public function __construct($level = null)
{
if (!$level) {
$level = nc_search::get_setting('LogLevel') | nc_search::LOG_ERROR | nc_search::LOG_CRAWLER_REQUEST | nc_search::LOG_INDEXING_BEGIN_END;
}
$this->level = $level;
}
public function filter(array $terms)
{
for ($i = 0, $max = sizeof($terms); $i < $max; $i++) {
$terms[$i] = mb_convert_case($terms[$i], nc_search::get_setting('FilterStringCase'), 'UTF-8');
}
return $terms;
}
/**
*
* @param nc_search_indexer $indexer
* @throws nc_search_exception
* @return boolean true when task is finished
*/
public function loop(nc_search_indexer $indexer)
{
$cycle_number = 0;
$delay = (int) nc_search::get_setting('CrawlerDelay');
while (true) {
// сохранять задачу каждые X циклов
if ($cycle_number % nc_search::get_setting('IndexerSaveTaskEveryNthCycle') == 0) {
$indexer->save_task();
}
switch ($indexer->next()) {
case nc_search_indexer::TASK_FINISHED:
return true;
// we're done
// we're done
case nc_search_indexer::TASK_STEP_FINISHED:
$delay && sleep($delay);
break;
case nc_search_indexer::TASK_STEP_SKIPPED:
break;
default:
throw new nc_search_exception("Incorrect return value from nc_search_indexer::next()");
}
$cycle_number++;
}
}
/**
* Если есть мета-тэг robots или с именем бота и значенем 'noindex',
* документ не индексируется (внимание: атрибуты case-sensitive)
*
* @return boolean
*/
public function should_index()
{
if (!nc_search::should('ObeyMetaNoindex')) {
return true;
}
$xpath_query = '//meta[((@name="robots") or (@name="' . nc_search::get_setting('CrawlerUserAgent') . '")) and (contains(@content, "noindex"))]';
return $this->xpath($xpath_query)->length == 0;
}
/**
*
*/
public function __construct()
{
// skip numbers?
$this->ignore_numbers = nc_search::should('IgnoreNumbers');
$this->tokenizer_regexp = $this->ignore_numbers ? $this->regexp_alpha : $this->regexp_alnum;
// max terms
$max_terms = nc_search::get_setting('MaxTermsPerField');
if ($max_terms > 0) {
$this->max_terms = $max_terms + 2;
$this->max_chunks = $max_terms * 2 + 2;
}
}
public function filter(array $terms)
{
$min_length = nc_search::get_setting('MinWordLength');
if ($min_length < 2) {
return $terms;
}
$result = array();
for ($i = 0, $max = sizeof($terms); $i < $max; $i++) {
if (mb_strlen($terms[$i], 'UTF-8') >= $min_length) {
$result[] = $terms[$i];
}
}
return $result;
}
/**
* @param string $default_operator "AND"|"OR" (case-sensitive); defaults
* to the 'DefaultBooleanOperator' setting.
* @param bool $ignore_numbers defaults to 'IgnoreNumbers'
*/
public function __construct($default_operator = null, $ignore_numbers = null)
{
if (!$default_operator || $default_operator != "AND" && $default_operator != "OR") {
$default_operator = nc_search::get_setting("DefaultBooleanOperator");
}
$this->default_operator = $default_operator;
if ($ignore_numbers === null) {
$ignore_numbers = nc_search::should('IgnoreNumbers');
}
if ($ignore_numbers) {
$this->term_chars = "\\pL";
}
$this->ignore_numbers = $ignore_numbers;
}
/**
* Запланировать запуск переиндексирования области или правила в указанное время
* @param string $area_string
* @param integer $timestamp
*/
public static function schedule_indexing($area_string, $timestamp)
{
// Если данная область уже поставлена в очередь на более раннее или ближайшее
// время, не нужно добавлять ещё раз
$interval = $timestamp + nc_search::get_setting('MinScheduleInterval');
$intent = nc_search::load('nc_search_scheduler_intent', "SELECT * FROM `%t%`" . " WHERE `StartTime` <= {$interval}" . " AND `AreaString` = '" . nc_search_util::db_escape($area_string) . "'")->first();
// type is ignored
if ($intent) {
// уже есть такое расписание!
