anotherjames · January 13, 2025 14:32 · anotherjames · Jan 13, 2025
diff --git a/latinvalidation.module.php b/latinvalidation.module.php
 <?php

 use Drupal\Component\Utility\Html;
 use Drupal\Core\Form\FormStateInterface;
 use Drupal\filter\FilterPluginCollection;

 /**
 * Implements hook_form_alter().
 */
 function latinvalidation_form_alter(&$form, FormStateInterface $form_state, $form_id) {
  // Attach validation to main contact form for non-latin characters.
  if (!empty($form['#webform_id']) && $form['#webform_id'] == 'get_in_touch') {
    $form['elements']['message']['#element_validate'][] = 'latinvalidation_latin_chars_element_validate';
  }
 }

 /**
 * Element validation handler that checks a value is mostly Latin characters.
 */
 function latinvalidation_latin_chars_element_validate($element, FormStateInterface $form_state) {
  if (!empty($element['#value'])) {
    // We've also received a lot of submissions with a common pattern using the
    // format [url=http...]...[/url]. Let's convert those to links before HTML
    // is stripped.
    $meaningful_text = preg_replace('/\[url=([^]]+)]/', '<a href="$1">', str_replace('[/url]', '</a>', $element['#value']));

    // Ignore HTML tags (since tags are usually links).
    $meaningful_text = strip_tags($meaningful_text);

    // Convert any URLs in plain text to links, using core's URL filter, then
    // filter out links entirely.
    $plugin_id = 'filter_url';
    $filterPluginManager = \Drupal::getContainer()
      ->get('plugin.manager.filter');
    if ($filterPluginManager->getDefinition($plugin_id, FALSE)) {
      // Use a filter plugin collection, just like
      // \Drupal\filter\Entity\FilterFormat::filters().
      $collection = new FilterPluginCollection($filterPluginManager, [
        $plugin_id => [],
      ]);
      /** @var \Drupal\filter\Plugin\Filter\FilterUrl $filter */
      $filter = $collection->get($plugin_id);

      $meaningful_text = $filter->prepare($meaningful_text, '');
      $meaningful_text = $filter
        ->process($meaningful_text, '')
        ->getProcessedText();

      $dom = Html::load($meaningful_text);
      $xpath = new \DOMXPath($dom);
      /** @var \DOMElement $dom_element */
      foreach ($xpath->query('//a') as $dom_element) {
        $dom_element->parentNode->removeChild($dom_element);
      }
      $meaningful_text = Html::decodeEntities(Html::serialize($dom));
    }

    // Ignore whitespace.
    $meaningful_text = preg_replace('/\s/', '', $meaningful_text);

    if (mb_strlen($meaningful_text)) {
      // When available, grapheme_strlen() could be used instead (e.g. to count
      // composite emojis as single characters), but the regex below doesn't
      // count like that anyway, it counts like the simpler mb_strlen() does. We
      // don't need to be too accurate as we're just using a vague threshold to
      // limit probably-unwanted submissions. All this inaccuracy means is that
      // certain emojis (those that are really combinations of emojis) and any
      // other multibyte-but-single-grapheme characters are 'unfairly' punished.
      // Meanwhile, do ignore punctuation (well, anything in the \p{Common}
      // class for regular expressions).
      $total_length = mb_strlen($meaningful_text) - preg_match_all('/\\p{Common}/u', $meaningful_text, $matches);

      if ($total_length) {
        $count = preg_match_all('/[^\\p{Common}\\p{Latin}]/u', $meaningful_text, $matches);
        $nonlatin_proportion = $count / $total_length;
        // We picked a fixed proportion of 20%, but this could be configurable.
        if ($nonlatin_proportion > 0.2) {
          $form_state->setError($element, t('We may have trouble trying to understand your message due to the language and/or symbols used in it. Please use plain English instead. You may prefer to <a href="/contact">contact us via phone or email</a>.'));
        }
      }
      else {
        $form_state->setError($element, t('Please include your message for us. You may prefer to <a href="/contact">contact us via phone or email</a>.'));
      }
    }
    else {
      $form_state->setError($element, t('Please include your message for us. You may prefer to <a href="/contact">contact us via phone or email</a>.'));
    }
  }
 }
	<?php

	use Drupal\Component\Utility\Html;
	use Drupal\Core\Form\FormStateInterface;
	use Drupal\filter\FilterPluginCollection;

