Spaces:
Sleeping
Sleeping
File size: 68,960 Bytes
9d6015f f41653d 9d6015f f41653d 9d6015f f41653d f266ceb f41653d 9d6015f f41653d b4b0135 f41653d b4b0135 f41653d cd9e047 25899d8 cd9e047 25899d8 cd9e047 f41653d 9d6015f f41653d 9d6015f f266ceb aaaba76 f266ceb aaaba76 f266ceb aaaba76 f266ceb aaaba76 f266ceb aaaba76 f266ceb aaaba76 f266ceb 9d6015f f41653d 9d6015f f41653d 9d6015f f41653d cd9e047 f41653d 9d6015f f41653d aaaba76 f41653d aaaba76 f41653d 9d6015f f41653d 9d6015f f41653d cd9e047 f41653d 9d6015f f41653d 9d6015f f41653d aaaba76 f41653d aaaba76 28f9d0d aaaba76 28f9d0d 9d6015f 28f9d0d 9d6015f f41653d 9d6015f f41653d 9d6015f cd9e047 f41653d f266ceb f41653d 5c130f4 f41653d 9d6015f 5c130f4 9d6015f 5c130f4 9d6015f 5c130f4 9d6015f cd9e047 5c130f4 9d6015f f41653d 9d6015f f41653d 9d6015f f41653d aaaba76 f41653d aaaba76 f266ceb aaaba76 f266ceb f41653d f266ceb 9d6015f f266ceb 6648afb f266ceb 9d6015f 6648afb f266ceb 9d6015f f41653d 9d6015f f41653d cd9e047 f41653d 9d6015f f41653d 9d6015f f41653d 9d6015f f41653d 9d6015f f41653d cd9e047 f41653d cd9e047 f41653d cd9e047 f41653d cd9e047 f41653d ffafd0f 1f8f6e9 f41653d 2eb5a6d 18aa80e 2eb5a6d f41653d bf45da8 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 |
