EP/myep
4
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: full anonymization of buspro xml dumps

Browse Source
This commit is contained in:
Björn Fromme
2026-06-21 16:49:54 +02:00
parent d77f92b6f0
commit 52ee6b26b1
6 changed files with 922 additions and 85 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/Command/BpnXmlAnonymizeCommand.php
+129 -9
View File
@@ -27,8 +27,8 @@ class BpnXmlAnonymizeCommand extends Command
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)')
->addArgument('infile', InputArgument::REQUIRED, 'Path to the input XML file or folder')
->addArgument('outfile', InputArgument::OPTIONAL, 'Path to the output XML file or folder (defaults to infile)')
;
}
@@ -39,19 +39,31 @@ class BpnXmlAnonymizeCommand extends Command
$inputFile = (string) $input->getArgument('infile');
$outputFile = $input->getArgument('outfile');
if (true === is_dir($inputFile)) {
$targetDirectory = null === $outputFile ? $inputFile : (string) $outputFile;
return $this->anonymizeDirectory($io, $inputFile, $targetDirectory);
}
if (false === is_file($inputFile)) {
$io->error(sprintf('Input file not found: %s', $inputFile));
return Command::FAILURE;
}
$targetFile = null === $outputFile ? $inputFile : (string) $outputFile;
return $this->anonymizeSingleFile($io, $inputFile, $targetFile);
}
private function anonymizeSingleFile(SymfonyStyle $io, string $inputFile, string $targetFile): int
{
if (false === is_readable($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);
@@ -85,12 +97,120 @@ class BpnXmlAnonymizeCommand extends Command
$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,
));
$identityCount = $this->xmlAnonymizer->getLastIdentityCount();
if (0 === $identityCount) {
$io->warning(sprintf('No anonymizable person data found in %s', $inputFile));
} else {
$io->success(sprintf(
'Anonymized %d persons from %s to %s',
$identityCount,
$inputFile,
$targetFile,
));
}
return Command::SUCCESS;
}
private function anonymizeDirectory(SymfonyStyle $io, string $inputDirectory, string $targetDirectory): int
{
if (false === is_readable($inputDirectory)) {
$io->error(sprintf('Input directory is not readable: %s', $inputDirectory));
return Command::FAILURE;
}
if (true === file_exists($targetDirectory) && false === is_dir($targetDirectory)) {
$io->error(sprintf('Output path is not a directory: %s', $targetDirectory));
return Command::FAILURE;
}
if (false === file_exists($targetDirectory) && false === mkdir($targetDirectory, 0777, true) && false === is_dir($targetDirectory)) {
$io->error(sprintf('Unable to create output directory: %s', $targetDirectory));
return Command::FAILURE;
}
$entries = scandir($inputDirectory);
if (false === $entries) {
$io->error(sprintf('Unable to list input directory: %s', $inputDirectory));
return Command::FAILURE;
}
$processed = [];
$fileCount = 0;
$personCount = 0;
foreach ($entries as $entry) {
if ('.' === $entry || '..' === $entry) {
continue;
}
$inputPath = $inputDirectory.DIRECTORY_SEPARATOR.$entry;
if (false === is_file($inputPath) || false === str_ends_with($entry, '.xml')) {
continue;
}
if (true === isset($processed[$inputPath])) {
continue;
}
$pairedInputPath = $this->derivePairedFilename($inputPath);
if (null !== $pairedInputPath && true === is_file($pairedInputPath) && false === isset($processed[$pairedInputPath])) {
$this->xmlAnonymizer->resetState();
$targetPath = $targetDirectory.DIRECTORY_SEPARATOR.$entry;
$result = $this->anonymizeFile($io, $inputPath, $targetPath, false);
if (Command::SUCCESS !== $result) {
return $result;
}
$processed[$inputPath] = true;
++$fileCount;
$pairedEntry = basename($pairedInputPath);
$pairedTargetPath = $targetDirectory.DIRECTORY_SEPARATOR.$pairedEntry;
$result = $this->anonymizeFile($io, $pairedInputPath, $pairedTargetPath, true);
if (Command::SUCCESS !== $result) {
return $result;
}
$processed[$pairedInputPath] = true;
++$fileCount;
$personCount += $this->xmlAnonymizer->getLastIdentityCount();
