Add a bin/files purge workflow

Summary:
We can lose file data through various means; one reasonable way is if files get deleted from disk with 'local-disk' storage. If data goes missing,

Ref T3265. Also, reduce some code duplication.

Test Plan:
Ran `bin/files purge`, `bin/files migrate`, `bin/files rebuild` with various args.

Deleted a file with "local-disk" storage, ran `bin/files purge`, made sure it got picked up.

Reviewers: dctrwatson, btrahan

Reviewed By: btrahan

CC: aran

Maniphest Tasks: T3265

Differential Revision: https://secure.phabricator.com/D6068

scripts/files/manage_files.php

@@ -17,8 +17,9 @@ $args->parseStandardArguments();
 $workflows = array(
   new PhabricatorFilesManagementEnginesWorkflow(),
   new PhabricatorFilesManagementMigrateWorkflow(),
-  new PhutilHelpArgumentWorkflow(),
   new PhabricatorFilesManagementRebuildWorkflow(),
+  new PhabricatorFilesManagementPurgeWorkflow(),
+  new PhutilHelpArgumentWorkflow(),
 );
 
 $args->parseWorkflows($workflows);

src/__phutil_library_map__.php

@@ -1014,6 +1014,7 @@ phutil_register_library_map(array(
     'PhabricatorFilesConfigOptions' => 'applications/files/config/PhabricatorFilesConfigOptions.php',
     'PhabricatorFilesManagementEnginesWorkflow' => 'applications/files/management/PhabricatorFilesManagementEnginesWorkflow.php',
     'PhabricatorFilesManagementMigrateWorkflow' => 'applications/files/management/PhabricatorFilesManagementMigrateWorkflow.php',
+    'PhabricatorFilesManagementPurgeWorkflow' => 'applications/files/management/PhabricatorFilesManagementPurgeWorkflow.php',
     'PhabricatorFilesManagementRebuildWorkflow' => 'applications/files/management/PhabricatorFilesManagementRebuildWorkflow.php',
     'PhabricatorFilesManagementWorkflow' => 'applications/files/management/PhabricatorFilesManagementWorkflow.php',
     'PhabricatorFlag' => 'applications/flag/storage/PhabricatorFlag.php',
@@ -2814,6 +2815,7 @@ phutil_register_library_map(array(
     'PhabricatorFilesConfigOptions' => 'PhabricatorApplicationConfigOptions',
     'PhabricatorFilesManagementEnginesWorkflow' => 'PhabricatorFilesManagementWorkflow',
     'PhabricatorFilesManagementMigrateWorkflow' => 'PhabricatorFilesManagementWorkflow',
+    'PhabricatorFilesManagementPurgeWorkflow' => 'PhabricatorFilesManagementWorkflow',
     'PhabricatorFilesManagementRebuildWorkflow' => 'PhabricatorFilesManagementWorkflow',
     'PhabricatorFilesManagementWorkflow' => 'PhutilArgumentWorkflow',
     'PhabricatorFlag' => 'PhabricatorFlagDAO',

src/applications/files/management/PhabricatorFilesManagementMigrateWorkflow.php

@@ -41,39 +41,8 @@ final class PhabricatorFilesManagementMigrateWorkflow
 
     $engine = PhabricatorFile::buildEngine($engine_id);
 
-    if ($args->getArg('all')) {
+    $iterator = $this->buildIterator($args);
-      if ($args->getArg('names')) {
+    if (!$iterator) {
-        throw new PhutilArgumentUsageException(
-          "Specify either a list of files or `--all`, but not both.");
-      }
-      $iterator = new LiskMigrationIterator(new PhabricatorFile());
-    } else if ($args->getArg('names')) {
-      $iterator = array();
-
-      foreach ($args->getArg('names') as $name) {
-        $name = trim($name);
-
-        $id = preg_replace('/^F/i', '', $name);
-        if (ctype_digit($id)) {
-          $file = id(new PhabricatorFile())->loadOneWhere(
-            'id = %d',
-            $id);
-          if (!$file) {
-            throw new PhutilArgumentUsageException(
-              "No file exists with id '{$name}'.");
-          }
-        } else {
-          $file = id(new PhabricatorFile())->loadOneWhere(
-            'phid = %d',
-            $name);
-          if (!$file) {
-            throw new PhutilArgumentUsageException(
-              "No file exists with PHID '{$name}'.");
-          }
-        }
-        $iterator[] = $file;
-      }
-    } else {
       throw new PhutilArgumentUsageException(
         "Either specify a list of files to migrate, or use `--all` ".
         "to migrate all files.");