if ($intent->get('start_time') > $timestamp) {
$intent->set('start_time', $timestamp);
// let's run it sooner
}
} else {
$intent = new nc_search_scheduler_intent(array('start_time' => $timestamp, 'type' => nc_search_scheduler_intent::ON_REQUEST, 'area_string' => $area_string));
}
$intent->save();
}
public function __construct()
{
require_once 'HTTP/Client.php';
// /netcat/require/lib
require_once 'HTTP/Request/Listener.php';
$headers = array('User-Agent' => nc_search::get_setting('CrawlerUserAgent'));
$this->http_client = new HTTP_Client(null, $headers);
$this->http_client->enableHistory(false);
$this->http_client->setMaxRedirects(nc_search::get_setting('CrawlerMaxRedirects'));
$max_doc_size = nc_search::get_setting('CrawlerMaxDocumentSize');
$db = nc_Core::get_object()->db;
$accepted_mime_types = $db->get_col("SELECT DISTINCT `ContentType`\n FROM `Search_Extension`\n WHERE `ExtensionInterface` = 'nc_search_document_parser'");
// $accepted_mime_types = nc_search::load_all('nc_search_extension_rule')
// ->where('extension_interface', 'nc_search_document_parser')
// ->each('get', 'content_type');
// $accepted_mime_types = array_unique($accepted_mime_types);
$listener = new nc_search_indexer_crawler_listener($max_doc_size, $accepted_mime_types);
$this->http_client->attach($listener, true);
}
/**
* Перед сохранением нужно прогнать список слов через фильтры
*/
public function save()
{
$mb_case = nc_search::get_setting('FilterStringCase');
$apply_filter = !$this->get('dont_filter');
$list = array();
foreach ($this->get('words') as $word) {
$word = trim($word);
if (strlen($word)) {
// пропустить пустые значения
// преобразовать регистр, если в дальнейшем не будут применены фильтры
$list[] = $apply_filter ? $word : mb_convert_case($word, $mb_case);
}
}
if ($apply_filter) {
$context = new nc_search_context(array('language' => $this->get('language')));
$list = nc_search_extension_manager::get('nc_search_language_filter', $context)->until_first('nc_search_language_filter_synonyms')->apply('filter', $list);
}
if (sizeof($list) < 2) {
throw new nc_search_data_exception(NETCAT_MODULE_SEARCH_ADMIN_SYNONYM_LIST_MUST_HAVE_AT_LEAST_TWO_WORDS);
}
$this->set('words', $list);
parent::save();
}
/**
* Установить свойство 'context' документа (фрагменты совпавшего с запросом текста)
* (не имеет отношения к nc_search_context)
*/
protected function document_set_context(nc_search_result_document $doc)
{
// служба поиска (напр. Гугль какой-нибудь), теоретически, может установить свойство context
if (!$doc->get('context') && $this->should_highlight()) {
$language = $this->get_query()->get('language');
$doc->set('context', $this->highlight($doc->get('content'), $language, nc_search::get_setting('ResultContextMaxNumberOfWords')));
$doc->set('title', $this->highlight($doc->get('title'), $language, nc_search::get_setting('ResultTitleMaxNumberOfWords')));
