xml_format.inc in Taxonomy import/export via XML 5.2

Include routines for the original XML parsing and taxonomy/term creation.

File

xml_format.inc

<?php

/**
 * @file Include routines for the original XML parsing and taxonomy/term
 * creation.
 */

/**
 * taxonomy_xml_xml_create
 * @param $vid
 *   Which vocabulary to generate the tree for.
 *
 * @param $parent
 *   The term ID under which to generate the tree. If 0, generate the tree
 *   for the entire vocabulary.
 *
 * @param $depth
 *   Internal use only.
 *
 * @param $max_depth
 *   The number of levels of the tree to return. Leave NULL to return all levels.
 *
 * @return
 *   The text of an XML document.
 */
function taxonomy_xml_xml_create($vid, $parent = 0, $depth = -1, $max_depth = NULL) {
  $output = '<?xml version="1.0" standalone="no"?>' . "\n";
  $output .= '<!DOCTYPE taxonomy SYSTEM "taxonomy.dtd">' . "\n";
  $tree = module_invoke('taxonomy', 'get_tree', $vid, $parent, $depth, $max_depth);
  if ($tree) {
    $vocabulary = module_invoke('taxonomy', 'get_vocabulary', $vid);
    $output .= "<vocabulary>\n";
    foreach ($vocabulary as $key => $value) {
      if (is_array($value)) {
        $output .= "<{$key}>" . check_plain(implode(',', $value)) . "</{$key}>";
      }
      else {
        $output .= "<{$key}>" . check_plain($value) . "</{$key}>";
      }
    }
    foreach ($tree as $term) {
      $synonyms = taxonomy_get_synonyms($term->tid);
      $output .= "<term>";
      foreach ($term as $key => $value) {
        if ($key == 'parents') {
          foreach ($value as $parent) {
            $output .= "<parent>" . check_plain($parent) . "</parent>";
          }
        }
        else {
          $output .= "<{$key}>" . check_plain($value) . "</{$key}>";
        }
      }
      if ($synonyms) {
        $output .= "<synonyms>";
        $output .= implode("\n", $synonyms);
        $output .= "</synonyms>";
      }
      $output .= "</term>";
    }
    $output .= "</vocabulary>\n";
  }
  return $output;
}

/**
 * Call-back function used by the XML parser.
 */
function taxonomy_xml_element_start($parser, $name, $attributes) {
  global $_tx_term, $_tx_element, $_tx_tag;
  switch ($name) {
    case 'vocabulary':
      $_tx_element = $name;
      break;
    case 'term':
      $_tx_element = $name;
      $_tx_term += 1;
  }
  $_tx_tag = $name;
}

/**
 * Call-back function used by the XML parser.
 */
function taxonomy_xml_element_end($parser, $name) {
  global $_tx_element;
  switch ($name) {
    case 'vocabulary':
    case 'term':
      $_tx_element = '';
  }
}

/**
 * Call-back function used by the XML parser.
 */
function taxonomy_xml_element_data($parser, $data) {
  global $_tx_vocabulary, $_tx_element, $_tx_terms, $_tx_term, $_tx_tag;
  switch ($_tx_element) {
    case 'term':
      if ($_tx_tag == 'parent') {
        if (trim($data)) {
          $_tx_terms[$_tx_term][$_tx_tag][] = $data;
        }
      }
      else {
        $_tx_terms[$_tx_term][$_tx_tag] .= trim($data);
      }
      break;
    default:
      $_tx_vocabulary[$_tx_tag] .= trim($data);
  }
}

/**
 * Initiate the parser on the custom XML schema.
 * 
 * This uses the XML callback parser with tag callbacks.
 * 
 * @see taxonomy_xml_element_start
 * @see taxonomy_xml_element_end
 * @see taxonomy_xml_element_data
 *
 */
function taxonomy_xml_xml_parse(&$data, $vid = 0, $parent_tid = NULL, $duplicate = 0) {
  global $_tx_terms, $_tx_vocabulary;

  // Unset the global variables before we use them:
  unset($GLOBALS['_tx_element'], $GLOBALS['_tx_term'], $GLOBALS['_tx_tag']);
  $_tx_terms = array();
  $_tx_vocabulary = array();
  $new_terms = array();