src/applications/files/management/PhabricatorFilesManagementPurgeWorkflow.php

@@ -0,0 +1,69 @@
+<?php
+
+final class PhabricatorFilesManagementPurgeWorkflow
+  extends PhabricatorFilesManagementWorkflow {
+
+  public function didConstruct() {
+    $this
+      ->setName('purge')
+      ->setSynopsis('Delete files with missing data.')
+      ->setArguments(
+        array(
+          array(
+            'name'      => 'all',
+            'help'      => 'Update all files.',
+          ),
+          array(
+            'name'      => 'dry-run',
+            'help'      => 'Show what would be updated.',
+          ),
+          array(
+            'name'      => 'names',
+            'wildcard'  => true,
+          ),
+        ));
+  }
+
+  public function execute(PhutilArgumentParser $args) {
+    $console = PhutilConsole::getConsole();
+
+    $iterator = $this->buildIterator($args);
+    if (!$iterator) {
+      throw new PhutilArgumentUsageException(
+        "Either specify a list of files to purge, or use `--all` ".
+        "to purge all files.");
+    }
+
+    $is_dry_run = $args->getArg('dry-run');
+
+    foreach ($iterator as $file) {
+      $fid = 'F'.$file->getID();
+
+      try {
+        $file->loadFileData();
+        $okay = true;
+      } catch (Exception $ex) {
+        $okay = false;
+      }
+
+      if ($okay) {
+        $console->writeOut(
+          "%s: File data is OK, not purging.\n",
+          $fid);
+      } else {
+        if ($is_dry_run) {
+          $console->writeOut(
+            "%s: Would purge (dry run).\n",
+            $fid);
+        } else {
+          $console->writeOut(
+            "%s: Purging.\n",
+            $fid);
+          $file->delete();
+        }
+      }
+    }
+
+    return 0;
+  }
+}

src/applications/files/management/PhabricatorFilesManagementRebuildWorkflow.php

@@ -13,12 +13,6 @@ final class PhabricatorFilesManagementRebuildWorkflow
             'name'      => 'all',
             'help'      => 'Update all files.',
           ),
-          array(
-            'name'      => 'id',
-            'wildcard'  => true,
-            'help'      => 'Update the given file. You can specify this flag '.
-                           'multiple times.',
-          ),
           array(
             'name'      => 'dry-run',
             'help'      => 'Show what would be updated.',
@@ -31,37 +25,18 @@ final class PhabricatorFilesManagementRebuildWorkflow
             'name'      => 'rebuild-dimensions',
             'help'      => 'Rebuild image dimension information.',
           ),
+          array(
+            'name'      => 'names',
+            'wildcard'  => true,
+          ),
         ));
   }
 
   public function execute(PhutilArgumentParser $args) {
     $console = PhutilConsole::getConsole();
 
-    if ($args->getArg('all')) {
+    $iterator = $this->buildIterator($args);
-      if ($args->getArg('id')) {
+    if (!$iterator) {
-        throw new PhutilArgumentUsageException(
-          "Specify either a list of files or `--all`, but not both.");
-      }
-      $iterator = new LiskMigrationIterator(new PhabricatorFile());
-    } else if ($args->getArg('id')) {
-      $iterator = array();
-
-      foreach ($args->getArg('id') as $id) {
-        $id = trim($id);
-
-        $id = preg_replace('/^F/i', '', $id);
-        if (ctype_digit($id)) {
-          $file = id(new PhabricatorFile())->loadOneWhere(
-            'id = %d',
-            $id);
-          if (!$file) {
-            throw new PhutilArgumentUsageException(
-              "No file exists with ID '{$id}'.");
-          }
-        }
-        $iterator[] = $file;
-      }
-    } else {
       throw new PhutilArgumentUsageException(
         "Either specify a list of files to update, or use `--all` ".
         "to update all files.");

src/applications/files/management/PhabricatorFilesManagementWorkflow.php

@@ -7,4 +7,47 @@ abstract class PhabricatorFilesManagementWorkflow
     return true;
   }
 
+  protected function buildIterator(PhutilArgumentParser $args) {
+    if ($args->getArg('all')) {
+      if ($args->getArg('names')) {
+        throw new PhutilArgumentUsageException(
+          "Specify either a list of files or `--all`, but not both.");
+      }
+      return new LiskMigrationIterator(new PhabricatorFile());
+    }
+
+    if ($args->getArg('names')) {
+      $iterator = array();
+
+      foreach ($args->getArg('names') as $name) {
+        $name = trim($name);
+
+        $id = preg_replace('/^F/i', '', $name);
+        if (ctype_digit($id)) {
+          $file = id(new PhabricatorFile())->loadOneWhere(
+            'id = %d',
+            $id);
+          if (!$file) {
+            throw new PhutilArgumentUsageException(
+              "No file exists with ID '{$name}'.");
+          }
+        } else {
+          $file = id(new PhabricatorFile())->loadOneWhere(
+            'phid = %d',
+            $name);
+          if (!$file) {
+            throw new PhutilArgumentUsageException(
+              "No file exists with PHID '{$name}'.");
+          }
+        }
+        $iterator[] = $file;
+      }
+
+      return $iterator;
+    }
+
+    return null;
+  }
+
+
 }

src/applications/files/storage/PhabricatorFile.php

@@ -429,7 +429,13 @@ final class PhabricatorFile extends PhabricatorFileDAO
     // If this is the only file using the storage, delete storage
     if (!$other_file) {
       $engine = $this->instantiateStorageEngine();
+      try {
         $engine->deleteFile($this->getStorageHandle());
+      } catch (Exception $ex) {
+        // In the worst case, we're leaving some data stranded in a storage
+        // engine, which is fine.
+        phlog($ex);
+      }
     }
 
     return $ret;