Spaces:
Running
Running
File size: 36,855 Bytes
07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 3647f22 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 3647f22 07ad116 3647f22 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 3647f22 07ad116 51e7f28 07ad116 3647f22 07ad116 3647f22 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 3647f22 07ad116 3647f22 07ad116 51e7f28 07ad116 3647f22 07ad116 3647f22 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 51e7f28 07ad116 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 |
# multimodular_modul version 7.0.py
"""
Multimodular Module β Multimodal SuperAgent v7.0
Upgrade from v6.0 -> v7.0 (in-place upgrade)
Features added:
- CHB (Close-to-Human Brain) is universal "middle-man" for all inputs/outputs.
- User-driven retrieval plan flow (AI generates queries; client fetches; submits results).
- Opportunistic Creative Skill Vault (media & text benchmarks).
- Versioned Fact Store (VFS) enhancements: freshness, controversy scoring, provenance.
- Real-time Global Brain (facts, skills, media) sync via WebSocket + optional HuggingFace dataset push.
- Automatic local backups (JSON + SQLite), downloadable via API endpoint.
- Self-upgrading modules: safe, signed modules can be auto-integrated into runtime.
* Auto-exec only when cryptographic signature verification + sandbox available (wasmtime or subprocess sandboxes).
* If signing or sandbox not present, modules are stored but not auto-executed.
- Universal FastAPI endpoints and CLI demo preserved from v6 with additions.
Security & Safety:
- No CAPTCHA/TLS evasion. Respect robots.txt & user-driven retrieval model.
- Self-upgrade requires signature verification (env GLOBAL_SYNC_SIGNING_PUBKEY).
- Default: local-only sync. Enable cloud via env variables (HUGGINGFACE_TOKEN, GLOBAL_SYNC_REPO).
- Media sync allowed; personal/private data must be filtered before upload.
"""
from __future__ import annotations
import os, sys, json, time, uuid, shutil, tempfile, hashlib, base64, logging
from dataclasses import dataclass, field, asdict
from typing import Any, Dict, List, Optional, Tuple
from pathlib import Path
import threading
import sqlite3
import zipfile
import hmac
import hashlib
import asyncio
# --------------------------
# Optional deps (feature unlocks)
# --------------------------
# pip install fastapi uvicorn pydantic requests websockets python-multipart cryptography wasmtime
try:
import fastapi
from fastapi import FastAPI, UploadFile, File, Form
from fastapi.responses import FileResponse, JSONResponse
from pydantic import BaseModel
FASTAPI_AVAILABLE = True
except Exception:
FASTAPI_AVAILABLE = False
try:
import requests
except Exception:
requests = None
try:
import websockets
except Exception:
websockets = None
try:
from cryptography.hazmat.primitives import serialization, hashes
from cryptography.hazmat.primitives.asymmetric import padding
CRYPTO_AVAILABLE = True
except Exception:
CRYPTO_AVAILABLE = False
try:
import wasmtime
WASM_AVAILABLE = True
except Exception:
WASM_AVAILABLE = False
# --------------------------
# Logging
# --------------------------
logging.basicConfig(level=logging.INFO, format="%(asctime)s %(levelname)s: %(message)s")
log = logging.getLogger("Multimodular_v7")
# --------------------------
# Config & Paths (edit env or constants)
# --------------------------
BASE_DIR = Path(os.getenv("MM_BASE_DIR", Path(__file__).parent.resolve()))
DATA_DIR = Path(os.getenv("MM_DATA_DIR", BASE_DIR / "mm_data"))
BACKUP_DIR = Path(os.getenv("MM_BACKUP_DIR", DATA_DIR / "backups"))
TMP_DIR = Path(os.getenv("MM_TMP_DIR", BASE_DIR / "tmp"))
CACHE_DIR = Path(os.getenv("MM_CACHE_DIR", BASE_DIR / "cache"))
for d in (DATA_DIR, BACKUP_DIR, TMP_DIR, CACHE_DIR):
