/**
* @desc 检测字典文件
*/
function checkDictionary()
{
global $db, $score;
require_once R_P . "lib/filterutil.class.php";
$classid = getCloseClass();
$db_dictlist = array();
$dict_folder = D_P . 'data/bbscache/';
//读取所有敏感词时间戳
$sql = "SELECT wordtime FROM pw_wordfb WHERE classid NOT IN ({$classid}) GROUP BY wordtime";
$query = $db->query($sql);
while ($rt = $db->fetch_array($query)) {
//字典名称
$title = $rt['wordtime'];
$db_dictlist[] = $title;
$bin_path = 'dict_' . $title . '.php';
$source_path = 'dict_' . $title . '.txt';
$bin_file = $dict_folder . $bin_path;
$source_file = $dict_folder . $source_path;
//查询该时间戳是否有字典,如果没有就插入记录
$sql = "SELECT id FROM pw_filter_dictionary WHERE title =" . $title;
$dictionary = $db->get_value($sql);
if (!$dictionary) {
//处理数据
$value = array('title' => $title, 'bin' => $bin_path, 'source' => $source_path);
$value = pwSqlSingle($value);
//插入记录
$sql = "INSERT INTO pw_filter_dictionary SET {$value} ";
$db->update($sql);
}
//查找字典所包含的敏感词
$querys = $db->query("SELECT word, type FROM pw_wordfb WHERE classid NOT IN ({$classid}) AND wordtime =" . pwEscape($title));
$content = "";
while ($value = $db->fetch_array($querys)) {
$weighing = $score[$value['type']];
$content .= "" . $value['word'] . "|" . $weighing . "\r\n";
}
//更新字典
if ($content) {
writeover($source_file, $content);
//文本形式字典
writeover($bin_file, '');
//二进制字典
//更新二进制字典
$trie = new Trie();
$trie->build($source_file, $bin_file);
}
}
foreach ($db_dictlist as $key => $value) {
$dict_str .= $dict_str ? ", " . pwEscape($value) : pwEscape($value);
}
if (!$dict_str) {
$dict_str = -1;
}
//删除多余字典记录
$sql = "DELETE FROM pw_filter_dictionary WHERE title NOT IN ({$dict_str})";
$db->update($sql);
$files = glob($dict_folder . 'dict_*.php');
$db_dictlist[] = 'all';
$db_dictlist = array_flip($db_dictlist);
foreach ($files as $value) {
$title = substr($value, strpos($value, '_') + 1);
$title = substr($title, 0, strpos($title, '.'));
if (!array_key_exists($title, $db_dictlist)) {
$bin_path = 'dict_' . $title . '.php';
$source_path = 'dict_' . $title . '.txt';
$bin_file = $dict_folder . $bin_path;
$source_file = $dict_folder . $source_path;
//删除多余文件
deldir($bin_file);
deldir($source_file);
}
}
}
/**
* @desc 生成总字典文件
*/
function setAllDictionary()
{
global $db, $score;
L::loadClass('filterutil', 'filter', false);
$bin_file = D_P . 'data/bbscache/dict_all.dat';
$source_file = D_P . 'data/bbscache/dict_all.txt';
if (!file_exists($bin_file) && !file_exists($source_file)) {
pwCache::setData($source_file, '');
//文本形式字典
pwCache::setData($bin_file, '');
//二进制字典
}
$classid = getCloseClass();
$classid = S::sqlImplode($classid);
$querys = $db->query("SELECT word, type, wordreplace FROM pw_wordfb WHERE classid NOT IN ({$classid})");
$content = "";
while ($value = $db->fetch_array($querys)) {
$weighing = $score[$value['type']];
$wordreplace = $value['type'] == 3 ? '|' . $value['wordreplace'] : '';
$value['word'] = strtolower($value['word']);
$content .= "" . $value['word'] . "|" . $weighing . $wordreplace . "\r\n";
}
pwCache::setData($source_file, $content);
//文本形式字典
pwCache::setData($bin_file, '');
//二进制字典
//更新二进制字典
$trie = new Trie();
$trie->build($source_file, $bin_file);
}
/**
* 构建字典
* @param $path 序列化后字典存放路径
* @return $return int 构造成功后返回的行数
*/
function buildDict($path = null)
{
if ($path == null) {
$path = array('bin' => $this->dict_bin_path, 'source' => $this->dict_source_path);
}
$trie = new Trie($path);
$return = $trie->build();
return $return;
}