# # app.py - MasterLLM v2.0 with Bedrock Fallback System
# """
# MasterLLM Pipeline Orchestrator v2.0
# - Bedrock (priority) + Gemini (fallback) for pipeline generation
# - Bedrock LangChain (priority) + CrewAI (fallback) for execution
# - MongoDB session management
# - Complete REST API
# - Gradio UI with fancy displays
# """
# import os
# import json
# import uuid
# from datetime import datetime
# from typing import List, Optional
# import gradio as gr
# from fastapi import FastAPI
# from fastapi.middleware.cors import CORSMiddleware
# from contextlib import asynccontextmanager
# import asyncio
# # Import our new services
# from services.pipeline_generator import generate_pipeline, format_pipeline_for_display
# from services.pipeline_executor import execute_pipeline_streaming
# from services.session_manager import session_manager
# from api_routes import router as api_router
# # ========================
# # BACKGROUND CLEANUP TASK
# # ========================
# async def periodic_cleanup():
# """Cleanup old sessions every hour"""
# while True:
# await asyncio.sleep(3600) # Run every hour
# try:
# removed = session_manager.cleanup_old_sessions(max_age_hours=24)
# if removed > 0:
# print(f"🧹 Cleaned up {removed} inactive sessions")
# except Exception as e:
# print(f"⚠️ Cleanup error: {e}")
# @asynccontextmanager
# async def lifespan(app: FastAPI):
# """Manage application lifecycle"""
# # Startup
# print("🚀 Starting MasterLLM v2.0...")
# task = asyncio.create_task(periodic_cleanup())
# yield
# # Shutdown
# task.cancel()
# session_manager.close()
# print("🛑 MasterLLM shut down gracefully")
# # ========================
# # FASTAPI APP
# # ========================
# app = FastAPI(
# title="MasterLLM v2.0 - AI Pipeline Orchestrator",
# description="Bedrock + Gemini fallback system with MongoDB sessions",
# version="2.0.0",
# lifespan=lifespan
# )
# # CORS Configuration
# app.add_middleware(
# CORSMiddleware,
# allow_origins=[os.getenv("FRONTEND_ORIGIN", "http://localhost:3000")],
# allow_credentials=True,
# allow_methods=["*"],
# allow_headers=["*"],
# )
# # Mount API routes
# app.include_router(api_router)
# # ========================
# # CONVERSATION STATE
# # ========================
# class ConversationState:
# INITIAL = "initial"
# PIPELINE_PROPOSED = "pipeline_proposed"
# PIPELINE_APPROVED = "pipeline_approved"
# EXECUTING = "executing"
# COMPLETED = "completed"
# ERROR = "error"
# # ========================
# # GRADIO UI HANDLERS
# # ========================
# def create_new_session():
# """Create a new session"""
# return session_manager.create_session()
# def handle_file_upload(file_path, session_id):
# """Handle file upload"""
# if not file_path:
# return None, json.dumps({
# "status": "error",
# "message": "No file uploaded"
# }, indent=2), session_id
# if not session_id:
# session_id = create_new_session()
# file_name = os.path.basename(file_path)
# # Update session
# session_manager.update_session(session_id, {
# "current_file": file_path,
# "state": ConversationState.INITIAL
# })
# # Add system message
# session_manager.add_message(
# session_id,
# "system",
# f"File uploaded: {file_name}"
# )
# status = {
# "status": "success",
# "message": f"File '{file_name}' uploaded successfully",
# "file_info": {
# "name": file_name,
# "path": file_path,
# "size_bytes": os.path.getsize(file_path) if os.path.exists(file_path) else 0
# },
# "next_action": "💬 Now tell me what you'd like to do with this document"
# }
# return file_path, json.dumps(status, indent=2), session_id
# def format_chat_history(history, new_user_msg, new_assistant_msg):
# """
# Convert chat history to new Gradio format (list of dicts with role/content)
# Handles both old format (tuples) and new format (dicts)
# """
# messages = []
# # Handle existing history - could be in old or new format
# if history:
# # Check if already in new format (list of dicts with 'role' and 'content')
# if isinstance(history[0], dict) and 'role' in history[0]:
# # Already in new format, just copy it
# messages = list(history)
# else:
# # Old format (list of tuples), convert it
# for item in history:
# if isinstance(item, (list, tuple)) and len(item) == 2:
# user_msg, bot_msg = item
# messages.append({"role": "user", "content": user_msg})
# messages.append({"role": "assistant", "content": bot_msg})
# # Add new messages
# messages.append({"role": "user", "content": new_user_msg})
# messages.append({"role": "assistant", "content": new_assistant_msg})
# return messages
# def chatbot_response_streaming(message: str, history: List, session_id: str, file_path: str = None):
# """
# Handle chat messages with streaming updates
# Uses Bedrock (priority) → Gemini (fallback) for both generation and execution
# """
# # Get or create session
# session = session_manager.get_session(session_id)
# if not session:
# session_id = create_new_session()
# session = session_manager.get_session(session_id)
# # Update file path if provided
# if file_path:
# session_manager.update_session(session_id, {"current_file": file_path})
# session = session_manager.get_session(session_id)
# # Add user message to session
