feat: anonymization command for request/response XML dumps

2026-03-04 10:47:20 +01:00
parent cd5d353158
commit 90fc66277e
5 changed files with 744 additions and 1 deletions
@@ -0,0 +1,158 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Command;
+
+use App\Service\BpnXmlAnonymizer;
+use Symfony\Component\Console\Attribute\AsCommand;
+use Symfony\Component\Console\Command\Command;
+use Symfony\Component\Console\Input\InputArgument;
+use Symfony\Component\Console\Input\InputInterface;
+use Symfony\Component\Console\Output\OutputInterface;
+use Symfony\Component\Console\Style\SymfonyStyle;
+
+#[AsCommand(
+    name: 'app:bpn:xml-anonymize',
+    description: 'Anonymize personal data in BPN XML request/response dumps',
+)]
+class BpnXmlAnonymizeCommand extends Command
+{
+    public function __construct(
+        private readonly BpnXmlAnonymizer $xmlAnonymizer,
+    ) {
+        parent::__construct();
+    }
+
+    protected function configure(): void
+    {
+        $this
+            ->addArgument('infile', InputArgument::REQUIRED, 'Path to the input XML file')
+            ->addArgument('outfile', InputArgument::OPTIONAL, 'Path to the output XML file (defaults to infile)')
+        ;
+    }
+
+    protected function execute(InputInterface $input, OutputInterface $output): int
+    {
+        $io = new SymfonyStyle($input, $output);
+
+        $inputFile = (string) $input->getArgument('infile');
+        $outputFile = $input->getArgument('outfile');
+
+        if (false === is_file($inputFile)) {
+            $io->error(sprintf('Input file not found: %s', $inputFile));
+
+            return Command::FAILURE;
+        }
+
+        if (false === is_readable($inputFile)) {
+            $io->error(sprintf('Input file is not readable: %s', $inputFile));
+
+            return Command::FAILURE;
+        }
+
+        $targetFile = null === $outputFile ? $inputFile : (string) $outputFile;
+        $this->xmlAnonymizer->resetState();
+
+        $processingResult = $this->anonymizeFile($io, $inputFile, $targetFile, false);
+        if (Command::SUCCESS !== $processingResult) {
+            return $processingResult;
+        }
+
+        $pairedInputFile = $this->derivePairedFilename($inputFile);
+        if (null !== $pairedInputFile && true === is_file($pairedInputFile)) {
+            if (false === is_readable($pairedInputFile)) {
+                $io->error(sprintf('Paired file is not readable: %s', $pairedInputFile));
+
+                return Command::FAILURE;
+            }
+
+            $pairedTargetFile = $this->derivePairedOutputFilename($targetFile, $inputFile, $pairedInputFile);
+            if (null === $pairedTargetFile) {
+                $io->error(sprintf(
+                    'Unable to derive paired output filename from: %s',
+                    $targetFile
+                ));
+
+                return Command::FAILURE;
+            }
+
+            $processingResult = $this->anonymizeFile($io, $pairedInputFile, $pairedTargetFile, true);
+            if (Command::SUCCESS !== $processingResult) {
+                return $processingResult;
+            }
+
+            $io->note(sprintf('Paired file anonymized: %s -> %s', $pairedInputFile, $pairedTargetFile));
+        }
+
+        $io->success(sprintf(
+            'Anonymized %d persons from %s to %s',
+            $this->xmlAnonymizer->getLastIdentityCount(),
+            $inputFile,
+            $targetFile,
+        ));
+
+        return Command::SUCCESS;
+    }
+
+    private function anonymizeFile(SymfonyStyle $io, string $inputFile, string $targetFile, bool $keepState): int
+    {
+        $xml = file_get_contents($inputFile);
+        if (false === $xml) {
+            $io->error(sprintf('Unable to read input file: %s', $inputFile));
+
+            return Command::FAILURE;
+        }
+
+        try {
+            $anonymizedXml = $this->xmlAnonymizer->anonymize($xml, false === $keepState);
