X-Git-Url: http://git.shiar.net/minimedit.git/blobdiff_plain/41811556ca2ca3ca29191bee012c35e5f40a3995..b5e83271a9acb565b3b0315fe90205edf90732b0:/article.inc.php

diff --git a/article.inc.php b/article.inc.php
index 336511e..a6a4ef8 100644
--- a/article.inc.php
+++ b/article.inc.php
@@ -82,7 +82,27 @@ class ArchiveArticle
 	}
 	function teaser()
 	{
-		if (preg_match('{<p>(.*?)</p>}s', $this->story, $bodyp)) {
+		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> \s* (.*?) </p>
+		}sx', $this->raw, $bodyp, PREG_OFFSET_CAPTURE)) {
+			# fallback paragraph contents following the page header
+			if ($bodyp[1][1] < 256) {
+				return $bodyp[1][0];
+			}
+		}
+
+		# starting paragraph for documents without title (assumed simple/partial)
+		if (strpos($this->raw, '<h2') === FALSE and preg_match('{
+			\A <p> \s* (.*?) </p>
+		}sx', $this->raw, $bodyp)) {
 			return $bodyp[1];
 		}
 	}