You are here

class TikaExtractor in Search API attachments 8

Same name and namespace in other branches
  1. 9.0.x src/Plugin/search_api_attachments/TikaExtractor.php \Drupal\search_api_attachments\Plugin\search_api_attachments\TikaExtractor

Provides tika extractor.

Plugin annotation


@SearchApiAttachmentsTextExtractor(
  id = "tika_extractor",
  label = @Translation("Tika Extractor"),
  description = @Translation("Adds Tika extractor support."),
)

Hierarchy

Expanded class hierarchy of TikaExtractor

File

src/Plugin/search_api_attachments/TikaExtractor.php, line 18

Namespace

Drupal\search_api_attachments\Plugin\search_api_attachments
View source
class TikaExtractor extends TextExtractorPluginBase {

  /**
   * Extract file with Tika library.
   *
   * @param \Drupal\file\Entity\File $file
   *   A file object.
   *
   * @return string
   *   The text extracted from the file.
   */
  public function extract(File $file) {
    $output = '';
    $filepath = $this
      ->getRealpath($file
      ->getFileUri());
    $tika = realpath($this->configuration['tika_path']);
    $java = $this->configuration['java_path'];

    // UTF-8 multibyte characters will be stripped by escapeshellargs() for the
    // default C-locale.
    // So temporarily set the locale to UTF-8 so that the filepath remains
    // valid.
    $backup_locale = setlocale(LC_CTYPE, '0');
    setlocale(LC_CTYPE, 'en_US.UTF-8');
    $param = '';
    if ($file
      ->getMimeType() != 'audio/mpeg') {
      $param = ' -Dfile.encoding=UTF8 -cp ' . escapeshellarg($tika);
    }

    // Force running the Tika jar headless.
    $param = ' -Djava.awt.headless=true ' . $param;
    $cmd = $java . $param . ' -jar ' . escapeshellarg($tika) . ' -t ' . escapeshellarg($filepath);
    if (strpos(ini_get('extension_dir'), 'MAMP/')) {
      $cmd = 'export DYLD_LIBRARY_PATH=""; ' . $cmd;
    }

    // Restore the locale.
    setlocale(LC_CTYPE, $backup_locale);

    // Support UTF-8 commands:
    // @see http://www.php.net/manual/en/function.shell-exec.php#85095
    shell_exec("LANG=en_US.utf-8");
    $output = shell_exec($cmd);
    if (is_null($output)) {
      throw new \Exception('Tika Exctractor is not available.');
    }
    return $output;
  }

  /**
   * {@inheritdoc}
   */
  public function buildConfigurationForm(array $form, FormStateInterface $form_state) {
    $form['java_path'] = [
      '#type' => 'textfield',
      '#title' => $this
        ->t('Path to java executable'),
      '#description' => $this
        ->t('Enter the path to java executable. Example: "java".'),
      '#default_value' => $this->configuration['java_path'],
      '#required' => TRUE,
    ];
    $form['tika_path'] = [
      '#type' => 'textfield',
      '#title' => $this
        ->t('Path to Tika .jar file'),
      '#description' => $this
        ->t('Enter the full path to tika executable jar file. Example: "/var/apache-tika/tika-app-1.8.jar".'),
      '#default_value' => $this->configuration['tika_path'],
      '#required' => TRUE,
    ];
    return $form;
  }

  /**
   * {@inheritdoc}
   */
  public function validateConfigurationForm(array &$form, FormStateInterface $form_state) {
    $values = $form_state
      ->getValue([
      'text_extractor_config',
    ]);
    $java_path = $values['java_path'];
    $tika_path = $values['tika_path'];

    // Check java path.
    exec($java_path, $output, $return_code);

    // $return_code = 127 if it fails. 1 instead.
    if ($return_code != 1) {
      $form_state
        ->setError($form['text_extractor_config']['java_path'], $this
        ->t('Invalid path or filename %path for java executable.', [
        '%path' => $java_path,
      ]));
      return;
    }

    // Check tika path.
    if (!file_exists($tika_path)) {
      $form_state
        ->setError($form['text_extractor_config']['tika_path'], $this
        ->t('Invalid path or filename %path for tika application jar.', [
        '%path' => $tika_path,
      ]));
    }
    else {
      $cmd = $java_path . ' -jar ' . escapeshellarg($tika_path) . ' -V';
      exec($cmd, $output, $return_code);