}
return $doc;
}
<?php
/**
* Входящие параметры:
* - term
* - language
*
* @global $catalogue
*/
$NETCAT_FOLDER = realpath("../../../../");
require_once "{$NETCAT_FOLDER}/vars.inc.php";
require $INCLUDE_FOLDER . "index.php";
// получение параметров
$input = trim($nc_core->input->fetch_get('term'));
if (!nc_search::should('EnableQuerySuggest') || nc_search::get_setting('SuggestMode') != 'titles' || mb_strlen($input) < nc_search::get_setting('SuggestionsMinInputLength')) {
die("[]");
}
$input = $nc_core->utf8->conv($nc_core->NC_CHARSET, 'utf-8', $input);
$language = $nc_core->input->fetch_get('language');
if (!$language) {
$language = $nc_core->lang->detect_lang(1);
}
// поиск подходящих заголовков is provider-dependent
$suggestions = nc_search::get_provider()->suggest_titles($input, $language, $catalogue);
if (!$nc_core->NC_UNICODE) {
$suggestions = $nc_core->utf8->array_utf2win($suggestions);
}
print nc_array_json($suggestions);
<?php
/* $Id: index.php 8366 2012-11-07 16:30:14Z aix $ */
/**
* Запуск переиндексации "в реальном времени"
*/
$NETCAT_FOLDER = realpath("../../../../");
require_once "{$NETCAT_FOLDER}/vars.inc.php";
$use_gzip_compression = false;
require_once "{$ADMIN_FOLDER}/function.inc.php";
require_once "../function.inc.php";
// замедление работы при необходимости
$delay = trim(nc_search::get_setting('IndexerInBrowserSlowdownDelay'));
// секунды
if ($delay) {
define('NC_SEARCH_INDEXER_DELAY_VALUE', (int) ($delay * 1000000));
// микросекунды
function nc_search_indexer_delay()
{
usleep(NC_SEARCH_INDEXER_DELAY_VALUE);
print " ";
}
register_tick_function('nc_search_indexer_delay');
declare (ticks=10000);
}
$input = nc_Core::get_object()->input;
$request = array('.page_title' => NETCAT_MODULE_SEARCH_ADMIN_INDEXING_TITLE, 'view' => 'indexing_on_request', 'rule_id' => $input->fetch_get('rule_id'), 'token' => $input->fetch_get('token'), 'continue' => $input->fetch_get('continue'));
nc_search_admin_controller::process_request($request);
/**
* @param string $string Text to tokenize
* @return array
*/
protected function tokenize_text($string)
{
// split words containing numbers into number+string parts
$string = preg_replace("/(\\pL)(\\d)/u", "\$1 \$2", $string);
$string = preg_replace("/(\\d)(\\pL)/u", "\$1 \$2", $string);
$delimiter = nc_search::should('IgnoreNumbers') ? '/[^\\pL]+/u' : '/[^\\pL\\d]+/u';
$max_terms = (int) nc_search::get_setting('MaxTermsPerField');
$tokens = preg_split($delimiter, $string, $max_terms);
return $tokens;
}
if (isset($_SERVER['REMOTE_ADDR'])) {
die("Access denied.");
}
$NETCAT_FOLDER = realpath(dirname(__FILE__) . "/../../../../");
putenv("DOCUMENT_ROOT={$NETCAT_FOLDER}");
putenv("HTTP_HOST=localhost");
putenv("REQUEST_URI=/");
require_once "{$NETCAT_FOLDER}/vars.inc.php";
require_once $ROOT_FOLDER . "connect_io.php";
$nc_core = nc_Core::get_object();
