remove-html.php

<?php

/**
 * Remove HTML tags from various i18n information object fields.
 *
 * To execute script (from AtoM root directory):
 *
 *   php symfony tools:run remove-html.php
 */
class i18nRemoveHtmlTags
{
  public static $columns = array(
    'title',
    'alternate_title',
    'edition',
    'extent_and_medium',
    'archival_history',
    'acquisition',
    'scope_and_content',
    'appraisal',
    'accruals',
    'arrangement',
    'access_conditions',
    'reproduction_conditions',
    'physical_characteristics',
    'finding_aids',
    'location_of_originals',
    'location_of_copies',
    'related_units_of_description',
    'institution_responsible_identifier',
    'rules',
    'sources',
    'revision_history'
  );

  public function execute()
  {
    $rowCount           = 0;
    $changedCount       = 0;
    $columnsChangedCount = 0;

    // Fetch all information object i18n rows
    $query = "SELECT * FROM information_object_i18n WHERE id != ". QubitInformationObject::ROOT_ID;
    $statement = QubitPdo::prepareAndExecute($query);

    while ($io = $statement->fetch(PDO::FETCH_OBJ))
    {
      // Process HTML in row's columns
      $columnsChanged = $this->processInformationObjectI18nHtml($io);

      // Update total column values changed
      if ($columnsChanged)
      {
        $changedCount++;
        $columnsChangedCount += $columnsChanged;
      }

      // Report progress
      $message = 'Processed information object '.$io->id;

      if ($columnsChanged)
      {
        $message .= ' ('. $columnsChanged . ' changes)';
      }

      print $message ."\n";
      $rowCount++;
    }

    // Report summary of processing
    $message = 'Processed '. $rowCount .' information objects.';

    if ($changedCount)
    {
      $message .= ' Changed '. $changedCount .' information objects';
      $message .= ' ('. $columnsChangedCount .' field values changed).';
    }

    print $message ."\n";
  }

  /**
   * Determine what information object i18n columns are populated
   * and update them.
   *
   * @param stdClass $io  row of information object i18n data
   *
   * @return integer  number of columns changed
   */
  private function processInformationObjectI18nHtml(&$io)
  {
    // Determine what column values contain HTML
    $columnValues = array();

    foreach(i18nRemoveHtmlTags::$columns as $column)
    {
      // Store column name/value for processing if it contains tags
      if ($io->{$column} && ($io->{$column} != strip_tags($io->{$column})))
      {
        $columnValues[$column] = $io->{$column};
      }
    }

    // Update database with transformed column values
    $this->transformHtmlInI18nTableColumns(
      'information_object_i18n', $io->id, $io->culture, $columnValues);

    return count($columnValues);
  }

  /**
   * Transform HTML column values into text and update specified i18n table row
   *
   * @param string  i18n table name
   * @param integer $id  ID of row in an i18n table
   * @param string $culture  culture code of a row in an i18n table
   * @param array $columnValues  key/value array of column/value data to process
   *
   * @return void
   */
  private function transformHtmlInI18nTableColumns($table, $id, $culture, $columnValues)
  {
    // Aseemble query and note parsed column values
    $values = array();

    $query = 'UPDATE '. $table .' SET ';

    foreach($columnValues as $column => $value)
    {
      // Only update if tags are found
      if ($value != strip_tags($value)) {
        $transformedValue = $this->transformHtmlToText($value);

        $query .= (count($values)) ? ', ' : '';

        $query .= $column ."=?";

        $values[] = $transformedValue;
      }
    }

    $query .= " WHERE id='". $id ."' AND culture='". $culture ."'";

    if (count($values))
    {
      QubitPdo::prepareAndExecute($query, $values);
    }
  }

  /**
   * Transform HTML into text using the Document Object Model
   *
   * @param string $html  HTML to transform into text
   *
   * @return string  transformed text
   */
  private function transformHtmlToText($html)
  {
    // Parse HTML
    $doc = new DOMDocument();
    $doc->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'));

    // Apply transformations
    $this->transformDocument($doc);

    // Convert to string and strip leading/trailing whitespace
    return trim(strip_tags($doc->saveXml($doc->documentElement)));
  }

  /**
   * Transform specific tags within a DOM document
   *
   * @param DOMDocument  $doc  DOM document
   *
   * @return void
   */
  private function transformDocument(&$doc)
  {
    // Create text representations of various HTML tags
    $this->transformDocumentLinks($doc);
    $this->transformDocumentLists($doc);
    $this->transformDocumentDescriptionLists($doc);
    $this->transformDocumentBreaks($doc);

