fix #226 - base path in AnonymiationConfig and related documentation

Author: pounard

docs/content/anonymization/core-anonymizers.md

@@ -19,3 +19,4 @@ This page list all *Anonymizers* provided by *DbToolsBundle*.
 <!--@include: ./core-anonymizers/lorem-ipsum.md-->
 <!--@include: ./core-anonymizers/address.md-->
 <!--@include: ./core-anonymizers/iban-bic.md-->
+<!--@include: ./core-anonymizers/file-resolution.md-->

docs/content/anonymization/core-anonymizers/file-resolution.md

@@ -0,0 +1,39 @@
+## File name resolution
+
+In various places you can configure relative file names in order to load data,
+here is how relative file names are resolved.
+**All relative file names will be considered relative to a given _base path_.**
+
+The default base path is always stable but depends upon your selected flavor.
+
+@todo examples
+
+@@@ symfony
+
+When parsing Symfony configuration, base path will always be the project
+directory, known as `%kernel.project_dir%` variable in Symfony configuration.
+This is the directory where your `composer.json` file.
+
+@todo examples
+
+@@@
+@@@ laravel
+
+When parsing Laravel configuration, base path will always be the project
+directory, as returned by the `base_path()` Laravel function.
+
+@todo examples
+
+@@@
+@@@ standalone docker
+
+When parsing configuration in the standalone CLI version or in docker context,
+base path will be currently being parsed Yaml file.
+
+:::tip
+If you set the `workdir` option in your configuration file, then it will
+override the file directory and use it as the base path.
+
+@todo link to `workdir` documentation
+:::
+@@@

src/Anonymization/Anonymizer/Core/StringPatternAnonymizer.php

@@ -192,15 +192,21 @@ private function getAnonymizer(string $anonymizer, ?Options $options = null, int
             return $ret;
         }
 
-        $config = new AnonymizerConfig($this->tableName, $this->columnName, $anonymizer, new Options());
+        $config = new AnonymizerConfig(
+            $this->tableName,
+            $this->columnName,
+            $anonymizer,
+            new Options(),
+            __DIR__, // @todo FIXME
+        );
 
         return $this->childAnonymizers[$key] = $this
             ->getAnonymizerRegistry()
             ->createAnonymizer(
                 $anonymizer,
                 $config,
                 $options ?? new Options(),
-                $this->databaseSession
+                $this->databaseSession,
             )
         ;
     }

src/Anonymization/Config/AnonymizerConfig.php

@@ -13,5 +13,11 @@ public function __construct(
         public readonly string $targetName,
         public readonly string $anonymizer,
         public readonly Options $options,
+        /**
+         * Root directory in which this anonymizer configuration was in the
+         * first place. This allows implementations to use it in order, for
+         * example, to load files using a relative path.
+         */
+        public readonly string $basePath,
     ) {}
 }

src/Anonymization/Config/Loader/ArrayLoader.php

@@ -14,7 +14,12 @@ class ArrayLoader implements LoaderInterface
 {
     public function __construct(
         private array $data,
-        private string $connectionName = 'default',
+        private string $connectionName,
+        /**
+         * Root directory from which the configuration was loaded. It allows
+         * later file loading (for example, when sources are CSV or TXT files).
+         */
+        private readonly string $basePath,
     ) {}
 
     #[\Override]
@@ -51,6 +56,7 @@ public function load(AnonymizationConfig $config): void
                     $target,
                     $targetConfig['anonymizer'],
                     new Options($targetConfig['options']),
+                    $this->basePath,
                 ));
             }
         }

src/Anonymization/Config/Loader/AttributesLoader.php

@@ -15,6 +15,11 @@ class AttributesLoader implements LoaderInterface
 {
     public function __construct(
         private EntityManagerProvider $entityManagerProvider,
+        /**
+         * Root directory from which the configuration was loaded. It allows
+         * later file loading (for example, when sources are CSV or TXT files).
+         */
+        private readonly string $basePath,
     ) {}
 
