Collapse repository URI normalization code into Arcanist

Summary: Ref T13546. Companion change to D21372. Move URI normalization code to Arcanist to we can more-often resolve remote URIs correctly.

Test Plan: Grepped for affected symbols.

Maniphest Tasks: T13546

Differential Revision: https://secure.phabricator.com/D21373

src/__phutil_library_map__.php

@@ -4653,8 +4653,6 @@ phutil_register_library_map(array(
     'PhabricatorRepositoryType' => 'applications/repository/constants/PhabricatorRepositoryType.php',
     'PhabricatorRepositoryURI' => 'applications/repository/storage/PhabricatorRepositoryURI.php',
     'PhabricatorRepositoryURIIndex' => 'applications/repository/storage/PhabricatorRepositoryURIIndex.php',
-    'PhabricatorRepositoryURINormalizer' => 'applications/repository/data/PhabricatorRepositoryURINormalizer.php',
-    'PhabricatorRepositoryURINormalizerTestCase' => 'applications/repository/data/__tests__/PhabricatorRepositoryURINormalizerTestCase.php',
     'PhabricatorRepositoryURIPHIDType' => 'applications/repository/phid/PhabricatorRepositoryURIPHIDType.php',
     'PhabricatorRepositoryURIQuery' => 'applications/repository/query/PhabricatorRepositoryURIQuery.php',
     'PhabricatorRepositoryURITestCase' => 'applications/repository/storage/__tests__/PhabricatorRepositoryURITestCase.php',
@@ -11407,8 +11405,6 @@ phutil_register_library_map(array(
       'PhabricatorConduitResultInterface',
     ),
     'PhabricatorRepositoryURIIndex' => 'PhabricatorRepositoryDAO',
-    'PhabricatorRepositoryURINormalizer' => 'Phobject',
-    'PhabricatorRepositoryURINormalizerTestCase' => 'PhabricatorTestCase',
     'PhabricatorRepositoryURIPHIDType' => 'PhabricatorPHIDType',
     'PhabricatorRepositoryURIQuery' => 'PhabricatorCursorPagedPolicyAwareQuery',
     'PhabricatorRepositoryURITestCase' => 'PhabricatorTestCase',

src/applications/repository/data/PhabricatorRepositoryURINormalizer.php

@@ -1,165 +0,0 @@
-<?php
-
-/**
- * Normalize repository URIs. For example, these URIs are generally equivalent
- * and all point at the same repository:
- *
- *   ssh://user@host/repo
- *   ssh://user@host/repo/
- *   ssh://user@host:22/repo
- *   user@host:/repo
- *   ssh://user@host/repo.git
- *
- * This class can be used to normalize URIs like this, in order to detect
- * alternate spellings of the same repository URI. In particular, the
- * @{method:getNormalizedPath} method will return:
- *
- *   repo
- *
- * ...for all of these URIs. Generally, usage looks like this:
- *
- *   $norm_a = new PhabricatorRepositoryURINormalizer($type, $uri_a);
- *   $norm_b = new PhabricatorRepositoryURINormalizer($type, $uri_b);
- *
- *   if ($norm_a->getNormalizedPath() == $norm_b->getNormalizedPath()) {
- *     // URIs appear to point at the same repository.
- *   } else {
- *     // URIs are very unlikely to be the same repository.
- *   }
- *
- * Because a repository can be hosted at arbitrarily many arbitrary URIs, there
- * is no way to completely prevent false negatives by only examining URIs
- * (that is, repositories with totally different URIs could really be the same).
- * However, normalization is relatively aggressive and false negatives should
- * be rare: if normalization says two URIs are different repositories, they
- * probably are.
- *
- * @task normal Normalizing URIs
- */
-final class PhabricatorRepositoryURINormalizer extends Phobject {
-
-  const TYPE_GIT = 'git';
-  const TYPE_SVN = 'svn';
-  const TYPE_MERCURIAL = 'hg';
-
-  private $type;
-  private $uri;
-
-  public function __construct($type, $uri) {
-    switch ($type) {
-      case self::TYPE_GIT:
-      case self::TYPE_SVN:
-      case self::TYPE_MERCURIAL:
-        break;
-      default:
-        throw new Exception(pht('Unknown URI type "%s"!', $type));
-    }
-
-    $this->type = $type;
-    $this->uri = $uri;
-  }
-
-  public static function getAllURITypes() {
-    return array(
-      self::TYPE_GIT,
-      self::TYPE_SVN,
-      self::TYPE_MERCURIAL,
-    );
-  }
-
-
-/* -(  Normalizing URIs  )--------------------------------------------------- */
-
-
-  /**
-   * @task normal
-   */
-  public function getPath() {
-    switch ($this->type) {
-      case self::TYPE_GIT:
-        $uri = new PhutilURI($this->uri);
-        return $uri->getPath();
-      case self::TYPE_SVN:
-      case self::TYPE_MERCURIAL:
-        $uri = new PhutilURI($this->uri);
-        if ($uri->getProtocol()) {
-          return $uri->getPath();
-        }
-
-        return $this->uri;
-    }
-  }
-
-  public function getNormalizedURI() {
-    return $this->getNormalizedDomain().'/'.$this->getNormalizedPath();
-  }
-
-
-  /**
-   * @task normal
-   */
-  public function getNormalizedPath() {
-    $path = $this->getPath();
-    $path = trim($path, '/');
-
-    switch ($this->type) {
-      case self::TYPE_GIT:
-        $path = preg_replace('/\.git$/', '', $path);
-        break;
-      case self::TYPE_SVN:
-      case self::TYPE_MERCURIAL:
-        break;
-    }
-
-    // If this is a Phabricator URI, strip it down to the callsign. We mutably
-    // allow you to clone repositories as "/diffusion/X/anything.git", for
-    // example.
-
-    $matches = null;
-    if (preg_match('@^(diffusion/(?:[A-Z]+|\d+))@', $path, $matches)) {
-      $path = $matches[1];
-    }
-
-    return $path;
-  }
-
-  public function getNormalizedDomain() {
-    $domain = null;
-
-    $uri = new PhutilURI($this->uri);
-    $domain = $uri->getDomain();
-
-    if (!strlen($domain)) {
-      return '<void>';
-    }
-
-    $domain = phutil_utf8_strtolower($domain);
-
-    // See T13435. If the domain for a repository URI is same as the install
-    // base URI, store it as a "<base-uri>" token instead of the actual domain
-    // so that the index does not fall out of date if the install moves.
-
-    $base_uri = PhabricatorEnv::getURI('/');
-    $base_uri = new PhutilURI($base_uri);
-    $base_domain = $base_uri->getDomain();
-    $base_domain = phutil_utf8_strtolower($base_domain);
-    if ($domain === $base_domain) {
-      return '<base-uri>';
-    }
-
-    // Likewise, store a token for the "SSH Host" domain so it can be changed
-    // without requiring an index rebuild.
-
-    $ssh_host = PhabricatorEnv::getEnvConfig('diffusion.ssh-host');
-    if (strlen($ssh_host)) {
-      $ssh_host = phutil_utf8_strtolower($ssh_host);
-      if ($domain === $ssh_host) {
-        return '<ssh-host>';
-      }
-    }
-
-    return $domain;
-  }
-
-
-}