  // parse the data:
  $xml_parser = drupal_xml_parser_create($data);
  xml_parser_set_option($xml_parser, XML_OPTION_CASE_FOLDING, false);

  //
  xml_set_element_handler($xml_parser, 'taxonomy_xml_element_start', 'taxonomy_xml_element_end');
  xml_set_character_data_handler($xml_parser, 'taxonomy_xml_element_data');
  if (!xml_parse($xml_parser, $data, 1)) {
    watchdog('error', t('Taxonomy_xml: failed to parse file: %error at line %line.', array(
      '%error' => xml_error_string(xml_get_error_code($xml_parser)),
      '%line' => xml_get_current_line_number($xml_parser),
    )));
    drupal_set_message(t('Failed to parse file: %error at line %line.', array(
      '%error' => xml_error_string(xml_get_error_code($xml_parser)),
      '%line' => xml_get_current_line_number($xml_parser),
    )), 'error');
  }
  xml_parser_free($xml_parser);

  // If an existing vocabulary has been chosen or has the same name as the vocabulary being added,
  // terms should be added to the existing vocabulary. Otherwise a new vocabulary should be created.
  if ($vid == 0) {
    $name = $_tx_vocabulary['name'];
    $_tx_vocabulary = (array) _taxonomy_xml_get_vocabulary_placeholder(trim($name));
  }
  else {
    $_tx_vocabulary = (array) module_invoke('taxonomy', 'get_vocabulary', $vid);
  }

  // Get the maximum depth of terms
  $term_depth = array();
  foreach ($_tx_terms as $term) {
    $term_depth[] = $term['depth'];
  }

  // Import terms in order of depth
  $new_tid = array();
  for ($i = 0; $i <= max($term_depth); $i++) {
    foreach ($_tx_terms as $term) {
      if ($term['depth'] != $i) {
        continue;
      }
      $term['vid'] = $_tx_vocabulary['vid'];
      $term['old_tid'] = $term['tid'];
      unset($term['tid']);
      if (is_array($term['parent'])) {
        foreach ($term['parent'] as $key => $value) {
          if ($value) {
            $term['parent'][$key] = $new_tid[$value];
          }
        }
      }
      $term_exists = false;
      if ($duplicate == 0) {
        $existing_terms = module_invoke('taxonomy', 'get_term_by_name', $term['name']);
        if (count($existing_terms > 0)) {
          foreach ($existing_terms as $existing_term) {
            if ($existing_term->vid == $term['vid']) {
              $term_exists = true;

              // Map the term tid from the imported XML file to the tid in term_data database table
              $new_tid[$term['old_tid']] = $existing_term->tid;
              $skipped_terms[$existing_term->name] = 1;
            }
          }
        }
      }

      // If the term doesn't already exist in this vocabulary, add it.
      if (!$term_exists) {
        taxonomy_save_term($term);

        // Map the term tid from the imported XML file to the tid in term_data database table
        $new_tid[$term['old_tid']] = $term['tid'];
        $new_terms[] = $term['name'];
      }
    }
  }
  $output .= t('Vocabulary %name: ', array(
    '%name' => $_tx_vocabulary['name'],
  ));
  if ($new_terms) {
    $output .= t('Added term(s) %terms. ', array(
      '%terms' => implode(', ', $new_terms),
    ));
  }
  else {
    $output .= t('No terms added. ');
  }
  if ($skipped_terms) {
    $output .= t('Ignored duplicate term(s) %terms. ', array(
      '%terms' => implode(', ', array_keys($skipped_terms)),
    ));
  }
  drupal_set_message($output);
  return count($new_terms);
}

Functions

Name	Description
taxonomy_xml_element_data	Call-back function used by the XML parser.
taxonomy_xml_element_end	Call-back function used by the XML parser.
taxonomy_xml_element_start	Call-back function used by the XML parser.
taxonomy_xml_xml_create	taxonomy_xml_xml_create
taxonomy_xml_xml_parse	Initiate the parser on the custom XML schema.