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

merge le monde and mediapart

This commit is contained in:
quenousimporte 2024-06-03 17:38:46 +02:00
parent 62383911cb
commit 1eb8d51d3c
2 changed files with 105 additions and 128 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

107
lemonde.php → epub.php
View File

@ -113,6 +113,109 @@
}
}
// Mediapart
if (isset($_POST['mediapart']) && $_POST['mediapart'])
{
$sessionid = $_POST['mpsessid'];
$feedurl = 'https://www.mediapart.fr/articles/feed';
$opts = [
'http' => [
'method' => "GET",
'header' => "Accept-language: en\nCookie: MPSESSID=" . $sessionid,
]
];
$context = stream_context_create($opts);
$epub = new TPEpubCreator();
$epub->temp_folder = 'temp/';
$epub->epub_file = 'epub/mediapart' . $date . '.epub';
$epub->title = 'Mediapart';
// 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 = '<h1>' . $title . '</h1>';
$result .= '<p>' . $author . '</p>';
$result .= '<p>' . $item->pubDate . '</p>';
$result .= '<p><b>' . $summary . '</b></p>';
$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 . '"<br>';
}
else
{
$node = $nodes->item(0);
$innerHTML = '';
foreach ($node->childNodes as $childNode){
$innerHTML .= $childNode->ownerDocument->saveHTML($childNode);
}
$outerHTML = $node->ownerDocument->saveHTML($node);
$textcontent = $node->textContent;
$result .= '<div>' . strip_tags($innerHTML, '<p><b><h2><i>') . '</div>';
$epub->AddPage($result, false, $title);
}
}
if ( ! $epub->error ) {
$epub->CreateEPUB();
if ( ! $epub->error ) {
echo 'Success: ' . $epub->epub_file . ' created.<br>';
}
else
{
echo $epub->error;
}
} else {
echo $epub->error;
}
}
// list existing files
$files = glob('epub/*');
foreach ($files as $file)
@ -123,8 +226,8 @@
?>
<form method="post">
<input input name="lemonde" type="checkbox"> Le Monde<br>
GetPublicationContentItems url: <input name="lmurl"><br>
<input input name="lemonde" type="checkbox">Le Monde. GetPublicationContentItems url: <input name="lmurl"><br>
<input input name="mediapart" type="checkbox">Mediapart.<br>
<input type="submit">
</form>
</body>

126
mediapart.php
View File

@ -1,126 +0,0 @@
<html>
<body>
<?php
require('settings.php');
if (isset($_POST['mpsessid']) && $_POST['mpsessid'])
{
$mp_sessionid = $_POST['mpsessid'];
echo 'Using ' . $mp_sessionid . '<br>';
}
/*ini_set('display_errors', '1');
ini_set('display_startup_errors', '1');
error_reporting(E_ALL);*/
require('TPEpubCreator.php');
$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 = 'mediapart.epub';
$epub->title = 'Mediapart';
// load feeds
$feed = file_get_contents($feedurl);
$xml = new SimpleXMLElement($feed);
$items = $xml->xpath("/rss/channel/item");
setlocale(LC_ALL, 'fr_FR.UTF8', 'fr_FR','fr','fr','fra','fr_FR@euro');
$localedate = strftime("%A %d %B %Y");
foreach ($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 = '<h1>' . $title . '</h1>';
$result .= '<p>' . $author . '</p>';
$result .= '<p>' . $item->pubDate . '</p>';
$result .= '<p><b>' . $summary . '</b></p>';
$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 . '"<br>';
}
else
{
$node = $nodes->item(0);
$innerHTML = '';
foreach ($node->childNodes as $childNode){
$innerHTML .= $childNode->ownerDocument->saveHTML($childNode);
}
$outerHTML = $node->ownerDocument->saveHTML($node);
$textcontent = $node->textContent;
$result .= '<div>' . strip_tags($innerHTML, '<p><b><h2><i>') . '</div>';
$epub->AddPage($result, false, $title);
}
//echo $result;
//break;
}
if ( ! $epub->error )
{
$epub->CreateEPUB();
if ( ! $epub->error ) {
echo 'Success: Download your book <a href="' . $epub->epub_file . '">here</a>.';
}
} else {
echo $epub->error;
}
?>
<form method="post">
Force MPSESSID: <input name="mpsessid">
<input type="submit">
</form>
</body>
</html>