function scrapeHtml()
{
$url = "http://nivent.nicovideo.jp/";
$items = array();
$html = file_get_html($url);
foreach ($html->find('div[id=list] div[class=box clearfix]') as $key => $html_item) {
//データの元
//var_dump($html_item);
$title = $html_item->find('div[class=infoUpper] strong a');
$link = $html_item->find('div[class=infoUpper] strong a');
$description = $html_item->find('div[class=infoLower] p[class=body]');
$thumbnail = $html->find('div[class=thumb] img');
//var_dump($link);
//アイテムを格納
$item = new Item();
$item->setTitle($title[0]->innertext);
$item->setLink("http://nivent.nicovideo.jp" . $link[0]->href);
$item->setDescription(createImgTag($thumbnail[$key]->src) . $description[0]->innertext);
$items[] = $item;
}
return $items;
}
post_date,
post_source_uri,
post_source_name,
post_category,
category_id,
category_name
FROM post
LEFT JOIN category ON category_id = post_category_id
WHERE post_valid=1
LIMIT 10');
while ($row = mysql_fetch_object($request)) {
// Creating a new feed item
$rssItem = new Item();
$rssItem->setTitle($row->post_title);
$rssItem->setDescription($row->post_description);
$rssItem->setLink('http://www.mywebsite.com/blog/post.php?id=' . $row->post_id);
$rssItem->setGuid('http://www.mywebsite.com/blog/post.php?id=' . $row->post_id, true);
$rssItem->setComments('http://www.mywebsite.com/blog/post.php?id=' . $row->post_id . '#comments');
$rssItem->setAuthor($row->post_author_email, $row->post_author_name);
$rssItem->setPubDate($row->post_date);
$rssItem->setSource($row->post_source_uri, $row->post_source_name);
$rssItem->setEnclosure('http://www.mywebsite.com/blog/images/nopicture.jpg', 2800, 'image/jpg');
$rssItem->setCategory('http://www.mywebsite.com/blog/category.php.idCat=' . $row->category_id, $row->category_name);
// Add the item to the feed
$rssFeed->appendItem($rssItem);
}
// Save the feed
$rssFeed->save();
// SQL connection closing
mysql_close();
// Send headers to the browser
/**
* @dataProvider toHtmlDataProvider
*
* @param array $liParams
* @param string $attrName
* @param string $attrValue
* @param string $innerText
*/
public function testToHtml($liParams, $attrName, $attrValue, $innerText)
{
$this->item->setLink($liParams, $innerText);
$this->assertTag(['tag' => 'li', 'attributes' => [$attrName => $attrValue], 'content' => $innerText], $this->item->toHtml());
}
foreach ($feeds as $feed) {
$channel = $feed->getElementsByTagName('channel')->item(0);
// the whole channel element
$channelTitle = $channel->getElementsByTagName('link')->item(0)->nodeValue;
// channel title
$items = $feed->getElementsByTagName('item');
// All items in this feed
foreach ($items as $key) {
$title = $key->getElementsByTagName('title')->item(0)->nodeValue;
$link = $key->getElementsByTagName('link')->item(0)->nodeValue;
$description = $key->getElementsByTagName('description')->item(0)->nodeValue;
$pubDate = $key->getElementsByTagName('pubDate')->item(0)->nodeValue;
$item = new Item();
$item->setMedia($channelTitle);
$item->setTitle($title);
$item->setLink($link);
$item->setDescription($description);
$item->setPubDate($pubDate);
$result[] = $item;
}
}
usort($result, 'compareByDate');
function compareByDate(Item $a, Item $b)
{
$a = strtotime($a->getPubDate());
$b = strtotime($b->getPubDate());
if ($a == $b) {
return 0;
}
return $a < $b ? 1 : -1;
}
/**
* @covers Debril\RssAtomBundle\Protocol\Parser\Item::setLink
*/
public function testSetLink()
{
$newLink = 'http://example.com/atom';
$this->object->setLink($newLink);
$this->assertEquals($newLink, $this->object->getLink());
}
