"""Conversational CLI for emergent Salience Runtime.""" from __future__ import annotations import argparse import os import sys from ..telemetry import BUS, ParameterEvent, TelemetryEvent, render_parameter_event, render_training_event from .filters import IngestionThresholds from .session import ConversationConfig, ConversationSession, IngestionConfig def parse_args() -> argparse.Namespace: parser = argparse.ArgumentParser(description="SalienceOS emergent conversation loop") parser.add_argument( "--corpus", action="append", default=None, help="Directory of text files to bootstrap from (can be repeated)", ) parser.add_argument( "--load-dv1", action="store_true", help="Preload every .txt/.md/.jsonl file under C:/DV1", ) parser.add_argument("--chunk-size", type=int, default=2048, help="Ingestion chunk size (bytes)") parser.add_argument( "--checkpoint-interval", type=int, default=100, help="Autosave after this many accepted chunks during ingest", ) parser.add_argument("--chunk-overlap", type=int, default=0, help="Overlap between successive chunks") parser.add_argument("--batch-size", type=int, default=1, help="Number of chunks to evaluate per batch") parser.add_argument( "--max-ingest-chunks", type=int, default=None, help="Optional limit on number of chunks to ingest per corpus directory", ) parser.add_argument( "--salience-filter", action="store_true", help="Enable salience-based filtering during ingestion", ) parser.add_argument("--min-uncertainty", type=float, default=0.0, help="Minimum uncertainty to accept chunk") parser.add_argument("--min-novelty", type=float, default=0.0, help="Minimum novelty to accept chunk") parser.add_argument("--max-drag", type=float, default=1.5, help="Maximum drag allowed for chunk") parser.add_argument( "--progress", dest="progress", action="store_true", default=True, help="Display ingestion progress (default)", ) parser.add_argument( "--no-progress", dest="progress", action="store_false", help="Disable ingestion progress output", ) parser.add_argument("--autosave", type=str, default=None, help="Path to persist conversation state") parser.add_argument("--response-tokens", type=int, default=48, help="Maximum tokens in generated responses") parser.add_argument("--steps", type=int, default=None, help="Optional max number of exchanges before exit") parser.add_argument("--quiet", action="store_true", help="Suppress meta reports") parser.add_argument("--adaptive-filter", action="store_true", help="Enable adaptive salience filtering") parser.add_argument("--adaptive-warmup", type=int, default=32) parser.add_argument("--adaptive-window", type=int, default=32) parser.add_argument("--adaptive-enable-ratio", type=float, default=0.3) parser.add_argument("--adaptive-disable-ratio", type=float, default=0.6) parser.add_argument("--adaptive-cooldown", type=int, default=32) parser.add_argument("--behavior-scaffolds", type=str, default=None, help="Path to behavior scaffold JSONL") parser.add_argument("--no-learning", action="store_true", help="Disable online learning during chat") parser.add_argument("--telemetry", action="store_true", help="Stream telemetry events to stderr") return parser.parse_args() def main() -> None: args = parse_args() thresholds = IngestionThresholds( enabled=args.salience_filter, min_uncertainty=args.min_uncertainty, min_novelty=args.min_novelty, max_drag=args.max_drag, ) ingestion_cfg = IngestionConfig( thresholds=thresholds, chunk_size=args.chunk_size, checkpoint_interval=args.checkpoint_interval, max_chunks=args.max_ingest_chunks, progress_enabled=args.progress, chunk_overlap=args.chunk_overlap, batch_size=args.batch_size, ) if args.adaptive_filter: ingestion_cfg.adaptive = ingestion_cfg.adaptive.__class__( enabled=True, warmup_chunks=args.adaptive_warmup, evaluation_window=args.adaptive_window, enable_ratio=args.adaptive_enable_ratio, disable_ratio=args.adaptive_disable_ratio, cooldown_chunks=args.adaptive_cooldown, ) config = ConversationConfig( response_tokens=args.response_tokens, auto_save_path=args.autosave, ingestion=ingestion_cfg, learning_enabled=not args.no_learning, behavior_scaffolds_path=args.behavior_scaffolds, ) session = ConversationSession(config=config) unsubscribe = None if args.telemetry: def _telemetry_sink(event: TelemetryEvent) -> None: if isinstance(event, ParameterEvent): print(render_parameter_event(event), file=sys.stderr) elif event.type == "training/step": for line in render_training_event(event): print(line, file=sys.stderr) elif event.type == "ingestion/chunk": from ..telemetry import render_ingestion_event print(render_ingestion_event(event), file=sys.stderr) unsubscribe = BUS.subscribe(_telemetry_sink) corpus_paths = [] if args.corpus: corpus_paths.extend(args.corpus) if args.load_dv1: corpus_paths.append(os.path.join("C:/DV1")) def progress_cb(payload: dict) -> None: if not args.progress: return meta = payload["metadata"] stats = payload["stats"] estimate = payload.get("estimate", {}) eta = estimate.get("eta", "unknown") total_chunks = estimate.get("chunks", 0) or 1 pct = min(100.0, 100.0 * stats["chunks_processed"] / total_chunks) line = ( f"[ingest] {pct:5.1f}% | file {meta['file_index']}/{estimate.get('files', '?')} | " f"chunk {meta['chunk_index']}/{meta['chunk_total']} | accepted={stats['chunks_accepted']} " f"rejected={stats['chunks_rejected']} | eta={eta}" ) print(line, end="\r", file=sys.stderr, flush=True) for path in corpus_paths: print(f"[bootstrapping] ingesting corpus at {path}") stats = session.ingest_directory(path, max_chunks=args.max_ingest_chunks, progress_cb=progress_cb) if args.progress: print("", file=sys.stderr) print( f"[ingest] completed {stats.chunks_processed} chunks " f"(accepted={stats.chunks_accepted}, rejected={stats.chunks_rejected})" ) print("SalienceOS Emergent Chat — type 'exit' to quit") exchanges = 0 while True: try: user_input = input("you> ").strip() except EOFError: break if user_input.lower() in {"exit", "quit"}: break session.process_user_input(user_input) snapshot = session.generate_response() if not args.quiet: print(f"[meta] {snapshot.meta_report}") print(f"agent> {snapshot.response}") exchanges += 1 if args.steps is not None and exchanges >= args.steps: break if args.autosave: saved = session.save_state(args.autosave) print(f"State saved to {saved}") if unsubscribe: unsubscribe() if __name__ == "__main__": main()