如何在 postgresql 中创建 n-gram

我希望在我的应用程序中具有搜索功能。

我使用 trigram 来实现它，并且工作正常。

问题是：

trigram 正在创建单词的 3 个字符组的序列。 我想要在单个对象中包含超过 3 个字符。 例如：

select show_trgm('abcpqrs');

这将返回：{" a"," ab","abc","bcp","cpq","pqr","qrs","rs "}

I want {" a"," ab","abc","bcp","cpq","pqr","qrs","rs ","abcd","bcpq","cpqr"...}

我怎样才能实现这个目标？

试试这个功能。

CREATE OR REPLACE FUNCTION myngram(mystr TEXT, n INT) RETURNS TEXT[]
AS $$
DECLARE
str VARCHAR;
arr TEXT[];
BEGIN

  str := lpad(mystr, n - 1 + char_length(mystr), ' ');
  str := rpad(str, n - 1 + char_length(str), ' ');

  arr := array[]::TEXT[];
  FOR i IN 1 .. char_length(str) - n + 1 LOOP
    arr := arr || substring(str from i for n);
  END LOOP;
  RETURN arr;
END
$$
LANGUAGE plpgsql;

我的测试结果。

testdb=# SELECT myngram('abcpqrs', 4);
                             myngram                             
-----------------------------------------------------------------
 {"   a","  ab"," abc",abcp,bcpq,cpqr,pqrs,"qrs ","rs  ","s   "}
(1 row)

testdb=# SELECT myngram('abcpqrs', 5);
                                       myngram                                       
-------------------------------------------------------------------------------------
 {"    a","   ab","  abc"," abcp",abcpq,bcpqr,cpqrs,"pqrs ","qrs  ","rs   ","s    "}
(1 row)