     #[\Override]
@@ -62,6 +67,7 @@ public function load(AnonymizationConfig $config): void
                         $anonymization->type . '_' . $key,
                         $anonymization->type,
                         new Options($anonymization->options),
+                        $this->basePath,
                     ));
                 }
             }
@@ -78,6 +84,7 @@ public function load(AnonymizationConfig $config): void
                                 $metadata->getColumnName($fieldName),
                                 $propertyConfig->type,
                                 new Options($propertyConfig->options),
+                                $this->basePath,
                             ));
                         }
                     }
@@ -109,6 +116,7 @@ public function load(AnonymizationConfig $config): void
                         $columnName,
                         $anonymization->type,
                         new Options($anonymization->options),
+                        $this->basePath,
                     ));
                 }
             }

src/Anonymization/Config/Loader/PhpFileLoader.php

@@ -11,8 +11,20 @@ class PhpFileLoader extends ArrayLoader
     public function __construct(
         private string $file,
         string $connectionName = 'default',
+        /**
+         * Root directory from which the configuration was loaded. It allows
+         * later file loading (for example, when sources are CSV or TXT files).
+         *
+         * If not set, the file directory will be used instead. If set, it will
+         * override the file directory.
+         */
+        ?string $basePath = null,
     ) {
-        parent::__construct([], $connectionName);
+        // @todo dirname() is purely soft logic, it does not imply any
+        //   syscalls, whereas realpah() may.
+        $basePath ??= \realpath(\dirname($file));
+
+        parent::__construct([], $connectionName, $basePath);
     }
 
     #[\Override]

src/Anonymization/Config/Loader/YamlLoader.php

@@ -11,8 +11,20 @@ class YamlLoader extends ArrayLoader
     public function __construct(
         private string $file,
         string $connectionName = 'default',
+        /**
+         * Root directory from which the configuration was loaded. It allows
+         * later file loading (for example, when sources are CSV or TXT files).
+         *
+         * If not set, the file directory will be used instead. If set, it will
+         * override the file directory.
+         */
+        ?string $basePath = null,
     ) {
-        parent::__construct([], $connectionName);
+        // @todo dirname() is purely soft logic, it does not imply any
+        //   syscalls, whereas realpah() may.
+        $basePath ??= \realpath(\dirname($file));
+
+        parent::__construct([], $connectionName, $basePath);
     }
 
     #[\Override]

src/Bridge/Laravel/DbToolsServiceProvider.php

@@ -144,6 +144,8 @@ function (AnonymizatorFactory $factory, Application $app): void {
                 /** @var Repository $config */
                 $config = $app->make('config');
 
+                $workdir = $config->get('db-tools.workdir', $app->basePath());
+
                 foreach ($config->get('db-tools.anonymization_files', []) as $connectionName => $file) {
                     // 0 is not a good index for extension, this fails for false and 0.
                     if (!($pos = \strrpos($file, '.'))) {
@@ -155,8 +157,8 @@ function (AnonymizatorFactory $factory, Application $app): void {
 
                     $ext = \substr($file, $pos + 1);
                     $loader = match ($ext) {
-                        'php' => new PhpFileLoader($file, $connectionName),
-                        'yml', 'yaml' => new YamlLoader($file, $connectionName),
+                        'php' => new PhpFileLoader($file, $connectionName, $workdir),
+                        'yml', 'yaml' => new YamlLoader($file, $connectionName, $workdir),
                         default => throw new ConfigurationException(\sprintf(
                             "File extension \"%s\" is unsupported (given path: \"%s\").",
                             $ext,
@@ -168,7 +170,7 @@ function (AnonymizatorFactory $factory, Application $app): void {
                 }
 
                 foreach ($config->get('db-tools.anonymization', []) as $connectionName => $array) {
-                    $factory->addConfigurationLoader(new ArrayLoader($array, $connectionName));
+                    $factory->addConfigurationLoader(new ArrayLoader($array, $connectionName, $workdir));
                 }
             }
         );

src/Bridge/Standalone/Bootstrap.php

@@ -237,10 +237,10 @@ public static function bootstrap(array $config = [], array $configFiles = [], ?L
         $anonymizatorFactory = new AnonymizatorFactory($databaseSessionRegistry, $anonymizerRegistry, $logger);
 
         foreach (($config['anonymization_files'] ?? []) as $connectionName => $file) {
-            $anonymizatorFactory->addConfigurationLoader(new YamlLoader($file, $connectionName));
+            $anonymizatorFactory->addConfigurationLoader(new YamlLoader($file, $connectionName, $config['workdir']));
         }
         foreach (($config['anonymization'] ?? []) as $connectionName => $array) {
-            $anonymizatorFactory->addConfigurationLoader(new ArrayLoader($array, $connectionName));
+            $anonymizatorFactory->addConfigurationLoader(new ArrayLoader($array, $connectionName, $config['workdir']));
         }
 
         $backupperFactory = new BackupperFactory($databaseSessionRegistry, $configRegistry, $logger);