d.mkdir(parents=True, exist_ok=True)
# Global sync config
GLOBAL_SYNC_ENABLED = os.getenv("MM_GLOBAL_SYNC_ENABLED", "false").lower() in ("1","true","yes")
HUGGINGFACE_TOKEN = os.getenv("HUGGINGFACE_TOKEN", None)
GLOBAL_SYNC_REPO = os.getenv("GLOBAL_SYNC_REPO", None) # e.g., "username/mm_global_brain"
GLOBAL_SYNC_SIGNING_PUBKEY = os.getenv("GLOBAL_SYNC_SIGNING_PUBKEY", None) # PEM public key for verifying modules
REALTIME_WS_PORT = int(os.getenv("MM_WS_PORT", "8765"))
# Auto-upgrade strictness: require signature & sandbox for auto-exec
AUTO_UPGRADE_REQUIRE_SIGN = True
AUTO_UPGRADE_REQUIRE_SANDBOX = True
# Backups
BACKUP_RETENTION = int(os.getenv("MM_BACKUP_RETENTION", "30")) # keep last N backups
# CHB confidence threshold
CHB_MIN_CONFIDENCE = float(os.getenv("CHB_MIN_CONFIDENCE", "0.85"))
# --------------------------
# Utilities
# --------------------------
def uid(prefix="id"):
return f"{prefix}_{uuid.uuid4().hex[:10]}"
def now_iso():
return time.strftime("%Y-%m-%dT%H:%M:%SZ", time.gmtime())
def sha256_b64(data: bytes) -> str:
return base64.urlsafe_b64encode(hashlib.sha256(data).digest()).decode()
def write_json(path: Path, data: Any):
tmp = path.with_suffix(".tmp")
with open(tmp, "w", encoding="utf-8") as f:
json.dump(data, f, ensure_ascii=False, indent=2)
tmp.replace(path)
def read_json(path: Path, default=None):
if not path.exists():
return default
try:
with open(path, "r", encoding="utf-8") as f:
return json.load(f)
except Exception:
return default
# --------------------------
# Database: local SQLite wrapper + JSON mirror
# --------------------------
class LocalDB:
def __init__(self, path: Path):
self.path = path
self.conn = sqlite3.connect(str(self.path))
self._init()
self.lock = threading.Lock()
def _init(self):
cur = self.conn.cursor()
cur.execute("""CREATE TABLE IF NOT EXISTS facts (
id TEXT PRIMARY KEY,
claim TEXT,
value TEXT,
confidence REAL,
sources TEXT,
first_seen TEXT,
last_seen TEXT,
controversy REAL,
staleness REAL
)""")
cur.execute("""CREATE TABLE IF NOT EXISTS skills (
id TEXT PRIMARY KEY,
tag TEXT,
kind TEXT,
meta TEXT,
score REAL,
added_at TEXT,
synced INTEGER DEFAULT 0
)""")
cur.execute("""CREATE TABLE IF NOT EXISTS modules (
id TEXT PRIMARY KEY,
name TEXT,
code TEXT,
meta TEXT,
verified INTEGER DEFAULT 0,
autointegrated INTEGER DEFAULT 0,
added_at TEXT
)""")
cur.execute("""CREATE TABLE IF NOT EXISTS backups (
id TEXT PRIMARY KEY,
path TEXT,
created_at TEXT
)""")
self.conn.commit()
def upsert_fact(self, claim, value, confidence, sources, controversy=0.0, staleness=0.0):
fid = sha256_b64(claim.encode())[:32]
now = now_iso()
with self.lock:
cur = self.conn.cursor()
cur.execute("SELECT id FROM facts WHERE id=?", (fid,))
if cur.fetchone():
cur.execute("""UPDATE facts SET value=?, confidence=?, sources=?, last_seen=?, controversy=?, staleness=? WHERE id=?""",
(value, float(confidence), json.dumps(sources), now, float(controversy), float(staleness), fid))
else:
cur.execute("""INSERT INTO facts (id,claim,value,confidence,sources,first_seen,last_seen,controversy,staleness)
VALUES (?,?,?,?,?,?,?,?,?)""",
(fid, claim, value, float(confidence), json.dumps(sources), now, now, float(controversy), float(staleness)))
self.conn.commit()
return fid
def add_skill(self, tag, kind, meta, score):
sid = uid("skill")
now = now_iso()
with self.lock:
self.conn.execute("INSERT INTO skills (id,tag,kind,meta,score,added_at) VALUES (?,?,?,?,?,?)",
(sid, tag, kind, json.dumps(meta), float(score), now))
self.conn.commit()
return sid
def add_module(self, name, code, meta, verified=0, autointegrated=0):
mid = uid("mod")
now = now_iso()
with self.lock:
self.conn.execute("INSERT INTO modules (id,name,code,meta,verified,autointegrated,added_at) VALUES (?,?,?,?,?,?,?)",
