meminbox 0.0.0-stage → 0.2.1

package/supabase/migrations/20261108000001_init.sql

@@ -0,0 +1,156 @@
+-- meminbox: initial schema
+-- Design rule: `documents` is the source of truth (human-readable Markdown).
+-- `document_chunks` (with embeddings) is a DERIVED index and can be dropped
+-- and rebuilt at any time with `meminbox reindex`.
+
+create extension if not exists vector;
+create extension if not exists pg_trgm;
+
+-- ---------------------------------------------------------------------------
+-- projects: a namespace for a body of knowledge (one repo, one domain, ...)
+-- ---------------------------------------------------------------------------
+create table if not exists public.projects (
+  id         uuid primary key default gen_random_uuid(),
+  slug       text not null unique,
+  name       text not null,
+  created_at timestamptz not null default now()
+);
+
+-- ---------------------------------------------------------------------------
+-- documents: the source of truth
+-- status:
+--   inbox   = written by an agent, not yet reviewed/promoted (the default)
+--   active  = trusted memory, returned by search by default
+--   archived = retired; kept for history, excluded from search
+-- provenance: who/what wrote this and from what, e.g.
+--   {"agent": "claude-code", "session": "...", "source": "conversation", "uri": "..."}
+-- ---------------------------------------------------------------------------
+create table if not exists public.documents (
+  id            uuid primary key default gen_random_uuid(),
+  project_id    uuid not null references public.projects(id) on delete cascade,
+  title         text not null,
+  content_md    text not null,
+  status        text not null default 'inbox' check (status in ('inbox', 'active', 'archived')),
+  source_agent  text,
+  provenance    jsonb not null default '{}'::jsonb,
+  version       int  not null default 1,
+  search_vector tsvector generated always as (
+                  to_tsvector('simple', coalesce(title, '') || ' ' || coalesce(content_md, ''))
+                ) stored,
+  created_at    timestamptz not null default now(),
+  updated_at    timestamptz not null default now()
+);
+
+create index if not exists documents_project_status_idx
+  on public.documents (project_id, status);
+create index if not exists documents_search_vector_idx
+  on public.documents using gin (search_vector);
+-- trigram index so substring search also works for Japanese text
+create index if not exists documents_content_trgm_idx
+  on public.documents using gin (content_md gin_trgm_ops);
+
+-- ---------------------------------------------------------------------------
+-- document_chunks: derived index for semantic search (rebuildable)
+-- ---------------------------------------------------------------------------
+create table if not exists public.document_chunks (
+  id          uuid primary key default gen_random_uuid(),
+  document_id uuid not null references public.documents(id) on delete cascade,
+  chunk_index int  not null,
+  content     text not null,
+  embedding   vector(1536),
+  created_at  timestamptz not null default now(),
+  unique (document_id, chunk_index)
+);
+
+create index if not exists document_chunks_document_idx
+  on public.document_chunks (document_id);
+create index if not exists document_chunks_embedding_idx
+  on public.document_chunks using hnsw (embedding vector_cosine_ops);
+
+-- ---------------------------------------------------------------------------
+-- audit_log: every write / promote / archive lands here
+-- ---------------------------------------------------------------------------
+create table if not exists public.audit_log (
+  id          uuid primary key default gen_random_uuid(),
+  actor       text,
+  action      text not null,
+  document_id uuid references public.documents(id) on delete set null,
+  metadata    jsonb not null default '{}'::jsonb,
+  created_at  timestamptz not null default now()
+);
+
+-- ---------------------------------------------------------------------------
+-- search helpers
+-- ---------------------------------------------------------------------------
+
+-- Keyword search: FTS rank, with an ILIKE fallback so Japanese substring
+-- queries (no word boundaries) still hit.
+create or replace function public.search_documents(
+  p_query  text,
+  p_limit  int default 10,
+  p_status text default 'active'
+)
+returns table (
+  id         uuid,
+  project_id uuid,
+  title      text,
+  content_md text,
+  status     text,
+  rank       real,
+  updated_at timestamptz
+)
+language sql stable
+as $$
+  select d.id, d.project_id, d.title, d.content_md, d.status,
+         ts_rank(d.search_vector, websearch_to_tsquery('simple', p_query)) as rank,
+         d.updated_at
+  from public.documents d
+  where d.status = p_status
+    and (
+      d.search_vector @@ websearch_to_tsquery('simple', p_query)
+      or d.title ilike '%' || p_query || '%'
+      or d.content_md ilike '%' || p_query || '%'
+    )
+  order by rank desc, d.updated_at desc
+  limit p_limit;
+$$;
+
+-- Vector search over the derived chunks.
+create or replace function public.match_chunks(
+  p_embedding vector(1536),
+  p_limit     int default 10,
+  p_status    text default 'active'
+)
+returns table (
+  document_id uuid,
+  project_id  uuid,
+  title       text,
+  chunk       text,
+  similarity  float
+)
+language sql stable
+as $$
+  select c.document_id, d.project_id, d.title, c.content,
+         1 - (c.embedding <=> p_embedding) as similarity
+  from public.document_chunks c
+  join public.documents d on d.id = c.document_id
+  where d.status = p_status
+    and c.embedding is not null
+  order by c.embedding <=> p_embedding
+  limit p_limit;
+$$;
+
+-- ---------------------------------------------------------------------------
+-- Row Level Security: enabled with NO public policies on purpose.
+-- All access in v0 goes through the service role key from the owner's own
+-- machines (MCP server / CLI). If a hosted/remote endpoint is added later,
+-- write explicit policies then — do not open these tables to anon.
+-- ---------------------------------------------------------------------------
+alter table public.projects        enable row level security;
+alter table public.documents       enable row level security;
+alter table public.document_chunks enable row level security;
+alter table public.audit_log       enable row level security;
+
+insert into public.projects (slug, name)
+values ('default', 'Default')
+on conflict (slug) do nothing;