PHP CLI script for removing specific XMLNodes from WPML Xliff files.

cleanup-xliffs.php

<?php
/**
 * Cleanup Xliffs
 * 
 * This file can be called through the php cli and clean up xliff files.
 * 
 * The script takes directories and/or xliff files and removes XML nodes which 
 * matches the XPath query.
 * 
 * @author Hampus Nordin <[email protected]>
 * @version 0.1
 */

/**
 * undocumented class
 *
 * @package hampusn
 * @author Hampus Nordin <[email protected]>
 **/
class Hampusn_Cleanup_Xliffs
{
  /**
   * Contains all arguments passed through the php cli.
   * This is a list of all directories and/or files which 
   * should be parsed and cleaned.
   *
   * @var array
   **/
  private $arguments = array();

  /**
   * A list of all options passed through the cli.
   * Can be used to override the default behaviour.
   *
   * @var array
   **/
  private $options = array(
    // Pattern to match files with in glob()
    'p' => '*.xliff',
    // Namespace to use with XPath query.
    'n' => 'x',
    // XPath query. The matches will be removed.
    'q' => '//x:trans-unit[starts-with(@resname,"field-_") and not(contains(@resname, "yoast"))]',
    // Verbose output
    'v' => 1,
  );

  /**
   * Will hold all files which should be parsed and cleaned.
   * The array is filled through Hampusn_Cleanup_Xliffs::parseArguments().
   *
   * @var array
   **/
  private $files = array();

  /**
   * The constructor.
   *
   * @return void
   * @author Hampus Nordin <[email protected]>
   **/
  public function __construct($arguments, $options = array())
  {
    $this->arguments = $arguments;
    $this->options = array_merge($this->options, $options);
  }

  /**
   * Execute the parsing and cleanup.
   *
   * @return void
   * @author Hampus Nordin <[email protected]>
   **/
  public function run()
  {
    $this->parseArguments();
    $this->removeDuplicates();

    $this->walkFiles(array($this, 'cleanFile'));
  }

  /**
   * Walk through all arguments and add all xliff files found.
   * The files needs to be readable.
   * 
   * TODO: consider refactoring this function and also support recursive walk.
   *
   * @return void
   * @author Hampus Nordin <[email protected]>
   **/
  private function parseArguments()
  {
    foreach ($this->arguments as $arg) {
      // Try to get the real path of the passed argument.
      // This allows the user to pass relative directories/files such as "./file.xliff"
      $realPath = realpath($arg);
      // Bail early if path cannot be resolved.
      if (!$realPath) {
        continue;
      }
      // Make sure the path is readable
      if (is_readable($realPath)) {
        // If it's a file add it directly to the files array and bail early.
        if (is_file($realPath)) {
          $this->files[] = $realPath;
          continue;
        }
        // Control directory.
        if (is_dir($realPath)) {
          // Add directory separator to the end of path if needed.
          $realPath = rtrim($realPath, DIRECTORY_SEPARATOR) . DIRECTORY_SEPARATOR;
          // Find files in directory.
          $matches = glob($realPath . $this->options['p']);
          // Loop through matches and make sure they are readable files.
          foreach ($matches as $match) {
            if (is_file($match) && is_readable($match)) {
              $this->files[] = $match;
            }
          }
        }
      } else {
        $this->log("Path not readable: $realPath", 0);
      }
    }
  }

  /**
   * Remove any duplicates from the files array.
   *
   * @return void
   * @author Hampus Nordin <[email protected]>
   **/
  private function removeDuplicates()
  {
    $this->files = array_unique($this->files);
  }


  /**
   * Walk through all files with passed function.
   * The passed function will get one parameter which is the real file path.
   *
   * @return void
   * @author Hampus Nordin <[email protected]>
   **/
  private function walkFiles($func)
  {
    foreach ($this->files as $file) {
      call_user_func($func, $file);      
    }
  }

  /**
   * Cleans the file by removing superfluous XML nodes.
   *
   * @return void
   * @author Hampus Nordin <[email protected]>
   **/
  private function cleanFile($file)
  {
    // Bail early if file is not writable.
    if (!is_writable($file)) {
      $this->log("File not writable: $file", 0);
      return;
    }
    // Create and load file as XML.
    $document = new DOMDocument;
    $document->load($file);

    $xpath = new DOMXpath($document);
    $xpath->registerNamespace($this->options['n'], $document->documentElement->namespaceURI);
    $query = $xpath->query($this->options['q']);

    if ($query->length) {
      $numRemoved = 0;

      foreach ($query as $node) {
        $node->parentNode->removeChild($node);
        $numRemoved++;
      }

      $this->log("Removed $numRemoved XMLNodes from file: $file", 1);
    } else {
      $this->log("Empty query for file: $file", 1);
    }


    $document->save($file);
  }

  /**
   * Echo to cli console.
   * 
   * Levels:
   * 0 Errors - Messages printed when something goes wrong.
   * 1 Verbose - Messages printed which might help debugging.
   *
   * @return void
   * @author Hampus Nordin <[email protected]>
   **/
  private function log($message, $level = 0)
  {
    if ($level >= $this->options['v']) {
      echo $message . PHP_EOL;
    }
  }
} // END class Hampusn_Cleanup_Xliffs

// $argv should only be populated if the script is run through the cli.
// The first value in $argv is the file being executed, so make sure 
// the arguments list is longer than 1.
if (!empty($argv) && count($argv) > 1) {
  // Remove the first value/the current file.
  array_shift($argv);
  // Get options.
  $options = getopt('p:n:q:v:');
  // Init
  $c = new Hampusn_Cleanup_Xliffs($argv, $options);
  // Execute
  $c->run();
}