      // $return_code = 1 if it fails. 0 instead.
      if ($return_code) {
        $form_state
          ->setError($form['text_extractor_config']['tika_path'], $this
          ->t('Tika could not be reached and executed.'));
      }
      else {
        $this
          ->getMessenger()
          ->addStatus(t('Tika can be reached and be executed'));
      }
    }
  }

  /**
   * {@inheritdoc}
   */
  public function submitConfigurationForm(array &$form, FormStateInterface $form_state) {
    $this->configuration['java_path'] = $form_state
      ->getValue([
      'text_extractor_config',
      'java_path',
    ]);
    $this->configuration['tika_path'] = $form_state
      ->getValue([
      'text_extractor_config',
      'tika_path',
    ]);
    parent::submitConfigurationForm($form, $form_state);
  }

}

Members

Namesort descending Modifiers Type Description Overrides
DependencySerializationTrait::$_entityStorages protected property An array of entity type IDs keyed by the property name of their storages.
DependencySerializationTrait::$_serviceIds protected property An array of service IDs keyed by property name used for serialization.
DependencySerializationTrait::__sleep public function 1
DependencySerializationTrait::__wakeup public function 2
MessengerTrait::messenger public function Gets the messenger. 29
MessengerTrait::setMessenger public function Sets the messenger.
PluginBase::$configuration protected property Configuration information passed into the plugin. 1
PluginBase::$pluginDefinition protected property The plugin implementation definition. 1
PluginBase::$pluginId protected property The plugin_id.
PluginBase::DERIVATIVE_SEPARATOR constant A string which is used to separate base plugin IDs from the derivative ID.
PluginBase::getBaseId public function Gets the base_plugin_id of the plugin instance. Overrides DerivativeInspectionInterface::getBaseId
PluginBase::getDerivativeId public function Gets the derivative_id of the plugin instance. Overrides DerivativeInspectionInterface::getDerivativeId
PluginBase::getPluginDefinition public function Gets the definition of the plugin implementation. Overrides PluginInspectionInterface::getPluginDefinition 3
PluginBase::getPluginId public function Gets the plugin_id of the plugin instance. Overrides PluginInspectionInterface::getPluginId
PluginBase::isConfigurable public function Determines if the plugin is configurable.
StringTranslationTrait::$stringTranslation protected property The string translation service. 1
StringTranslationTrait::formatPlural protected function Formats a string containing a count of items.
StringTranslationTrait::getNumberOfPlurals protected function Returns the number of plurals supported by a given language.
StringTranslationTrait::getStringTranslation protected function Gets the string translation service.
StringTranslationTrait::setStringTranslation public function Sets the string translation service to use. 2
StringTranslationTrait::t protected function Translates a string to the current language or to a given language.
TextExtractorPluginBase::$configFactory protected property Config factory service.
TextExtractorPluginBase::$messenger protected property The messenger. Overrides MessengerTrait::$messenger
TextExtractorPluginBase::$mimeTypeGuesser protected property Mime type guesser service.
TextExtractorPluginBase::$streamWrapperManager protected property Stream wrapper manager service.
TextExtractorPluginBase::calculateDependencies public function
TextExtractorPluginBase::CONFIGNAME constant Name of the config being edited.
TextExtractorPluginBase::create public static function Creates an instance of the plugin. Overrides ContainerFactoryPluginInterface::create 2
TextExtractorPluginBase::defaultConfiguration public function Gets default configuration for this plugin. Overrides ConfigurableInterface::defaultConfiguration
TextExtractorPluginBase::getConfiguration public function Gets this plugin's configuration. Overrides ConfigurableInterface::getConfiguration
TextExtractorPluginBase::getmessenger public function
TextExtractorPluginBase::getPdfMimeTypes public function Helper method to get the PDF MIME types.
TextExtractorPluginBase::getRealpath public function Helper method to get the real path from an uri.
TextExtractorPluginBase::setConfiguration public function Sets the configuration for this plugin instance. Overrides ConfigurableInterface::setConfiguration
TextExtractorPluginBase::__construct public function Constructs a \Drupal\Component\Plugin\PluginBase object. Overrides PluginBase::__construct 2
TikaExtractor::buildConfigurationForm public function Form constructor. Overrides PluginFormInterface::buildConfigurationForm
TikaExtractor::extract public function Extract file with Tika library. Overrides TextExtractorPluginBase::extract
TikaExtractor::submitConfigurationForm public function Form submission handler. Overrides TextExtractorPluginBase::submitConfigurationForm
TikaExtractor::validateConfigurationForm public function Form validation handler. Overrides TextExtractorPluginBase::validateConfigurationForm