    // Deal with paragraphs last, as other transformations create them
    $this->transformDocumentParasIntoNewlines($doc);
  }

  /**
   * Transform link tags into text
   *
   * @param DOMDocument  $doc  DOM document
   *
   * @return void
   */
  private function transformDocumentLinks(&$doc)
  {
    $linkList = $doc->getElementsByTagName('a');

    // Loop through each <a> tag and replace with text content
    while ($linkList->length > 0)
    {
      $linkNode = $linkList->item(0);

      $linkText = $linkNode->textContent;
      $linkHref = $linkNode->getAttribute('href');

      if ($linkHref)
      {
        if (0 === strpos(strtolower($linkHref), 'mailto:')) {
          $linkHref = removeFromStartOfString($linkHref, 'mailto:');
        }

        $linkText .= ' ['. $linkHref .']';
      }

      $newTextNode = $doc->createTextNode($linkText);
      $linkNode->parentNode->replaceChild($newTextNode, $linkNode);
    }
  }

  /**
   * Transform unordered list-related tags into text and enclose in
   * <p> tags
   *
   * @param DOMDocument  $doc  DOM document
   *
   * @return void
   */
  private function transformDocumentLists(&$doc)
  {
    $ulList = $doc->getElementsByTagName('ul');

    // Loop through each <ul> tag and change to a <p> tag
    while ($ulList->length > 0)
    {
      $listNode = $ulList->item(0);

      $newParaNode = $doc->createElement('p');

      // Assemble text representation of list
      $paraText = '';

      foreach($listNode->childNodes as $childNode)
      {
        $paraText .= '* '. $childNode->textContent ."\n";
      }

      // Set <p> element's text
      $newTextNode = $doc->createTextNode($paraText);
      $newParaNode->appendChild($newTextNode);

      $listNode->parentNode->replaceChild($newParaNode, $listNode);
    }
  }

  /**
   * Transform description list-related tags, removing description
   * terms and enclosing the description text in <p> tags
   *
   * @param DOMDocument  $doc  DOM document
   *
   * @return void
   */
  private function transformDocumentDescriptionLists(&$doc)
  {
    $termList = $doc->getElementsByTagName('dt');

    // Loop through each <dt> tag and remove it
    while ($termList->length > 0)
    {
      $termNode = $termList->item(0);
      $termNode->parentNode->removeChild($termNode);
    }

    $descriptionList = $doc->getElementsByTagName('dd');

    // Look through each <dd> element and change to a <p> element
    while ($descriptionList->length > 0)
    {
      $descriptionNode = $descriptionList->item(0);

      // Create <p> node with description's text
      $newParaNode = $doc->createElement('p');
      $newTextNode = $doc->createTextNode($descriptionNode->textContent);
      $newParaNode->appendChild($newTextNode);

      $descriptionNode->parentNode->replaceChild($newParaNode, $descriptionNode);
    }
  }

  /**
   * Transform break tags into newlines
   *
   * @param DOMDocument  $doc  DOM document
   *
   * @return void
   */
  private function transformDocumentBreaks(&$doc)
  {
    $breakList = $doc->getElementsByTagName('br');

    // Loop through each <p> and replace with text
    while($breakList->length)
    {
      $breakNode = $breakList->item(0);

      $newTextNode = $doc->createTextNode("\n");

      $breakNode->parentNode->replaceChild($newTextNode, $breakNode);
    }
  }

  /**
   * Transform paragraph tags into newlines
   *
   * @param DOMDocument  $doc  DOM document
   *
   * @return void
   */
  private function transformDocumentParasIntoNewlines(&$doc)
  {
    $paraList = $doc->getElementsByTagName('p');

    // Loop through each <p> and replace with text
    while($paraList->length)
    {
      $paraNode = $paraList->item(0);

      $paraText = "\n". $paraNode->textContent ."\n";
      $newTextNode = $doc->createTextNode($paraText);

      $paraNode->parentNode->replaceChild($newTextNode, $paraNode);
    }
  }

  /**
   * Remove text at the start of strings
   *
   * @param string  $string  text
   * @param string  $substring  text to remove
   *
   * @return string  processed text
   */
  private function removeFromStartOfString($string, $substring)
  {
    return substr($string, strlen($substring), strlen($string) - strlen($substring));
  }
}

$script = new i18nRemoveHtmlTags();
$script->execute();