$nc_core->modules->load_env('ru');
$lang = $nc_core->lang->detect_lang();
require_once $ADMIN_FOLDER . "lang/" . $lang . ".php";
error_reporting(E_PARSE | E_ERROR | E_WARNING | E_USER_ERROR | E_USER_WARNING);
// замедление работы при необходимости
$delay = trim(nc_search::get_setting('IndexerConsoleSlowdownDelay'));
// секунды
if ($delay) {
define('NC_SEARCH_INDEXER_DELAY_VALUE', (int) ($delay * 1000000));
// микросекунды
function nc_search_indexer_delay()
{
usleep(NC_SEARCH_INDEXER_DELAY_VALUE);
}
register_tick_function('nc_search_indexer_delay');
declare (ticks=10000);
}
while (@ob_end_flush()) {
}
// Поменяйте nc_search::LOG_CONSOLE на другое значение, если хотите получать
// больше или меньше информации о переиндексации
require_once $ROOT_FOLDER . "connect_io.php";
$nc_core = nc_Core::get_object();
$nc_core->modules->load_env('ru');
$lang = $nc_core->lang->detect_lang();
require_once $ADMIN_FOLDER . "lang/" . $lang . ".php";
error_reporting(E_PARSE | E_ERROR | E_WARNING | E_USER_ERROR | E_USER_WARNING);
// замедление работы при необходимости
$delay = trim(nc_search::get_setting('IndexerConsoleSlowdownDelay'));
// секунды
if ($delay) {
define('NC_SEARCH_INDEXER_DELAY_VALUE', (int) ($delay * 1000000));
// микросекунды
function nc_search_indexer_delay()
{
usleep(NC_SEARCH_INDEXER_DELAY_VALUE);
}
register_tick_function('nc_search_indexer_delay');
declare (ticks=10000);
}
while (@ob_end_flush()) {
}
nc_search::register_logger(new nc_search_logger_plaintext(nc_search::LOG_CONSOLE));
$remove_hung_tasks = !nc_search::should('IndexerConsoleRestartHungTasks');
$current_task = nc_search_indexer::get_current_task($remove_hung_tasks);
$continue = $current_task instanceof nc_search_indexer_task && $current_task->get('runner_type') == nc_search::INDEXING_CONSOLE_BATCH && ($current_task->get('is_idle') || nc_search::should('IndexerConsoleRestartHungTasks') && time() > $current_task->get('last_activity') + nc_search::get_setting("IndexerRemoveIdleTasksAfter"));
if ($continue) {
$indexer = new nc_search_indexer();
$indexer->resume($current_task, new nc_search_indexer_runner_batch());
} else {
nc_search_scheduler::run(nc_search::INDEXING_CONSOLE_BATCH);
}
} else {
?>
<fieldset>
<legend><?php
echo NETCAT_MODULE_SEARCH_ADMIN_INTERFACE_FORM_TEMPLATES;
?>
</legend>
<textarea class="code"
name="s[SearchFormTemplate]"><?php
echo htmlspecialchars(nc_search::get_setting('SearchFormTemplate'));
?>
</textarea>
</fieldset>
<fieldset>
<legend><?php
echo NETCAT_MODULE_SEARCH_ADMIN_INTERFACE_ADVANCED_FORM_TEMPLATE;
?>
</legend>
<textarea class="code"
name="s[AdvancedSearchFormTemplate]"><?php
echo htmlspecialchars(nc_search::get_setting('AdvancedSearchFormTemplate'));
?>
</textarea>
</fieldset>
<?php
}
?>
</form>
}
nc_search::save_setting($k, $v);
}
// check SearchProvider
if ($search_provider_changed) {
