thumb: indicate 1 month caching of generated output

[minimedit.git] / article.inc.php

diff --git a/article.inc.php b/article.inc.php
index bc8dbff21be1fc99969daebb3bc0bbfa2ea5d2c4..3c65a97d871b7fb8245023785a40f74646dc73ad 100644 (file)
--- 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 <meta \s+ property="( [^"]+ )" \s+ content="( [^"]* )" > \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('{<h2>(.*?)</h2>\s*}', $this->raw, 2, PREG_SPLIT_DELIM_CAPTURE);
+               }
        }
 
        function __get($col)
@@ -27,23 +44,9 @@ class ArchiveArticle
                return $this->$col = $this->$col();  # run method and cache
        }
 
-       function file()
-       {
-               if (!file_exists($this->page)) return;
-               return fopen($this->page, 'r');
-       }
-
-       function rawtitle()
-       {
-               return fgets($this->file);
-       }
-       function title()
-       {
-               return preg_replace('{<h2>(.*)</h2>\s*}', '\1', $this->rawtitle);
-       }
        function safetitle()
        {
-               return trim(strip_tags($this->title));
+               return trim($this->meta['og:title'] ?? strip_tags($this->title));
        }
        function name()
        {
@@ -74,12 +77,6 @@ class ArchiveArticle
                return showdate($this->dateparts);
        }
 
-       function body()
-       {
-               if (!$this->file) return;
-               $this->rawtitle;
-               return fread($this->file, filesize($this->page) ?: 1);
-       }
        function story()
        {
                if ( preg_match('{
@@ -94,13 +91,26 @@ class ArchiveArticle
                return $this->body;
        }
 
-       function raw()
-       {
-               return $this->rawtitle . $this->body;
-       }
        function teaser()
        {
-               if (preg_match('{<p>(.*?)</p>}s', $this->story, $bodyp)) {
+               if ($override = @$this->meta['og:description']) {
+                       # prefer specific page description if found in metadata
+                       return $override;
+               }
+
+               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] < 256) {
+                               return $bodyp[1][0];
+                       }
+               }
+
+               # starting paragraph for documents without title (assumed simple/partial)
+               if (strpos($this->raw, '<h2') === FALSE and preg_match('{
+                       \A <p> \s* (.*?) </p>
+               }sx', $this->raw, $bodyp)) {
                        return $bodyp[1];
                }
        }
@@ -113,6 +123,11 @@ class ArchiveArticle
        }
        function image()
        {
+               if ($override = @$this->meta['og:image']) {
+                       # prefer specific page image if found in metadata
+                       return $override;
+               }
+
                if ( preg_match('/\bsrc="([^"]*)"/', $this->img, $src) ) {
                        return $src[1];
                }