KingbaseES 全文检索功能介绍
KingbaseES 内置的缺省的分词解析器采用空格分词，因为中文的词语之间没有空格分割，所以这种方法并不适用于中文。要支持中文的全文检索需要额外的中文分词插件：zhparser and sys_jieba，其中zhparser 支持 GBK 和 UTF8 字符集，sys_jieba 支持 UTF8 字符集。
一、默认空格分词
1、tsvector
test=# SELECT to_tsvector('English','Try not to become a man of success, but rather try to become a man of value'); to_tsvector ---------------------------------------------------------------------- 'becom':4,13 'man':6,15 'rather':10 'success':8 'tri':1,11 'valu':17 (1 row) test=# SELECT to_tsvector('simple','Try not to become a man of success, but rather try to become a man of value'); to_tsvector --------------------------------------------------------------------------------------------------------------------- 'a':5,14 'become':4,13 'but':9 'man':6,15 'not':2 'of':7,16 'rather':10 'success':8 'to':3,12 'try':1,11 'value':17 (1 row) test=# SELECT to_tsvector('Try not to become a man of success, but rather try to become a man of value'); to_tsvector --------------------------------------------------------------------------------------------------------------------- 'a':5,14 'become':4,13 'but':9 'man':6,15 'not':2 'of':7,16 'rather':10 'success':8 'to':3,12 'try':1,11 'value':17 (1 row)
这里可以看到，如果词干分析器是english ，会采取词干标准化的过程；而simple 只是转换成小写。默认是 simple。
test=# show default_text_search_config; default_text_search_config ---------------------------- pg_catalog.simple (1 row)
2、标准化过程
标准化过程会完成以下操作：
- 总是把大写字母换成小写的
- 也经常移除后缀（比如英语中的s,es和ing等），这样可以搜索同一个字的各种变体，而不是乏味地输入所有可能的变体。
- 数字表示词位在原始字符串中的位置，比如“man"出现在第6和15的位置上。
- to_tesvetor的默认配置的文本搜索是“英语“。它会忽略掉英语中的停用词（stopword，译注：也就是am is are a an等单词)。
3、tsvector搜索
test=# SELECT to_tsvector('Try not to become a man of success, but rather try to become a man of value') @@ 'become'; ?column? ---------- t (1 row) test=# SELECT to_tsvector('Try not to become a man of success, but rather try to become a man of value') @@ 'becom'; ?column? ---------- f (1 row)
test=# select 'become'::tsquery,to_tsquery('become'),to_tsquery('english','become'); tsquery | to_tsquery | to_tsquery ----------+------------+------------ 'become' | 'become' | 'becom' (1 row)
to_tsquery 也会进行标准化转换，在搜索时必须用 to_tsquery，确保数据不会因为标准化转换而搜索不到。
4、逻辑操作
test=# SELECT to_tsvector('Try not to become a man of success, but rather try to become a man of value') @@ to_tsquery('become'); ?column? ---------- t (1 row) test=# SELECT to_tsvector('Try not to become a man of success, but rather try to become a man of value') @@ to_tsquery('!become'); ?column? ---------- f (1 row) test=# SELECT to_tsvector('Try not to become a man of success, but rather try to become a man of value') @@ to_tsquery('tri & become'); ?column? ---------- t (1 row) test=# SELECT to_tsvector('Try not to become a man of success, but rather try to become a man of value') @@ to_tsquery('Try & !becom'); ?column? ---------- f (1 row) test=# SELECT to_tsvector('Try not to become a man of success, but rather try to become a man of value') @@ to_tsquery('Try | !become'); ?column? ---------- t (1 row)
5、可以用 :* 表示某词开始字符
test=# SELECT to_tsvector('Try not to become a man of success, but rather try to become a man of value') @@ to_tsquery('bec:*'); ?column? ---------- t (1 row)
6、其他语言支持
test=# SELECT to_tsvector('simple','Try not to become a man of success, but rather try to become a man of value'); to_tsvector --------------------------------------------------------------------------------------------------------------------- 'a':5,14 'become':4,13 'but':9 'man':6,15 'not':2 'of':7,16 'rather':10 'success':8 'to':3,12 'try':1,11 'value':17 (1 row) test=# SELECT to_tsvector('english','Try not to become a man of success, but rather try to become a man of value') ; to_tsvector ---------------------------------------------------------------------- 'becom':4,13 'man':6,15 'rather':10 'success':8 'tri':1,11 'valu':17 (1 row) ^ test=# SELECT to_tsvector('french','Try not to become a man of success, but rather try to become a man of value') ; to_tsvector ----------------------------------------------------------------------------------------------------------------- 'a':5,14 'becom':4,13 'but':9 'man':6,15 'not':2 'of':7,16 'rath':10 'success':8 'to':3,12 'try':1,11 'valu':17 (1 row) ^ test=# SELECT to_tsvector('french'::regconfig,'Try not to become a man of success, but rather try to become a man of value') ; to_tsvector ----------------------------------------------------------------------------------------------------------------- 'a':5,14 'becom':4,13 'but':9 'man':6,15 'not':2 'of':7,16 'rath':10 'success':8 'to':3,12 'try':1,11 'valu':17 (1 row)
