Allow parsing of rare extra-broken non-UTF8 messages.

scripts/daemon/phabricator_daemon_launcher.php

@@ -192,13 +192,20 @@ switch (isset($argv[1]) ? $argv[1] : 'help') {
       throw new Exception('Unknown commit.');
     }
 
+    $workers = array();
+
+
     switch ($repo->getVersionControlSystem()) {
       case PhabricatorRepositoryType::REPOSITORY_TYPE_GIT:
-        $worker = new PhabricatorRepositoryGitCommitChangeParserWorker(
+        $workers[] = new PhabricatorRepositoryGitCommitMessageParserWorker(
+          $commit->getID());
+        $workers[] = new PhabricatorRepositoryGitCommitChangeParserWorker(
           $commit->getID());
         break;
       case PhabricatorRepositoryType::REPOSITORY_TYPE_SVN:
-        $worker = new PhabricatorRepositorySvnCommitChangeParserWorker(
+        $workers[] = new PhabricatorRepositorySvnCommitMessageParserWorker(
+          $commit->getID());
+        $workers[] = new PhabricatorRepositorySvnCommitChangeParserWorker(
           $commit->getID());
         break;
       default:
@@ -207,7 +214,10 @@ switch (isset($argv[1]) ? $argv[1] : 'help') {
 
     ExecFuture::pushEchoMode(true);
 
+    foreach ($workers as $worker) {
+      echo "Running ".get_class($worker)."...\n";
       $worker->doWork();
+    }
 
     echo "Done.\n";
 

src/applications/repository/worker/base/PhabricatorRepositoryCommitParserWorker.php

@@ -50,4 +50,38 @@ abstract class PhabricatorRepositoryCommitParserWorker
     PhabricatorRepository $repository,
     PhabricatorRepositoryCommit $commit);
 
+  /**
+   * This method is kind of awkward here but both the SVN message and
+   * change parsers use it.
+   */
+  protected function getSVNLogXMLObject($uri, $revision) {
+
+    try {
+      list($xml) = execx(
+        'svn log --xml --limit 1 --non-interactive %s@%d',
+        $uri,
+        $revision);
+    } catch (CommandException $ex) {
+      // HTTPS is generally faster and more reliable than svn+ssh, but some
+      // commit messages with non-UTF8 text can't be retrieved over HTTPS, see
+      // Facebook rE197184 for one example. Make an attempt to fall back to
+      // svn+ssh if we've failed outright to retrieve the message.
+      $fallback_uri = new PhutilURI($uri);
+      if ($fallback_uri->getProtocol() != 'https') {
+        throw $ex;
+      }
+      $fallback_uri->setProtocol('svn+ssh');
+      list($xml) = execx(
+        'svn log --xml --limit 1 --non-interactive %s@%d',
+        $fallback_uri,
+        $revision);
+    }
+
+    // Subversion may send us back commit messages which won't parse because
+    // they have non UTF-8 garbage in them. Slam them into valid UTF-8.
+    $xml = phutil_utf8ize($xml);
+
+    return new SimpleXMLElement($xml);
+  }
+
 }

src/applications/repository/worker/base/__init__.php

@@ -10,6 +10,8 @@ phutil_require_module('phabricator', 'applications/repository/storage/commit');
 phutil_require_module('phabricator', 'applications/repository/storage/repository');
 phutil_require_module('phabricator', 'infrastructure/daemon/workers/worker');
 
+phutil_require_module('phutil', 'future/exec');
+phutil_require_module('phutil', 'parser/uri');
 phutil_require_module('phutil', 'utils');
 
 

src/applications/repository/worker/commitchangeparser/svn/PhabricatorRepositorySvnCommitChangeParserWorker.php

@@ -48,12 +48,8 @@ class PhabricatorRepositorySvnCommitChangeParserWorker
 
     // Pull the top-level path changes out of "svn log". This is pretty
     // straightforward; just parse the XML log.
-    list($xml) = execx(
-      'svn log --verbose --xml --limit 1 --non-interactive %s@%d',
-      $uri,
-      $svn_commit);
+    $log = $this->getSVNLogXMLObject($uri, $svn_commit);
 
-    $log = new SimpleXMLElement($xml);
     $entry = $log->logentry[0];
 
     if (!$entry->paths) {

src/applications/repository/worker/commitmessageparser/svn/PhabricatorRepositorySvnCommitMessageParserWorker.php

@@ -25,16 +25,8 @@ class PhabricatorRepositorySvnCommitMessageParserWorker
 
     $uri = $repository->getDetail('remote-uri');
 
-    list($xml) = execx(
-      'svn log --xml --limit 1 --non-interactive %s@%d',
-      $uri,
-      $commit->getCommitIdentifier());
+    $log = $this->getSVNLogXMLObject($uri, $commit->getCommitIdentifier());
 
-    // Subversion may send us back commit messages which won't parse because
-    // they have non UTF-8 garbage in them. Slam them into valid UTF-8.
-    $xml = phutil_utf8ize($xml);
-
-    $log = new SimpleXMLElement($xml);
     $entry = $log->logentry[0];
 
     $author = (string)$entry->author;

src/applications/repository/worker/commitmessageparser/svn/__init__.php

@@ -9,8 +9,5 @@
 phutil_require_module('phabricator', 'applications/repository/worker/commitmessageparser/base');
 phutil_require_module('phabricator', 'infrastructure/daemon/workers/storage/task');
 
-phutil_require_module('phutil', 'future/exec');
-phutil_require_module('phutil', 'utils');
-
 
 phutil_require_source('PhabricatorRepositorySvnCommitMessageParserWorker.php');