109 lines
3.8 KiB
PL/PgSQL
109 lines
3.8 KiB
PL/PgSQL
-- 备选方案：不使用自定义函数的全文搜索索引
|
||
-- 如果IMMUTABLE函数方法失败，可以使用此备选方案
|
||
|
||
-- ===================================================================
|
||
-- 方案1: 使用表达式索引 (推荐)
|
||
-- ===================================================================
|
||
|
||
-- 删除可能存在的有问题的索引
|
||
DROP INDEX IF EXISTS public.idx_contents_fts_title;
|
||
DROP INDEX IF EXISTS public.idx_contents_fts_content;
|
||
|
||
-- 删除可能有问题的函数
|
||
DROP FUNCTION IF EXISTS public.content_title_to_tsvector(TEXT);
|
||
DROP FUNCTION IF EXISTS public.content_body_to_tsvector(TEXT);
|
||
|
||
-- 方法1: 使用简单的表达式索引
|
||
CREATE INDEX IF NOT EXISTS idx_contents_fts_title_simple
|
||
ON public.ak_contents USING gin((to_tsvector('simple', title)));
|
||
|
||
CREATE INDEX IF NOT EXISTS idx_contents_fts_content_simple
|
||
ON public.ak_contents USING gin((to_tsvector('simple', content)));
|
||
|
||
-- ===================================================================
|
||
-- 方案2: 使用预计算列 (最佳性能)
|
||
-- ===================================================================
|
||
|
||
-- 添加预计算的tsvector列
|
||
DO $$
|
||
BEGIN
|
||
-- 检查列是否已存在
|
||
IF NOT EXISTS (
|
||
SELECT 1 FROM information_schema.columns
|
||
WHERE table_name = 'ak_contents'
|
||
AND column_name = 'title_search_vector'
|
||
) THEN
|
||
ALTER TABLE public.ak_contents
|
||
ADD COLUMN title_search_vector tsvector;
|
||
END IF;
|
||
|
||
IF NOT EXISTS (
|
||
SELECT 1 FROM information_schema.columns
|
||
WHERE table_name = 'ak_contents'
|
||
AND column_name = 'content_search_vector'
|
||
) THEN
|
||
ALTER TABLE public.ak_contents
|
||
ADD COLUMN content_search_vector tsvector;
|
||
END IF;
|
||
END
|
||
$$;
|
||
|
||
-- 更新现有数据的搜索向量
|
||
UPDATE public.ak_contents SET
|
||
title_search_vector = to_tsvector('simple', COALESCE(title, '')),
|
||
content_search_vector = to_tsvector('simple', COALESCE(content, ''))
|
||
WHERE title_search_vector IS NULL OR content_search_vector IS NULL;
|
||
|
||
-- 创建触发器自动更新搜索向量
|
||
CREATE OR REPLACE FUNCTION public.update_content_search_vectors()
|
||
RETURNS trigger AS $$
|
||
BEGIN
|
||
NEW.title_search_vector := to_tsvector('simple', COALESCE(NEW.title, ''));
|
||
NEW.content_search_vector := to_tsvector('simple', COALESCE(NEW.content, ''));
|
||
RETURN NEW;
|
||
END;
|
||
$$ LANGUAGE plpgsql;
|
||
|
||
-- 创建触发器
|
||
DROP TRIGGER IF EXISTS trig_update_content_search_vectors ON public.ak_contents;
|
||
CREATE TRIGGER trig_update_content_search_vectors
|
||
BEFORE INSERT OR UPDATE ON public.ak_contents
|
||
FOR EACH ROW EXECUTE FUNCTION public.update_content_search_vectors();
|
||
|
||
-- 在预计算列上创建索引
|
||
CREATE INDEX IF NOT EXISTS idx_contents_title_search_vector
|
||
ON public.ak_contents USING gin(title_search_vector);
|
||
|
||
CREATE INDEX IF NOT EXISTS idx_contents_content_search_vector
|
||
ON public.ak_contents USING gin(content_search_vector);
|
||
|
||
-- ===================================================================
|
||
-- 查询示例
|
||
-- ===================================================================
|
||
|
||
-- 使用方案1的查询方式 (表达式索引)
|
||
/*
|
||
SELECT id, title FROM ak_contents
|
||
WHERE to_tsvector('simple', title) @@ to_tsquery('simple', 'AI');
|
||
|
||
SELECT id, title FROM ak_contents
|
||
WHERE to_tsvector('simple', content) @@ to_tsquery('simple', 'technology');
|
||
*/
|
||
|
||
-- 使用方案2的查询方式 (预计算列)
|
||
/*
|
||
SELECT id, title FROM ak_contents
|
||
WHERE title_search_vector @@ to_tsquery('simple', 'AI');
|
||
|
||
SELECT id, title FROM ak_contents
|
||
WHERE content_search_vector @@ to_tsquery('simple', 'technology');
|
||
|
||
-- 组合搜索并按相关性排序
|
||
SELECT
|
||
id, title,
|
||
ts_rank(title_search_vector, query) + ts_rank(content_search_vector, query) as rank
|
||
FROM ak_contents, to_tsquery('simple', 'AI & technology') query
|
||
WHERE title_search_vector @@ query OR content_search_vector @@ query
|
||
ORDER BY rank DESC;
|
||
*/
|