github/AutoGPT
1
1
Fork 0
mirror of https://github.com/Significant-Gravitas/AutoGPT.git synced 2026-04-08 03:00:28 -04:00
Code Issues Packages Projects Releases Wiki Activity

fix: use unqualified vector type for operator compatibility

Browse Source
This commit is contained in:
Swifty
2026-01-09 17:58:39 +01:00
parent e69f14353e
commit 9cfe70e554
3 changed files with 6 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
autogpt_platform/backend/backend/api/features/store/embeddings.py
View File

@@ -115,12 +115,12 @@ async def store_embedding(
"searchableText", "contentHash", "createdAt", "updatedAt"
)
VALUES (
gen_random_uuid(), $1, $2::public.vector,
gen_random_uuid(), $1, $2::vector,
$3, $4, NOW(), NOW()
)
ON CONFLICT ("storeListingVersionId")
DO UPDATE SET
"embedding" = $2::public.vector,
"embedding" = $2::vector,
"searchableText" = $3,
"contentHash" = $4,
"updatedAt" = NOW()

4
autogpt_platform/backend/backend/api/features/store/hybrid_search.py
View File

@@ -152,7 +152,7 @@ async def hybrid_search(
SELECT
sa.*,
-- Semantic score: cosine similarity (1 - distance)
COALESCE(1 - (sle.embedding <=> {embedding_param}::public.vector), 0) as semantic_score,
COALESCE(1 - (sle.embedding <=> {embedding_param}::vector), 0) as semantic_score,
-- Lexical score: ts_rank_cd normalized
COALESCE(ts_rank_cd(sa.search, plainto_tsquery('english', {query_param})), 0) as lexical_raw,
-- Category match: 1 if query term appears in categories, else 0
@@ -227,7 +227,7 @@ async def hybrid_search(
WITH search_scores AS (
SELECT
sa.slug,
COALESCE(1 - (sle.embedding <=> {embedding_param}::public.vector), 0) as semantic_score,
COALESCE(1 - (sle.embedding <=> {embedding_param}::vector), 0) as semantic_score,
COALESCE(ts_rank_cd(sa.search, plainto_tsquery('english', {query_param})), 0) as lexical_raw,
CASE
WHEN EXISTS (

4
autogpt_platform/backend/migrations/20251216181700_add_store_embeddings/migration.sql
View File

@@ -8,7 +8,7 @@ CREATE EXTENSION IF NOT EXISTS vector;
CREATE TABLE "StoreListingEmbedding" (
"id" TEXT NOT NULL DEFAULT gen_random_uuid(),
"storeListingVersionId" TEXT NOT NULL,
"embedding" public.vector(1536), -- OpenAI text-embedding-3-small produces 1536 dimensions
"embedding" vector(1536), -- OpenAI text-embedding-3-small produces 1536 dimensions
"searchableText" TEXT, -- The text that was embedded (for debugging/recomputation)
"contentHash" TEXT, -- MD5 hash of searchable text for change detection
"createdAt" TIMESTAMP(3) NOT NULL DEFAULT CURRENT_TIMESTAMP,
@@ -25,7 +25,7 @@ CREATE UNIQUE INDEX "StoreListingEmbedding_storeListingVersionId_key"
-- Using cosine distance (vector_cosine_ops) which is standard for text embeddings
CREATE INDEX "StoreListingEmbedding_embedding_idx"
ON "StoreListingEmbedding"
USING hnsw ("embedding" public.vector_cosine_ops);
USING hnsw ("embedding" vector_cosine_ops);
-- Index on content hash for fast lookup during change detection
CREATE INDEX "StoreListingEmbedding_contentHash_idx"