X-Git-Url: http://git.shiar.nl/minimedit.git/blobdiff_plain/6fb9cdf658e15a2f774e0ad61b41fa7d8a9fa88d..bd940f48a9b1c6fb45aef4860d3f23b8094d5ba4:/article.inc.php

diff --git a/article.inc.php b/article.inc.php
index 2f5b873..93bbe3b 100644
--- a/article.inc.php
+++ b/article.inc.php
@@ -16,8 +16,8 @@ function showdate($parts)
 
 class ArchiveArticle
 {
-	public $raw, $preface, $title, $body;
-	public $meta = [];
+	public $raw, $title, $body; # file contents
+	public $meta = [];  # head metadata properties
 
 	function __construct($path)
 	{
@@ -41,8 +41,11 @@ class ArchiveArticle
 			$this->meta = array_combine($meta[1], $meta[2]); # [property => content]
 		}
 
-		@list ($this->preface, $this->title, $this->body) =
-			preg_split('{<h2>(.*?)</h2>\s*}s', $this->raw, 2, PREG_SPLIT_DELIM_CAPTURE);
+		// find significant contents
+		$this->body = preg_replace('{<aside\b.*?</aside>}s', '', $this->raw);
+		if (preg_match('{<h2>(.*?)</h2>\s*(.*)}s', $this->body, $titlematch)) {
+			list (, $this->title, $this->body) = $titlematch;
+		}
 	}
 
 	function __get($col)
@@ -129,11 +132,9 @@ class ArchiveArticle
 		}
 
 		# paragraph contents following the page header if any
-		$offset = strpos($this->raw, '</h2>');
-		$offset = $offset ? $offset + 5 : 0;
 		if (preg_match('{
-			\G (?> \s+ | <p\sclass="nav\b.*?</p> | <div [^>]*> | \[\[[^]]*\]\] )* <p> \s* (.*?) </p>
-		}sx', $this->raw, $bodyp, 0, $offset)) {
+			\G (?> \s+ | <div [^>]*> | \[\[[^]]*\]\] )* <p> \s* (.*?) </p>
+		}sx', $this->body, $bodyp, 0)) {
 			return $bodyp[1];
 		}
 	}