$io->note(sprintf('Paired file anonymized: %s -> %s', $pairedInputPath, $pairedTargetPath));
continue;
}
$this->xmlAnonymizer->resetState();
$targetPath = $targetDirectory.DIRECTORY_SEPARATOR.$entry;
$result = $this->anonymizeFile($io, $inputPath, $targetPath, false);
if (Command::SUCCESS !== $result) {
return $result;
}
$processed[$inputPath] = true;
++$fileCount;
$personCount += $this->xmlAnonymizer->getLastIdentityCount();
}
$io->text(sprintf('Processed %d XML files', $fileCount));
if (0 === $personCount) {
$io->warning(sprintf('No anonymizable person data found in %s', $inputDirectory));
} else {
$io->success(sprintf(
'Anonymized %d persons from %s to %s',
$personCount,
$inputDirectory,
$targetDirectory,
));
}
return Command::SUCCESS;
}
src/Service/BpnXmlAnonymizer.php
+293 -12
View File
@@ -10,11 +10,41 @@ use Faker\Generator;
class BpnXmlAnonymizer
{
private const SAFE_EMAIL_DOMAIN = 'example.com';
private const SALUTATIONS = ['Herr', 'Frau', 'Divers'];
private const GENDERS = ['M', 'W', 'D'];
private const COUNTRY_CODES = ['AT', 'BE', 'CH', 'DE', 'DK', 'ES', 'FR', 'IT', 'NL', 'NO', 'PL', 'SE'];
private Generator $faker;
/**
* @var array<string, array{firstName: string, lastName: string, street: string, postalCode: string, city: string, email: string, mobilePhone: string}>
* @var array<string, array{
* firstName: string,
* lastName: string,
* salutation: string,
* title: string,
* gender: string,
* nationality: string,
* birthDate: string,
* street: string,
* postalCode: string,
* city: string,
* district: string,
* country: string,
* email: string,
* mobilePhone: string,
* phone: string,
* newsletter: string,
* iban: string,
* userName: string,
* requestKey: string,
* requestPassword: string,
* personId: string,
* addressId: string,
* customerId: string,
* height: string,
* weight: string,
* shoeSize: string
* }>
*/
private array $identities = [];
@@ -40,7 +70,7 @@ class BpnXmlAnonymizer
}
$xpath = new \DOMXPath($document);
$personNodes = $xpath->query('//anmelder | //teilnehmerliste/teilnehmer | //kundennamen/kunde');
$personNodes = $xpath->query('//anfrage | //anmelder | //teilnehmerliste/teilnehmer | //kundennamen/kunde | //adressdaten');
if (false === $personNodes) {
throw new \RuntimeException('Unable to query person nodes.');
@@ -54,9 +84,17 @@ class BpnXmlAnonymizer
$personId = $this->resolvePersonId($personNode, (int) $index);
$identity = $this->getOrCreateIdentity($personId);
if ('anfrage' === $personNode->tagName) {
$this->replaceRequestFields($personNode, $identity);
continue;
}
$this->replaceNameFields($personNode, $identity);
$this->replaceEmailFields($personNode, $identity);
$this->replaceAddressFields($personNode, $identity);
$this->replaceDemographicFields($personNode, $identity);
$this->replaceFinancialFields($personNode, $identity);
$this->replaceMetadataFields($personNode, $identity);
}
$result = $document->saveXML();
@@ -80,14 +118,47 @@ class BpnXmlAnonymizer
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;
if ('anfrage' === $personNode->tagName || 'adressdaten' === $personNode->tagName) {
$emailNode = $this->getDirectChild($personNode, 'email');
if (null !== $emailNode) {
$emailValue = trim($emailNode->textContent);
if ('' !== $emailValue) {
return sprintf('email:%s', mb_strtolower($emailValue));
}
}
if ('adressdaten' === $personNode->tagName) {
$communicationNode = $this->getDirectChild($personNode, 'kommunikation');
if (null !== $communicationNode) {
$nestedEmailNode = $this->getDirectChild($communicationNode, 'email');
if (null !== $nestedEmailNode) {
$emailValue = trim($nestedEmailNode->textContent);
if ('' !== $emailValue) {
return sprintf('email:%s', mb_strtolower($emailValue));
}
}
}
}
}
$currentNode = $personNode;
while ($currentNode instanceof \DOMElement) {
foreach (['personid', 'idperson', 'idadresseperson'] as $tagName) {
$child = $this->getDirectChild($currentNode, $tagName);
if (null !== $child) {
$value = trim($child->textContent);
