/**
* Скопировать свойства поля документа в свойства поля индекса
* @param nc_search_field $field
* @return nc_search_provider_index_field
*/
public function copy_options(nc_search_field $field)
{
foreach ($this->properties as $o => $v) {
if ($o != 'id') {
$this->set($o, $field->get($o));
}
}
return $this;
}
/**
* @param $doc_id
* @param nc_search_field $field
* @param nc_search_context $doc_context
* @return string
*/
protected function store_field($doc_id, nc_search_field $field, nc_search_context $doc_context)
{
$text = $field->get('value');
if (strlen($text) == 0) {
return '';
}
// empty fields are not stored
$content = '';
$raw_data = '';
$is_stored = $field->get('is_stored') || $field->get('is_sortable');
// 'is_indexed' → store in the `Content` field
if ($field->get('is_indexed')) {
$content = "";
$filters = nc_search_extension_manager::get('nc_search_language_filter', $doc_context)->except('nc_search_language_filter_case')->stop_on(array());
// Processing in chunks - compromise between performance (less
// method call overhead) and memory usage
$n = 0;
$current_position = 0;
$text_length = strlen($text);
$batch_length = $this->text_batch_length;
while ($current_position < $text_length) {
if ($text_length < $batch_length) {
$batch = $text;
$current_position = $text_length;
} else {
$space_position = strpos($text, " ", min($text_length, $current_position + $batch_length));
if (!$space_position) {
$batch = substr($text, $current_position);
$current_position = $text_length;
} else {
$batch = substr($text, $current_position, $space_position - $current_position);
$current_position = $space_position + 1;
}
}
$tokens = $this->tokenize_text(mb_convert_case($batch, nc_search::get_setting('FilterStringCase'), 'UTF-8'));
if ($field->get('is_normalized')) {
// apply filters
$tokens = $filters->apply('filter', $tokens);
}
$content .= ($n ? ' ' : '') . join(' ', $this->get_term_codes($tokens, true));
$n++;
}
}
// 'is_stored' → store raw text as well
if ($is_stored) {
$raw_data = $text;
}
// save data in the DB
$this->store_index_data($this->get_field_table_name($field), $doc_id, $content, $raw_data);
return $content;
}
/**
*
* @param nc_search_field $field
* @throws nc_search_exception
*/
protected function extract_field_value(nc_search_field $field)
{
$source_name = $field->get('query_scope');
$recipient_name = $field->get('name');
if (!isset($this->parts[$source_name])) {
throw new nc_search_exception("Cannot extract field value: '{$source_name}' part does not exist");
}
$source = $this->parts[$source_name];
if (!isset($this->parts[$recipient_name])) {
$this->parts[$recipient_name] = new nc_search_document_parser_html_fragment();
}
$recipient = $this->parts[$recipient_name];
$this->execute_queries($field->get('query'), $source, $recipient, $field->get('remove_from_parent'), $field->get('query_use_first_matched'));
// filter
if ($field->get('filter_content')) {
$this->execute_queries($field->get('filter_content'), $recipient, null, true, false);
}
}
/**
* Добавляет поле (для индекса).
*
* Внимание, может перезаписать существующее поле без предупреждения
* (it’s a feature / by design)
*
* @param nc_search_field $field
* @return nc_search_document
*/
public function add_field(nc_search_field $field)
{
$field_name = $field->get('name');
$this->fields[$field_name] = $field;
if (!isset($this->field_mapping[$field_name])) {
// required for option_to_field and vice versa
$this->field_mapping[$field_name] = array("name" => $field_name);
}
return $this;
}
/**
* Получить составной ключ по определяющим свойствами поля (используется для
* индексации элементов в коллекции данного типа с целью быстрого поиска полей)
*
* @param nc_search_field|nc_search_provider_index_field $field
* @return string
*/
public static function get_key_options($field)
{
return $field->get('name') . "#" . sprintf("%.2F", $field->get('weight')) . '#' . $field->get('type') . '#' . (int) $field->get('is_sortable');
}
