X-Git-Url: https://git.cweiske.de/phorkie.git/blobdiff_plain/1ee5f44ac9a3125f16318ca1d24a7f7e8010a169..2a665e0f4eeb45c9d57df7dc7f9d853465b5fea9:/src/phorkie/ForkRemote.php

diff --git a/src/phorkie/ForkRemote.php b/src/phorkie/ForkRemote.php
index 31b4839..f8e319a 100644
--- a/src/phorkie/ForkRemote.php
+++ b/src/phorkie/ForkRemote.php
@@ -27,84 +27,12 @@ class ForkRemote
 
     public function parse()
     {
-        if ($this->url == '') {
-            $this->error = 'Empty fork URL';
-            return false;
-        }
-
-        $arUrl  = parse_url($this->url);
-        $scheme = isset($arUrl['scheme']) ? $arUrl['scheme'] : '';
-
-        if ($scheme == 'https' && isset($arUrl['host'])
-            && $arUrl['host'] == 'gist.github.com'
-        ) {
-            $this->arGitUrls[][] = 'git://gist.github.com/'
-                . ltrim($arUrl['path'], '/') . '.git';
-            return true;
-        }
-
-        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);
-        }
-
-        $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"]');
-        $titles = $sx->xpath('/html/head/title');
-        $pageTitle = $this->cleanPageTitle((string) reset($titles));
-
-        $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 = $str;
-            } else if ($pageTitle != '') {
-                $title = $pageTitle;
-            } else {
-                $title = 'Unnamed repository #' . ++$anonymous;
-            }
-            $url = (string)$elem['href'];
-            if ($this->isSupported($url)) {
-                ++$count;
-                $this->arGitUrls[$title][] = $url;
-            }
-        }
-
-        if ($count > 0) {
-            return true;
-        }
+        $hp = new HtmlParser();
+        $ret = $hp->extractGitUrls($this->url);
+        $this->arGitUrls = $hp->getGitUrls();
+        $this->error = $hp->error;
 
-        $this->error = 'No git:// clone URL found';
-        return false;
+        return $ret;
     }
 
     /**
@@ -149,30 +77,5 @@ class ForkRemote
     {
         $this->url = $url;
     }
-
-    public function isSupported($url)
-    {
-        $scheme = parse_url($url, PHP_URL_SCHEME);
-        return $scheme == 'git'
-            || $scheme == 'http' || $scheme == 'https';
-    }
-
-    /**
-     * Remove application names from HTML page titles
-     *
-     * @param string $title HTML page title
-     *
-     * @return string Cleaned HTML page title
-     */
-    protected function cleanPageTitle($title)
-    {
-        $title = trim($title);
-        if (substr($title, -9) == '- phorkie') {
-            $title = trim(substr($title, 0, -9));
-        }
-
-        return $title;
-    }
 }
-
 ?>