if ('' !== $value) {
return $value;
}
}
}
$parentNode = $currentNode->parentNode;
if (false === $parentNode instanceof \DOMElement) {
break;
}
$currentNode = $parentNode;
}
$fingerprintKey = $this->buildFingerprintKey($personNode);
@@ -161,7 +232,34 @@ class BpnXmlAnonymizer
}
/**
* @return array{firstName: string, lastName: string, street: string, postalCode: string, city: string, email: string, mobilePhone: string}
* @return array{
* firstName: string,
* lastName: string,
* salutation: string,
* title: string,
* gender: string,
* nationality: string,
* birthDate: string,
* street: string,
* postalCode: string,
* city: string,
* district: string,
* country: string,
* email: string,
* mobilePhone: string,
* phone: string,
* newsletter: string,
* iban: string,
* userName: string,
* requestKey: string,
* requestPassword: string,
* personId: string,
* addressId: string,
* customerId: string,
* height: string,
* weight: string,
* shoeSize: string
* }
*/
private function getOrCreateIdentity(string $personId): array
{
@@ -169,25 +267,63 @@ class BpnXmlAnonymizer
return $this->identities[$personId];
}
$identityIndex = \count($this->identities) + 1;
$firstName = $this->faker->firstName();
$lastName = $this->faker->lastName();
$salutation = 'Divers';
$title = '';
$gender = 'D';
$nationality = 'XX';
$birthDate = '01.01.1980';
$street = sprintf('%s %s', $this->faker->streetName(), $this->faker->buildingNumber());
$postalCode = $this->faker->postcode();
$city = $this->faker->city();
$mobilePhone = $this->faker->numerify('01#########');
$district = '';
$country = 'XX';
$mobilePhone = sprintf('01%09d', $identityIndex);
$phone = sprintf('02%09d', $identityIndex);
$newsletter = 'False';
$iban = sprintf('DE%020d', $identityIndex);
$userName = sprintf('anon-user-%d', $identityIndex);
$requestKey = str_pad((string) $identityIndex, 32, '0', STR_PAD_LEFT);
$requestPassword = str_pad((string) ($identityIndex + 1), 32, '0', STR_PAD_LEFT);
$personIdValue = (string) (700000 + $identityIndex);
$addressIdValue = (string) (800000 + $identityIndex);
$customerIdValue = (string) (900000 + $identityIndex);
$height = '170';
$weight = '70';
$shoeSize = '42';
$emailLocal = sprintf('%s.%s', $this->slug($firstName), $this->slug($lastName));
$email = sprintf('%s@%s', $emailLocal, self::SAFE_EMAIL_DOMAIN);
$email = sprintf('%s@%s', $userName, self::SAFE_EMAIL_DOMAIN);
$this->identities[$personId] = [
'firstName' => $firstName,
'lastName' => $lastName,
'salutation' => $salutation,
'title' => $title,
'gender' => $gender,
'nationality' => $nationality,
'birthDate' => $birthDate,
'street' => $street,
'postalCode' => $postalCode,
'city' => $city,
'district' => $district,
'country' => $country,
'email' => strtolower($email),
'mobilePhone' => $mobilePhone,
'phone' => $phone,
'newsletter' => $newsletter,
'iban' => $iban,
'userName' => $userName,
'requestKey' => $requestKey,
'requestPassword' => $requestPassword,
'personId' => $personIdValue,
'addressId' => $addressIdValue,
'customerId' => $customerIdValue,
'height' => $height,
'weight' => $weight,
'shoeSize' => $shoeSize,
];
return $this->identities[$personId];
@@ -215,6 +351,10 @@ class BpnXmlAnonymizer
if (true === $personNode->hasAttribute('vorname')) {
$personNode->setAttribute('vorname', $identity['firstName']);
}
if (true === $personNode->hasAttribute('id') && 'kunde' === $personNode->tagName) {
$personNode->setAttribute('id', $identity['customerId']);
}
}
/**
@@ -224,6 +364,13 @@ class BpnXmlAnonymizer
{
$communicationNode = $this->getDirectChild($personNode, 'kommunikation');
if (null === $communicationNode) {
if ('anfrage' === $personNode->tagName) {
$emailNode = $this->getDirectChild($personNode, 'email');
if (null !== $emailNode) {
$emailNode->nodeValue = $identity['email'];
}
}
return;
}
@@ -236,6 +383,16 @@ class BpnXmlAnonymizer
if (null !== $mobileNode) {
$mobileNode->nodeValue = $identity['mobilePhone'];
}
$phoneNode = $this->getDirectChild($communicationNode, 'telefonprivat');
