- 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;
- }