epub_file))
{
unlink($epub->epub_file);
}
if (!$epub->error)
{
$epub->CreateEPUB();
if (!$epub->error)
{
echo 'Success: ' . $epub->epub_file . ' created.
';
}
else
{
echo $epub->error;
}
}
else
{
echo $epub->error;
}
}
function lm_download_image($base_url, $url_part, $id)
{
$image_url = preg_replace('/GetPublicationContentItems-.*\.json/', $url_part . $id . '.jpg', $base_url);
$temp_img_path = './temp/' . $id . '.jpg';
$temp_content = file_get_contents($image_url);
file_put_contents($temp_img_path, $temp_content);
return $temp_img_path;
}
function inner_html($node)
{
$innerHTML = '';
foreach ($node->childNodes as $childNode)
{
$innerHTML .= $childNode->ownerDocument->saveHTML($childNode);
}
return $innerHTML;
}
date_default_timezone_set('Europe/Paris');
$date = (new DateTime('today'))->format('Ymd');
// Le Monde
if (isset($_POST['lemonde']) && $_POST['lemonde'])
{
// parameters hardcoded for now
$includeimages = false;
$includepages = true;
$url = $_POST['lmurl'];
// extract url from curl command
if (str_starts_with($url, 'curl '))
{
$url = explode("'", $url)[1];
echo 'extracted url from curl command:
' . $url . '
';
}
$epub = new TPEpubCreator();
$epub->temp_folder = 'temp/';
$epub->epub_file = 'epub/lemonde.epub';
$epub->title = 'Le Monde ' . $date ;
// cache json in case url expires
$tempjsonpath = 'temp/' . hash('md5', $url) . '.json';
$json = '';
if (file_exists($tempjsonpath))
{
$json = file_get_contents($tempjsonpath);
}
else
{
$json = file_get_contents($url);
file_put_contents($tempjsonpath, $json);
}
$publication = json_decode($json);
$content = array_filter($publication->Content, function($item) { return $item->Category == 'Le Monde'; });
usort($content, function ($a, $b) { return $a->PageNumber - $b->PageNumber; });
if ($includepages)
{
$pageindex = json_decode(file_get_contents(str_replace('GetPublicationContentItems', 'GetPublicationPages', $url)));
}
$page = 0;
foreach ($content as $article)
{
if ($includepages && $article->PageNumber > $page)
{
$page = $article->PageNumber;
$pageobj = array_values(array_filter($pageindex->Page, function($p)
{
global $page;
return $page == $p->PageNumber;
}))[0];
$pageid = $pageobj->PublicationPageID;
$path = lm_download_image($url, 'Preview-MEDIUM-', $pageid);
$epub->AddPage('
', false, 'Page ' . $page, true);
// Add page 1 as cover
if ($page == 1)
{
$epub->AddImage($path, 'image/jpeg', true);
}
}
$articlebody = array_filter($article->ContentItem, function($item) { return $item->ContentType == 'text/xml'; });
$articlebody = array_values($articlebody)[0];
if ($articlebody->Title && $articlebody->HtmlText)
{
$pagecontent = ''. strip_tags($articlebody->Title) . '
';
$pagecontent .= 'Page ' . $article->PageNumber . '
';
$author = array_filter($article->ContentItem, function($item) { return $item->ContentType == 'author/xml'; });
$author = array_values($author)[0];
if ($author->Author)
{
$pagecontent .= $author->Author;
}
if ($articlebody->Introduction)
{
$pagecontent .= '' . $articlebody->Introduction . '';
}
if ($includeimages)
{
$images = array_values(array_filter($article->ContentItem, function($item) { return $item->ContentType == 'graphic/jpeg' || $item->ContentType == 'image/jpeg'; }));
foreach ($images as $image)
{
$path = lm_download_image($url, 'Image-MEDIUM-', $image->ContentItemId);
$pagecontent .= '
';
}
}
$pagecontent .= $articlebody->HtmlText;