(mid, name, code, json.dumps(meta), int(verified), int(autointegrated), now))
self.conn.commit()
return mid
def list_facts(self):
cur = self.conn.cursor(); cur.execute("SELECT * FROM facts"); rows=cur.fetchall()
cols=[c[0] for c in cur.description]
return [dict(zip(cols, r)) for r in rows]
def list_skills(self):
cur = self.conn.cursor(); cur.execute("SELECT * FROM skills"); rows=cur.fetchall()
cols=[c[0] for c in cur.description]
return [dict(zip(cols, r)) for r in rows]
def list_modules(self):
cur = self.conn.cursor(); cur.execute("SELECT * FROM modules"); rows=cur.fetchall()
cols=[c[0] for c in cur.description]
return [dict(zip(cols, r)) for r in rows]
def mark_module_verified(self, module_id, verified=1):
with self.lock:
self.conn.execute("UPDATE modules SET verified=? WHERE id=?", (int(verified), module_id))
self.conn.commit()
def mark_module_autointegrated(self, module_id, val=1):
with self.lock:
self.conn.execute("UPDATE modules SET autointegrated=? WHERE id=?", (int(val), module_id))
self.conn.commit()
def add_backup(self, path):
bid = uid("bak")
now = now_iso()
with self.lock:
self.conn.execute("INSERT INTO backups (id,path,created_at) VALUES (?,?,?)", (bid, str(path), now))
self.conn.commit()
self._prune_backups()
return bid
def _prune_backups(self):
cur = self.conn.cursor(); cur.execute("SELECT id,path,created_at FROM backups ORDER BY created_at DESC")
rows = cur.fetchall()
if len(rows) <= BACKUP_RETENTION: return
for r in rows[BACKUP_RETENTION:]:
pid, p, _ = r
try:
if os.path.exists(p): os.remove(p)
except Exception:
pass
self.conn.execute("DELETE FROM backups WHERE id=?", (pid,))
self.conn.commit()
# --------------------------
# VFS and Creative Skill Vault (JSON + SQLite)
# --------------------------
class VFS:
def __init__(self, db: LocalDB):
self.db = db
def store_fact(self, claim:str, value:str, sources:List[Dict[str,Any]], confidence:float, controversy:float=0.0):
# staleness computed from source dates (simple)
staleness = 0.0
for s in (sources or []):
dt = s.get("date")
if dt:
try:
# naive parse as ISO; compute days
t = time.mktime(time.strptime(dt[:19], "%Y-%m-%dT%H:%M:%S"))
age_days = max(0, (time.time() - t)/86400.0)
staleness = max(staleness, min(1.0, age_days/365.0))
except Exception:
continue
fid = self.db.upsert_fact(claim, value, confidence, sources, controversy, staleness)
return fid
def query(self, q:str):
# naive substring search
res = self.db.list_facts()
qlow = q.lower()
return [r for r in res if qlow in (r.get("claim") or "").lower() or qlow in (r.get("value") or "").lower()]
class CreativeSkillVault:
def __init__(self, db: LocalDB):
self.db = db
def add_benchmark(self, tag:str, kind:str, meta:Dict[str,Any], score:float):
sid = self.db.add_skill(tag, kind, meta, score)
return sid
def top_by_tag(self, tag:str, k:int=5):
all_skills = self.db.list_skills()
filtered = [s for s in all_skills if s.get("tag")==tag]
filtered.sort(key=lambda x: x.get("score",0), reverse=True)
return filtered[:k]
# --------------------------
# Global Sync: Hugging Face push & WebSocket real-time (simple)
# --------------------------
class GlobalSync:
def __init__(self, db: LocalDB, hf_token:Optional[str]=None, repo:Optional[str]=None):
self.db = db
self.hf_token = hf_token
self.repo = repo
self.ws_clients = set()
self.ws_server_task = None
self.loop = None
self.lock = threading.Lock()
# --- push facts/skills package to Hugging Face dataset via simple HTTP (requires token & repo)
def push_to_hf(self, package:Dict[str,Any]) -> Tuple[bool,str]:
if not (self.hf_token and self.repo and requests):
return False, "huggingface not configured or requests missing"
# Minimal implementation: upload JSON file to HF repo via API
try:
url = f"https://huggingface.co/api/repos/create"
