Distill.hs

module Kioku.Cli.Commands.Distill
  ( CandidateSource (..),
    DistillOptions (..),
    distillOptionsParser,
    runDistill,
  )
where

import Data.Text qualified as Text
import Kioku.AI.Config (AIFeature (CandidateEmbedding))
import Kioku.App (runAppIO, withNoopAppEnv)
import Kioku.Cli.AIConfig (aiConfigOption, loadAIRuntime)
import Kioku.Cli.Context (cliMemoryContext)
import Kioku.Cli.Options (boundedIntReader)
import Kioku.Distill.L1 (L1Outcome (..), L1RunMode (..), L1Summary (..), distillSessionL1, recallCandidates, scopedScanCandidates)
import Kioku.Distill.Runtime (newDistillRuntime)
import Kioku.Id (SessionId, idText, parseId)
import Kioku.Memory.Embedding (resolveEmbeddingConfig)
import Kioku.Recall.Capability (detectVectorCapability)
import Kiroku.Store.Connection (defaultConnectionSettings)
import Options.Applicative
import System.Environment (lookupEnv)

data CandidateSource = CandidateScan | CandidateRecall
  deriving stock (Eq, Show)

data DistillOptions = DistillOptions
  { sessionId :: !SessionId,
    candidateSource :: !CandidateSource,
    candidateLimit :: !Int,
    force :: !Bool,
    aiConfig :: !(Maybe FilePath)
  }
  deriving stock (Eq, Show)

distillOptionsParser :: Parser DistillOptions
distillOptionsParser =
  hsubparser $
    command
      "session"
      ( info
          sessionOptionsParser
          (progDesc "Run one L1 distillation pass for a session")
      )

sessionOptionsParser :: Parser DistillOptions
sessionOptionsParser =
  DistillOptions
    <$> argument
      (eitherReader parseSessionId)
      (metavar "SESSION_ID")
    <*> option
      (eitherReader parseCandidateSource)
      ( long "candidates"
          <> metavar "scan|recall"
          <> value CandidateScan
          <> help "Candidate lookup source"
      )
    <*> option
      (boundedIntReader "LIMIT" 1 50)
      ( long "limit"
          <> metavar "N"
          <> value 5
          <> help "Maximum merge candidates per extracted atom (1-50)"
      )
    <*> switch
      ( long "force"
          <> help "Re-run even when the session has no turns newer than the last successful pass"
      )
    <*> aiConfigOption

runDistill :: DistillOptions -> IO ()
runDistill opts = do
  connStr <- requireEnv "PG_CONNECTION_STRING"
  ai <- loadAIRuntime True opts.aiConfig
  let rt = newDistillRuntime ai Nothing
  context <- cliMemoryContext
  withNoopAppEnv (defaultConnectionSettings (Text.pack connStr)) \env -> do
    result <- runAppIO env do
      finder <- case (opts.candidateSource, resolveEmbeddingConfig ai CandidateEmbedding) of
        (CandidateRecall, Right _) -> do
          capability <- detectVectorCapability 1536
          pure (recallCandidates ai capability opts.candidateLimit)
        _ -> pure (scopedScanCandidates opts.candidateLimit)
      distillSessionL1 context (runMode opts) rt finder opts.sessionId
    case result of
      Left storeErr -> ioError (userError ("kioku distill store error: " <> show storeErr))
      Right (Left l1Err) -> ioError (userError ("kioku distill error: " <> show l1Err))
      Right (Right (L1Distilled summary)) -> printSummary opts.sessionId summary
      Right (Right L1SkippedUpToDate) ->
        putStrLn "Session already distilled (no new turns); use --force to re-run."

runMode :: DistillOptions -> L1RunMode
runMode opts
  | opts.force = IgnoreWatermark
  | otherwise = RespectWatermark

-- | Strict: only a @kioku_session@ id is accepted here. The lenient parser this used to call
-- would take a @kioku_memory@ id, throw its prefix away, and rebrand the UUID — so the pass ran
-- against a session that does not exist and reported nothing wrong.
parseSessionId :: String -> Either String SessionId
parseSessionId raw =
  case parseId (Text.pack raw) of
    Left err -> Left (Text.unpack err)
    Right sid -> Right sid

parseCandidateSource :: String -> Either String CandidateSource
parseCandidateSource = \case
  "scan" -> Right CandidateScan
  "recall" -> Right CandidateRecall
  other -> Left ("unknown candidate source: " <> other)

printSummary :: SessionId -> L1Summary -> IO ()
printSummary sid summary =
  putStrLn $
    "Distilled session "
      <> Text.unpack (idText sid)
      <> ": extracted="
      <> show summary.extracted
      <> " stored="
      <> show summary.stored
      <> " merged="
      <> show summary.merged
      <> " skipped="
      <> show summary.skipped

requireEnv :: String -> IO String
requireEnv name = do
  found <- lookupEnv name
  case found of
    Just envValue -> pure envValue
    Nothing -> ioError (userError (name <> " is not set"))