// debug
// echo '' . $pagecontent . '
----------------------
';
$epub->AddPage($pagecontent, false, strip_tags($articlebody->Title), true);
}
}
write_epub($epub);
}
// Mediapart
if (isset($_POST['mediapart']) && $_POST['mediapart'])
{
$feedurl = 'https://www.mediapart.fr/articles/feed';
$opts = [
'http' => [
'method' => "GET",
'header' => "Accept-language: en\nCookie: MPSESSID=" . $mp_sessionid,
]
];
$context = stream_context_create($opts);
$epub = new TPEpubCreator();
$epub->temp_folder = 'temp/';
$epub->epub_file = 'epub/mediapart.epub';
$epub->title = 'Mediapart ' . $date;
// load feeds
$feed = file_get_contents($feedurl);
$xml = new SimpleXMLElement($feed);
$items = $xml->xpath("/rss/channel/item");
foreach ($items as $item)
{
$title = $item->title;
$category = $item->xpath('dc:subject')[0];
$author = $item->xpath('dc:creator')[0];
$summary = $item->description;
$article = file_get_contents($item->link, false, $context);
$doc = new DOMDocument();
$doc->loadHTML($article);
$finder = new DomXPath($doc);
// strip images
$toremove = $finder->query('//svg');
foreach ($toremove as $elt)
{
$elt->parentNode->removeChild($elt);
}
$toremove = $finder->query('//figure');
foreach ($toremove as $elt)
{
$elt->parentNode->removeChild($elt);
}
$toremove = $finder->query('//span[@class="screen-reader-only"]');
foreach ($toremove as $elt)
{
$elt->parentNode->removeChild($elt);
}
$result = '' . $title . '
';
$result .= '' . $author . '
';
$result .= '' . $item->pubDate . '
';
$result .= '' . $summary . '
';
$nodes = $finder->query('//div[contains(@class, "paywall-restricted-content")]');
if (!$nodes->length)
{
// articles accès libre
$nodes = $finder->query('//div[contains(@class, "news__body__center__article")]');
}
if (!$nodes->length)
{
echo 'warning: could not get content of "' . $title . '"
';
}
else
{
$node = $nodes->item(0);
$innerHTML = '';
foreach ($node->childNodes as $childNode){
$innerHTML .= $childNode->ownerDocument->saveHTML($childNode);
}
$result .= '' . strip_tags($innerHTML, '
') . '
' . $title . '
';
$result .= '' . $author . '
';
$result .= '' . $item->pubDate . '
';
$result .= '' . $summary . '
';
$nodes = $finder->query('//section[@name="articleBody"]');
if (!$nodes->length)
{
echo 'warning: could not get content of "' . $title . '"
';
}
else
{
$node = $nodes->item(0);
$innerHTML = '';
foreach ($node->childNodes as $childNode){
$innerHTML .= $childNode->ownerDocument->saveHTML($childNode);
}
$result .= '' . strip_tags($innerHTML, '
') . '
' . $title . '
';
// authors
$result .= '';
$creators = ($data->initialState->{ $item->bylines[0]->id })->creators;
foreach ($creators as $creator)
{
$author = $data->initialState->{ $creator->id };
$result .= $author->displayName . ' ';
}
$result .= '
';
$result .= '' . $item->lastMajorModification . '
';
$result .= '' . $summary . '
';
$article = file_get_contents($item->url, false, $context);
$doc = new DOMDocument();
$doc->loadHTML($article);
$finder = new DomXPath($doc);
$nodes = $finder->query('//section[@name="articleBody"]');
if (!$nodes->length)
{
echo 'warning: could not get content of "' . $title . '"
';
}
else
{
$node = $nodes->item(0);
$innerHTML = '';
foreach ($node->childNodes as $childNode){
$innerHTML .= $childNode->ownerDocument->saveHTML($childNode);
}
$result .= '' . strip_tags($innerHTML, '
') . '
Generate epub
Fichiers générés
Revue de presse