-
Notifications
You must be signed in to change notification settings - Fork 29
Expand file tree
/
Copy pathsetup_vector_store.sql
More file actions
258 lines (239 loc) · 7.08 KB
/
Copy pathsetup_vector_store.sql
File metadata and controls
258 lines (239 loc) · 7.08 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
-- Enable the pgvector extension
CREATE EXTENSION IF NOT EXISTS vector;
-------------------------------------------------------------------------------
-- 1. langchain_docs (ID as text)
-------------------------------------------------------------------------------
DROP TABLE IF EXISTS "langchain_docs";
CREATE TABLE "langchain_docs" (
id TEXT PRIMARY KEY,
content TEXT,
metadata JSONB,
embedding VECTOR(1536)
);
CREATE INDEX IF NOT EXISTS "langchain_docs_embedding_idx"
ON "langchain_docs"
USING ivfflat (embedding vector_l2_ops)
WITH (lists = 100);
CREATE INDEX IF NOT EXISTS "langchain_docs_metadata_idx"
ON "langchain_docs"
USING GIN (metadata);
CREATE INDEX IF NOT EXISTS "langchain_docs_content_idx"
ON "langchain_docs"
USING GIN (to_tsvector('english', COALESCE(content, '')));
-------------------------------------------------------------------------------
-- 2. langchain_example (ID as text)
-------------------------------------------------------------------------------
DROP TABLE IF EXISTS "langchain_example";
CREATE TABLE "langchain_example" (
id TEXT PRIMARY KEY,
content TEXT,
metadata JSONB,
embedding VECTOR(1536)
);
CREATE INDEX IF NOT EXISTS "langchain_example_embedding_idx"
ON "langchain_example"
USING ivfflat (embedding vector_l2_ops)
WITH (lists = 100);
CREATE INDEX IF NOT EXISTS "langchain_example_metadata_idx"
ON "langchain_example"
USING GIN (metadata);
CREATE INDEX IF NOT EXISTS "langchain_example_content_idx"
ON "langchain_example"
USING GIN (to_tsvector('english', COALESCE(content, '')));
-------------------------------------------------------------------------------
-- 3. documents (ID as bigserial - auto increments)
-------------------------------------------------------------------------------
DROP TABLE IF EXISTS "documents";
CREATE TABLE "documents" (
id BIGSERIAL PRIMARY KEY,
content TEXT,
metadata JSONB,
embedding VECTOR(1536)
);
CREATE INDEX IF NOT EXISTS "documents_embedding_idx"
ON "documents"
USING ivfflat (embedding vector_l2_ops)
WITH (lists = 100);
CREATE INDEX IF NOT EXISTS "documents_metadata_idx"
ON "documents"
USING GIN (metadata);
CREATE INDEX IF NOT EXISTS "documents_content_idx"
ON "documents"
USING GIN (to_tsvector('english', COALESCE(content, '')));
-------------------------------------------------------------------------------
-- Functions for langchain_docs
-------------------------------------------------------------------------------
-- Vector-based similarity on langchain_docs
CREATE OR REPLACE FUNCTION match_documents (
query_embedding VECTOR(1536),
match_count INT,
filter JSONB DEFAULT '{}'
)
RETURNS TABLE (
id TEXT,
content TEXT,
metadata JSONB,
similarity FLOAT
)
LANGUAGE plpgsql
AS $$
BEGIN
RETURN QUERY
SELECT
langchain_docs.id,
langchain_docs.content,
langchain_docs.metadata,
1 - (langchain_docs.embedding <=> query_embedding) AS similarity
FROM langchain_docs
WHERE metadata @> filter
ORDER BY langchain_docs.embedding <=> query_embedding
LIMIT match_count;
END;
$$;
-- Vector-based similarity on langchain_example