+        } catch (\InvalidArgumentException|\RuntimeException $exception) {
+            $io->error(sprintf('Failed to anonymize %s: %s', $inputFile, $exception->getMessage()));
+
+            return Command::FAILURE;
+        }
+
+        $writeResult = file_put_contents($targetFile, $anonymizedXml);
+        if (false === $writeResult) {
+            $io->error(sprintf('Unable to write output file: %s', $targetFile));
+
+            return Command::FAILURE;
+        }
+
+        return Command::SUCCESS;
+    }
+
+    private function derivePairedFilename(string $filename): ?string
+    {
+        if (str_ends_with($filename, '_request.xml')) {
+            return substr($filename, 0, -12).'_response.xml';
+        }
+
+        if (str_ends_with($filename, '_response.xml')) {
+            return substr($filename, 0, -13).'_request.xml';
+        }
+
+        return null;
+    }
+
+    private function derivePairedOutputFilename(string $outputFile, string $inputFile, string $pairedInputFile): ?string
+    {
+        if (str_ends_with($inputFile, '_request.xml') && str_ends_with($pairedInputFile, '_response.xml')) {
+            if (false === str_ends_with($outputFile, '_request.xml')) {
+                return null;
+            }
+
+            return substr($outputFile, 0, -12).'_response.xml';
+        }
+
+        if (str_ends_with($inputFile, '_response.xml') && str_ends_with($pairedInputFile, '_request.xml')) {
+            if (false === str_ends_with($outputFile, '_response.xml')) {
+                return null;
+            }
+
+            return substr($outputFile, 0, -13).'_request.xml';
+        }
+
+        return null;
+    }
+}
@@ -0,0 +1,301 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Service;
+
+use Faker\Factory;
+use Faker\Generator;
+
+class BpnXmlAnonymizer
+{
+    private const SAFE_EMAIL_DOMAIN = 'example.com';
+
+    private Generator $faker;
+
+    /**
+     * @var array<string, array{firstName: string, lastName: string, street: string, postalCode: string, city: string, email: string, mobilePhone: string}>
+     */
+    private array $identities = [];
+
+    private int $fallbackIdentityCounter = 0;
+
+    public function __construct()
+    {
+        $this->faker = Factory::create('de_DE');
+    }
+
+    public function anonymize(string $xml, bool $resetState = true): string
+    {
+        if (true === $resetState) {
+            $this->resetState();
+        }
+
+        $document = new \DOMDocument();
+        $document->preserveWhiteSpace = true;
+        $document->formatOutput = false;
+
+        if (false === @$document->loadXML($xml)) {
+            throw new \InvalidArgumentException('Invalid XML input.');
+        }
+
+        $xpath = new \DOMXPath($document);
+        $personNodes = $xpath->query('//anmelder | //teilnehmerliste/teilnehmer | //kundennamen/kunde');
+
+        if (false === $personNodes) {
+            throw new \RuntimeException('Unable to query person nodes.');
+        }
+
+        foreach ($personNodes as $index => $personNode) {
+            if (false === $personNode instanceof \DOMElement) {
+                continue;
+            }
+
+            $personId = $this->resolvePersonId($personNode, (int) $index);
+            $identity = $this->getOrCreateIdentity($personId);
+
+            $this->replaceNameFields($personNode, $identity);
+            $this->replaceEmailFields($personNode, $identity);
+            $this->replaceAddressFields($personNode, $identity);
+        }
+
+        $result = $document->saveXML();
+        if (false === $result) {
+            throw new \RuntimeException('Unable to serialize anonymized XML.');
+        }
+
+        return $result;
+    }
+
+    public function getLastIdentityCount(): int
+    {
+        return \count($this->identities);
+    }
+
+    public function resetState(): void
+    {
+        $this->identities = [];
+        $this->fallbackIdentityCounter = 0;
+    }
+