if (null !== $phoneNode) {
$phoneNode->nodeValue = $identity['phone'];
}
$newsletterNode = $this->getDirectChild($communicationNode, 'newsletter');
if (null !== $newsletterNode) {
$newsletterNode->nodeValue = $identity['newsletter'];
}
}
/**
@@ -262,6 +419,130 @@ class BpnXmlAnonymizer
if (null !== $cityNode) {
$cityNode->nodeValue = $identity['city'];
}
$districtNode = $this->getDirectChild($addressNode, 'ortsteil');
if (null !== $districtNode) {
$districtNode->nodeValue = $identity['district'];
}
$countryNode = $this->getDirectChild($addressNode, 'land');
if (null !== $countryNode) {
$countryNode->nodeValue = $identity['country'];
}
$addressIdNode = $this->getDirectChild($addressNode, 'id');
if (null !== $addressIdNode) {
$addressIdNode->nodeValue = $identity['addressId'];
}
}
private function replaceDemographicFields(\DOMElement $personNode, array $identity): void
{
$salutationNode = $this->getDirectChild($personNode, 'anrede');
if (null !== $salutationNode) {
$salutationNode->nodeValue = $identity['salutation'];
}
$titleNode = $this->getDirectChild($personNode, 'titel');
if (null !== $titleNode) {
$titleNode->nodeValue = $identity['title'];
}
$genderNode = $this->getDirectChild($personNode, 'geschlecht');
if (null !== $genderNode) {
$genderNode->nodeValue = $identity['gender'];
}
$nationalityNode = $this->getDirectChild($personNode, 'nationalitaet');
if (null !== $nationalityNode) {
$nationalityNode->nodeValue = $identity['nationality'];
}
$birthDateNode = $this->getDirectChild($personNode, 'geburtsdatum');
if (null !== $birthDateNode) {
$birthDateNode->nodeValue = $identity['birthDate'];
}
$heightNode = $this->getDirectChild($personNode, 'sonstiges1');
if (null !== $heightNode) {
$heightNode->nodeValue = $identity['height'];
}
$weightNode = $this->getDirectChild($personNode, 'sonstiges2');
if (null !== $weightNode) {
$weightNode->nodeValue = $identity['weight'];
}
$shoeSizeNode = $this->getDirectChild($personNode, 'sonstiges3');
if (null !== $shoeSizeNode) {
$shoeSizeNode->nodeValue = $identity['shoeSize'];
}
}
private function replaceFinancialFields(\DOMElement $personNode, array $identity): void
{
$bankNode = $this->getDirectChild($personNode, 'bankverbindung');
if (null === $bankNode) {
return;
}
$ibanNode = $this->getDirectChild($bankNode, 'iban');
if (null !== $ibanNode) {
$ibanNode->nodeValue = $identity['iban'];
}
}
private function replaceMetadataFields(\DOMElement $personNode, array $identity): void
{
if (true === $personNode->hasAttribute('id')) {
if ('kunde' === $personNode->tagName) {
$personNode->setAttribute('id', $identity['customerId']);
} elseif ('teilnehmer' === $personNode->tagName || 'anmelder' === $personNode->tagName) {
$personNode->setAttribute('id', $identity['personId']);
}
}
}
private function replaceRequestFields(\DOMElement $personNode, array $identity): void
{
$userNode = $this->getDirectChild($personNode, 'user');
if (null !== $userNode) {
$userNode->nodeValue = $identity['userName'];
}
$keyNode = $this->getDirectChild($personNode, 'key');
if (null !== $keyNode) {
$keyNode->nodeValue = $identity['requestKey'];
}
$emailNode = $this->getDirectChild($personNode, 'email');
if (null !== $emailNode) {
$emailNode->nodeValue = $identity['email'];
}
$passwordNode = $this->getDirectChild($personNode, 'passwort');
if (null !== $passwordNode) {
$passwordNode->nodeValue = $identity['requestPassword'];
}
}
private function findAncestorDirectChild(\DOMElement $personNode, string $ancestorTagName, string $childTagName): ?\DOMElement
{
$currentNode = $personNode;
while ($currentNode instanceof \DOMElement) {
if ($ancestorTagName === $currentNode->tagName) {
return $this->getDirectChild($currentNode, $childTagName);
}
$parentNode = $currentNode->parentNode;
if (false === $parentNode instanceof \DOMElement) {
break;
}
$currentNode = $parentNode;
}
return null;
}
private function getDirectChild(\DOMElement $parent, string $name): ?\DOMElement