From f0a0679f7ec71d6ae4969d39da4d6be84147fcb4 Mon Sep 17 00:00:00 2001
From: Mischa POSLAWSKY <perl@shiar.org>
Date: Fri, 28 Jun 2019 00:17:10 +0200
Subject: [PATCH] page: prefer meta description as article teaser

---
 article.inc.php | 16 ++++++++++++++--
 1 file changed, 14 insertions(+), 2 deletions(-)

diff --git a/article.inc.php b/article.inc.php
index 336511e..e3a3ee8 100644
--- a/article.inc.php
+++ b/article.inc.php
@@ -82,8 +82,20 @@ class ArchiveArticle
 	}
 	function teaser()
 	{
-		if (preg_match('{<p>(.*?)</p>}s', $this->story, $bodyp)) {
-			return $bodyp[1];
+		if (preg_match('{
+			<meta \s+ name="description" [^>]* content="([^">]*)"
+		}x', $this->preface, $meta)) {
+			# prefer specific page description if found (assume before title)
+			#TODO: strip from body contents
+			return $meta[1];
+		}
+		if (preg_match('{
+			</h2> (?: \s+ | <p\sclass="nav\b.*?</p> | <div[^>]*> )* <p> (.*?) </p>
+		}sx', $this->raw, $bodyp, PREG_OFFSET_CAPTURE)) {
+			# fallback paragraph contents following the page header
+			if ($bodyp[1][1] < 256) {
+				return $bodyp[1][0];
+			}
 		}
 	}
 
-- 
2.30.0