diff --git a/xmlsitemap.xmlsitemap.inc b/xmlsitemap.xmlsitemap.inc index 0befbca..f617afa 100644 --- a/xmlsitemap.xmlsitemap.inc +++ b/xmlsitemap.xmlsitemap.inc @@ -194,6 +194,11 @@ class XMLSitemapWriter extends XMLWriter { if (is_array($content)) { $this->startElement($name); $xml_content = format_xml_elements($content); + if (module_exists('xmlsitemap_image')) { + if ($content['images'] > 0) { + $xml_content = xmlsitemap_image_write_images_records_to_xml($content); + } + } // Remove additional spaces from the output. $xml_content = str_replace(array(" <", ">\n"), array("<", ">"), $xml_content); $this->writeRaw($xml_content); diff --git a/xmlsitemap_image/xmlsitemap_image.info b/xmlsitemap_image/xmlsitemap_image.info new file mode 100644 index 0000000..6b64a0b --- /dev/null +++ b/xmlsitemap_image/xmlsitemap_image.info @@ -0,0 +1,5 @@ +name = XML sitemap for images +description = Allow xmlsitemap module to get images tag on sitemap +package = XML sitemap +core = 7.x +dependencies[] = xmlsitemap \ No newline at end of file diff --git a/xmlsitemap_image/xmlsitemap_image.module b/xmlsitemap_image/xmlsitemap_image.module new file mode 100644 index 0000000..67a25bc --- /dev/null +++ b/xmlsitemap_image/xmlsitemap_image.module @@ -0,0 +1,370 @@ + 'Image settings', + 'description' => 'settings for the image sitemap', + 'page callback' => 'drupal_get_form', + 'page arguments' => array('xmlsitemap_image_admin'), + 'access arguments' => array('administer site configuration'), + 'type' => MENU_LOCAL_TASK, + ); + return $items; +} + +/** + * Alter the content added to an XML sitemap for an individual element. + * + * This hooks is called when the module is generating the XML content for the + * sitemap and allows other modules to alter existing or add additional XML data + * for any element by adding additional key value paris to the $element array. + * + * @param array $element + * The element that will be converted to XML for the link. + * @param array $link + * An array of properties providing context about the link that we are + * generating an XML element for. + * @param object $sitemap + * The sitemap that is currently being generated. + */ +function xmlsitemap_image_xmlsitemap_element_alter(array &$element, array $link, $sitemap) { + $allowed_content_types = array_filter(variable_get('xmlsitemap_image_content_types', '')); + $image_count = 0; + // If the subtype is in the $allowed_content_types array, load the node. + if (!empty($link['subtype']) && in_array($link['subtype'], $allowed_content_types)) { + // Load the node. + $current_node = node_load($link['id']); + // Here we collect the field names for checking. + $checking_fields = array(); + // Let's get the available image fields of the content type. + $image_fields = _xmlsitemap_image_get_imagefield_by_node_type($link['subtype']); + // Handle the allowed image fields. + $allowed_image_fields = variable_get('xmlsitemap_image_image_fields', ''); + if (!empty($allowed_image_fields)) { + $allowed_image_fields = array_map('trim', explode(',', $allowed_image_fields)); + $checking_fields = $allowed_image_fields; + } + // If allowed image fields not provided, use the image fields of content type. + // @todo: remove duplicates. + if (!empty($image_fields)) { + $checking_fields += $image_fields; + } + // Handle the excluded fields. + $excluded_image_fields = variable_get('xmlsitemap_image_excluded_image_fields', ''); + // If excluded field are provided, remove thme from checking array. + if (!empty($excluded_image_fields)) { + $excluded_image_fields = array_map('trim', explode(',', $excluded_image_fields)); + $checking_fields = array_filter($checking_fields, + function ($a) use ($excluded_image_fields) { + return !in_array($a, $excluded_image_fields); + } + ); + } + $imgtypes = array('image/png', 'image/jpg', 'image/gif', 'image/jpeg'); + // If we still have field to be checked, start the process. + if (!empty($checking_fields)) { + foreach ($checking_fields as $field_name) { + if (!empty($current_node->$field_name)) { + foreach ($current_node->$field_name[$current_node->language] as $image) { + if (!empty($image["filemime"]) && in_array($image["filemime"], $imgtypes)) { + $element[] = _xmlsitemap_image_handle_imagefield($image); + $image_count++; + } + if (!empty($image['value'])) { + $items = _xmlsitemap_image_handle_textfield($image); + if (!empty($items)) { + $image_count = count($items); + foreach ($items as $item) { + $element[] = $item; + } + } + } + } + } + } + } + } + $element['images'] = $image_count; +} + +/** + * Helper function to build output from image fields. + * + * @param array $image + * + * @return array + */ +function _xmlsitemap_image_handle_imagefield(array $image) { + $title = !empty($image['field_file_image_title_text']) ? $image['field_file_image_title_text'] : (!empty($image['title']) ? $image['title'] : $image['filename']); + $caption = !empty($image['field_file_image_caption_text']) ? $image['field_file_image_caption_text'] : (!empty($image['alt']) ? $image['alt'] : $image['filename']); + return array( + "key" => "image:image", + "value" => array( + "image:loc" => file_create_url($image["uri"]), + "image:title" => $title, + "image:caption" => $caption, + ), + ); +} + +/** + * Helper function to build output from textfields. + * + * @param array $image + * + * @return array + */ +function _xmlsitemap_image_handle_textfield(array $image) { + $images = array(); + $checking_contents = array(); + // The site uses a Drupal-core provided textfield. + if (array_key_exists('safe_value', $image)) { + $checking_contents[] = $image['value']; + } + // The site uses somewhat custom textfield. + if (array_key_exists('revision_id', $image)) { + // Cooperation with Paragraphs module. + if (module_exists('paragraphs')) { + $y = paragraphs_item_load($image['value']); + } + // Cooperation with Field collection module. + if (module_exists('field_collection_items')) { + $y = field_collection_item_load($image['value']); + } + if (!empty($y)) { + foreach ((array) $y as $key => $value) { + if (is_array($value) && !empty($value) && array_key_exists($y->langcode(), $value)) { + foreach ($value[$y->langcode()] as $item) { + $checking_contents[] = $item['value']; + } + } + } + } + } + + // Let's go through the contents. + if (!empty($checking_contents)) { + /* + * The regex pattern will look for an image tag and create individual capture groups + * for 'class'(1), 'src'(2), 'alt'(3), 'title'(4), 'width'(5) and 'height'(6). + * The numbers within the brackets stand for the capture group number + * respectively the key within the PHP array. The array key 0 will contain + * the whole tag with all attributes (the full img link) + * Original: https://www.lexo.ch/blog/2014/05/fetch-img-tags-with-all-attributes-from-a-php-string-using-a-regex-pattern-and-capture-groups/ + * Extended by title option. + */ + $pattern = '//si'; + $excluded_classes = variable_get('xmlsitemap_image_excluded_image_classes', ''); + if (!empty($excluded_classes)) { + $excluded_classes = array_map('trim', explode(',', $excluded_classes)); + } + foreach ($checking_contents as $content) { + preg_match_all($pattern, $content, $matches); + if (!empty($matches)) { + $current_host = $_SERVER['REQUEST_SCHEME'] . '://' . $_SERVER['SERVER_NAME']; + foreach ($matches[0] as $key => $match) { + $skip = FALSE; + // If it's not fully qualified URL, we extend it. + if (!preg_match('/^(http:|https)/', $matches[2][$key])) { + $matches[2][$key] = $current_host . $matches[2][$key]; + } + else { + // If the given URL is not came from the current site. + if (strstr($matches[2][$key], $current_host) === FALSE) { + $skip = TRUE; + } + // Let's give a chance for developers on dev enviroments. + if (module_exists('stage_file_proxy') && $skip) { + $origin_host = variable_get('stage_file_proxy_origin', ''); + if (strstr($matches[2][$key], $origin_host) != FALSE) { + $skip = FALSE; + } + } + } + // Check excluded classes. + if (!$skip && !empty($excluded_classes) && !empty($matches[1][$key])) { + $current_image_classes = array_map('trim', explode(' ', $matches[1][$key])); + $same = array_intersect($excluded_classes, $current_image_classes); + if (!empty($same)) { + $skip = TRUE; + } + } + // And finally check the availability of image URL. + if (!$skip) { + $file_headers = @get_headers($matches[2][$key]); + // Only the available images can be in the sitemap. + if ($file_headers[0] !== 'HTTP/1.1 200 OK') { + $skip = TRUE; + } + } + if ($skip) continue; + + $images[] = array( + "key" => "image:image", + "value" => array( + "image:loc" => $matches[2][$key], + "image:title" => $matches[3][$key], + "image:caption" => $matches[4][$key], + ), + ); + } + } + } + } + return $images; +} + +/** + * Alter the attributes used for the root element of the XML sitemap. + * + * For example add an xmlns:image attribute: + * + * + * @param array $attributes + * An associative array of attributes to use in the root element of an XML + * sitemap. + * @param object $sitemap + * The sitemap that is currently being generated. + */ +function xmlsitemap_image_xmlsitemap_root_attributes_alter(&$attributes, $sitemap) { + $attributes['xmlns:image'] = 'http://www.google.com/schemas/sitemap-image/1.1'; +} + +/** + * Implements hook_admin_settings() for module settings configuration. + */ +function xmlsitemap_image_admin() { + $form = array(); + $types = node_type_get_types(); + $node_type_options = array(); + foreach ($types as $type) { + $imagefields = ''; + // Provides availabe image fields for admin. + $fields = _xmlsitemap_image_get_imagefield_by_node_type($type->type); + if (!empty($fields)) { + $imagefields = ' (' . implode(', ', $fields) . ')'; + } + $node_type_options[$type->type] = $type->name . $imagefields; + } + + $form['xmlsitemap_image_content_types'] = array( + '#type' => 'checkboxes', + '#title' => t("Add