$new_provider = $input['SearchProvider'];
if (@class_exists($new_provider)) {
try {
$provider = new $new_provider();
if ($provider instanceof nc_search_provider) {
$provider->first_run();
} else {
nc_print_status(NETCAT_MODULE_SEARCH_ADMIN_SETTINGS_INCORRECT_PROVIDER_CLASS, 'error', array($new_provider));
}
} catch (Exception $e) {
nc_print_status(NETCAT_MODULE_SEARCH_ADMIN_SETTINGS_PROVIDER_CLASS_INITIALIZATION_ERROR, 'error', array($new_provider, $e->getMessage()));
}
} else {
nc_print_status(NETCAT_MODULE_SEARCH_ADMIN_SETTINGS_PROVIDER_CLASS_NOT_FOUND, 'error', array($new_provider));
}
}
// done saving
nc_print_status(NETCAT_MODULE_SEARCH_ADMIN_SETTINGS_SAVED, 'ok');
}
$settings = array('ComponentID', 'SearchProvider', 'IndexerSecretKey', 'IndexerNormalizeLinks', 'IndexerSaveTaskEveryNthCycle', 'IndexerRemoveIdleTasksAfter', 'IndexerTimeThreshold', 'IndexerMemoryThreshold', 'IndexerConsoleMemoryThreshold', 'IndexerConsoleTimeThreshold', 'IndexerConsoleDocumentsPerSession', 'IndexerConsoleSlowdownDelay', 'IndexerConsoleRestartHungTasks', 'IndexerInBrowserSlowdownDelay', 'MinScheduleInterval', 'CrawlerMaxRedirects', 'NumberOfEntriesPerSitemap', 'MaxTermsPerQuery', 'MaxTermsPerField', 'ZendSearchLucene_MaxBufferedDocs', 'ZendSearchLucene_MaxMergeDocs', 'ZendSearchLucene_MergeFactor', 'PhpMorphy_LoadDictsDuringIndexing', 'DatabaseIndex_LoadAllCodesForIndexing', 'DatabaseIndex_MaxSimilarityCandidates', 'DatabaseIndex_MaxRewriteTerms', 'DatabaseIndex_UseUtf8Levenshtein', 'DatabaseIndex_MaxProximityTerms', 'DatabaseIndex_MaxProximityDistance', 'DatabaseIndex_AlwaysGetTotalCount', 'DatabaseIndex_OptimizationFrequency');
$form_description = array();
foreach ($settings as $s) {
$form_description[$s] = array('type' => 'string', 'caption' => $s, 'value' => nc_search::get_setting($s));
}
$form = new nc_a2f($form_description, "settings");
echo "<form class='settings system_settings' method='POST'>", "<input type='hidden' name='view' value='systemsettings' />", $form->render("<div>", "", "</div>", ""), "</form>";
<legend><?php
echo NETCAT_MODULE_SEARCH_ADMIN_SETTING_QUERY_FEATURES;
?>
</legend>
<div class="setting">
<?php
echo NETCAT_MODULE_SEARCH_ADMIN_SETTING_QUERY_DEFAULT_OPERATOR;
?>
:
<select name="s[DefaultBooleanOperator]">
<option value="AND"><?php
echo NETCAT_MODULE_SEARCH_ADMIN_SETTING_QUERY_DEFAULT_OPERATOR_AND;
?>
</option>
<option value="OR"<?php
echo nc_search::get_setting('DefaultBooleanOperator') == 'OR' ? ' selected' : '';
?>
>
<?php
echo NETCAT_MODULE_SEARCH_ADMIN_SETTING_QUERY_DEFAULT_OPERATOR_OR;
?>
</option>
</select>
</div>
<?php
echo $this->setting_cb('AllowTermBoost', NETCAT_MODULE_SEARCH_ADMIN_SETTING_QUERY_TERM_BOOST);
?>
<?php
echo $this->setting_cb('AllowProximitySearch', NETCAT_MODULE_SEARCH_ADMIN_SETTING_QUERY_PROXIMITY_SEARCH);
?>
<?php
/**
* Логгер может иметь собственный уровень сообщений об ошибках.
