Files
accounted/supabase/migrations/20240101000033_ai_chat_schema.sql
T
2026-02-21 17:14:08 +01:00

128 lines
4.4 KiB
PL/PgSQL

-- Migration 033: AI Chat Schema
-- Creates tables for the AI chat assistant extension:
-- chat_sessions, chat_messages, knowledge_documents, and match_documents RPC
-- Enable pgvector for embedding storage
create extension if not exists vector with schema extensions;
-- ============================================================
-- chat_sessions
-- ============================================================
create table public.chat_sessions (
id uuid primary key default gen_random_uuid(),
user_id uuid references auth.users on delete cascade not null,
title text,
created_at timestamptz not null default now(),
updated_at timestamptz not null default now()
);
alter table public.chat_sessions enable row level security;
create policy "chat_sessions_select" on public.chat_sessions
for select using (auth.uid() = user_id);
create policy "chat_sessions_insert" on public.chat_sessions
for insert with check (auth.uid() = user_id);
create policy "chat_sessions_update" on public.chat_sessions
for update using (auth.uid() = user_id);
create policy "chat_sessions_delete" on public.chat_sessions
for delete using (auth.uid() = user_id);
create index idx_chat_sessions_user_created on public.chat_sessions (user_id, created_at desc);
create trigger chat_sessions_updated_at
before update on public.chat_sessions
for each row execute function public.update_updated_at_column();
-- ============================================================
-- chat_messages
-- ============================================================
create table public.chat_messages (
id uuid primary key default gen_random_uuid(),
session_id uuid references public.chat_sessions on delete cascade not null,
user_id uuid references auth.users on delete cascade not null,
role text not null check (role in ('user', 'assistant')),
content text not null,
sources jsonb,
created_at timestamptz not null default now()
);
alter table public.chat_messages enable row level security;
create policy "chat_messages_select" on public.chat_messages
for select using (auth.uid() = user_id);
create policy "chat_messages_insert" on public.chat_messages
for insert with check (auth.uid() = user_id);
create policy "chat_messages_update" on public.chat_messages
for update using (auth.uid() = user_id);
create policy "chat_messages_delete" on public.chat_messages
for delete using (auth.uid() = user_id);
create index idx_chat_messages_session on public.chat_messages (session_id, created_at);
-- ============================================================
-- knowledge_documents
-- ============================================================
create table public.knowledge_documents (
id uuid primary key default gen_random_uuid(),
source_file text not null,
title text not null,
section_title text,
content text not null,
content_hash text unique not null,
embedding extensions.vector(1536),
metadata jsonb default '{}',
created_at timestamptz not null default now()
);
alter table public.knowledge_documents enable row level security;
-- Knowledge documents are shared — any authenticated user can read
create policy "knowledge_documents_select" on public.knowledge_documents
for select using (true);
create index idx_knowledge_documents_hash on public.knowledge_documents (content_hash);
-- ============================================================
-- match_documents RPC (vector similarity search)
-- ============================================================
create or replace function public.match_documents(
query_embedding extensions.vector,
match_count int default 5,
match_threshold float default 0.7
)
returns table (
id uuid,
source_file text,
title text,
section_title text,
content text,
metadata jsonb,
similarity float
)
language plpgsql
security definer
set search_path = public, extensions
as $$
begin
return query
select
kd.id,
kd.source_file,
kd.title,
kd.section_title,
kd.content,
kd.metadata,
1 - (kd.embedding <=> query_embedding)::float as similarity
from public.knowledge_documents kd
where 1 - (kd.embedding <=> query_embedding) >= match_threshold
order by kd.embedding <=> query_embedding
limit match_count;
end;
$$;
grant execute on function public.match_documents(extensions.vector, int, float) to authenticated;