# session_manager.add_message(session_id, "user", message)
# current_state = session.get("state", ConversationState.INITIAL)
# # ========================
# # STATE: INITIAL - Generate Pipeline
# # ========================
# if current_state == ConversationState.INITIAL:
# # Check if file is uploaded
# if not session.get("current_file"):
# response = {
# "status": "error",
# "message": "Please upload a document first",
# "action": "📁 Click 'Upload Document' to begin"
# }
# response_text = f"```json\n{json.dumps(response, indent=2)}\n```"
# session_manager.add_message(session_id, "assistant", response_text)
# yield format_chat_history(history, message, response_text)
# return
# try:
# # Generate pipeline using Bedrock → Gemini fallback
# yield format_chat_history(history, message, "🤖 Generating pipeline with AI...\n⏳ Trying Bedrock first...")
# pipeline = generate_pipeline(
# user_input=message,
# file_path=session.get("current_file"),
# prefer_bedrock=True
# )
# # Save proposed pipeline to session
# session_manager.update_session(session_id, {
# "proposed_pipeline": pipeline,
# "state": ConversationState.PIPELINE_PROPOSED
# })
# # Format for display
# formatted_display = format_pipeline_for_display(pipeline)
# # Create response with both fancy display and JSON
# response_text = formatted_display + f"\n\n```json\n{json.dumps(pipeline, indent=2)}\n```"
# session_manager.add_message(session_id, "assistant", response_text)
# yield format_chat_history(history, message, response_text)
# return
# except Exception as e:
# error_response = {
# "status": "error",
# "message": "Failed to generate pipeline",
# "error": str(e),
# "action": "Please try rephrasing your request"
# }
# response_text = f"```json\n{json.dumps(error_response, indent=2)}\n```"
# session_manager.add_message(session_id, "assistant", response_text)
# yield format_chat_history(history, message, response_text)
# return
# # ========================
# # STATE: PIPELINE_PROPOSED - Handle Approval/Rejection
# # ========================
# elif current_state == ConversationState.PIPELINE_PROPOSED:
# user_input = message.lower().strip()
# # APPROVE - Execute the pipeline
# if "approve" in user_input or "yes" in user_input:
# session_manager.update_session(session_id, {"state": ConversationState.EXECUTING})
# plan = session.get("proposed_pipeline", {})
# # Initial status
# initial_status = {
# "status": "executing",
# "message": "🚀 Starting pipeline execution...",
# "pipeline": plan.get("pipeline_name", "unknown"),
# "executor": "Attempting Bedrock LangChain first",
# "steps": []
# }
# accumulated_response = f"```json\n{json.dumps(initial_status, indent=2)}\n```"
# yield format_chat_history(history, message, accumulated_response)
# steps_completed = []
# final_payload = None
# executor_used = "unknown"
# try:
# # Execute pipeline with Bedrock → CrewAI fallback
# for event in execute_pipeline_streaming(
# pipeline=plan,
# file_path=session.get("current_file"),
# session_id=session_id,
# prefer_bedrock=True
# ):
# event_type = event.get("type")
# # Info events (fallback notifications, etc.)
# if event_type == "info":
# info_status = {
# "status": "info",
# "message": event.get("message"),
# "executor": event.get("executor", "unknown")
# }
# accumulated_response = f"```json\n{json.dumps(info_status, indent=2)}\n```"
# yield format_chat_history(history, message, accumulated_response)
# # Step updates
# elif event_type == "step":
# step_info = {
# "step": event.get("step", 0),
# "tool": event.get("tool", "processing"),
# "status": event.get("status", "running"),
# "executor": event.get("executor", "unknown")
# }
# # Add observation if available (tool output)
# if "observation" in event:
# step_info["observation"] = event.get("observation")
# # Add tool input if available
# if "input" in event:
# step_info["input"] = event.get("input")
# steps_completed.append(step_info)
# executor_used = event.get("executor", executor_used)
# # Create more informative status message
# status_message = f"📍 Step {event.get('step', 0)}: {event.get('tool', 'processing')}"
# if event.get('status') == 'completed' and 'observation' in event:
# obs_preview = str(event.get('observation'))[:100]
# status_message += f" ✅\n Output: {obs_preview}..."
# elif event.get('status') == 'executing':
# status_message += " ⏳"
# progress_status = {
# "status": "executing",
# "message": status_message,
# "pipeline": plan.get("pipeline_name", ""),
# "executor": executor_used,
# "current_step": step_info,
# "steps_completed": steps_completed
# }
# accumulated_response = f"```json\n{json.dumps(progress_status, indent=2)}\n```"
# yield format_chat_history(history, message, accumulated_response)
# # Final result
# elif event_type == "final":
# final_payload = event.get("data")
# executor_used = event.get("executor", executor_used)
# # Error
# elif event_type == "error":
# error_result = {
# "status": "failed",
# "error": event.get("error"),
# "steps_completed": steps_completed,
# "executor": event.get("executor", "unknown")
# }
# final_response = f"```json\n{json.dumps(error_result, indent=2)}\n```"
# session_manager.update_session(session_id, {"state": ConversationState.INITIAL})
