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

Compare commits

base: quenousimporte:158fa4c51ac63f3175e56d8aa538c838c8155923
Branches Tags
quenousimporte:main
..
compare: quenousimporte:bf56f2e120e61a77a789db57d19f319250c8a340
Branches Tags
quenousimporte:main

No commits in common. "158fa4c51ac63f3175e56d8aa538c838c8155923" and "bf56f2e120e61a77a789db57d19f319250c8a340" have entirely different histories.
158fa4c51a ... bf56f2e120

1 changed files with 146 additions and 57 deletions
Show Stats Expand all files Collapse all files

203
epub.php
View File

@ -97,54 +97,16 @@
date_default_timezone_set('Europe/Paris'); date_default_timezone_set('Europe/Paris');
$date = (new DateTime('today'))->format('Ymd'); $date = (new DateTime('today'))->format('Ymd');
// Cafeyn // Socialter
if (isset($_POST['cafeyn']) && $_POST['cafeyn']) /*if (isset($_POST['socialter']) && $_POST['socialter'])
{ {
$url = $_POST['cafeynurl']; // parameters hardcoded for now
$includeimages = false;
$includepages = true;
$imagesonly = false;
$imagesize = $imagesonly ? "XLARGE" : "MEDIUM";
// extract url from curl command $url = $_POST['socurl'];
if (str_starts_with($url, 'curl '))
{
$url = explode("'", $url)[1];
echo '<p>extracted url from curl command:</p><p>' . $url . '</p>';
}
// 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);
$pubname = $publication->metadata->title;
$epub = new TPEpubCreator();
$epub->temp_folder = 'temp/';
$epub->epub_file = 'epub/' . $pubname . '.epub';
$epub->title = $pubname . ' ' . $date ;
$list = (array)$publication->articles;
foreach ($list as $entry)
{
$articleurl = str_replace('material', 'article/' . $entry->hash . '.json', $url);
$articlejson = file_get_contents($articleurl);
$article = json_decode($articlejson);
$epub->AddPage($article->title, false, $article->title);
}
write_epub($epub);
}
// Le Monde
if (isset($_POST['lemonde']) && $_POST['lemonde'])
{
$url = $_POST['lmurl'];
// extract url from curl command // extract url from curl command
if (str_starts_with($url, 'curl ')) if (str_starts_with($url, 'curl '))
@ -155,8 +117,8 @@
$epub = new TPEpubCreator(); $epub = new TPEpubCreator();
$epub->temp_folder = 'temp/'; $epub->temp_folder = 'temp/';
$epub->epub_file = 'epub/lemonde.epub'; $epub->epub_file = 'epub/socialter.epub';
$epub->title = 'Le Monde ' . $date ; $epub->title = 'Socialter ' . $date ;
// cache json in case url expires // cache json in case url expires
$tempjsonpath = 'temp/' . hash('md5', $url) . '.json'; $tempjsonpath = 'temp/' . hash('md5', $url) . '.json';
@ -175,7 +137,10 @@
$content = array_filter($publication->Content, function($item) { return $item->Category == 'Le Monde'; }); $content = array_filter($publication->Content, function($item) { return $item->Category == 'Le Monde'; });
usort($content, function ($a, $b) { return $a->PageNumber - $b->PageNumber; }); usort($content, function ($a, $b) { return $a->PageNumber - $b->PageNumber; });
$pageindex = json_decode(file_get_contents(str_replace('GetPublicationContentItems', 'GetPublicationPages', $url))); if ($includepages)
{
$pageindex = json_decode(file_get_contents(str_replace('GetPublicationContentItems', 'GetPublicationPages', $url)));
}
$page = 0; $page = 0;
foreach ($content as $article) foreach ($content as $article)
@ -190,7 +155,7 @@
}))[0]; }))[0];
$pageid = $pageobj->PublicationPageID; $pageid = $pageobj->PublicationPageID;
$path = lm_download_image($url, 'Preview-MEDIUM-', $pageid); $path = lm_download_image($url, 'Preview-' . $imagesize . '-', $pageid);
if ($page == 1) if ($page == 1)
{ {
@ -222,9 +187,133 @@
{ {
$pagecontent .= '<b>' . $articlebody->Introduction . '</b>'; $pagecontent .= '<b>' . $articlebody->Introduction . '</b>';
} }
if ($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 .= '<p><img style="width: 100%" src="' . $path . '"></p>';
}
}
$pagecontent .= $articlebody->HtmlText; $pagecontent .= $articlebody->HtmlText;
$epub->AddPage($pagecontent, false, strip_tags($articlebody->Title), true); if (!$imagesonly)
{
$epub->AddPage($pagecontent, false, strip_tags($articlebody->Title), true);
}
}
}
write_epub($epub);
}*/
// Le Monde
if (isset($_POST['lemonde']) && $_POST['lemonde'])
{
// parameters hardcoded for now
$includeimages = false;
$includepages = true;
$imagesonly = false;
$imagesize = $imagesonly ? "XLARGE" : "MEDIUM";
$url = $_POST['lmurl'];
// extract url from curl command
if (str_starts_with($url, 'curl '))
{
$url = explode("'", $url)[1];
echo '<p>extracted url from curl command:</p><p>' . $url . '</p>';
}
$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-' . $imagesize . '-', $pageid);
if ($page == 1)
{
// cover
$epub->AddImage($path, 'image/jpeg', true);
}
else
{
$epub->AddPage('<img style="width: 100%" src="' . $path . '">', false, 'Page ' . $page, true);
}
}
$articlebody = array_filter($article->ContentItem, function($item) { return $item->ContentType == 'text/xml'; });
$articlebody = array_values($articlebody)[0];
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];
if ($author->Author)
{
$pagecontent .= $author->Author;
}
if ($articlebody->Introduction)
{
$pagecontent .= '<b>' . $articlebody->Introduction . '</b>';
}
if ($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 .= '<p><img style="width: 100%" src="' . $path . '"></p>';
}
}
$pagecontent .= $articlebody->HtmlText;
if (!$imagesonly)
{
$epub->AddPage($pagecontent, false, strip_tags($articlebody->Title), true);
}
} }
} }
write_epub($epub); write_epub($epub);
@ -240,7 +329,7 @@
$settings = file_get_contents('settings.php'); $settings = file_get_contents('settings.php');
$settings = str_replace($old, $mp_sessionid, $settings); $settings = str_replace($old, $mp_sessionid, $settings);
file_put_contents('settings.php', $settings); file_put_contents('settings.php', $settings);
echo '<div>Updated mpsessid in settings.</div>'; echo '<div>Updated mpsessid in settings</div>';
} }
$feedurl = 'https://www.mediapart.fr/articles/feed'; $feedurl = 'https://www.mediapart.fr/articles/feed';
@ -488,15 +577,15 @@
<b>Le Monde</b> <b>Le Monde</b>
<br> <br>
<input id="lemonde" name="lemonde" type="checkbox"> <input id="lemonde" name="lemonde" type="checkbox">
<label for="lemonde">From "GetPublicationContentItems" json url: </label> <label for="lemonde">From json url: </label>
<input name="lmurl"><br> <input name="lmurl"><br>
<br> <br>
<b>Cafeyn</b> <b>Socialter</b>
<br> <br>
<input id="cafeyn" name="cafeyn" type="checkbox"> <input id="socialter" name="socialter" type="checkbox">
<label for="cafeyn">From "material" json url: </label> <label for="socialter">From json url: </label>
<input name="cafeynurl"><br> <input name="socurl"><br>
<br> <br>
<b>Mediapart</b> <b>Mediapart</b>