l_iPropertyID"]);
unset($l_oExtractor);
if ($a_bIsPreview) {
return htmlspecialchars_decode(trim($l_aDetail[0]));
}
}
// check prices
if (!(stristr($l_aPropResults["{$l_iPropertyID}"], '{%PRICE%}') === FALSE)) {
$l_aPropResults["{$l_iPropertyID}"] = $this->process_price($l_iPropertyID, $l_aPropResults["{$l_iPropertyID}"]);
}
}
if ($a_bSave) {
try {
$this->save_listing($a_sListingURL, $l_aPropResults);
} catch (Exception $e) {
$this->m_oTheLogger->write_log("**EXCEPTION: Saving - " . $e->getMessage());
$this->m_bHasFails = true;
return false;
}
}
return true;
}
/**
Complex string extractor
@var $extractType = {%} delimited extract type string
example: left(x){%}numerics{%}price{%}plaintext
@var $originalValue = page source after loading or not set on error
@return = http return code (200 = OK)
*/
protected function extract($extractType, &$originalValue)
{
$l_oExtractor = new CExtractor();
$l_oExtractor->reset_values($extractType, htmlspecialchars_decode(trim($originalValue)));
$res = $l_oExtractor->get_result();
unset($l_oExtractor);
return $res;
}