no issue - allow cleanup command to remove deprecated column and tabl…

…e names from previous versions
makinacorpus · Nov 21, 2024 · 2bea7f3 · 2bea7f3
1 parent 4a7c94c
commit 2bea7f3
Show file tree

Hide file tree

Showing 2 changed files with 66 additions and 15 deletions.
diff --git a/src/Anonymization/Anonymizator.php b/src/Anonymization/Anonymizator.php
@@ -12,13 +12,12 @@
 use MakinaCorpus\DbToolsBundle\Helper\Output\NullOutput;
 use MakinaCorpus\DbToolsBundle\Helper\Output\OutputInterface;
 use MakinaCorpus\QueryBuilder\DatabaseSession;
+use MakinaCorpus\QueryBuilder\Vendor;
 use MakinaCorpus\QueryBuilder\Error\Server\DatabaseObjectDoesNotExistError;
 use MakinaCorpus\QueryBuilder\Query\Update;
 use MakinaCorpus\QueryBuilder\Schema\Read\Column;
 use MakinaCorpus\QueryBuilder\Schema\Read\Index;
-use MakinaCorpus\QueryBuilder\Schema\Read\Table;
 use MakinaCorpus\QueryBuilder\Type\Type;
-use MakinaCorpus\QueryBuilder\Vendor;
 use Psr\Log\LoggerAwareInterface;
 use Psr\Log\LoggerAwareTrait;
 use Psr\Log\NullLogger;
@@ -27,6 +26,21 @@ class Anonymizator implements LoggerAwareInterface
 {
     use LoggerAwareTrait;
 
+    /**
+     * Used for the garbage collection/cleanup procedure, removes tables with older names.
+     */
+    protected const DEPRECATED_JOIN_ID = [
+        '_anonymize_id', // @todo Remove in 3.0
+        '_anonymizer_id', // @todo Remove in 3.0
+    ];
+
+    /**
+     * Used for the garbage collection/cleanup procedure, removes tables with older names.
+     */
+    protected const DEPRECATED_TEMP_TABLE_PREFIX = [
+        'anonymizer_sample_', // @todo Remove in 3.0
+    ];
+
     private OutputInterface $output;
 
     public function __construct(
@@ -268,6 +282,10 @@ public function clean(): void
     }
 
     /**
+     * @param bool $withDeprecated
+     *   Also collect garbage from previous versions. This is not default
+     *   because it might conflict with legit project tables. We leave this
+     *   responsability to the user.
      *
      * @return array<array<string, string>>
      *   Each item is an array structured as such:
@@ -277,23 +295,39 @@ public function clean(): void
      *     'table' => 'table_name' (only for columns and indexes)
      *   ]
      */
-    public function collectGarbage(): array
+    public function collectGarbage(bool $withDeprecated = false): array
     {
         $schemaManager = $this->databaseSession->getSchemaManager();
         $garbage = [];
 
+        $prefixes = [AbstractAnonymizer::TEMP_TABLE_PREFIX];
+        // For backward compatibilty, removes legacy table names as well.
+        if ($withDeprecated) {
+            foreach (self::DEPRECATED_TEMP_TABLE_PREFIX as $prefix) {
+                $prefixes[] = $prefix;
+            }
+        }
+
         foreach ($schemaManager->listTables() as $tableName) {
-            if (\str_starts_with($tableName, AbstractAnonymizer::TEMP_TABLE_PREFIX)) {
-                $garbage[] = ['type' => 'table', 'name' => $tableName];
-            } else {
-                $garbage = \array_merge($garbage, $this->collectGarbageInto($tableName));
+            $found = false;
+            foreach ($prefixes as $prefix) {
+                if (\str_starts_with($tableName, $prefix)) {
+                    $garbage[] = ['type' => 'table', 'name' => $tableName];
+                    $found = true;
+                    break;
+                }
+            }
+
+            // If table is not marked for deletion, lookup for added columns.
+            if (!$found) {
+                $garbage = \array_merge($garbage, $this->collectGarbageInto($tableName, $withDeprecated));
             }
         }
 
         return $garbage;
     }
 
-    protected function collectGarbageInto(string $table): array
+    protected function collectGarbageInto(string $table, bool $withDeprecated = false): array
     {
         $schemaManager = $this->databaseSession->getSchemaManager();
         $garbage = [];
@@ -320,15 +354,26 @@ protected function collectGarbageInto(string $table): array
             }
         }
 
+        $names = [AbstractAnonymizer::JOIN_ID];
+        // For backward compatibilty, removes legacy column names as well.
+        if ($withDeprecated) {
+            foreach (self::DEPRECATED_JOIN_ID as $name) {
+                $names[] = $name;
+            }
+        }
+
         foreach ($table->getColumns() as $column) {
             \assert($column instanceof Column);
 
-            if (AbstractAnonymizer::JOIN_ID === $column->getName()) {
-                $garbage[] = [
-                    'type' => 'column',
-                    'name' => $column->getName(),
-                    'table' => $table->getName(),
-                ];
+            foreach ($names as $name) {
+                if ($name === $column->getName()) {
+                    $garbage[] = [
+                        'type' => 'column',
+                        'name' => $column->getName(),
+                        'table' => $table->getName(),
+                    ];
+                    break;
+                }
             }
         }
 

diff --git a/src/Command/Anonymization/CleanCommand.php b/src/Command/Anonymization/CleanCommand.php
@@ -35,6 +35,12 @@ protected function configure(): void
                 InputOption::VALUE_OPTIONAL,
                 'A doctrine connection name. If not given, use default connection'
             )
+            ->addOption(
+                'deprecated',
+                'd',
+                InputOption::VALUE_NONE,
+                'Also attempt to cleanup tables and columns whose name have changed from the previous versions'
+            )
             ->addOption(
                 'force',
                 null,
@@ -70,7 +76,7 @@ protected function execute(InputInterface $input, OutputInterface $output): int
             ->setOutput(new ConsoleOutput($io))
         ;
 
-        $garbage = $anonymizator->collectGarbage();
+        $garbage = $anonymizator->collectGarbage((bool) $input->getOption('deprecated'));
 
         if (!$garbage) {
             $io->success("There is no left-overs to clean, exiting.");