/**
* 把用户输入的字符串转换为 UTF-8 编码
* @param string $buf 要转换字符串
* @return string 转换后的字符串
*/
public static function convertIn($buf)
{
if (self::$charset !== null) {
return XS::convert($buf, 'UTF-8', self::$charset);
}
return $buf;
}
public function testSetCharset()
{
$text = '全文检索';
$tk = $this->object;
$words = $tk->getResult($text);
$this->assertEquals($words[0]['word'], $text);
$gbk = XS::convert($text, 'GBK', 'UTF-8');
$tk->setCharset('GBK');
$words = $tk->getResult($gbk);
$this->assertEquals($words[0]['word'], $gbk);
}
$input = dirname(__FILE__) . '/skel';
$dir = dir($input);
while (($entry = $dir->read()) !== false) {
if ($entry === '.' || $entry === '..') {
continue;
}
if (substr($entry, -3) === '.in') {
echo "正在生成 " . substr($entry, 0, -3) . " ...\n";
$file = $output . '/' . substr($entry, 0, -3);
if (file_exists($file)) {
copy($file, $file . '.bak');
}
$content = file_get_contents($input . '/' . $entry);
$content = strtr($content, $vars);
if ($vars['@charset@'] !== 'UTF-8') {
$content = XS::convert($content, $vars['@charset@'], 'UTF-8');
}
file_put_contents($file, $content);
} else {
echo "正在复制 " . $entry . " ...\n";
$file = $output . '/' . $entry;
if (is_dir($input . '/' . $entry)) {
XSUtil::copyDir($input . '/' . $entry, $file);
} else {
if (file_exists($file)) {
copy($file, $file . '.bak');
}
copy($input . '/' . $entry, $file);
}
}
}
function xs_output_encoding($buf)
{
return XS::convert($buf, $GLOBALS['oe'], 'UTF-8');
}
public static function setUpBeforeClass()
{
self::$data = array('pid' => 1234, 'subject' => "Hello, 测试标题", 'message' => "您好，这儿是真正的测试内容\n另起一行用英文\n\nHello, the world!", 'chrono' => time());
self::$data_gbk = XS::convert(self::$data, 'GBK', 'UTF-8');
}
public function testConvert()
{
$str = 'english string only';
$this->assertEquals(XS::convert($str, 'GBK', 'UTF-8'), $str);
$str = 'english 和中文混合';
$str1 = XS::convert($str, 'GBK', 'UTF-8');
$this->assertEquals(XS::convert($str1, 'UTF-8', 'GBK'), $str);
}
/**
* 初始始化高亮替换数据
*/
private function initHighlight()
{
$terms = array();
$tmps = $this->terms($this->_highlight, false);
for ($i = 0; $i < count($tmps); $i++) {
if (strlen($tmps[$i]) !== 6 || ord(substr($tmps[$i], 0, 1)) < 0xc0) {
$terms[] = XS::convert($tmps[$i], $this->_charset, 'UTF-8');
continue;
}
// auto fixed duality in libscws
// ABC => AB,BC => ABC,BC,AB
// ABCD => AB,BC,CD => CD,ABC,BC,AB
// ABCDE => AB,BC,CD,DE => CDE,DE,CD,ABC,BC,AB
for ($j = $i + 1; $j < count($tmps); $j++) {
if (strlen($tmps[$j]) !== 6 || substr($tmps[$j], 0, 3) !== substr($tmps[$j - 1], 3, 3)) {
break;
}
}
if (($k = $j - $i) === 1) {
$terms[] = XS::convert($tmps[$i], $this->_charset, 'UTF-8');
} else {
$i = $j - 1;
while ($k--) {
$j--;
if ($k & 1) {
$terms[] = XS::convert(substr($tmps[$j - 1], 0, 3) . $tmps[$j], $this->_charset, 'UTF-8');
}
$terms[] = XS::convert($tmps[$j], $this->_charset, 'UTF-8');
}
}
}
$pattern = $replace = $pairs = array();
foreach ($terms as $term) {
if (!preg_match('/[a-zA-Z]/', $term)) {
$pairs[$term] = '<em>' . $term . '</em>';
} else {
$pattern[] = '/' . strtr($term, array('+' => '\\+', '/' => '\\/')) . '/i';
$replace[] = '<em>$0</em>';
}
}
$this->_highlight = array();
if (count($pairs) > 0) {
$this->_highlight['pairs'] = $pairs;
}
if (count($pattern) > 0) {
