From 44017ffd35bda38b75e5f633e812de81407f2394 Mon Sep 17 00:00:00 2001
From: quenousimporte <quenousimporte@riseup.net>
Date: Sun, 7 Jul 2024 15:12:56 +0200
Subject: [PATCH] start to add socialter

---
 epub.php | 121 ++++++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 119 insertions(+), 2 deletions(-)

diff --git a/epub.php b/epub.php
index d16984f..4b3a8d0 100644
--- a/epub.php
+++ b/epub.php
@@ -97,6 +97,118 @@
 	date_default_timezone_set('Europe/Paris');
 	$date = (new DateTime('today'))->format('Ymd');
 
+	// Socialter
+	if (isset($_POST['socialter']) && $_POST['socialter'])
+	{
+		// parameters hardcoded for now
+		$includeimages = false;
+		$includepages = true;
+		$imagesonly = false;
+		$imagesize = $imagesonly ? "XLARGE" : "MEDIUM";
+
+		$url = $_POST['socurl'];
+
+		// extract url from curl command
+		if (str_starts_with($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/socialter.epub';
+		$epub->title = 'Socialter ' . $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);
+	}
+
+
 	// Le Monde
 	if (isset($_POST['lemonde']) && $_POST['lemonde'])
 	{
@@ -464,8 +576,13 @@
 	<input id="lemonde" name="lemonde" type="checkbox">
 	<label for="lemonde">From json url: </label>
 	<input name="lmurl"><br>
-	<!--Update MPSESSID: <input name="mpsessid"><br>
-	Update NYT-S: <input name="nyts"><br>-->
+	<br>
+
+	<b>Socialter</b>
+	<br>
+	<input id="socialter" name="socialter" type="checkbox">
+	<label for="socialter">From json url: </label>
+	<input name="socurl"><br>
 	<br>
 
 	<b>Mediapart</b>