CREATE OR REPLACE FUNCTION match_example_documents (
query_embedding VECTOR(1536),
match_count INT,
filter JSONB DEFAULT '{}'
)
RETURNS TABLE (
id TEXT,
content TEXT,
metadata JSONB,
similarity FLOAT
)
LANGUAGE plpgsql
AS $$
BEGIN
RETURN QUERY
SELECT
langchain_example.id,
langchain_example.content,
langchain_example.metadata,
1 - (langchain_example.embedding <=> query_embedding) AS similarity
FROM langchain_example
WHERE metadata @> filter
ORDER BY langchain_example.embedding <=> query_embedding
LIMIT match_count;
END;
$$;
-- Hybrid search (keyword + vector) on langchain_docs
CREATE OR REPLACE FUNCTION kw_match_documents (
query_text TEXT,
query_embedding VECTOR(1536),
match_count INT,
filter JSONB DEFAULT '{}'
)
RETURNS TABLE (
id TEXT,
content TEXT,
metadata JSONB,
similarity FLOAT,
text_similarity FLOAT
)
LANGUAGE plpgsql
AS $$
BEGIN
RETURN QUERY
SELECT
langchain_docs.id,
langchain_docs.content,
langchain_docs.metadata,
1 - (langchain_docs.embedding <=> query_embedding) AS similarity,
ts_rank(
to_tsvector('english', COALESCE(langchain_docs.content, '')),
plainto_tsquery('english', query_text)
) AS text_similarity
FROM langchain_docs
WHERE metadata @> filter
AND (
to_tsvector('english', COALESCE(langchain_docs.content, ''))
@@ plainto_tsquery('english', query_text)
OR 1 - (langchain_docs.embedding <=> query_embedding) > 0.7
)
ORDER BY
(
(1 - (langchain_docs.embedding <=> query_embedding)) * 0.8
+ ts_rank(
to_tsvector('english', COALESCE(langchain_docs.content, '')),
plainto_tsquery('english', query_text)
) * 0.2
) DESC
LIMIT match_count;
END;
$$;
-------------------------------------------------------------------------------
-- New Functions for the "documents" table
-------------------------------------------------------------------------------
-- 1) Vector-based similarity on the "documents" table
CREATE OR REPLACE FUNCTION match_main_documents (
query_embedding VECTOR(1536),
match_count INT,
filter JSONB DEFAULT '{}'
)
RETURNS TABLE (
id BIGINT,
content TEXT,
metadata JSONB,
similarity FLOAT
)
LANGUAGE plpgsql
AS $$
BEGIN
RETURN QUERY
SELECT
documents.id,
documents.content,
documents.metadata,
1 - (documents.embedding <=> query_embedding) AS similarity
FROM documents
WHERE metadata @> filter
ORDER BY documents.embedding <=> query_embedding
LIMIT match_count;
END;
$$;
-- 2) Hybrid search (keyword + vector) on the "documents" table
CREATE OR REPLACE FUNCTION kw_match_main_documents (
query_text TEXT,
query_embedding VECTOR(1536),
match_count INT,
filter JSONB DEFAULT '{}'
)
RETURNS TABLE (
id BIGINT,
content TEXT,
metadata JSONB,
similarity FLOAT,
text_similarity FLOAT
)
LANGUAGE plpgsql
AS $$
BEGIN
RETURN QUERY
SELECT
documents.id,
documents.content,
documents.metadata,
1 - (documents.embedding <=> query_embedding) AS similarity,
ts_rank(
to_tsvector('english', COALESCE(documents.content, '')),
plainto_tsquery('english', query_text)
) AS text_similarity
FROM documents
WHERE metadata @> filter
AND (
to_tsvector('english', COALESCE(documents.content, ''))
@@ plainto_tsquery('english', query_text)
OR 1 - (documents.embedding <=> query_embedding) > 0.7
)
ORDER BY
(
(1 - (documents.embedding <=> query_embedding)) * 0.8
+ ts_rank(
to_tsvector('english', COALESCE(documents.content, '')),
plainto_tsquery('english', query_text)
) * 0.2
) DESC
LIMIT match_count;
END;
$$;