* @param int $level
*/
public function __construct($level = null)
{
$this->level = $level ? $level : nc_search::get_setting('LogLevel');
}
/**
*
*/
protected function get_disallowed_areas()
{
$disallowed = array();
// (1) robots.txt
if (nc_search::should('CrawlerObeyRobotsTxt')) {
$disallowed = $this->get_robots_txt_area_parts();
}
// (2) Settings (ExcludeUrlRegexps)
$regexps = preg_split("/\\s*\n/u", nc_search::get_setting('ExcludeUrlRegexps'), -1, PREG_SPLIT_NO_EMPTY);
foreach ($regexps as $regexp) {
$regexp = "@" . addcslashes($regexp, "@") . "@u";
$disallowed[] = new nc_search_area_regexp(array('regexp' => $regexp));
}
// done
return new nc_search_area($disallowed);
}
protected function get_max_cycles_number()
{
return (int) nc_search::get_setting('IndexerConsoleDocumentsPerSession');
}
header("Content-type: text/xml");
//$NETCAT_FOLDER = realpath("../../../");
$NETCAT_FOLDER = join(strstr(__FILE__, "/") ? "/" : "\\", array_slice(preg_split("/[\\/\\\\]+/", __FILE__), 0, -4)) . (strstr(__FILE__, "/") ? "/" : "\\");
require_once "{$NETCAT_FOLDER}/vars.inc.php";
//require ($INCLUDE_FOLDER."index.php");
require $ROOT_FOLDER . "connect_io.php";
$nc_core->modules->load_env();
print '<?xml version="1.0" encoding="UTF-8"?>';
// bark before the cat tries to meow
$scheme = isset($_SERVER['HTTPS']) && $_SERVER['HTTPS'] != 'off' ? 'https' : 'http';
$url_prefix = "{$scheme}://{$_SERVER['HTTP_HOST']}";
$site = $nc_core->catalogue->get_by_host_name($_SERVER['HTTP_HOST']);
// never trust a cat
$site_id = $site['Catalogue_ID'];
$start = $nc_core->input->fetch_get("start");
$max_num_urls = nc_search::get_setting('NumberOfEntriesPerSitemap');
if (!strlen($start)) {
// если результатов слишком много, выдать sitemapindex
$num_urls = $db->get_var("SELECT COUNT(*)\n FROM `Search_Document` \n WHERE `Catalogue_ID` = {$site_id} AND `IncludeInSitemap`=1");
if ($num_urls > $max_num_urls) {
$url = "{$url_prefix}{$_SERVER['REQUEST_URI']}?start=";
print '<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n";
for ($i = 0, $last = ceil($num_urls / $max_num_urls); $i < $last; $i++) {
print "<sitemap><loc>" . $url . $i * $max_num_urls . "</loc></sitemap>\n";
}
print "</sitemapindex>\n";
die;
}
}
$start = (int) $start;
$entries = $db->get_results("SELECT `Path`, \n `SitemapChangefreq`, \n `SitemapPriority`,\n DATE_FORMAT(`LastModified`, '%Y-%m-%dT%T') AS `LastModified`\n FROM `Search_Document`\n WHERE `Catalogue_ID` = {$site_id} AND `IncludeInSitemap`=1\n LIMIT {$max_num_urls} OFFSET {$start}", ARRAY_A);
<?php
/* $Id: netcat_cron.php 8456 2012-11-23 10:42:55Z aix $ */
/**
* Запуск из "крона" неткета
*/
$NETCAT_FOLDER = realpath(dirname(__FILE__) . "/../../../../");
require_once "{$NETCAT_FOLDER}/vars.inc.php";
require_once $ROOT_FOLDER . "connect_io.php";
$nc_core = nc_Core::get_object();
$nc_core->modules->load_env('ru');
$lang = $nc_core->lang->detect_lang();
require_once $ADMIN_FOLDER . "lang/" . $lang . ".php";
error_reporting(E_PARSE | E_ERROR | E_WARNING | E_USER_ERROR | E_USER_WARNING);
while (@ob_end_flush()) {
}
$secret_key = nc_Core::get_object()->input->fetch_get("secret_key");
if ($secret_key != nc_search::get_setting('IndexerSecretKey')) {
$file = __FILE__;
nc_search::log(nc_search::LOG_ERROR, "Attempt to access '{$file}' with a wrong secret key '{$secret_key}' from {$_SERVER['REMOTE_ADDR']}");
die("Access denied.");
}
nc_search::register_logger(new nc_search_logger_plaintext());
nc_search_scheduler::run(nc_search::INDEXING_NC_CRON);
/**
*
* @param nc_search_indexer $indexer
* @throws nc_search_exception
* @return boolean is task finished
*/
public function loop(nc_search_indexer $indexer)
{
$cycle_number = 0;
$save_cycles = nc_search::get_setting('IndexerSaveTaskEveryNthCycle');
while (true) {
// stop prematurely:
if (!$this->check_connection()) {
$indexer->cancel();
return true;
// nobody listens anyway
}
if ($this->interrupt_if_needed($indexer, $cycle_number)) {
return false;
}
// сохранять задачу каждые X циклов
if ($cycle_number % $save_cycles == 0) {