# Note: full implementation requires use of hf_hub or dataset APIs; here we do a simple placeholder
# We recommend using huggingface_hub library in production.
return False, "HF push requires huggingface_hub implementation; configure HF client"
except Exception as e:
return False, str(e)
# --- broadcast to connected WebSocket clients (realtime)
async def ws_broadcast(self, message:Dict[str,Any]):
if websockets is None:
return
data = json.dumps(message)
clients = list(self.ws_clients)
for ws in clients:
try:
await ws.send(data)
except Exception:
try:
self.ws_clients.remove(ws)
except Exception:
pass
# --- start a simple websocket server to accept other CHBs / clients that want live updates
def start_ws_server(self, host="0.0.0.0", port=REALTIME_WS_PORT):
if websockets is None:
log.warning("websockets library missing; realtime sync disabled")
return
async def handler(websocket, path):
log.info("WS client connected")
self.ws_clients.add(websocket)
try:
async for msg in websocket:
# accept 'ping' or 'submit' messages
try:
data = json.loads(msg)
typ = data.get("type")
if typ == "submit_skill":
payload = data.get("payload")
# minimal processing: store skill locally and broadcast
tag = payload.get("tag","global")
kind = payload.get("kind","image")
meta = payload.get("meta",{})
score = float(payload.get("score", 0.5))
self.db.add_skill(tag, kind, meta, score)
await self.ws_broadcast({"type":"skill_added","tag":tag,"kind":kind,"meta":meta,"score":score})
except Exception:
pass
except Exception:
pass
finally:
try: self.ws_clients.remove(websocket)
except Exception: pass
log.info("WS client disconnected")
log.info("Starting WebSocket server on %s:%d", host, port)
self.loop = asyncio.new_event_loop()
asyncio.set_event_loop(self.loop)
start_server = websockets.serve(handler, host, port)
self.ws_server_task = self.loop.run_until_complete(start_server)
try:
self.loop.run_forever()
except Exception:
pass
def run_ws_in_thread(self, host="0.0.0.0", port=REALTIME_WS_PORT):
t = threading.Thread(target=self.start_ws_server, args=(host,port), daemon=True)
t.start()
return t
# --------------------------
# ModuleManager: Verify & Sandbox auto-integration of learned modules
# --------------------------
class ModuleManager:
def __init__(self, db: LocalDB, signing_pubkey_pem:Optional[str]=None):
self.db = db
self.signing_pubkey_pem = signing_pubkey_pem
self.sandbox_available = WASM_AVAILABLE # prefer WASM sandbox if available
self.lock = threading.Lock()
def verify_signature(self, code: bytes, signature_b64: str) -> bool:
if not (CRYPTO_AVAILABLE and self.signing_pubkey_pem):
log.warning("Crypto or public key not available, cannot verify signature")
return False
try:
pub = serialization.load_pem_public_key(self.signing_pubkey_pem.encode())
sig = base64.b64decode(signature_b64)
pub.verify(sig, code, padding.PKCS1v15(), hashes.SHA256())
return True
except Exception as e:
log.warning("signature verification failed: %s", e)
return False
def sandbox_run_wasm(self, wasm_bytes: bytes, func_name: str="run", inputs: Optional[dict]=None, timeout: int=5) -> Tuple[bool,str]:
if not WASM_AVAILABLE:
return False, "wasm runtime not available"
try:
# create store & module
engine = wasmtime.Engine()
module = wasmtime.Module(engine, wasm_bytes)
store = wasmtime.Store(engine)
instance = wasmtime.Instance(store, module, [])
# This is a very conservative pattern β real WASM modules need standard interface; here we just attempt safe run if exposes memory/free functions
# For safety, we do not invoke arbitrary functions unless module authors follow the expected interface
# We'll attempt to call an exported function named 'run' that returns int
if hasattr(instance.exports, func_name):
fn = instance.exports.__getattr__(func_name)
try:
res = fn()
return True, f"wasm-run-res:{res}"
except Exception as e:
return False, f"wasm-run-exc:{e}"
else:
return False, "wasm module lacks 'run' export"
except Exception as e:
return False, f"wasm-failed:{e}"
def sandbox_run_subprocess(self, code_str: str, timeout: int=5) -> Tuple[bool,str]:
