pegas-memory/schema.sql

-- Pegas Memory: PostgreSQL — источник истины
-- Запускать в atom-db (:5434), БД content_atom (или отдельная БД alicorn)

CREATE EXTENSION IF NOT EXISTS pg_trgm;
CREATE EXTENSION IF NOT EXISTS btree_gin;

-- ============================================================
-- ИСТОЧНИКИ
-- ============================================================
CREATE TABLE IF NOT EXISTS pm_document (
  id            BIGSERIAL PRIMARY KEY,
  kind          TEXT NOT NULL,           -- web|pdf|epub|text|code_repo|manual
  uri           TEXT,
  title         TEXT,
  fetched_at    TIMESTAMPTZ,
  content_hash  TEXT,                    -- дедуп целого источника
  authority     REAL DEFAULT 0.5,        -- репутация источника
  meta          JSONB DEFAULT '{}',
  created_at    TIMESTAMPTZ DEFAULT now(),
  UNIQUE (uri)
);

-- ============================================================
-- ЧАНКИ (сырые сегменты — доказательства, НЕ для retrieval напрямую)
-- ============================================================
CREATE TABLE IF NOT EXISTS pm_chunk (
  id         BIGSERIAL PRIMARY KEY,
  doc_id     BIGINT REFERENCES pm_document(id) ON DELETE CASCADE,
  seq        INT,
  content    TEXT NOT NULL,
  locator    JSONB DEFAULT '{}',         -- page/§/URL-якорь/строка
  created_at TIMESTAMPTZ DEFAULT now()
);
CREATE INDEX IF NOT EXISTS pm_chunk_doc_idx ON pm_chunk (doc_id, seq);

-- ============================================================
-- АТОМЫ ЗНАНИЯ (основная единица retrieval)
-- ============================================================
CREATE TABLE IF NOT EXISTS pm_atom (
  id           BIGSERIAL PRIMARY KEY,
  type         TEXT NOT NULL,            -- text|definition|logical_rule|code|table|fact
  claim        TEXT,                     -- короткое каноническое утверждение
  content      TEXT NOT NULL,            -- полное значение (для code — исходник/сигнатура)
  content_hash TEXT NOT NULL,            -- дедуп
  confidence   REAL DEFAULT 0.5,
  status       TEXT DEFAULT 'candidate', -- candidate|verified|stale|contested
  doc_id       BIGINT REFERENCES pm_document(id) ON DELETE SET NULL,
  locator      JSONB DEFAULT '{}',
  metadata     JSONB DEFAULT '{}',       -- logical_rule: pattern/conclusion; code: repo/path
  created_at   TIMESTAMPTZ DEFAULT now(),
  updated_at   TIMESTAMPTZ DEFAULT now()
);
CREATE UNIQUE INDEX IF NOT EXISTS pm_atom_hash_idx ON pm_atom (content_hash);
CREATE INDEX IF NOT EXISTS pm_atom_type_status_idx ON pm_atom (type, status);
CREATE INDEX IF NOT EXISTS pm_atom_claim_idx ON pm_atom USING gin (claim gin_trgm_ops);

-- ============================================================
-- СВЯЗИ МЕЖДУ АТОМАМИ (граф → Memgraph)
-- ============================================================
CREATE TABLE IF NOT EXISTS pm_atom_link (
  id         BIGSERIAL PRIMARY KEY,
  src        BIGINT REFERENCES pm_atom(id) ON DELETE CASCADE,
  dst        BIGINT REFERENCES pm_atom(id) ON DELETE CASCADE,
  kind       TEXT NOT NULL,              -- inference|cause|reference|contradicts|depends_on|calls|defines
  confidence REAL DEFAULT 0.5,
  source     TEXT,                       -- manual|extraction|inference
  inferred   BOOLEAN DEFAULT false,      -- выведено инференцией (не явно)
  created_at TIMESTAMPTZ DEFAULT now(),
  UNIQUE (src, dst, kind)
);
CREATE INDEX IF NOT EXISTS pm_link_src_idx ON pm_atom_link (src);
CREATE INDEX IF NOT EXISTS pm_link_dst_idx ON pm_atom_link (dst);

