lm refactor

This commit is contained in:
quenousimporte 2024-06-03 17:03:19 +02:00
parent 9d4bd89162
commit 2a8aec8d73
2 changed files with 40 additions and 21 deletions

View File

@ -3,14 +3,15 @@
<?php
require 'settings.php';
$url = '';
$coverdate = '';
$date = (new DateTime('today'))->format('Ymd');
if (isset($_POST['url']))
{
require('TPEpubCreator.php');
$url = $_POST['url'];
$coverdate = $_POST['date'];
// extract url from curl command
if (str_starts_with($url, 'curl '))
@ -21,11 +22,15 @@
$epub = new TPEpubCreator();
$epub->temp_folder = 'temp/';
$epub->epub_file = 'lemonde' . $coverdate . '.epub';
$epub->title = 'Le Monde ' . $coverdate ;
$epub->epub_file = 'epub/lemonde' . $date . '.epub';
$epub->title = 'Le Monde ' . $date ;
$coverurl = 'https://www.lemonde.fr/thumbnail/journal/'. $coverdate .'/1000/1490';
$epub->AddImage( $coverurl, 'image/jpeg', true );
if ($lm_includecover)
{
// todo get correct cover according to date and time
$coverurl = 'https://www.lemonde.fr/thumbnail/journal/'. $date .'/1000/1490';
$epub->AddImage( $coverurl, 'image/jpeg', true );
}
// cache json in case url expires
$tempjsonpath = 'temp/' . hash('md5', $url) . '.json';
@ -52,6 +57,7 @@
if ($articlebody->Title && $articlebody->HtmlText)
{
$pagecontent = '<h1>'. strip_tags($articlebody->Title) . '</h1>';
$pagecontent .= '<p>Page ' . $article->PageNumber . '</p>';
$author = array_filter($article->ContentItem, function($item) { return $item->ContentType == 'author/xml'; });
$author = array_values($author)[0];
@ -65,23 +71,26 @@
$pagecontent .= '<b>' . $articlebody->Introduction . '</b>';
}
/*$images = array_values(array_filter($article->ContentItem, function($item) { return $item->ContentType == 'graphic/jpeg' || $item->ContentType == 'image/jpeg'; }));
foreach ($images as $image)
if ($lm_includeimages)
{
$imageid = $image->ContentItemId;
$imageurl = preg_replace('/GetPublicationContentItems-.*\.json/', 'Image-MEDIUM-' . $imageid . '.jpg', $url);
$tempcontent = file_get_contents($imageurl);
file_put_contents('temp/' . $imageid, $tempcontent);
//$epub->AddImage('temp/' . $imageid, false, false );
$pagecontent .= '<div><img src="' . $imageurl . '"></div>';
if ($image->HtmlText)
$images = array_values(array_filter($article->ContentItem, function($item) { return $item->ContentType == 'graphic/jpeg' || $item->ContentType == 'image/jpeg'; }));
foreach ($images as $image)
{
$pagecontent .= $image->HtmlText;
$imageid = $image->ContentItemId;
$imageurl = preg_replace('/GetPublicationContentItems-.*\.json/', 'Image-MEDIUM-' . $imageid . '.jpg', $url);
$tempcontent = file_get_contents($imageurl);
file_put_contents('temp/' . $imageid, $tempcontent);
//$epub->AddImage('temp/' . $imageid, false, false );
$pagecontent .= '<div><img src="' . $imageurl . '"></div>';
if ($image->HtmlText)
{
$pagecontent .= $image->HtmlText;
}
}
}*/
}
$pagecontent .= $articlebody->HtmlText;
$epub->AddPage($pagecontent, false, strip_tags($articlebody->Title));
@ -104,12 +113,18 @@
}
}
// list existing files
$files = glob('epub/*');
foreach ($files as $file)
{
echo '<a href="' . $file '.">' . $file . '</a><br>';
}
?>
<form method="post">
<p>Go to <a href="https://journal.lemonde.fr" target="_blank">https://journal.lemonde.fr</a>, hit f12, open edition, and get GetPublicationContentItems json url.</p>
Json url: <input name="url" value="<?php echo $url; ?>"><br>
Date: <input name="date" value="<?php echo $coverdate ? $coverdate : (new DateTime('tomorrow'))->format('Ymd'); /* or today if before 1pm */?>"><br>
<input type="submit">
</form>
</body>

4
settings.php.sample Normal file
View File

@ -0,0 +1,4 @@
<?php
$lm_includeimages = false;
$lm_includecover = false;
?>