function wiki_to_db($url, $label, $award, $bdd, $first)
{
$url = $url;
$html = connect_to($url);
$html = str_get_html($html);
// 2. get category
$h2 = $html->find('h2');
foreach ($h2 as $h2_node) {
$title = $h2_node->find('span', 1);
if (isset($title)) {
$title = $title->plaintext;
if (strcmp($title, $label) == 0) {
// 3. find ul list of books
$list = find_ul_tag($h2_node);
if (isset($list)) {
$award_books = parse_list_books($list, $first);
} else {
return;
}
// 4. insert books in db
foreach ($award_books as $book) {
$id = $book[0];
$title = $book[1];
$author = $book[2];
$date = $book[3];
$award = $award;
$amazon_best_url = "";
$amazon_price = "";
$cheapest_price = "";
$ship_price = "";
$ISBN = "";
// step 5 : insert in db
$request = 'INSERT INTO literary_award VALUES(\'' . $id . '\', \'' . $title . '\', \'' . $author . '\', \'' . $date . '\', \'' . $award . '\', \'' . $amazon_best_url . '\', \'' . $amazon_price . '\', \'' . $cheapest_price . '\', \'' . $ship_price . '\', \'' . $ISBN . '\')';
$bdd->exec($request);
echo $id . "\n";
}
break;
}
}
}
$html->clear();
unset($html);
}
function get_amazon_price($ISBN)
{
$amazon_url = "http://www.amazon.fr/dp/" . $ISBN;
$html = connect_to($amazon_url);
$html = str_get_html($html);
// 0. get the price
$amazon_price = $html->find('.priceLarge', 0);
if (isset($amazon_price)) {
$amazon_price = $amazon_price->plaintext;
$amazon_price = trim(str_replace('EUR ', '', $amazon_price));
} else {
$amazon_price = 0;
}
$html->clear();
unset($html);
return $amazon_price;
}
function reading_list(){
// step 0 : delete feeds from db
//1. connect to BDD
$pdo_options[PDO::ATTR_ERRMODE] = PDO::ERRMODE_EXCEPTION;
$bdd = new PDO(DB_PMBA, DB_USER, DB_PWD);
$bdd->exec("SET CHARACTER SET utf8");
$count = 0;
// 2. clean
$bdd->exec('DELETE FROM reading_list');
$url = "http://personalmba.com/best-business-books/";
$pmba_url = "http://personalmba.com";
$html = connect_to($url,$book);
$html = str_get_html($html);
// 2. get category
$h2 = $html->find('h2[id]');
foreach($h2 as $category){
$en_category = $category->innertext;
$list_books = $category->next_sibling()->find('li');
foreach($list_books as $book){
//3. info
$title = $book->find('a',0)->plaintext;
$amazon_com_url = $pmba_url.$book->find('a',0)->href;
$author = $book->plaintext;
$id = get_id($author);
$author = get_author($author);
$en_review = $pmba_url.$book->find('a',1)->href;
$fr_category = "";
$fr_review = "";
$amazon_best_url = "";
$amazon_price = "";
$cheapest_price = "";
$ship_price = "";
$ISBN = "";
// step 4 : insert in db
$request = 'INSERT INTO reading_list VALUES(\''.$id.'\', \''.$title.'\', \''.$author.'\', \''.$en_category.'\', \''.$fr_category.
'\', \''.$en_review.'\', \''.$fr_review.'\', \''.$amazon_com_url.'\', \''.$amazon_best_url.'\', \''.$amazon_price.'\', \''.$cheapest_price.'\', \''.$ship_price.'\', \''.$ISBN.'\')';
$bdd->exec($request);
echo $id."\n";
}
}
$html->clear();
unset($html);
// and now we're done; close it
$bdd = null;
}
