X-Git-Url: http://git.shiar.nl/minimedit.git/blobdiff_plain/41811556ca2ca3ca29191bee012c35e5f40a3995..1df0d6467575a4664c03f02ecc471870b4cfb93b:/article.inc.php diff --git a/article.inc.php b/article.inc.php index 336511e..6fcb640 100644 --- a/article.inc.php +++ b/article.inc.php @@ -17,13 +17,23 @@ function showdate($parts) class ArchiveArticle { public $raw, $preface, $title, $body; + public $meta = []; function __construct($path) { - $this->page = $path; - $this->link = preg_replace('{(?:/index)?\.html$}', '', $path); + $this->page = preg_replace('{^\.(?:/|$)}', '', $path); + $this->link = preg_replace('{(?:/index)?\.html$}', '', $this->page); if (file_exists($this->page)) { $this->raw = file_get_contents($this->page); + + if (preg_match_all('{ + \G \s* + }x', $this->raw, $meta)) { + $matchlen = array_sum(array_map('strlen', $meta[0])); + $this->raw = substr($this->raw, $matchlen); # delete matched contents + $this->meta = array_combine($meta[1], $meta[2]); # [property => content] + } + @list ($this->preface, $this->title, $this->body) = preg_split('{
(.*?)
}s', $this->story, $bodyp)) { + if ($override = @$this->meta['og:description']) { + # prefer specific page description if found in metadata + return $override; + } + + if (preg_match('{ + (?: \s+ ||
\s* (.*?)
+ }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, '