-- ============================================================
-- ЦИТИРОВАНИЕ (доказательная база)
-- ============================================================
CREATE TABLE IF NOT EXISTS pm_citation (
  id          BIGSERIAL PRIMARY KEY,
  atom_id     BIGINT REFERENCES pm_atom(id) ON DELETE CASCADE,
  doc_id      BIGINT REFERENCES pm_document(id) ON DELETE CASCADE,
  locator     JSONB DEFAULT '{}',
  quoted_text TEXT,
  created_at  TIMESTAMPTZ DEFAULT now()
);
CREATE INDEX IF NOT EXISTS pm_citation_atom_idx ON pm_citation (atom_id);

-- ============================================================
-- СИМВОЛЫ КОДА (tree-sitter)
-- ============================================================
CREATE TABLE IF NOT EXISTS pm_code_symbol (
  id          BIGSERIAL PRIMARY KEY,
  repo        TEXT,
  path        TEXT,
  language    TEXT,
  symbol_name TEXT,
  kind        TEXT,                      -- function|class|method|type|const
  signature   TEXT,                      -- индексируется для retrieval
  body_hash   TEXT,                      -- content-addressed
  line_start  INT,
  line_end    INT,
  created_at  TIMESTAMPTZ DEFAULT now()
);
CREATE INDEX IF NOT EXISTS pm_symbol_name_idx ON pm_code_symbol USING gin (symbol_name gin_trgm_ops);
CREATE INDEX IF NOT EXISTS pm_symbol_lang_idx ON pm_code_symbol (language, kind);

-- ============================================================
-- ЗАВИСИМОСТИ КОДА (граф кода: calls/imports/inherits)
-- ============================================================
CREATE TABLE IF NOT EXISTS pm_code_dependency (
  id         BIGSERIAL PRIMARY KEY,
  repo       TEXT,
  path       TEXT,
  language   TEXT,
  src        TEXT,                       -- имя символа-источника
  dst        TEXT,                       -- имя цели
  kind       TEXT NOT NULL,              -- calls|imports|inherits
  created_at TIMESTAMPTZ DEFAULT now(),
  UNIQUE (repo, path, src, dst, kind)
);
CREATE INDEX IF NOT EXISTS pm_code_dep_src_idx ON pm_code_dependency (src);
CREATE INDEX IF NOT EXISTS pm_code_dep_kind_idx ON pm_code_dependency (kind);

-- ============================================================
-- ПАМЯТЬ АГЕНТА (episodic) — ОТДЕЛЬНО от базы знаний
-- ============================================================
CREATE TABLE IF NOT EXISTS pm_episodic_trace (
  id          BIGSERIAL PRIMARY KEY,
  session_id  TEXT,
  step        INT,
  kind        TEXT,                      -- plan|action|observation|reflection
  content     TEXT NOT NULL,
  created_at  TIMESTAMPTZ DEFAULT now()
);
CREATE INDEX IF NOT EXISTS pm_episodic_session_idx ON pm_episodic_trace (session_id, step);

-- ============================================================
-- ПРОБЕЛЫ ДЛЯ ИССЛЕДОВАНИЯ
-- ============================================================
CREATE TABLE IF NOT EXISTS pm_research_gap (
  id          BIGSERIAL PRIMARY KEY,
  question    TEXT NOT NULL,
  reason      TEXT,                      -- low_coverage|low_confidence|stale|contested|user_query
  priority    REAL DEFAULT 0.5,
  status      TEXT DEFAULT 'open',       -- open|researching|resolved|dismissed
  created_at  TIMESTAMPTZ DEFAULT now()
);

-- ============================================================
-- OUTBOX (CDC → Qdrant + Memgraph)
-- ============================================================
CREATE TABLE IF NOT EXISTS pm_outbox (
  id           BIGSERIAL PRIMARY KEY,
  entity_type  TEXT NOT NULL,            -- atom|link|doc
  entity_id    BIGINT NOT NULL,
  op           TEXT NOT NULL,            -- upsert|delete
  created_at   TIMESTAMPTZ DEFAULT now(),
  processed    BOOLEAN DEFAULT false
);
CREATE INDEX IF NOT EXISTS pm_outbox_pending_idx ON pm_outbox (processed, id);