# Very limited subprocess sandbox: write file, run in subprocess with restricted env and timeout.
# NOTE: this is not fully secure against malicious code. Use real OS-level sandboxing for production.
tmp = Path(TMP_DIR) / f"module_{uid()}.py"
tmp.write_text(code_str, encoding="utf-8")
import subprocess, shlex
try:
p = subprocess.run([sys.executable, str(tmp)], stdout=subprocess.PIPE, stderr=subprocess.PIPE, timeout=timeout, check=False)
out = p.stdout.decode()[:4000]
err = p.stderr.decode()[:2000]
return True, out + ("\nERR:\n" + err if err else "")
except subprocess.TimeoutExpired:
return False, "timeout"
except Exception as e:
return False, f"exec-error:{e}"
finally:
try: tmp.unlink()
except Exception: pass
def integrate_module(self, name: str, code: str, signature_b64: Optional[str]=None, autointegrate: bool=True) -> Dict[str,Any]:
# Store module first
meta = {"name": name, "signature": bool(signature_b64), "autointegrate": bool(autointegrate)}
mid = self.db.add_module(name, code, meta, verified=0, autointegrated=0)
# Verify signature if present and required
if AUTO_UPGRADE_REQUIRE_SIGN:
if not signature_b64 or not self.verify_signature(code.encode(), signature_b64):
return {"ok": False, "reason": "signature_missing_or_invalid", "module_id": mid}
# Sandbox-run tests
ran_ok = False; run_info = None
if self.sandbox_available and AUTO_UPGRADE_REQUIRE_SANDBOX:
# expect code to be WASM base64 (prefer) or python code string. Detect if code is base64 wasm by heuristic.
try:
# try decode base64, check for wasm magic
raw = base64.b64decode(code)
if raw[:4] == b"\x00asm":
ok, info = self.sandbox_run_wasm(raw)
ran_ok, run_info = ok, info
else:
# treat as python source
ok, info = self.sandbox_run_subprocess(code)
ran_ok, run_info = ok, info
except Exception as e:
ran_ok, run_info = False, f"sandbox-error:{e}"
else:
# sandbox not available; do not autointegrate (security)
ran_ok, run_info = False, "sandbox-not-available"
# If sandboxed OK and autointegrate allowed, mark module autointegrated and (optionally) import into runtime
if ran_ok and autointegrate:
self.db.mark_module_verified(mid, verified=1)
# For safety, we will NOT exec arbitrary Python into this process.
# Instead, save module to disk as a safe package and mark autointegrated. A separate process can load it.
self.db.mark_module_autointegrated(mid, val=1)
return {"ok": True, "module_id": mid, "sandbox_result": run_info}
else:
return {"ok": False, "module_id": mid, "sandbox_result": run_info}
# --------------------------
# CHB β Universal middleman (upgrades v6 behaviour)
# --------------------------
class CHB:
def __init__(self, db: LocalDB, vfs: VFS, csv: CreativeSkillVault, module_mgr: ModuleManager, global_sync: GlobalSync):
self.db = db
self.vfs = vfs
self.csv = csv
self.module_mgr = module_mgr
self.global_sync = global_sync
# lightweight internal state
self.min_conf = CHB_MIN_CONFIDENCE
def perceive(self, incoming: Dict[str,Any]) -> Dict[str,Any]:
# normalize inputs (text,image,audio,video,plan_results)
kind = "text"
if incoming.get("image") or incoming.get("image_path"): kind="image"
if incoming.get("audio") or incoming.get("audio_path"): kind="audio"
if incoming.get("video") or incoming.get("video_path"): kind="video"
if incoming.get("plan_results"): kind="plan_results"
return {"kind": kind, "payload": incoming}
def plan(self, text: str) -> Dict[str,Any]:
# produce a user-driven retrieval plan
queries = []
queries.append({"q": text, "type":"web", "max_results":5})
# image/video heuristics
if any(k in text.lower() for k in ["image","design","render","photo","logo","illustration","concept"]):
queries.append({"q": text + " high quality concept art", "type":"image", "max_results":8})
if any(k in text.lower() for k in ["video","footage","tour","demo","walkthrough"]):
queries.append({"q": text + " video", "type":"video", "max_results":4})
