From f2df190a6c4a0128eb351bc398cffd0edf8d6096 Mon Sep 17 00:00:00 2001
From: Mischa POSLAWSKY <perl@shiar.org>
Date: Sun, 17 May 2020 05:17:38 +0200
Subject: [PATCH] article: find teaser paragraph from variable offset

Simplify matching by starting at optional first </h2>.
No longer ignores (short) preceding titles such as on Lijtweg home;
more reliable otherwise.
---
 article.inc.php | 17 +++++------------
 1 file changed, 5 insertions(+), 12 deletions(-)

diff --git a/article.inc.php b/article.inc.php
index dd3d5b2..dbc545f 100644
--- a/article.inc.php
+++ b/article.inc.php
@@ -128,19 +128,12 @@ class ArchiveArticle
 			return $override;
 		}
 
+		# paragraph contents following the page header if any
+		$offset = strpos($this->raw, '</h2>');
+		$offset = $offset ? $offset + 5 : 0;
 		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] < 512) {
-				return $bodyp[1][0];
-			}
-		}
-
-		# starting paragraph for documents without title (assumed simple/partial)
-		if (strpos($this->raw, '<h2') === FALSE and preg_match('{
-			\A (?: <div [^>]*> \s* )* <p> \s* (.*?) </p>
-		}sx', $this->raw, $bodyp)) {
+			\G (?: \s+ | <p\sclass="nav\b.*?</p> | <div [^>]*> )* <p> \s* (.*?) </p>
+		}sx', $this->raw, $bodyp, 0, $offset)) {
 			return $bodyp[1];
 		}
 	}
-- 
2.30.0