	/**
	* Implements hook_form_alter().
	*/
	function latinvalidation_form_alter(&$form, FormStateInterface $form_state, $form_id) {
	// Attach validation to main contact form for non-latin characters.
	if (!empty($form['#webform_id']) && $form['#webform_id'] == 'get_in_touch') {
	$form['elements']['message']['#element_validate'][] = 'latinvalidation_latin_chars_element_validate';
	}
	}

	/**
	* Element validation handler that checks a value is mostly Latin characters.
	*/
	function latinvalidation_latin_chars_element_validate($element, FormStateInterface $form_state) {
	if (!empty($element['#value'])) {
	// We've also received a lot of submissions with a common pattern using the
	// format [url=http...]...[/url]. Let's convert those to links before HTML
	// is stripped.
	$meaningful_text = preg_replace('/\[url=([^]]+)]/', '<a href="$1">', str_replace('[/url]', '</a>', $element['#value']));

	// Ignore HTML tags (since tags are usually links).
	$meaningful_text = strip_tags($meaningful_text);

	// Convert any URLs in plain text to links, using core's URL filter, then
	// filter out links entirely.
	$plugin_id = 'filter_url';
	$filterPluginManager = \Drupal::getContainer()
	->get('plugin.manager.filter');
	if ($filterPluginManager->getDefinition($plugin_id, FALSE)) {
	// Use a filter plugin collection, just like
	// \Drupal\filter\Entity\FilterFormat::filters().
	$collection = new FilterPluginCollection($filterPluginManager, [
	$plugin_id => [],
	]);
	/** @var \Drupal\filter\Plugin\Filter\FilterUrl $filter */
	$filter = $collection->get($plugin_id);

	$meaningful_text = $filter->prepare($meaningful_text, '');
	$meaningful_text = $filter
	->process($meaningful_text, '')
	->getProcessedText();

	$dom = Html::load($meaningful_text);
	$xpath = new \DOMXPath($dom);
	/** @var \DOMElement $dom_element */
	foreach ($xpath->query('//a') as $dom_element) {
	$dom_element->parentNode->removeChild($dom_element);
	}
	$meaningful_text = Html::decodeEntities(Html::serialize($dom));
	}

	// Ignore whitespace.
	$meaningful_text = preg_replace('/\s/', '', $meaningful_text);

	if (mb_strlen($meaningful_text)) {
	// When available, grapheme_strlen() could be used instead (e.g. to count
	// composite emojis as single characters), but the regex below doesn't
	// count like that anyway, it counts like the simpler mb_strlen() does. We
	// don't need to be too accurate as we're just using a vague threshold to
	// limit probably-unwanted submissions. All this inaccuracy means is that
	// certain emojis (those that are really combinations of emojis) and any
	// other multibyte-but-single-grapheme characters are 'unfairly' punished.
	// Meanwhile, do ignore punctuation (well, anything in the \p{Common}
	// class for regular expressions).
	$total_length = mb_strlen($meaningful_text) - preg_match_all('/\\p{Common}/u', $meaningful_text, $matches);

	if ($total_length) {
	$count = preg_match_all('/[^\\p{Common}\\p{Latin}]/u', $meaningful_text, $matches);
	$nonlatin_proportion = $count / $total_length;
	// We picked a fixed proportion of 20%, but this could be configurable.
	if ($nonlatin_proportion > 0.2) {
	$form_state->setError($element, t('We may have trouble trying to understand your message due to the language and/or symbols used in it. Please use plain English instead. You may prefer to <a href="/contact">contact us via phone or email</a>.'));
	}
	}
	else {
	$form_state->setError($element, t('Please include your message for us. You may prefer to <a href="/contact">contact us via phone or email</a>.'));
	}
	}
	else {
	$form_state->setError($element, t('Please include your message for us. You may prefer to <a href="/contact">contact us via phone or email</a>.'));
	}
	}
	}
No results found