+    private function resolvePersonId(\DOMElement $personNode, int $nodeIndex): string
+    {
+        foreach (['personid', 'idperson', 'idadresseperson'] as $tagName) {
+            $child = $this->getDirectChild($personNode, $tagName);
+            if (null !== $child) {
+                $value = trim($child->textContent);
+                if ('' !== $value) {
+                    return $value;
+                }
+            }
+        }
+
+        $fingerprintKey = $this->buildFingerprintKey($personNode);
+        if (null !== $fingerprintKey) {
+            return $fingerprintKey;
+        }
+
+        if (true === $personNode->hasAttribute('id')) {
+            $attributeValue = trim((string) $personNode->getAttribute('id'));
+            if ('' !== $attributeValue) {
+                return sprintf('attr-id:%s', $attributeValue);
+            }
+        }
+
+        ++$this->fallbackIdentityCounter;
+
+        return sprintf('fallback:%d:%d', $nodeIndex, $this->fallbackIdentityCounter);
+    }
+
+    private function buildFingerprintKey(\DOMElement $personNode): ?string
+    {
+        $nameValues = [];
+        $nameNode = $this->getDirectChild($personNode, 'name');
+        if (null !== $nameNode) {
+            $nameValues[] = trim($nameNode->textContent);
+        }
+        if (true === $personNode->hasAttribute('name')) {
+            $nameValues[] = trim((string) $personNode->getAttribute('name'));
+        }
+
+        $firstNameNode = $this->getDirectChild($personNode, 'vorname');
+        if (null !== $firstNameNode) {
+            $nameValues[] = trim($firstNameNode->textContent);
+        }
+        if (true === $personNode->hasAttribute('vorname')) {
+            $nameValues[] = trim((string) $personNode->getAttribute('vorname'));
+        }
+
+        $communicationNode = $this->getDirectChild($personNode, 'kommunikation');
+        if (null !== $communicationNode) {
+            $emailNode = $this->getDirectChild($communicationNode, 'email');
+            if (null !== $emailNode) {
+                $nameValues[] = trim($emailNode->textContent);
+            }
+        }
+
+        $addressNode = $this->getDirectChild($personNode, 'anschrift');
+        if (null !== $addressNode) {
+            foreach (['strasse', 'plz', 'ort'] as $fieldName) {
+                $fieldNode = $this->getDirectChild($addressNode, $fieldName);
+                if (null !== $fieldNode) {
+                    $nameValues[] = trim($fieldNode->textContent);
+                }
+            }
+        }
+
+        $normalizedParts = [];
+        foreach ($nameValues as $value) {
+            if ('' === $value) {
+                continue;
+            }
+
+            $normalizedParts[] = mb_strtolower($value);
+        }
+
+        if (0 === \count($normalizedParts)) {
+            return null;
+        }
+
+        return sprintf('fp:%s', md5(implode('|', $normalizedParts)));
+    }
+
+    /**
+     * @param string $personId
+     *
+     * @return array{firstName: string, lastName: string, street: string, postalCode: string, city: string, email: string, mobilePhone: string}
+     */
+    private function getOrCreateIdentity(string $personId): array
+    {
+        if (true === isset($this->identities[$personId])) {
+            return $this->identities[$personId];
+        }
+
+        $firstName = $this->faker->firstName();
+        $lastName = $this->faker->lastName();
+
+        $street = sprintf('%s %s', $this->faker->streetName(), $this->faker->buildingNumber());
+        $postalCode = $this->faker->postcode();
+        $city = $this->faker->city();
+        $mobilePhone = $this->faker->numerify('01#########');
+
+        $emailLocal = sprintf('%s.%s', $this->slug($firstName), $this->slug($lastName));
+        $email = sprintf('%s@%s', $emailLocal, self::SAFE_EMAIL_DOMAIN);
+
+        $this->identities[$personId] = [
+            'firstName' => $firstName,
+            'lastName' => $lastName,