plan = {"id": uid("plan"), "queries": queries, "created_at": now_iso(), "instructions": "Run these queries locally on user's device and return structured results (web/images/videos/audio)."}
return plan
def verify(self, plan_results: Optional[Dict[str,Any]], local_tool_outputs: Optional[List[Dict[str,Any]]]=None) -> Dict[str,Any]:
# compute reliability, controversy, citations
citations = []
reliability = 0.4
controversy = 0.0
if plan_results:
web = plan_results.get("web") or []
domains = set()
for r in web:
u = r.get("url") or r.get("link") or ""
domains.add(u.split("/")[2] if "/" in u else u)
citations.append({"title": r.get("title"), "url": u, "date": r.get("date")})
reliability = min(1.0, 0.2 + 0.1*len(domains))
# controversy: if two top results contradict in short text heuristics
if len(web) >= 2:
s0 = web[0].get("snippet","").lower()
s1 = web[1].get("snippet","").lower()
if any(w in s1 for w in ["not", "contradict", "dispute"]) or any(w in s0 for w in ["not","contradict","dispute"]):
controversy = 0.5
# tool outputs contribution
tool_bonus = 0.0
for t in (local_tool_outputs or []):
if t.get("tool") == "image" and t.get("ok"): tool_bonus += 0.2
if t.get("tool") == "math" and not t.get("result","").lower().startswith("math error"): tool_bonus += 0.2
confidence = min(1.0, reliability*0.6 + tool_bonus)
return {"confidence": confidence, "reliability": reliability, "controversy": controversy, "citations": citations}
def opportunistic_learning(self, plan_results: Dict[str,Any]):
# extract images/videos/audio and store as skill benchmarks if quality high
images = plan_results.get("images", []) or []
for im in images:
path = im.get("path") or im.get("url")
# naive quality score
score = float(im.get("quality_score", 0.6))
tag = im.get("tags",[ "web" ])[0] if im.get("tags") else "web"
meta = {"source": path, "desc": im.get("caption") or im.get("alt") or "", "origin":"user_client"}
self.csv.add_benchmark(tag, "image", meta, score)
videos = plan_results.get("videos", []) or []
for v in videos:
path = v.get("path") or v.get("url")
score = float(v.get("quality_score", 0.6))
tag = v.get("tags",[ "web" ])[0] if v.get("tags") else "web"
meta = {"source": path, "desc": v.get("caption") or "", "origin":"user_client"}
self.csv.add_benchmark(tag, "video", meta, score)
audios = plan_results.get("audios", []) or []
for a in audios:
path = a.get("path") or a.get("url")
score = float(a.get("quality_score", 0.6))
tag = a.get("tags",[ "web" ])[0] if a.get("tags") else "web"
meta = {"source": path, "desc": a.get("caption") or "", "origin":"user_client"}
self.csv.add_benchmark(tag, "audio", meta, score)
# push to global sync immediately
if GLOBAL_SYNC_ENABLED:
payload = {"type":"skill_update", "time": now_iso(), "added": len(images)+len(videos)+len(audios)}
# best-effort: broadcast via websocket
if self.global_sync:
loop = asyncio.new_event_loop()
try:
loop.run_until_complete(self.global_sync.ws_broadcast(payload))
except Exception:
pass
def handle_plan_results(self, plan_id: str, plan_results: Dict[str,Any], local_tool_outputs: Optional[List[Dict[str,Any]]]=None):
# verify, opportunistic learn, store facts in VFS
v = self.verify(plan_results, local_tool_outputs)
# store simple fact example: top web title as a fact
web = plan_results.get("web", []) or []
if web:
top = web[0]
claim = top.get("title","").strip()
value = top.get("snippet","").strip()
sources = [{"url": top.get("url") or top.get("link"), "title": top.get("title"), "date": top.get("date")}]
self.vfs.store_fact(claim, value, sources, float(v.get("confidence",0.4)), controversy=v.get("controversy",0.0))
# opportunistic learning
self.opportunistic_learning(plan_results)
return v
# --------------------------
# Auto-backup & Export
# --------------------------
class BackupManager:
def __init__(self, db: LocalDB, data_dir: Path, backup_dir: Path):
