lm refactor
This commit is contained in:
parent
9d4bd89162
commit
2a8aec8d73
57
lemonde.php
57
lemonde.php
|
@ -3,14 +3,15 @@
|
||||||
|
|
||||||
<?php
|
<?php
|
||||||
|
|
||||||
|
require 'settings.php';
|
||||||
|
|
||||||
$url = '';
|
$url = '';
|
||||||
$coverdate = '';
|
$date = (new DateTime('today'))->format('Ymd');
|
||||||
if (isset($_POST['url']))
|
if (isset($_POST['url']))
|
||||||
{
|
{
|
||||||
require('TPEpubCreator.php');
|
require('TPEpubCreator.php');
|
||||||
|
|
||||||
$url = $_POST['url'];
|
$url = $_POST['url'];
|
||||||
$coverdate = $_POST['date'];
|
|
||||||
|
|
||||||
// extract url from curl command
|
// extract url from curl command
|
||||||
if (str_starts_with($url, 'curl '))
|
if (str_starts_with($url, 'curl '))
|
||||||
|
@ -21,11 +22,15 @@
|
||||||
|
|
||||||
$epub = new TPEpubCreator();
|
$epub = new TPEpubCreator();
|
||||||
$epub->temp_folder = 'temp/';
|
$epub->temp_folder = 'temp/';
|
||||||
$epub->epub_file = 'lemonde' . $coverdate . '.epub';
|
$epub->epub_file = 'epub/lemonde' . $date . '.epub';
|
||||||
$epub->title = 'Le Monde ' . $coverdate ;
|
$epub->title = 'Le Monde ' . $date ;
|
||||||
|
|
||||||
$coverurl = 'https://www.lemonde.fr/thumbnail/journal/'. $coverdate .'/1000/1490';
|
if ($lm_includecover)
|
||||||
$epub->AddImage( $coverurl, 'image/jpeg', true );
|
{
|
||||||
|
// 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
|
// cache json in case url expires
|
||||||
$tempjsonpath = 'temp/' . hash('md5', $url) . '.json';
|
$tempjsonpath = 'temp/' . hash('md5', $url) . '.json';
|
||||||
|
@ -52,6 +57,7 @@
|
||||||
if ($articlebody->Title && $articlebody->HtmlText)
|
if ($articlebody->Title && $articlebody->HtmlText)
|
||||||
{
|
{
|
||||||
$pagecontent = '<h1>'. strip_tags($articlebody->Title) . '</h1>';
|
$pagecontent = '<h1>'. strip_tags($articlebody->Title) . '</h1>';
|
||||||
|
$pagecontent .= '<p>Page ' . $article->PageNumber . '</p>';
|
||||||
|
|
||||||
$author = array_filter($article->ContentItem, function($item) { return $item->ContentType == 'author/xml'; });
|
$author = array_filter($article->ContentItem, function($item) { return $item->ContentType == 'author/xml'; });
|
||||||
$author = array_values($author)[0];
|
$author = array_values($author)[0];
|
||||||
|
@ -65,23 +71,26 @@
|
||||||
$pagecontent .= '<b>' . $articlebody->Introduction . '</b>';
|
$pagecontent .= '<b>' . $articlebody->Introduction . '</b>';
|
||||||
}
|
}
|
||||||
|
|
||||||
/*$images = array_values(array_filter($article->ContentItem, function($item) { return $item->ContentType == 'graphic/jpeg' || $item->ContentType == 'image/jpeg'; }));
|
if ($lm_includeimages)
|
||||||
foreach ($images as $image)
|
|
||||||
{
|
{
|
||||||
$imageid = $image->ContentItemId;
|
$images = array_values(array_filter($article->ContentItem, function($item) { return $item->ContentType == 'graphic/jpeg' || $item->ContentType == 'image/jpeg'; }));
|
||||||
$imageurl = preg_replace('/GetPublicationContentItems-.*\.json/', 'Image-MEDIUM-' . $imageid . '.jpg', $url);
|
foreach ($images as $image)
|
||||||
|
|
||||||
$tempcontent = file_get_contents($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;
|
$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;
|
$pagecontent .= $articlebody->HtmlText;
|
||||||
$epub->AddPage($pagecontent, false, strip_tags($articlebody->Title));
|
$epub->AddPage($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">
|
<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>
|
<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>
|
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">
|
<input type="submit">
|
||||||
</form>
|
</form>
|
||||||
</body>
|
</body>
|
||||||
|
|
|
@ -0,0 +1,4 @@
|
||||||
|
<?php
|
||||||
|
$lm_includeimages = false;
|
||||||
|
$lm_includecover = false;
|
||||||
|
?>
|
Loading…
Reference in New Issue