Issue 8700: Resolvable activities

[friendica.git] / src / Network / Probe.php

diff --git a/src/Network/Probe.php b/src/Network/Probe.php
index 5872ae587b53fe84efd4248699a98a2d9cf5c064..28f58c467621ef7f72e4e66b948c756fcae6701c 100644 (file)
--- a/src/Network/Probe.php
+++ b/src/Network/Probe.php
@@ -436,9 +436,7 @@ class Probe
                        $data['url'] = $uri;
                }
 
-               if (!empty($data['photo']) && !empty($data['baseurl'])) {
-                       $data['baseurl'] = Network::getUrlMatch(Strings::normaliseLink($data['baseurl']), Strings::normaliseLink($data['photo']));
-               } elseif (empty($data['photo'])) {
+               if (empty($data['photo'])) {
                        $data['photo'] = DI::baseUrl() . '/images/person-300.jpg';
                }
 
@@ -1763,50 +1761,87 @@ class Probe
        }
 
        /**
-        * Check page for feed link
+        * Checks HTML page for RSS feed link
         *
-        * @param string $url Page link
-        *
-        * @return string feed link
+        * @param string $url  Page link
+        * @param string $body Page body string
+        * @return string|false Feed link or false if body was invalid HTML document
         */
-       private static function getFeedLink($url)
+       public static function getFeedLink(string $url, string $body)
        {
-               $curlResult = Network::curl($url);
-               if (!$curlResult->isSuccess()) {
-                       return false;
-               }
-
                $doc = new DOMDocument();
-               if (!@$doc->loadHTML($curlResult->getBody())) {
+               if (!@$doc->loadHTML($body)) {
                        return false;
                }
 
-               $xpath = new DomXPath($doc);
+               $xpath = new DOMXPath($doc);
 
-               //$feeds = $xpath->query("/html/head/link[@type='application/rss+xml']");
-               $feeds = $xpath->query("/html/head/link[@type='application/rss+xml' and @rel='alternate']");
-               if (!is_object($feeds)) {
-                       return false;
-               }
+               $feedUrl = $xpath->evaluate('string(/html/head/link[@type="application/rss+xml" and @rel="alternate"]/@href)');
 
-               if ($feeds->length == 0) {
-                       return false;
+               $feedUrl = $feedUrl ? self::ensureAbsoluteLinkFromHTMLDoc($feedUrl, $url, $xpath) : '';
+
+               return $feedUrl;
+       }
+
+       /**
+        * Return an absolute URL in the context of a HTML document retrieved from the provided URL.
+        *
+        * Loosely based on RFC 1808
+        *
+        * @see https://tools.ietf.org/html/rfc1808
+        *
+        * @param string   $href  The potential relative href found in the HTML document
+        * @param string   $base  The HTML document URL
+        * @param DOMXPath $xpath The HTML document XPath
+        * @return string
+        */
+       private static function ensureAbsoluteLinkFromHTMLDoc(string $href, string $base, DOMXPath $xpath)
+       {
+               if (filter_var($href, FILTER_VALIDATE_URL)) {
+                       return $href;
                }
 
-               $feed_url = "";
+               $base = $xpath->evaluate('string(/html/head/base/@href)') ?: $base;
 
-               foreach ($feeds as $feed) {
-                       $attr = [];
-                       foreach ($feed->attributes as $attribute) {
-                               $attr[$attribute->name] = trim($attribute->value);
-                       }
+               $baseParts = parse_url($base);
 
-                       if (empty($feed_url) && !empty($attr['href'])) {
-                               $feed_url = $attr["href"];
+               // Naked domain case (scheme://basehost)
+               $path = $baseParts['path'] ?? '/';
+
+               // Remove the filename part of the path if it exists (/base/path/file)
+               $path = implode('/', array_slice(explode('/', $path), 0, -1));
+
+               $hrefParts = parse_url($href);
+
+               // Root path case (/path) including relative scheme case (//host/path)
+               if ($hrefParts['path'] && $hrefParts['path'][0] == '/') {
+                       $path = $hrefParts['path'];
+               } else {
+                       $path = $path . '/' . $hrefParts['path'];
+
+                       // Resolve arbitrary relative path
+                       // Lifted from https://www.php.net/manual/en/function.realpath.php#84012
+                       $parts = array_filter(explode('/', $path), 'strlen');
+                       $absolutes = array();
+                       foreach ($parts as $part) {
+                               if ('.' == $part) continue;
+                               if ('..' == $part) {
+                                       array_pop($absolutes);
+                               } else {
+                                       $absolutes[] = $part;
+                               }
                        }
+
+                       $path = '/' . implode('/', $absolutes);
                }
 
-               return $feed_url;
+               // Relative scheme case (//host/path)
+               $baseParts['host'] = $hrefParts['host'] ?? $baseParts['host'];
+               $baseParts['path'] = $path;
+               unset($baseParts['query']);
+               unset($baseParts['fragment']);
+
+               return Network::unparseURL($baseParts);
        }
 
        /**
@@ -1833,7 +1868,7 @@ class Probe
                                return false;
                        }
 
-                       $feed_url = self::getFeedLink($url);
+                       $feed_url = self::getFeedLink($url, $feed);
 
                        if (!$feed_url) {
                                return false;