$indexer->save_task();
}
switch ($indexer->next()) {
case nc_search_indexer::TASK_FINISHED:
return true;
// we're done
// we're done
case nc_search_indexer::TASK_STEP_FINISHED:
if ($this->delay) {
if ($this->interrupt_if_needed($indexer, $cycle_number)) {
return false;
}
sleep($this->delay);
}
break;
case nc_search_indexer::TASK_STEP_SKIPPED:
break;
default:
throw new nc_search_exception("Incorrect return value from nc_search_indexer::next()");
}
$cycle_number++;
}
}
/**
* @param string $term1
* @param float $min_similarity
* @return array of similar term codes (array("____") if no similar terms were found)
*/
protected function get_similar_terms($term1, $min_similarity)
{
$max_candidates = (int) nc_search::get_setting("DatabaseIndex_MaxSimilarityCandidates");
$max_results = (int) nc_search::get_setting("DatabaseIndex_MaxRewriteTerms");
$use_utf_levenshtein = (bool) nc_search::get_setting("DatabaseIndex_UseUtf8Levenshtein");
$term_length = mb_strlen($term1, 'UTF-8');
$max_distance = intval((1 - $min_similarity) * $term_length);
// == floor()
$min_length = $term_length - $max_distance;
$max_length = $term_length + $max_distance;
// проверять совпадение в PHP до 10 раз быстрее, чем делать это хранимой
// функцией в MySQL
$query = "SELECT `Term`, `Code`\n FROM `{$this->term_table_name}`\n WHERE `Length` BETWEEN {$min_length} AND {$max_length}\n LIMIT {$max_candidates}";
$terms = $this->get_db()->get_results($query, ARRAY_A);
$similar = array();
if ($terms) {
foreach ($terms as $row) {
// Функция levenshtein() не UTF-8-aware и производит неправильные
// результаты в случае, если есть замена однобайтовой буквы на
// многобайтовую, например levenshtein("Z", "Я") == 2, а не 1.
// Но всё же используется именно эта функция, поскольку она более чем
// в два раза быстрее кода на PHP, а в этом цикле может обрабатываться
// большое количество ($this->max_similarity_candidates) терминов
$distance = $use_utf_levenshtein ? $this->levenshtein_utf8($term1, $row['Term']) : levenshtein($term1, $row['Term']);
$terms_similarity = 1 - $distance / min($term_length, mb_strlen($row['Term'], 'UTF-8'));
if ($terms_similarity >= $min_similarity) {
$similar[] = $row['Code'];
}
if (sizeof($similar) >= $max_results) {
break;
}
}
}
if (!sizeof($similar)) {
$similar[] = "____";
}
// haven't found any similar terms!
return $similar;
}
/**
* Получить путь до раздела поиска на сайте с указанным идентификатором.
* @global nc_db $db
* @throws Exception @see nc_catalogue::get_by_id()
* @param integer $site_id
* @param boolean $with_host
* @return string
*/
public function get_search_url($site_id, $with_host = true)
{
if ($with_host && isset($this->paths[$site_id])) {
return $this->paths[$site_id];
}
global $db, $nc_core;
$folder_data = $db->get_row("SELECT sub.`Hidden_URL` AS `path`, sub.`Subdivision_ID` AS `id`\n FROM `Subdivision` AS `sub`, `Sub_Class` AS `c`\n WHERE c.`Class_ID` = " . (int) nc_search::get_setting("ComponentID") . "\n AND c.`Subdivision_ID` = sub.`Subdivision_ID`\n AND sub.`Catalogue_ID` = " . (int) $site_id . "\n LIMIT 1", ARRAY_A);
if ($folder_data) {
if (nc_module_check_by_keyword('routing')) {
$path = nc_routing::get_folder_path($folder_data['id']);
} else {
$path = $nc_core->SUB_FOLDER . $folder_data['path'];
}
if ($with_host) {
$host = $nc_core->catalogue->get_by_id($site_id, 'Domain');
if ($host) {
$path = "http://{$host}{$path}";
}
}
$this->paths[$site_id] = $path;
} else {
$this->paths[$site_id] = false;
}
return $this->paths[$site_id];
}
/**
* @param string $option
* @return mixed
*/
public function get($option)
{
if ($option == 'saved_content' && ($max_length = nc_search::get_setting('MaxDocumentPreviewTextLengthInKbytes'))) {
$max_length *= 1024;
$content = $this->properties['intact_content'];
if (strlen($content) > $max_length) {
$content = substr($content, 0, strrpos($content, " ", $max_length - strlen($content)));
}
return $content;
}
return parent::get($option);
}
/**
* Работает ли в данный момент переиндексация?
