feat: full anonymization of buspro xml dumps

src/Command/BpnXmlAnonymizeCommand.php

@@ -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;
     }