self.db = db
self.data_dir = data_dir
self.backup_dir = backup_dir
self.lock = threading.Lock()
def create_backup(self) -> str:
with self.lock:
ts = now_iso().replace(":", "-")
out_path = self.backup_dir / f"mm_backup_{ts}.zip"
with zipfile.ZipFile(out_path, "w", zipfile.ZIP_DEFLATED) as zf:
# include SQLite DB file if exists
try:
dbf = Path(self.db.path)
if dbf.exists():
zf.write(str(dbf), arcname=dbf.name)
except Exception:
pass
# include JSON DB mirrors (facts/skills/modules)
# try to export via DB list functions
try:
facts = self.db.list_facts()
skills = self.db.list_skills()
mods = self.db.list_modules()
zf.writestr("facts.json", json.dumps(facts, ensure_ascii=False, indent=2))
zf.writestr("skills.json", json.dumps(skills, ensure_ascii=False, indent=2))
zf.writestr("modules.json", json.dumps(mods, ensure_ascii=False, indent=2))
except Exception:
pass
bid = self.db.add_backup(str(out_path))
return str(out_path)
def download_backup_path(self) -> Optional[str]:
# return latest backup path
cur = self.db.conn.cursor()
cur.execute("SELECT id,path,created_at FROM backups ORDER BY created_at DESC LIMIT 1")
r = cur.fetchone()
if not r: return None
return r[1]
# --------------------------
# Main Orchestrator (upgrades v6.SuperAgent)
# --------------------------
class SuperAgentV7:
def __init__(self):
self.db_path = DATA_DIR / "multimodular_v7.db"
self.db = LocalDB(self.db_path)
self.vfs = VFS(self.db)
self.csv = CreativeSkillVault(self.db)
self.global_sync = GlobalSync(self.db, hf_token=HUGGINGFACE_TOKEN, repo=GLOBAL_SYNC_REPO)
# module manager uses GLOBAL_SYNC_SIGNING_PUBKEY if present
self.module_mgr = ModuleManager(self.db, signing_pubkey_pem=GLOBAL_SYNC_SIGNING_PUBKEY)
self.chb = CHB(self.db, self.vfs, self.csv, self.module_mgr, self.global_sync)
self.backup_mgr = BackupManager(self.db, DATA_DIR, BACKUP_DIR)
# start realtime ws server in background for incoming updates
if websockets is not None:
try:
self.global_sync.run_ws_in_thread()
except Exception as e:
log.warning("ws server thread failed: %s", e)
# --- user-driven search plan creation (CHB.plan)
def plan_search(self, text: str) -> Dict[str,Any]:
return self.chb.plan(text)
# --- client submits results
def submit_plan_results(self, plan_id: str, results: Dict[str,Any]) -> Dict[str,Any]:
# Accept results; pass to CHB for verify & learning
v = self.chb.handle_plan_results(plan_id, results)
# create backup after major update
try:
bp = self.backup_mgr.create_backup()
log.info("backup created: %s", bp)
except Exception:
log.exception("backup failed")
# push to global sync (best-effort)
if GLOBAL_SYNC_ENABLED:
try:
payload = {"type":"fact_skill_update", "ts": now_iso(), "payload": {"summary":"update","plan_id": plan_id}}
# attempt HF push or WS broadcast
# HF push requires implementation using huggingface_hub; we leave a placeholder
asyncio.run(self.global_sync.ws_broadcast(payload))
except Exception:
pass
return {"ok": True, "verify": v}
# --- expose facts/skills
def search_facts(self, q: str):
return self.vfs.query(q)
def top_skills(self, tag: str, k: int=5):
return self.csv.top_by_tag(tag, k)
def add_module(self, name: str, code: str, signature_b64: Optional[str]=None, autointegrate: bool=True):
return self.module_mgr.integrate_module(name, code, signature_b64, autointegrate)
def download_latest_backup(self) -> Optional[str]:
return self.backup_mgr.download_backup_path()
# direct chat route that funnels through CHB
def chat(self, text: str, plan_results: Optional[Dict[str,Any]]=None):
# For direct chat, if user provided plan_results (client retrieval), handle them
if plan_results:
v = self.chb.handle_plan_results(uid("plan"), plan_results)
return {"status":"ok", "verify": v}
# Simple path: CHB will create plan if needed and answer (we re-use simple plan + verify)
plan = self.chb.plan(text)