$this->_highlight['pattern'] = $pattern;
$this->_highlight['replace'] = $replace;
}
}
public function del($term, $field = null)
{
$field = $field === null ? $this->xs->getFieldId() : $this->xs->getField($field);
$cmds = array();
$terms = is_array($term) ? array_unique($term) : array($term);
$terms = XS::convert($terms, 'UTF-8', $this->xs->getDefaultCharset());
foreach ($terms as $term) {
$cmds[] = new XSCommand(XS_CMD_INDEX_REMOVE, 0, $field->vno, strtolower($term));
}
if ($this->_bufSize > 0) {
$this->appendBuffer(implode('', $cmds));
} elseif (count($cmds) == 1) {
$this->execCommand($cmds[0], XS_CMD_OK_RQST_FINISHED);
} else {
$cmd = array('cmd' => XS_CMD_INDEX_EXDATA, 'buf' => implode('', $cmds));
$this->execCommand($cmd, XS_CMD_OK_RQST_FINISHED);
}
return $this;
}
private function applySetting($text)
{
self::$_server->reopen();
foreach ($this->_setting as $key => $cmd) {
self::$_server->execCommand($cmd);
}
return XS::convert($text, 'UTF-8', $this->_charset);
}
/**
* 智能字符集编码转换
* 将 XS 内部用的 UTF-8 与指定的文档编码按需相互转换
* 索引文档: ... -> UTF-8, 搜索结果文档: ... <-- UTF-8
* @param string $value 要转换的字符串
* @return string 转好的字符串
* @see setCharset
*/
private function autoConvert($value)
{
// Is the value need to convert
if ($this->_charset === null || $this->_charset == 'UTF-8' || !is_string($value) || !preg_match('/[\\x81-\\xfe]/', $value)) {
return $value;
}
// _meta === null ? index document : search result document
$from = $this->_meta === null ? $this->_charset : 'UTF-8';
$to = $this->_meta === null ? 'UTF-8' : $this->_charset;
return XS::convert($value, $to, $from);
}
public function testGetAddIndex()
{
$this->assertNull($this->doc2->getAddIndex('subject2'));
$this->assertNull($this->doc2->getAddIndex('subject'));
$subject = new XSFieldMeta('subject');
$this->doc2->addIndex('subject', 'hello the world');
$this->doc2->addIndex($subject, XS::convert('您好世界', 'GBK', 'UTF-8'));
$this->assertEquals(self::$data['subject'], $this->doc2->subject);
$this->assertEquals("hello the world\n您好世界", $this->doc2->getAddIndex($subject));
}
private function applySetting($text)
{
self::$_server->reopen();
foreach ($this->_setting as $key => $cmd) {
if (is_array($cmd)) {
foreach ($cmd as $_cmd) {
self::$_server->execCommand($_cmd);
}
} else {
self::$_server->execCommand($cmd);
}
}
return XS::convert($text, 'UTF-8', self::$_charset);
}
private function autoConvert($value)
{
if ($this->_charset === null || $this->_charset == 'UTF-8' || !is_string($value) || !preg_match('/[\\x81-\\xfe]/', $value)) {
return $value;
}
$from = $this->_meta === null ? $this->_charset : 'UTF-8';
$to = $this->_meta === null ? 'UTF-8' : $this->_charset;
return XS::convert($value, $to, $from);
}
public function testCharset()
{
$xs = self::$xs;
$search = $xs->search;
$query = 'subject:测试';
$docs = $search->search($query);
$this->assertEquals(3, count($docs));
$this->assertEquals('测试第二篇', $docs[0]->subject);
$this->assertEquals('GBK', $xs->getDefaultCharset());
$search->setCharset($xs->getDefaultCharset());
$this->assertEquals(0, $search->count($query));
$docs = $search->search(XS::convert($query, 'GBK', 'UTF-8'));
$this->assertEquals(3, count($docs));
$this->assertEquals(XS::convert('测试第二篇', 'GBK', 'UTF-8'), $docs[0]->subject);
}