simple并不忽略禁用词表，它也不会试着去查找单词的词根。使用simple时，空格分割的每一组字符都是一个语义；simple 只做了小写转换；对于数据来说，simple文本搜索配置项很实用。
二、中文检索
在开始介绍中文检索前，我们先来看个例子：
test=# select to_tsvector('人大金仓致力于提供高可靠的数据库产品'); to_tsvector ------------------------------------------ '人大金仓致力于提供高可靠的数据库产品':1
因为内置的分词器是按空格分割的，而中文间没有空格，因此，整句话就被看做一个分词。
1、创建中文搜索插件
create extension zhparser; create text search configuration zhongwen_parser (parser = zhparser); alter text search configuration zhongwen_parser add mapping for n,v,a,i,e,l,j with simple;
上面 for 后面的字母表示分词的token，上面的token映射只映射了名词(n)，动词(v)，形容词(a)，成语(i)，叹词(e)，缩写(j) 和习用语(l)6种，这6种以外的token全部被屏蔽。词典使用的是内置的simple词典。具体的token 如下：
test=# select ts_token_type('zhparser'); ts_token_type ------------------------ (97,a,adjective) (98,b,differentiation) (99,c,conjunction) (100,d,adverb) (101,e,exclamation) (102,f,position) (103,g,root) (104,h,head) (105,i,idiom) (106,j,abbreviation) (107,k,tail) (108,l,tmp) (109,m,numeral) (110,n,noun) (111,o,onomatopoeia) (112,p,prepositional) (113,q,quantity) (114,r,pronoun) (115,s,space) (116,t,time) (117,u,auxiliary) (118,v,verb) (119,w,punctuation) (120,x,unknown) (121,y,modal) (122,z,status) (26 rows)
2、查看pg_ts_config
创建text search configuration 后，可以在视图pg_ts_config 看到如下信息：
test=# select * from pg_ts_config; oid | cfgname | cfgnamespace | cfgowner | cfgparser -------+-----------------+--------------+----------+----------- 3748 | simple | 11 | 10 | 3722 13265 | arabic | 11 | 10 | 3722 13267 | danish | 11 | 10 | 3722 13269 | dutch | 11 | 10 | 3722 13271 | english | 11 | 10 | 3722 13273 | finnish | 11 | 10 | 3722 13275 | french | 11 | 10 | 3722 13277 | german | 11 | 10 | 3722 13279 | hungarian | 11 | 10 | 3722 13281 | indonesian | 11 | 10 | 3722 13283 | irish | 11 | 10 | 3722 13285 | italian | 11 | 10 | 3722 13287 | lithuanian | 11 | 10 | 3722 13289 | nepali | 11 | 10 | 3722 13291 | norwegian | 11 | 10 | 3722 13293 | portuguese | 11 | 10 | 3722 13295 | romanian | 11 | 10 | 3722 13297 | russian | 11 | 10 | 3722 13299 | spanish | 11 | 10 | 3722 13301 | swedish | 11 | 10 | 3722 13303 | tamil | 11 | 10 | 3722 13305 | turkish | 11 | 10 | 3722 16390 | parser_name | 2200 | 10 | 16389 24587 | zhongwen_parser | 2200 | 10 | 16389
3、使用中文分词
test=# select to_tsvector('zhongwen_parser','人大金仓致力于提供高可靠的数据库产品'); to_tsvector ------------------------------------------------------------------ '产品':7 '人大':1 '可靠':5 '提供':3 '数据库':6 '致力于':2 '高':4
4、contains 函数
test=# \df+ contains List of functions Schema | Name | Result data type | Argument data types | Type | Volatility | Parallel | Owner | Security | Access privileges | Language | Source code | Description --------+----------+------------------+---------------------+------+------------+----------+--------+----------+-------------------+----------+------------------------------------------+------------- sys | contains | boolean | text, text | func | immutable | safe | system | invoker | | sql | select to_tsvector($1) @@ to_tsquery($2) | sys | contains | boolean | text, text, integer | func | immutable | safe | system | invoker | | sql | select to_tsvector($1) @@ to_tsquery($2) | sys | contains | boolean | text, tsquery | func | immutable | safe | system | invoker | | sql | select $1::tsvector @@ $2 | sys | contains | boolean | tsvector, text | func | immutable | safe | system | invoker | | sql | select $1 @@ $2::tsquery | sys | contains | boolean | tsvector, tsquery | func | immutable | safe | system | invoker | | sql | select $1 @@ $2 |
默认contains 函数使用的是空格分词解析器，因此，无法使用contains 进行中文判断
test=# select contains('人大金仓致力于提供高可靠的数据库产品','产品'); contains ---------- f
如果需要使用中文分词，可以设置默认的文本解析器：
alter system set default_text_search_config='zhongwen_parser'; --上面创建的
从以上的 to_tsvector 解析结果可以看到，对于特定的单字符无法解析成词，需要设置参数
test=# select contains('人大金仓致力于提供高可靠的数据库产品','产'); contains ---------- f (1 row) test=# alter system set zhparser.multi_zmain = 'true'; ALTER SYSTEM test=# select sys_reload_conf(); sys_reload_conf ----------------- t (1 row) test=# select contains('人大金仓致力于提供高可靠的数据库产品','产'); contains ---------- t
KINGBASE研究院
