while ($result = $scws->get_result()) {
foreach ($result as $tmp) {
$words[] = $tmp['word'];
}
}
$result2 = implode(' ', $words);
if ($result2 === $data['result']) {
$success++;
echo "PASS!\n";
} else {
$failure++;
echo "FAILURE!\n";
echo "----------------------------------------\n";
echo "ORGINAL TEXT: {$text}\n";
echo "EXPECTED RESULT: {$data['result']}\n";
echo "ACTUAL RESULT: {$result2}\n";
echo "========================================\n";
}
$start++;
}
$scws->close();
//5. show result report
$start -= 1;
echo "// -------------------------------------\n";
echo "// TEST result report\n";
echo "// " . scws_version() . "\n";
echo "// -------------------------------------\n";
echo "// Total test: {$start}\n";
printf("// Passed Num: %d (%.2f%%)\n", $success, 100 * $success / $start);
printf("// Failed Num: %d (%.2f%%)\n", $failure, 100 * $failure / $start);
echo "// -------------------------------------\n";
} else {
if ($showa) {
printf("%s/%s ", $tmp['word'], $tmp['attr']);
} else {
printf("%s ", $tmp['word']);
}
}
}
flush();
}
}
$cws->close();
$time_end = get_microtime();
$time = $time_end - $time_start;
?>
</textarea>
<small>
分词耗时: <?php
echo $time;
?>
秒
<a href="../">返回scws主页</a>或直接<a href="?source" target="_blank">查看源码</a> Powered by <?php
echo scws_version();
?>
</small>
</td>
</tr>
</table>
</body>
</html>
//
set_time_limit(0);
error_reporting(0);
ini_set('display_errors', '0');
// show source
if (isset($_SERVER['QUERY_STRING']) && !strncasecmp($_SERVER['QUERY_STRING'], 'source', 6)) {
header("Content-Type: text/html; charset=utf-8");
highlight_file(__FILE__);
exit(0);
}
// get data
if (!isset($_POST['data'])) {
$charset = 'utf-8';
$respond = 'plain';
$apiurl = 'http://' . $_SERVER['HTTP_HOST'] . $_SERVER['SCRIPT_NAME'];
$version2 = scws_version();
include 'header.inc.php';
echo <<<__EOF__
<div class="block">
<h2>SCWS(简易中文分词) 基于HTTP/POST的分词 API 使用说明</h2>
<ol>
<li>API 地址：<a href="{$apiurl}">{$apiurl}</a></li>
<li>
\t请求方式：仅支持 POST，推荐采用纯 PHP 实现的 <a href="http://www.hightman.cn/bbs/showthread.php?tid=838" target="_blank">HTTP client 库</a>
</li>
<li>请求的参数变量及含义：
\t<ul>
\t\t<li><em>data</em> 需要分词的字符串(*必须*)</li>
\t\t<li><em>respond</em> 响应结果格式(其值为: php/json/xml, 默认为 php，其中 php是指用php序列化后的结果)</li>
\t\t<li><em>charset</em> 待分词的字符串编码(gbk/utf8，默认是utf8)</li>
\t\t<li><em>ignore</em> 是否忽略标点符号(yes/no，默认为 no)</li>