[Summary view] [Print] [Text view]

   1  <?php
   2  
   3  /**
   4   * Normalize repository URIs. For example, these URIs are generally equivalent
   5   * and all point at the same repository:
   6   *
   7   *   ssh://user@host/repo
   8   *   ssh://user@host/repo/
   9   *   ssh://user@host:22/repo
  10   *   user@host:/repo
  11   *   ssh://user@host/repo.git
  12   *
  13   * This class can be used to normalize URIs like this, in order to detect
  14   * alternate spellings of the same repository URI. In particular, the
  15   * @{method:getNormalizedPath} method will return:
  16   *
  17   *   repo
  18   *
  19   * ...for all of these URIs. Generally, usage looks like this:
  20   *
  21   *   $norm_a = new PhabricatorRepositoryURINormalizer($type, $uri_a);
  22   *   $norm_b = new PhabricatorRepositoryURINormalizer($type, $uri_b);
  23   *
  24   *   if ($norm_a->getNormalizedPath() == $norm_b->getNormalizedPath()) {
  25   *     // URIs appear to point at the same repository.
  26   *   } else {
  27   *     // URIs are very unlikely to be the same repository.
  28   *   }
  29   *
  30   * Because a repository can be hosted at arbitrarly many arbitrary URIs, there
  31   * is no way to completely prevent false negatives by only examining URIs
  32   * (that is, repositories with totally different URIs could really be the same).
  33   * However, normalization is relatively agressive and false negatives should
  34   * be rare: if normalization says two URIs are different repositories, they
  35   * probably are.
  36   *
  37   * @task normal Normalizing URIs
  38   */
  39  final class PhabricatorRepositoryURINormalizer extends Phobject {
  40  
  41    const TYPE_GIT = 'git';
  42    const TYPE_SVN = 'svn';
  43    const TYPE_MERCURIAL = 'hg';
  44  
  45    private $type;
  46    private $uri;
  47  
  48    public function __construct($type, $uri) {
  49      switch ($type) {
  50        case self::TYPE_GIT:
  51        case self::TYPE_SVN:
  52        case self::TYPE_MERCURIAL:
  53          break;
  54        default:
  55          throw new Exception(pht('Unknown URI type "%s"!'));
  56      }
  57  
  58      $this->type = $type;
  59      $this->uri = $uri;
  60    }
  61  
  62  
  63  /* -(  Normalizing URIs  )--------------------------------------------------- */
  64  
  65  
  66    /**
  67     * @task normal
  68     */
  69    public function getPath() {
  70      switch ($this->type) {
  71        case self::TYPE_GIT:
  72          $uri = new PhutilURI($this->uri);
  73          if ($uri->getProtocol()) {
  74            return $uri->getPath();
  75          }
  76  
  77          $uri = new PhutilGitURI($this->uri);
  78          if ($uri->getDomain()) {
  79            return $uri->getPath();
  80          }
  81  
  82          return $this->uri;
  83        case self::TYPE_SVN:
  84        case self::TYPE_MERCURIAL:
  85          $uri = new PhutilURI($this->uri);
  86          if ($uri->getProtocol()) {
  87            return $uri->getPath();
  88          }
  89  
  90          return $this->uri;
  91      }
  92    }
  93  
  94  
  95    /**
  96     * @task normal
  97     */
  98    public function getNormalizedPath() {
  99      $path = $this->getPath();
 100      $path = trim($path, '/');
 101  
 102      switch ($this->type) {
 103        case self::TYPE_GIT:
 104          $path = preg_replace('/\.git$/', '', $path);
 105          break;
 106        case self::TYPE_SVN:
 107        case self::TYPE_MERCURIAL:
 108          break;
 109      }
 110  
 111      // If this is a Phabricator URI, strip it down to the callsign. We mutably
 112      // allow you to clone repositories as "/diffusion/X/anything.git", for
 113      // example.
 114  
 115      $matches = null;
 116      if (preg_match('@^(diffusion/[A-Z]+)@', $path, $matches)) {
 117        $path = $matches[1];
 118      }
 119  
 120      return $path;
 121    }
 122  
 123  }