# no client retrieval performed: CHB can still reply with local knowledge (fallback)
# For v7 we simply return plan and ask client to run it OR CHB will attempt local infer (fallback)
return {"status":"ok", "plan": plan, "hint": "Run the plan on client and submit results via submit_plan_results"}
# --------------------------
# FastAPI endpoints (added/extended)
# --------------------------
if FASTAPI_AVAILABLE:
app = FastAPI(title="Multimodular SuperAgent v7.0")
AGENT = SuperAgentV7()
class PlanIn(BaseModel):
text: str
@app.post("/v1/plan_search")
async def api_plan_search(inp: PlanIn):
plan = AGENT.plan_search(inp.text)
return {"ok": True, "plan": plan}
@app.post("/v1/submit_results")
async def api_submit_results(plan_id: str = Form(...), results: str = Form(...)):
try:
payload = json.loads(results)
except Exception:
return JSONResponse({"ok": False, "error": "invalid_json"}, status_code=400)
out = AGENT.submit_plan_results(plan_id, payload)
return out
@app.post("/v1/facts/search")
async def api_facts_search(q: str = Form(...)):
res = AGENT.search_facts(q)
return {"ok": True, "results": res}
@app.post("/v1/skills/top")
async def api_skills_top(tag: str = Form(...), k: int = Form(5)):
res = AGENT.top_skills(tag, k)
return {"ok": True, "results": res}
@app.post("/v1/module/add")
async def api_module_add(name: str = Form(...), code: str = Form(...), signature_b64: Optional[str] = Form(None)):
out = AGENT.add_module(name, code, signature_b64)
return out
@app.get("/v1/backup/download")
async def api_backup_download():
p = AGENT.download_latest_backup()
if not p or not os.path.exists(p):
return JSONResponse({"ok": False, "error": "no_backup"}, status_code=404)
return FileResponse(p, media_type="application/zip", filename=os.path.basename(p))
@app.post("/v1/chat")
async def api_chat(text: str = Form(...), plan_results: Optional[str] = Form(None)):
if plan_results:
try:
pr = json.loads(plan_results)
except Exception:
return JSONResponse({"ok": False, "error": "invalid_plan_results"}, status_code=400)
out = AGENT.chat(text, pr)
else:
out = AGENT.chat(text, None)
return out
else:
app = None
AGENT = SuperAgentV7()
# --------------------------
# CLI demo & execution
# --------------------------
def demo_run():
print("Multimodular SuperAgent v7.0 - Demo")
a = AGENT
# 1) plan search
plan = a.plan_search("futuristic electric motorcycle neon blue lights battery tech 2025")
print("Plan:", plan)
# Simulate client retrieval
sample_results = {
"web": [
{"title":"Solid-state battery prototype reaches 500 Wh/kg", "url":"https://example.org/article", "snippet":"Researchers at X report 500 Wh/kg...", "date": now_iso()}
],
"images":[
{"path": str(BASE_DIR / "demo_motorcycle.png"), "quality_score": 0.92, "caption":"Futuristic motorcycle concept", "tags":["motorcycle","futuristic"]}
],
"videos": [],
"audios": []
}
# ensure demo image exists
try:
from PIL import Image, ImageDraw
img = Image.new("RGB", (640,480), (20,20,30))
d = ImageDraw.Draw(img); d.text((20,20), "Demo motorcycle", fill=(200,200,255))
img.save(BASE_DIR / "demo_motorcycle.png")
except Exception:
pass
res = a.submit_plan_results(plan["id"], sample_results)
print("Submit results ->", res)
# show facts
facts = a.search_facts("solid-state")
print("Facts:", facts)
# download backup
path = a.download_latest_backup()
print("Backup path:", path)
if __name__ == "__main__":
import argparse
ap = argparse.ArgumentParser()
ap.add_argument("--demo", action="store_true")
ap.add_argument("--runserver", action="store_true")
ap.add_argument("--port", type=int, default=8000)
args = ap.parse_args()
if args.demo:
demo_run()
elif args.runserver and FASTAPI_AVAILABLE:
import uvicorn
uvicorn.run("multimodular_modul version 7.0:app", host="0.0.0.0", port=args.port, reload=False)
else:
print("Run with --demo or --runserver. FASTAPI available:", FASTAPI_AVAILABLE)
|