* @param bool $remove_hung_tasks
* @return false|nc_search_indexer_task
*/
public static function get_current_task($remove_hung_tasks = true)
{
$tasks = nc_search::load_all('nc_search_indexer_task', true);
if (!sizeof($tasks)) {
return false;
}
// не подвисли ли мы?
$task = $tasks->first();
if ($remove_hung_tasks && time() > $task->get('last_activity') + nc_search::get_setting("IndexerRemoveIdleTasksAfter")) {
$task->delete();
$db = nc_Core::get_object()->db;
$db->query("TRUNCATE TABLE `Search_Link`");
$db->query("TRUNCATE TABLE `Search_LinkReferrer`");
nc_search::log(nc_search::LOG_ERROR, "Indexer task was last active at " . strftime("%Y-%m-%d %H:%M:%S", (int) $task->get('last_activity')) . ". Task removed.");
return false;
}
return $task;
}
/**
* @param string $input
* @param string $language
* @param integer $site_id
* @return array
*/
public function suggest_titles($input, $language, $site_id)
{
$suggestions = array();
// собственно подсказки
$titles = array();
$limit = nc_search::get_setting('NumberOfSuggestions');
// поиск в индексе (то есть будут варианты после обработки фильтрами - базовая форма)
if (nc_search::should('SearchTitleBaseformsForSuggestions')) {
$last_space = strrpos($input, " ");
$as_phrase = nc_search::should('SearchTitleAsPhraseForSuggestions');
$b1 = $as_phrase ? '"' : '(';
$b2 = $as_phrase ? '"' : ')';
/* @todo сделать проверку на то, что последнее слово является правильным/полным? */
$query_string = "(title:{$b1}{$input}{$b2}" . ($last_space ? " OR title:{$b1}" . trim(substr($input, 0, $last_space)) . $b2 : '') . ") AND site_id:{$site_id}";
$query = new nc_search_query($query_string);
$query->set('limit', $limit)->set('options_to_fetch', array('title', 'site_id', 'path'))->set('language', $language);
$documents = $this->find($query, false);
foreach ($documents as $doc) {
$suggestions[] = array("label" => $doc->get('title'), "url" => $doc->get('url'));
$titles[] = '"' . nc_search_util::db_escape($doc->get('title')) . '"';
}
$titles = array_unique($titles);
}
// поиск точного соответствия в таблице с документами
// по-хорошему следовало бы сначала сделать запрос к БД, а потом к индексу, однако
// в случае запроса к индексу не получится так же просто отфильтровать уже совпавшие запросы
$query = "SELECT `Catalogue_ID`, `Path`, `Title` FROM `%t%` " . ' WHERE `Title` LIKE "' . nc_search_util::db_escape($input) . '%" ' . ($titles ? " AND `Title` NOT IN (" . join(", ", $titles) . ") " : "") . " ORDER BY `Title` " . " LIMIT {$limit}";
$documents = new nc_search_result();
$documents->select_from_database($query);
foreach ($documents as $doc) {
array_unshift($suggestions, array("label" => $doc->get('title'), "url" => $doc->get('url')));
}
$suggestions = array_slice($suggestions, 0, $limit);
return $suggestions;
}
