public function process($content, Filter $filter = null)
{
if (!$content) {
return;
}
$content = str_replace("<br/>", " ", $content);
$this->loadHTML($content);
$timeToday = strtotime(date("Y-m-d") . " 23:59:59");
$dateYesterday = $timeToday - 24 * 3600;
$ads = array();
$tables = $this->getElementsByTagName("table");
$tableOffers = null;
foreach ($tables as $table) {
if (false !== strpos($table->getAttribute("id"), "offers_table")) {
$tableOffers = $table;
break;
}
}
if (!$tableOffers) {
return array();
}
$adNodes = $tableOffers->getElementsByTagName("td");
foreach ($adNodes as $adNode) {
if (false === strpos($adNode->getAttribute("class"), "offer")) {
continue;
}
$ad = new Ad();
$ad->setUrgent(false);
// aucun indicateur pour savoir si c'est un pro ou non.
$ad->setProfessional(false);
// permet d'éliminer les annonces déjà envoyées.
// @todo pour le moment, pas possible. Les IDs ne semblent pas
// numérique et incrémentals.
// if ($filter && $m[1] <= $filter->getMinId()) {
// continue;
// }
$rows = $adNode->getElementsByTagName("tr");
if (0 == $rows->length) {
continue;
}
$columns = $adNode->getElementsByTagName("td");
$row2_p = $rows->item(1)->getElementsByTagName("p");
// analyse de la date
$dateStr = preg_replace("#\\s+#", " ", trim($row2_p->item(1)->nodeValue));
if (!$dateStr) {
continue;
}
$aDate = explode(' ', $dateStr);
if (false !== strpos($dateStr, 'Сегодня')) {
// aujourd'hui
$time = strtotime(date("Y-m-d") . " 00:00:00");
} elseif (false !== strpos($dateStr, 'Вчера')) {
$time = strtotime(date("Y-m-d") . " 00:00:00");
$time = strtotime("-1 day", $time);
} else {
if (!isset(self::$months[$aDate[1]])) {
continue;
}
$time = strtotime(date("Y") . "-" . self::$months[$aDate[1]] . "-" . $aDate[0]);
}
$timeStr = $aDate[count($aDate) - 1];
if (false !== ($pos = mb_strpos($dateStr, ":"))) {
$time += (int) mb_substr($dateStr, $pos - 2, 2) * 3600;
$time += (int) mb_substr($dateStr, $pos + 1, 2) * 60;
if ($timeToday < $time) {
$time = strtotime("-1 year", $time);
}
}
$ad->setDate($time);
// image
$img = $columns->item(0)->getElementsByTagName("img");
if ($img->length) {
$ad->setThumbnailLink(str_replace("94x72", "644x461", $img->item(0)->getAttribute("src")));
}
// titre + lien
$link = $adNode->getElementsByTagName("h3")->item(0)->getElementsByTagName("a")->item(0);
if ($link) {
$ad->setTitle(trim($link->nodeValue));
$ad->setLink($link->getAttribute("href"));
}
// urgent
if (false !== strpos($adNode->nodeValue, "Срочно")) {
$ad->setUrgent(true);
}
// lieu
$ad->setCity(trim($row2_p->item(0)->nodeValue));
// catégorie
$ad->setCategory(trim($columns->item(1)->getElementsByTagName("p")->item(0)->nodeValue));
if (!preg_match("#ID([^.]+)\\.html#", $ad->getLink(), $m)) {
continue;
}
$ad->setId(base_convert($m[1], 32, 10));
$priceColumn = trim($columns->item(2)->nodeValue);
if (preg_match('#(?<price>[0-9\\s]+)\\s+(?<currency>грн|\\$|€)#imsU', $priceColumn, $m)) {
$ad->setPrice((int) str_replace(" ", "", $m["price"]))->setCurrency($m["currency"]);
}
if ($filter && !$filter->isValid($ad)) {
continue;
}
$ads[$ad->getId()] = $ad;
}
return $ads;
}
public function process($content, Filter $filter = null, $scheme = "http")
{
if (!$content) {
return;
}
$this->scheme = $scheme;
$this->loadHTML($content);
$timeToday = strtotime(date("Y-m-d") . " 23:59:59");
$dateYesterday = $timeToday - 24 * 3600;
$divsAd = $this->getElementsByTagName("div");
$ads = array();
foreach ($divsAd as $result) {
if (false === strpos($result->getAttribute("class"), "lbc")) {
continue;
}
$ad = new Ad();
$ad->setProfessional(false)->setUrgent(false);
$parent = $result->parentNode;
if ($parent->tagName == "a") {
$a = $parent;
} else {
$aTags = $result->getElementsByTagName("a");
if (!$aTags->length) {
continue;
}
$a = $aTags->item(0);
}
if (!preg_match('/([0-9]+)\\.htm.*/', $a->getAttribute("href"), $m)) {
continue;
}
// permet d'éliminer les annonces déjà envoyées.
if ($filter && $m[1] <= $filter->getMinId()) {
continue;
}
$ad->setLink($this->formatLink($a->getAttribute("href")))->setId($m[1]);
foreach ($result->getElementsByTagName("div") as $node) {
if ($node->hasAttribute("class")) {
$class = $node->getAttribute("class");
if ($class == "date") {
$dateStr = preg_replace("#\\s+#", " ", trim($node->nodeValue));
$aDate = explode(' ', $dateStr);
if (false !== strpos($dateStr, 'Aujourd')) {
$time = strtotime(date("Y-m-d") . " 00:00:00");
} elseif (false !== strpos($dateStr, 'Hier')) {
$time = strtotime(date("Y-m-d") . " 00:00:00");
$time = strtotime("-1 day", $time);
} else {
if (!isset(self::$months[$aDate[1]])) {
continue;
}
$time = strtotime(date("Y") . "-" . self::$months[$aDate[1]] . "-" . $aDate[0]);
}
$aTime = explode(":", $aDate[count($aDate) - 1]);
$time += (int) $aTime[0] * 3600 + (int) $aTime[1] * 60;
if ($timeToday < $time) {
$time = strtotime("-1 year", $time);
}
$ad->setDate($time);
} elseif ($class == "title") {
$ad->setTitle(trim($node->nodeValue));
} elseif ($class == "image") {
$img = $node->getElementsByTagName("img");
if ($img->length > 0) {
$img = $img->item(0);
$ad->setThumbnailLink($this->formatLink($img->getAttribute("src")));
}
} elseif ($class == "placement") {
$placement = $node->nodeValue;
if (false !== strpos($placement, "/")) {
$placement = explode("/", $placement);
$ad->setCountry(trim($placement[1]))->setCity(trim($placement[0]));
} else {
$ad->setCountry(trim($placement));
}
} elseif ($class == "category") {
$category = $node->nodeValue;
if (false !== strpos($category, "(pro)")) {
$ad->setProfessional(true);
}
$ad->setCategory(trim(str_replace("(pro)", "", $category)));
} elseif ($class == "price") {
if (preg_match("#[0-9 ]+#", $node->nodeValue, $m)) {
$ad->setPrice((int) str_replace(" ", "", trim($m[0])));
}
} elseif ($class == "urgent") {
$ad->setUrgent(true);
}
}
}
$h2Tags = $result->getElementsByTagName("h2");
if ($h2Tags->length) {
$h2 = $h2Tags->item(0);
$ad->setTitle(trim($h2->nodeValue));
}
if ($filter && !$filter->isValid($ad)) {
continue;
}
if ($ad->getDate()) {
$ads[$ad->getId()] = $ad;
}
}
return $ads;
}