quenousimporte
/
presse
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Compare commits

base: quenousimporte:a7a1095487700270b56b33c974f2638c30d53d53
Branches Tags
quenousimporte:main
..
compare: quenousimporte:c6f604c28b9fac6f2b74e3f15fadeae933c88f35
Branches Tags
quenousimporte:main

No commits in common. "a7a1095487700270b56b33c974f2638c30d53d53" and "c6f604c28b9fac6f2b74e3f15fadeae933c88f35" have entirely different histories.
a7a1095487 ... c6f604c28b

1 changed files with 17 additions and 116 deletions
Show Stats Expand all files Collapse all files

133
epub.php
View File

@ -150,6 +150,16 @@
return $temp_img_path;
}
function inner_html($node)
{
$innerHTML = '';
foreach ($node->childNodes as $childNode)
{
$innerHTML .= $childNode->ownerDocument->saveHTML($childNode);
}
return $innerHTML;
}
function get_json($url)
{
// cache json in case url expires
@ -168,7 +178,7 @@
}
date_default_timezone_set('Europe/Paris');
$today = (new DateTime('today'))->format('Ymd');
$date = (new DateTime('today'))->format('Ymd');
if (isset($_POST['emptycache']) && $_POST['emptycache'])
{
@ -201,7 +211,7 @@
$epub = new TPEpubCreator();
$epub->temp_folder = 'temp/';
$epub->epub_file = 'epub/' . $pubname . '.epub';
$epub->title = $pubname . ' ' . $today ;
$epub->title = $pubname . ' ' . $date ;
$list = (array)$publication->articles;
$page = 1;
@ -286,7 +296,7 @@
$epub = new TPEpubCreator();
$epub->temp_folder = 'temp/';
$epub->epub_file = 'epub/lemonde.epub';
$epub->title = 'Le Monde ' . $today;
$epub->title = 'Le Monde ' . $date ;
$publication = get_json($url);
$content = array_filter($publication->Content, function($item) { return $item->Category == 'Le Monde'; });
@ -372,7 +382,7 @@
$epub = new TPEpubCreator();
$epub->temp_folder = 'temp/';
$epub->epub_file = 'epub/mediapart.epub';
$epub->title = 'Mediapart - RSS ' . $today;
$epub->title = 'Mediapart - RSS ' . $date;
// load feeds
$feed = file_get_contents($feedurl);
@ -402,7 +412,7 @@
$epub = new TPEpubCreator();
$epub->temp_folder = 'temp/';
$epub->epub_file = 'epub/mediaparthome.epub';
$epub->title = 'Mediapart - A la une ' . $today;
$epub->title = 'Mediapart - A la une ' . $date;
$home = file_get_contents($homeurl, false, $context);
$doc = new DOMDocument();
@ -436,7 +446,7 @@
$epub = new TPEpubCreator();
$epub->temp_folder = 'temp/';
$epub->epub_file = 'epub/newyorktimes.epub';
$epub->title = 'The New York Times ' . $today;
$epub->title = 'The New York Times ' . $date;
// load feeds
$feed = file_get_contents($feedurl);
@ -492,7 +502,7 @@
$epub = new TPEpubCreator();
$epub->temp_folder = 'temp/';
$epub->epub_file = 'epub/newyorktimestoday.epub';
$epub->title = 'The New York Times today\'s paper ' . $today;
$epub->title = 'The New York Times today\'s paper ' . $date;
// grab today's paper
$start = strpos($todaypage, "window.__preloadedData = ") + 25;
@ -545,110 +555,6 @@
}
write_epub($epub);
}
// Politis
if (isset($_POST['politis']) && $_POST['politis'])
{
// cookies?
/*$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/politis.epub';
$epub->title = 'Politis - RSS ' . $today;
// load feeds
$feedurl = 'https://www.politis.fr/flux-rss-politis-fr/';
$feed = file_get_contents($feedurl);
$xml = new SimpleXMLElement($feed);
$items = $xml->xpath("/rss/channel/item");
// toc-based to test
/*
action=get_sommaire
id_produit=310761
nonce=89e48e1696
https://www.politis.fr/wp-admin/admin-ajax.php
*/
foreach ($items as $item)
{
if (new DateTime($item->pubDate) < new DateTime('-7 days'))
{
continue;
}
// restrict to last week
$url = $item->link;
$article = file_get_contents($url); //, false, $context);
$doc = new DOMDocument();
$doc->loadHTML($article);
$finder = new DomXPath($doc);
$title = $finder->query('//h1')->item(0)->textContent;
$author = $finder->query('//div[@class="auteur_date"]/span[@class="nom"]')->item(0)->textContent;
$date = $finder->query('//div[@class="auteur_date"]/span[@class="date"]')->item(0)->textContent;
$summary = $finder->query('//div[@class="extrait"]')->item(0)->textContent;
// strip images
/*$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);
}*/
$toremove = $finder->query('//script');
foreach ($toremove as $elt)
{
$elt->parentNode->removeChild($elt);
}
$result = '<h1>' . $title . '</h1>';
$result .= '<p>' . $author . '</p>';
$result .= '<p>' . $date . '</p>';
$result .= '<p><b>' . $summary . '</b></p>';
$nodes = $finder->query('//main/section[@class="section_top" or @class="section_content"]');
if (!$nodes->length)
{
echo 'warning: could not get content of "' . $title . '"<br>';
}
else
{
// top
$node = $nodes->item(0);
$nodehtml = get_clean_html($node);
$result .= $nodehtml;
// content
$node = $nodes->item(1);
$nodehtml = get_clean_html($node);
$result .= $nodehtml;
$epub->AddPage($result, false, $title);
}
}
write_epub($epub);
}
?>
<h1>Generate epub</h1>
@ -685,11 +591,6 @@
<input id="nyttoday" name="nyttoday" type="checkbox">
<label for="nyttoday">From today's paper embedded script</label>
<h2>Politis</h2>
<input id="politis" name="politis" type="checkbox">
<label for="politis">From RSS (last 7 days)</label>
<br>
<h2>Admin</h2>
<input id="emptycache" name="emptycache" type="checkbox" checked>
<label for="emptycache">Empty cache</label>