-
Notifications
You must be signed in to change notification settings - Fork 3
Open
Description
If this makes sense, and purely for compatibility, do you think it would be a good idea to make match_documents in the format as expected by n8n? this would allow to re-use same function for n8n workflows. Minor difference in implementation, but makes it usable with n8n out of the box
Minor differences are arguments in the function:
n8n-compatible:
create function match_documents (
query_embedding vector(1536),
match_count int default null,
filter jsonb DEFAULT '{}'
) returns table (
id bigint,
content text,
metadata jsonb,
similarity float
)
current mind-matrix: https://github.com/khwerhahn/MindMatrix/blob/main/sql/setup.sql#L70-L81
CREATE OR REPLACE FUNCTION public.match_documents(
query_embedding vector(1536),
search_vault_id TEXT,
match_count INT
)
RETURNS TABLE (
id BIGINT,
obsidian_id TEXT,
content TEXT,
metadata JSONB,
similarity FLOAT
)
Documentation reference for function structure:
https://supabase.com/docs/guides/ai/langchain?database-method=sql
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
No labels