分享一个PHP中文分词

3. /**
4. * 中文分词处理方法
5. *+---------------------------------
6. * @param stirng $string 要处理的字符串
7. * @param boolers $sort=false 根据value进行倒序
8. * @param Numbers $top=0 返回指定数量，默认返回全部
9. *+---------------------------------
10. * @return void
11. */
12. function scws($text, $top = 5, $return_array = false, $sep = ',') {
13. include('./pscws4/pscws4.php');//去下面给的网址把pscws4下载下来
14. $cws = new pscws4('utf-8');
15. $cws -> set_charset('utf-8');
16. $cws -> set_dict('./pscws4/etc/dict.utf8.xdb');
17. $cws -> set_rule('./pscws4/etc/rules.utf8.ini');
18. //$cws->set_multi(3);
19. $cws -> set_ignore(true);
20. //$cws->set_debug(true);
21. //$cws->set_duality(true);
22. $cws -> send_text($text);
23. $ret = $cws -> get_tops($top, 'r,v,p');
24. $result = null;
25. foreach ($ret as $value) {
26. if (false === $return_array) {
27. $result .= $sep . $value['word'];
28. } else {
29. $result[] = $value['word'];
30. }
31. }
32. return false === $return_array ? substr($result, 1) : $result;
33. }
34. print_r(scws('青花夔龙纹香炉'));