X-Git-Url: http://git.shiar.net/minimedit.git/blobdiff_plain/aa4a9fd115357f8d8781220e1f1345df774d6893..a09479494778e395ffe2dc75eb858ab3537ef14a:/article.inc.php

diff --git a/article.inc.php b/article.inc.php
index dd3d5b2..6c3ca93 100644
--- a/article.inc.php
+++ b/article.inc.php
@@ -16,7 +16,7 @@ function showdate($parts)
 
 class ArchiveArticle
 {
-	public $raw, $preface, $title, $body;
+	public $raw, $title, $body; # file contents
 	public $meta = [];
 
 	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)
@@ -128,19 +131,10 @@ class ArchiveArticle
 			return $override;
 		}
 
+		# paragraph contents following the page header if any
 		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+ | <div [^>]*> | \[\[[^]]*\]\] )* <p> \s* (.*?) </p>
+		}sx', $this->body, $bodyp, 0)) {
 			return $bodyp[1];
 		}
 	}