+            'street' => $street,
+            'postalCode' => $postalCode,
+            'city' => $city,
+            'email' => strtolower($email),
+            'mobilePhone' => $mobilePhone,
+        ];
+
+        return $this->identities[$personId];
+    }
+
+    /**
+     * @param array{firstName: string, lastName: string, street: string, postalCode: string, city: string, email: string, mobilePhone: string} $identity
+     */
+    private function replaceNameFields(\DOMElement $personNode, array $identity): void
+    {
+        $nameNode = $this->getDirectChild($personNode, 'name');
+        if (null !== $nameNode) {
+            $nameNode->nodeValue = $identity['lastName'];
+        }
+
+        $firstNameNode = $this->getDirectChild($personNode, 'vorname');
+        if (null !== $firstNameNode) {
+            $firstNameNode->nodeValue = $identity['firstName'];
+        }
+
+        if (true === $personNode->hasAttribute('name')) {
+            $personNode->setAttribute('name', $identity['lastName']);
+        }
+
+        if (true === $personNode->hasAttribute('vorname')) {
+            $personNode->setAttribute('vorname', $identity['firstName']);
+        }
+    }
+
+    /**
+     * @param array{firstName: string, lastName: string, street: string, postalCode: string, city: string, email: string, mobilePhone: string} $identity
+     */
+    private function replaceEmailFields(\DOMElement $personNode, array $identity): void
+    {
+        $communicationNode = $this->getDirectChild($personNode, 'kommunikation');
+        if (null === $communicationNode) {
+            return;
+        }
+
+        $emailNode = $this->getDirectChild($communicationNode, 'email');
+        if (null !== $emailNode) {
+            $emailNode->nodeValue = $identity['email'];
+        }
+
+        $mobileNode = $this->getDirectChild($communicationNode, 'telefonmobil');
+        if (null !== $mobileNode) {
+            $mobileNode->nodeValue = $identity['mobilePhone'];
+        }
+    }
+
+    /**
+     * @param array{firstName: string, lastName: string, street: string, postalCode: string, city: string, email: string, mobilePhone: string} $identity
+     */
+    private function replaceAddressFields(\DOMElement $personNode, array $identity): void
+    {
+        $addressNode = $this->getDirectChild($personNode, 'anschrift');
+        if (null === $addressNode) {
+            return;
+        }
+
+        $streetNode = $this->getDirectChild($addressNode, 'strasse');
+        if (null !== $streetNode) {
+            $streetNode->nodeValue = $identity['street'];
+        }
+
+        $postalCodeNode = $this->getDirectChild($addressNode, 'plz');
+        if (null !== $postalCodeNode) {
+            $postalCodeNode->nodeValue = $identity['postalCode'];
+        }
+
+        $cityNode = $this->getDirectChild($addressNode, 'ort');
+        if (null !== $cityNode) {
+            $cityNode->nodeValue = $identity['city'];
+        }
+    }
+
+    private function getDirectChild(\DOMElement $parent, string $name): ?\DOMElement
+    {
+        foreach ($parent->childNodes as $childNode) {
+            if (false === $childNode instanceof \DOMElement) {
+                continue;
+            }
+
+            if ($childNode->tagName === $name) {
+                return $childNode;
+            }
+        }
+
+        return null;
+    }
+
+    private function slug(string $value): string
+    {
+        $normalizedValue = iconv('UTF-8', 'ASCII//TRANSLIT//IGNORE', $value);
+        if (false === $normalizedValue) {
+            $normalizedValue = $value;
+        }
+
+        $normalizedValue = strtolower($normalizedValue);
+        $normalizedValue = preg_replace('/[^a-z0-9]+/', '.', $normalizedValue);
+        $normalizedValue = trim((string) $normalizedValue, '.');
+
+        if ('' === $normalizedValue) {
+            return 'user';
+        }
+
+        return $normalizedValue;
+    }
+}