How to get the first letter of Chinese pinyin in PostgreSQL database sql query

I. Introduction

In the actual development process, fuzzy queries are often used. Fuzzy searches are based on a certain keyword, usually name like '%123%'. But what if the user does not remember a certain keyword and only knows the first letter? What about fuzzy queries? For example, if you query name like '%I am Chinese%', how can you perform a fuzzy search based on the initials wszgr?

2. Create a custom function

Because the system's built-in like query can no longer satisfy the fuzzy search in this scenario, a custom function needs to be used here to convert the Chinese characters in a certain field in the database into the first letter, and then carry out fuzzy search based on the letters.

Create a custom function CnFirstChar to convert Chinese strings into first letters.


CREATE OR REPLACE FUNCTION CnFirstChar(s character varying)
  RETURNS character varying AS
$BODY$
declare
  retval character varying;
  c character varying;
  l integer;
  b bytea;  
  w integer;
begin
l=length(s);
retval='';
while l>0 loop
  c=left(s,1);
  b=convert_to(c,'GB18030')::bytea;
  if get_byte(b,0)<127 then
    retval=retval || upper(c);
  elsif length(b)=2 then
    begin
    w=get_byte(b,0)*256+get_byte(b,1);
    --汉字GBK编码按拼音排序，按字符数来查找，基于概率来说，效率应该比这个强:)
    if w between 45217 and 45252 then --"A";45217;45252;35
      retval=retval || 'a';
	elsif w between 45253 and 45760 then --"B";45253;45760;507
      retval=retval || 'b';
	elsif w between 45761 and 46317 then --"C";45761;46317;556
      retval=retval || 'c';
	elsif w between 46318 and 46825 then --"D";46318;46825;507
      retval=retval || 'd';
	elsif w between 46826 and 47009 then--"E";46826;47009;183
      retval=retval || 'e';
	elsif w between 47010 and 47296 then--"F";47010;47296;286
      retval=retval || 'f';
	elsif w between 47297 and 47613 then --"G";47297;47613;316
      retval=retval || 'g';
	elsif w between 47614 and 48118 then --"H";47614;48118;504
      retval=retval || 'h';
	elsif w between 48119 and 49061 then --"J";48119;49061;942
      retval=retval || 'j';
	elsif w between 49062 and 49323 then--"K";49062;49323;261
      retval=retval || 'k';
	elsif w between 49324 and 49895 then --"L";49324;49895;571
      retval=retval || 'l';
	elsif w between 49896 and 50370 then --"M";49896;50370;474
      retval=retval || 'm';
	elsif w between 50371 and 50613 then --"N";50371;50613;242
      retval=retval || 'n';
    elsif w between 50614 and 50621 then --"O";50614;50621;7
      retval=retval || 'o';
	elsif w between 50622 and 50905 then--"P";50622;50905;283
      retval=retval || 'p';
	elsif w between 50906 and 51386 then --"Q";50906;51386;480
      retval=retval || 'q';
	elsif w between 51387 and 51445 then--"R";51387;51445;58
      retval=retval || 'r';
	elsif w between 51446 and 52216 then --"S";51446;52217;
      retval=retval || 's';
	elsif w between 52218 and 52697 then --"T";52218;52697;479
      retval=retval || 't';
	elsif w between 52698 and 52979 then--"W";52698;52979;281
      retval=retval || 'w';
	elsif w between 52980 and 53688 then --"X";52980;53688;
      retval=retval || 'x';
	elsif w between 53689 and 54480 then --"Y";53689;54480;791
      retval=retval || 'y';
	elsif w between 54481 and 55289 then --"Z";54481;55289;808
      retval=retval || 'z';
    end if;
    end;
  end if;
  s=substring(s,2,l-1);
  l=l-1;
end loop;
return retval;
end;
$BODY$
  LANGUAGE plpgsql IMMUTABLE;

3. SQL fuzzy search uses custom functions

The above custom functions have been created in the postgresql database and can be used directly during SQL queries.

Insert image description here


select * from user  where CnFirstChar(user_name) like '%wszgr%';