X-Git-Url: http://git.shiar.net/minimedit.git/blobdiff_plain/1abb286e88defee799e43cdd296914f74f722bde..293b5ef7d25a3c284b056272102fb04d684d82af:/article.inc.php?ds=sidebyside diff --git a/article.inc.php b/article.inc.php index bc8dbff..3c65a97 100644 --- a/article.inc.php +++ b/article.inc.php @@ -16,10 +16,27 @@ function showdate($parts) class ArchiveArticle { + public $raw, $preface, $title, $body; + public $meta = []; + function __construct($path) { $this->page = $path; $this->link = preg_replace('{(?:/index)?\.html$}', '', $path); + if (file_exists($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, '