src/applications/repository/data/__tests__/PhabricatorRepositoryURINormalizerTestCase.php

@@ -1,81 +0,0 @@
-<?php
-
-final class PhabricatorRepositoryURINormalizerTestCase
-  extends PhabricatorTestCase {
-
-  public function testGitURINormalizer() {
-    $cases = array(
-      'ssh://user@domain.com/path.git' => 'path',
-      'https://user@domain.com/path.git' => 'path',
-      'git@domain.com:path.git' => 'path',
-      'ssh://user@gitserv002.com/path.git' => 'path',
-      'ssh://htaft@domain.com/path.git' => 'path',
-      'ssh://user@domain.com/bananas.git' => 'bananas',
-      'git@domain.com:bananas.git' => 'bananas',
-      'user@domain.com:path/repo' => 'path/repo',
-      'user@domain.com:path/repo/' => 'path/repo',
-      'file:///path/to/local/repo.git' => 'path/to/local/repo',
-      '/path/to/local/repo.git' => 'path/to/local/repo',
-      'ssh://something.com/diffusion/X/anything.git' => 'diffusion/X',
-      'ssh://something.com/diffusion/X/' => 'diffusion/X',
-    );
-
-    $type_git = PhabricatorRepositoryURINormalizer::TYPE_GIT;
-
-    foreach ($cases as $input => $expect) {
-      $normal = new PhabricatorRepositoryURINormalizer($type_git, $input);
-      $this->assertEqual(
-        $expect,
-        $normal->getNormalizedPath(),
-        pht('Normalized Git path for "%s".', $input));
-    }
-  }
-
-  public function testDomainURINormalizer() {
-    $base_domain = 'base.phabricator.example.com';
-    $ssh_domain = 'ssh.phabricator.example.com';
-
-    $env = PhabricatorEnv::beginScopedEnv();
-    $env->overrideEnvConfig('phabricator.base-uri', 'http://'.$base_domain);
-    $env->overrideEnvConfig('diffusion.ssh-host', $ssh_domain);
-
-    $cases = array(
-      '/' => '<void>',
-      '/path/to/local/repo.git' => '<void>',
-      'ssh://user@domain.com/path.git' => 'domain.com',
-      'ssh://user@DOMAIN.COM/path.git' => 'domain.com',
-      'http://'.$base_domain.'/diffusion/X/' => '<base-uri>',
-      'ssh://'.$ssh_domain.'/diffusion/X/' => '<ssh-host>',
-      'git@'.$ssh_domain.':bananas.git' => '<ssh-host>',
-    );
-
-    $type_git = PhabricatorRepositoryURINormalizer::TYPE_GIT;
-
-    foreach ($cases as $input => $expect) {
-      $normal = new PhabricatorRepositoryURINormalizer($type_git, $input);
-
-      $this->assertEqual(
-        $expect,
-        $normal->getNormalizedDomain(),
-        pht('Normalized domain for "%s".', $input));
-    }
-  }
-
-  public function testSVNURINormalizer() {
-    $cases = array(
-      'file:///path/to/repo' => 'path/to/repo',
-      'file:///path/to/repo/' => 'path/to/repo',
-    );
-
-    $type_svn = PhabricatorRepositoryURINormalizer::TYPE_SVN;
-
-    foreach ($cases as $input => $expect) {
-      $normal = new PhabricatorRepositoryURINormalizer($type_svn, $input);
-      $this->assertEqual(
-        $expect,
-        $normal->getNormalizedPath(),
-        pht('Normalized SVN path for "%s".', $input));
-    }
-  }
-
-}

