X-Git-Url: https://git.cweiske.de/phorkie.git/blobdiff_plain/c85fb1700ce938f28da420af9636e246912a996b..2a665e0f4eeb45c9d57df7dc7f9d853465b5fea9:/src/phorkie/ForkRemote.php

diff --git a/src/phorkie/ForkRemote.php b/src/phorkie/ForkRemote.php
index 524255a..f8e319a 100644
--- a/src/phorkie/ForkRemote.php
+++ b/src/phorkie/ForkRemote.php
@@ -3,6 +3,11 @@ namespace phorkie;
 
 class ForkRemote
 {
+    /**
+     * Contains error message when parse() failed
+     */
+    public $error;
+
     protected $url;
 
     /**
@@ -17,72 +22,24 @@ class ForkRemote
 
     public function __construct($url)
     {
-        $this->url = $url;
+        $this->url = trim($url);
     }
 
     public function parse()
     {
-        $scheme = parse_url($this->url, PHP_URL_SCHEME);
-        switch ($scheme) {
-        case 'git':
-            //clearly a git url
-            $this->arGitUrls = array(array($this->url));
-            return true;
-
-        case 'ssh':
-            //FIXME: maybe loosen this when we know how to skip the
-            //"do you trust this server" question of ssh
-            $this->error = 'ssh:// URLs are not supported';
-            return false;
-
-        case 'http':
-        case 'https':
-            return $this->extractUrlsFromHtml($this->url);
-        }
+        $hp = new HtmlParser();
+        $ret = $hp->extractGitUrls($this->url);
+        $this->arGitUrls = $hp->getGitUrls();
+        $this->error = $hp->error;
 
-        $this->error = 'Unknown URLs scheme: ' . $scheme;
-        return false;
-    }
-
-    protected function extractUrlsFromHtml($url)
-    {
-        //HTML is not necessarily well-formed, and Gitorious has many problems
-        // in this regard
-        //$sx = simplexml_load_file($url);
-        libxml_use_internal_errors(true);
-        $sx = simplexml_import_dom(\DomDocument::loadHtmlFile($url));
-        $elems = $sx->xpath('//*[@rel="vcs-git"]');
-
-        $count = $anonymous = 0;
-        foreach ($elems as $elem) {
-            if (!isset($elem['href'])) {
-                continue;
-            }
-            $str = (string)$elem;
-            if (isset($elem['title'])) {
-                //<link href=".." rel="vcs-git" title="title" />
-                $title = (string)$elem['title'];
-            } else if ($str != '') {
-                //<a href=".." rel="vcs-git">title</a>
-                $title = $str;
-            } else {
-                $title = 'Unnamed repository #' . ++$anonymous;
-            }
-            $url = (string)$elem['href'];
-            if ($this->isSupported($url)) {
-                ++$count;
-                $this->arGitUrls[$title][] = $url;
-            }
-        }
-
-        return $count > 0;
+        return $ret;
     }
 
     /**
      * Iterate through all git urls and return one if there is only
      * one supported one.
      *
-     * @return mixed Boolean false or string
+     * @return mixed Boolean false or array with keys "url" and "title"
      */
     public function getUniqueGitUrl()
     {
@@ -90,7 +47,7 @@ class ForkRemote
         foreach ($this->arGitUrls as $title => $arUrls) {
             foreach ($arUrls as $url) {
                 $nFound++;
-                $uniqueUrl = $url;
+                $uniqueUrl = array('url' => $url, 'title' => $title);
             }
         }
 
@@ -105,10 +62,20 @@ class ForkRemote
         return $this->arGitUrls;
     }
 
-    public function isSupported($url)
+    /**
+     * Get the URL from which the git URL was derived, often
+     * the HTTP URL.
+     *
+     * @return string
+     */
+    public function getUrl()
+    {
+        return $this->url;
+    }
+
+    public function setUrl($url)
     {
-        return parse_url($url, PHP_URL_SCHEME) == 'git';
+        $this->url = $url;
     }
 }
-
 ?>