content types to be checked for images"), + '#options' => $node_type_options, + '#default_value' => variable_get('xmlsitemap_image_content_types', ''), + '#required' => TRUE, + '#description' => t('You can find the name of available image fields next to the name of content type in parenthesis.'), + ); + + $form['xmlsitemap_image_image_fields'] = array( + '#type' => 'textfield', + '#title' => t("Add comma separated list of image field names you want to INCLUDE in the sitemap eg.: field_article_image. If the field is empty all image field will be checked in the allowed content type(s). Text fields can be added too, the tags will be processed from the content."), + '#default_value' => variable_get('xmlsitemap_image_image_fields', ''), + ); + + $form['xmlsitemap_image_excluded_image_fields'] = array( + '#type' => 'textfield', + '#title' => t("Add comma separated list of image field names you want to EXCLUDE from the sitemap. eg.: field_notimportant_image, field_excludable_image"), + '#default_value' => variable_get('xmlsitemap_image_excluded_image_fields', ''), + ); + + $form['xmlsitemap_image_excluded_image_classes'] = array( + '#type' => 'textfield', + '#title' => t("Add comma separated list of image classes you want to EXCLUDE from the sitemap."), + '#default_value' => variable_get('xmlsitemap_image_excluded_image_classes', 'hidden'), + ); + // We only have to define form elements for the actual settings elements. + // The system_settings_form() function will take care of creating the form + // page, adding a submit button, and saving the settings. + return system_settings_form($form); +} + +/** + * Implements hook_menu_alter(). + */ +function xmlsitemap_image_menu_alter(&$items) { + $items['sitemap.xsl'] = array( + 'page callback' => 'xmlsitemap_image_xmlsitemap_output_xsl', + 'access callback' => TRUE, + 'type' => MENU_CALLBACK, + ); +} + +/** + * Alter the output of xmlsitemap.xsl + */ +function xmlsitemap_image_xmlsitemap_output_xsl() { + // Read the XSL content from the file. + $module_path = drupal_get_path('module', 'xmlsitemap'); + $xsl_content = file_get_contents(drupal_get_path('module', 'xmlsitemap') . '/xsl/xmlsitemap.xsl'); + + // Make sure the strings in the XSL content are translated properly. + $replacements = array( + 'Sitemap file' => t('Sitemap file'), + 'Generated by the Drupal XML sitemap module.' => t('Generated by the Drupal XML sitemap module.', array('@link-xmlsitemap' => 'http://drupal.org/project/xmlsitemap')), + 'Number of sitemaps in this index' => t('Number of sitemaps in this index'), + 'Click on the table headers to change sorting.' => t('Click on the table headers to change sorting.'), + 'Sitemap URL' => t('Sitemap URL'), + 'Last modification date' => t('Last modification date'), + 'Number of URLs in this sitemap' => t('Number of URLs in this sitemap'), + 'URL location' => t('URL location'), + 'Change frequency' => t('Change frequency'), + 'Priority' => t('Priority'), + '[jquery]' => base_path() . 'misc/jquery.js', + '[jquery-tablesort]' => base_path() . $module_path . '/xsl/jquery.tablesorter.min.js', + '[xsl-js]' => base_path() . $module_path . '/xsl/xmlsitemap.xsl.js', + '[xsl-css]' => base_path() . $module_path . '/xsl/xmlsitemap.xsl.css', + ); + $xsl_content = strtr($xsl_content, $replacements); + + // Output the XSL content. + drupal_add_http_header('Content-type', 'application/xml; charset=utf-8'); + drupal_add_http_header('X-Robots-Tag', 'noindex, follow'); + print $xsl_content; +} + +/** + * Helper function to get image fields by node types. + * + * @param string $node_type + * + * @return mixed + */ +function _xmlsitemap_image_get_imagefield_by_node_type($node_type) { + $query = db_select('field_config', 'fc'); + $query->join('field_config_instance', 'fci', 'fc.id = fci.field_id'); + $query->fields('fc', array('field_name')) + ->condition('fc.type', 'image') + ->condition('fc.module', 'image') + ->condition('fci.bundle', $node_type); + $fields = $query->execute()->fetchAllKeyed(0, 0); + return $fields; +} + +/** + * Write the images xml elements into xml content. + * + * @param $content + * @param $xml_content + */ +function xmlsitemap_image_write_images_records_to_xml($content) { + $multiimgstr = ""; + foreach ($content as $keycontent => $valuecontent) { + if (is_array($valuecontent)) { + foreach ($valuecontent as $imagekey => $imagesvalues){ + $multiimgstr .= format_xml_elements($imagesvalues); + } + unset($content[$keycontent]); + } + } + // We have to regenerate the original $xml_content value. + $xml_content = format_xml_elements($content); + if (!empty($multiimgstr)) { + $xml_content .= $multiimgstr; + } + return $xml_content; +} diff --git a/xsl/xmlsitemap.xsl b/xsl/xmlsitemap.xsl index a5e6412..005700c 100644 --- a/xsl/xmlsitemap.xsl +++ b/xsl/xmlsitemap.xsl @@ -93,6 +93,7 @@ URL location Last modification date Change frequency + Images Priority @@ -113,6 +114,7 @@ +