From 7c2d9c92d3178c0986f69ec75e63880ad4d99b99 Mon Sep 17 00:00:00 2001
From: quenousimporte <quenousimporte@riseup.net>
Date: Wed, 12 Jun 2024 10:54:18 +0200
Subject: [PATCH] le monde add each page preview + cover

---
 epub.php | 48 ++++++++++++++++++++++++++++++++++++++----------
 1 file changed, 38 insertions(+), 10 deletions(-)

diff --git a/epub.php b/epub.php
index dfb7ed2..e56b0b6 100644
--- a/epub.php
+++ b/epub.php
@@ -41,9 +41,10 @@
 	// Le Monde
 	if (isset($_POST['lemonde']) && $_POST['lemonde'])
 	{
-		// hardcoded for now
+		// parameters hardcoded for now
 		$includeimages = true;
-		$includecover = false;
+		$includepages = true;
+
 		$url = $_POST['lmurl'];
 
 		// extract url from curl command
@@ -58,13 +59,6 @@
 		$epub->epub_file = 'epub/lemonde.epub';
 		$epub->title = 'Le Monde ' . $date ;
 
-		if ($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';
 		$json = '';
@@ -82,8 +76,38 @@
 		$content = array_filter($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;
+
+				// SMALL / MEDIUM / XLARGE
+				$pageurl = preg_replace('/GetPublicationContentItems-.*\.json/', 'Preview-MEDIUM-' . $pageid . '.jpg', $url);
+				$tempimgpath = './temp/' . $pageid . '.jpg';
+				$tempcontent = file_get_contents($pageurl);
+				file_put_contents($tempimgpath, $tempcontent);
+				$epub->AddPage('<img src="' . $tempimgpath . '">', false, 'Page ' . $page, true);
+
+				// Add page 1 as cover
+				if ($page == 1)
+				{
+					$epub->AddImage($tempimgpath, 'image/jpeg', true);
+				}
+			}
+
 			$articlebody = array_filter($article->ContentItem, function($item) { return $item->ContentType == 'text/xml'; });
 			$articlebody = array_values($articlebody)[0];
 
@@ -117,11 +141,15 @@
 						$tempimgpath = './temp/' . $imageid . '.jpg';
 						$tempcontent = file_get_contents($imageurl);
 						file_put_contents($tempimgpath, $tempcontent);
-						$pagecontent .= '<div><img src="' . $tempimgpath . '"></div>';
+						$pagecontent .= '<p><img src="' . $tempimgpath . '"></p>';
 					}
 				}
 
 				$pagecontent .= $articlebody->HtmlText;
+
+				// debug
+				// echo '<pre style="white-space: pre-wrap;">' . $pagecontent . '<br>----------------------</pre>';
+
 				$epub->AddPage($pagecontent, false, strip_tags($articlebody->Title), true);
 			}
 		}