src/applications/repository/engine/PhabricatorRepositoryDiscoveryEngine.php

@@ -290,13 +290,13 @@ final class PhabricatorRepositoryDiscoveryEngine
     $remote_root = (string)($xml->entry[0]->repository[0]->root[0]);
     $expect_root = $repository->getSubversionPathURI();
 
-    $normal_type_svn = PhabricatorRepositoryURINormalizer::TYPE_SVN;
+    $normal_type_svn = ArcanistRepositoryURINormalizer::TYPE_SVN;
 
-    $remote_normal = id(new PhabricatorRepositoryURINormalizer(
+    $remote_normal = id(new ArcanistRepositoryURINormalizer(
       $normal_type_svn,
       $remote_root))->getNormalizedPath();
 
-    $expect_normal = id(new PhabricatorRepositoryURINormalizer(
+    $expect_normal = id(new ArcanistRepositoryURINormalizer(
       $normal_type_svn,
       $expect_root))->getNormalizedPath();
 

src/applications/repository/query/PhabricatorRepositoryQuery.php

@@ -689,10 +689,13 @@ final class PhabricatorRepositoryQuery
     // or an `svn+ssh` URI, we could deduce how to normalize it. However, this
     // would be more complicated and it's not clear if it matters in practice.
 
-    $types = PhabricatorRepositoryURINormalizer::getAllURITypes();
+    $domain_map = PhabricatorRepositoryURI::getURINormalizerDomainMap();
+
+    $types = ArcanistRepositoryURINormalizer::getAllURITypes();
     foreach ($this->uris as $uri) {
       foreach ($types as $type) {
-        $normalized_uri = new PhabricatorRepositoryURINormalizer($type, $uri);
+        $normalized_uri = new ArcanistRepositoryURINormalizer($type, $uri);
+        $normalized_uri->setDomainMap($domain_map);
         $normalized_uris[] = $normalized_uri->getNormalizedURI();
       }
     }

src/applications/repository/storage/PhabricatorRepositoryURI.php

@@ -196,19 +196,22 @@ final class PhabricatorRepositoryURI
 
     $map = array(
       PhabricatorRepositoryType::REPOSITORY_TYPE_GIT =>
-        PhabricatorRepositoryURINormalizer::TYPE_GIT,
+        ArcanistRepositoryURINormalizer::TYPE_GIT,
       PhabricatorRepositoryType::REPOSITORY_TYPE_SVN =>
-        PhabricatorRepositoryURINormalizer::TYPE_SVN,
+        ArcanistRepositoryURINormalizer::TYPE_SVN,
       PhabricatorRepositoryType::REPOSITORY_TYPE_MERCURIAL =>
-        PhabricatorRepositoryURINormalizer::TYPE_MERCURIAL,
+        ArcanistRepositoryURINormalizer::TYPE_MERCURIAL,
     );
 
     $type = $map[$vcs];
     $display = (string)$this->getDisplayURI();
 
-    $normal_uri = new PhabricatorRepositoryURINormalizer($type, $display);
+    $normalizer = new ArcanistRepositoryURINormalizer($type, $display);
 
-    return $normal_uri->getNormalizedURI();
+    $domain_map = self::getURINormalizerDomainMap();
+    $normalizer->setDomainMap($domain_map);
+
+    return $normalizer->getNormalizedURI();
   }
 
   public function getDisplayURI() {
@@ -735,4 +738,27 @@ final class PhabricatorRepositoryURI
     return array();
   }
 
+  public static function getURINormalizerDomainMap() {
+    $domain_map = array();
+
+    // See T13435. If the domain for a repository URI is same as the install
+    // base URI, store it as a "<base-uri>" token instead of the actual domain
+    // so that the index does not fall out of date if the install moves.
+
+    $base_uri = PhabricatorEnv::getURI('/');
+    $base_uri = new PhutilURI($base_uri);
+    $base_domain = $base_uri->getDomain();
+    $domain_map['<base-uri>'] = $base_domain;
+
+    // Likewise, store a token for the "SSH Host" domain so it can be changed
+    // without requiring an index rebuild.
+
+    $ssh_host = PhabricatorEnv::getEnvConfig('diffusion.ssh-host');
+    if (strlen($ssh_host)) {
+      $domain_map['<ssh-host>'] = $ssh_host;
+    }
+
+    return $domain_map;
+  }
+
 }