# session_manager.add_message(session_id, "assistant", final_response)
# yield format_chat_history(history, message, final_response)
# return
# # Process final result
# if final_payload:
# session_manager.update_session(session_id, {
# "pipeline_result": final_payload,
# "state": ConversationState.INITIAL
# })
# # Save execution to MongoDB
# session_manager.save_pipeline_execution(
# session_id=session_id,
# pipeline=plan,
# result=final_payload,
# file_path=session.get("current_file"),
# executor=executor_used
# )
# # Format final response
# final_display = {
# "status": "completed",
# "executor": executor_used,
# "pipeline": plan.get("pipeline_name"),
# "result": final_payload,
# "summary": {
# "total_steps": len(steps_completed),
# "completed_successfully": len([s for s in steps_completed if s.get("status") == "completed"])
# }
# }
# final_response = f"```json\n{json.dumps(final_display, indent=2)}\n```"
# else:
# final_response = f"```json\n{json.dumps({'status': 'completed', 'steps': steps_completed, 'executor': executor_used}, indent=2)}\n```"
# session_manager.update_session(session_id, {"state": ConversationState.INITIAL})
# session_manager.add_message(session_id, "assistant", final_response)
# yield format_chat_history(history, message, final_response)
# return
# except Exception as e:
# error_result = {
# "error": str(e),
# "status": "failed",
# "message": "Pipeline execution failed",
# "steps_completed": steps_completed
# }
# final_response = f"```json\n{json.dumps(error_result, indent=2)}\n```"
# session_manager.update_session(session_id, {"state": ConversationState.INITIAL})
# session_manager.add_message(session_id, "assistant", final_response)
# yield format_chat_history(history, message, final_response)
# return
# # REJECT - Cancel the pipeline
# elif "reject" in user_input or "no" in user_input:
# session_manager.update_session(session_id, {
# "state": ConversationState.INITIAL,
# "proposed_pipeline": None
# })
# response_data = {
# "status": "rejected",
# "message": "Pipeline rejected by user",
# "action": "💬 Please provide a new instruction"
# }
# response = f"```json\n{json.dumps(response_data, indent=2)}\n```"
# session_manager.add_message(session_id, "assistant", response)
# yield format_chat_history(history, message, response)
# return
# # EDIT - Request modifications
# elif "edit" in user_input or "modify" in user_input:
# current_pipeline = session.get("proposed_pipeline", {})
# edit_help = {
# "status": "edit_mode",
# "message": "To modify the plan, describe your changes",
# "current_plan": current_pipeline,
# "examples": [
# "Add summarization at the end",
# "Remove table extraction",
# "Only process pages 1-3",
# "Translate to French instead of Spanish"
# ],
# "action": "Describe your changes, or say 'approve' to run as-is"
# }
# response = f"```json\n{json.dumps(edit_help, indent=2)}\n```"
# session_manager.add_message(session_id, "assistant", response)
# yield format_chat_history(history, message, response)
# return
# # Try to modify pipeline based on user input
# else:
# if len(message.strip()) > 5:
# try:
# original_plan = session.get("proposed_pipeline", {})
# edit_context = f"Original: {original_plan.get('pipeline_name')}. User wants: {message}"
# # Generate new pipeline with modification
# new_pipeline = generate_pipeline(
# user_input=edit_context,
# file_path=session.get("current_file"),
# prefer_bedrock=True
# )
# session_manager.update_session(session_id, {
# "proposed_pipeline": new_pipeline,
# "state": ConversationState.PIPELINE_PROPOSED
# })
# formatted = format_pipeline_for_display(new_pipeline)
# response = formatted + f"\n\n```json\n{json.dumps(new_pipeline, indent=2)}\n```"
# session_manager.add_message(session_id, "assistant", response)
# yield format_chat_history(history, message, response)
# return
# except Exception as e:
# error_response = {
# "status": "edit_failed",
# "error": str(e),
# "message": "Could not modify the plan",
# "action": "Try 'approve' to run as-is, or 'reject' to start over"
# }
# response = f"```json\n{json.dumps(error_response, indent=2)}\n```"
# session_manager.add_message(session_id, "assistant", response)
# yield format_chat_history(history, message, response)
# return
# # Default waiting message
# response_data = {
# "status": "waiting_for_confirmation",
# "message": "Please type 'approve', 'reject', or describe changes",
# "hint": "You can also say 'edit' for modification hints"
# }
# response = f"```json\n{json.dumps(response_data, indent=2)}\n```"
# session_manager.add_message(session_id, "assistant", response)
# yield format_chat_history(history, message, response)
# return
# # Default fallback
# response = json.dumps({"status": "ready", "message": "Ready for your next instruction"}, indent=2)
# session_manager.add_message(session_id, "assistant", response)
# yield format_chat_history(history, message, response)
# # ========================
# # GRADIO UI
# # ========================
# # Simple Blocks initialization for HF Spaces compatibility (older Gradio version)
# with gr.Blocks(title="MasterLLM v2.0 - AI Pipeline Orchestrator") as demo:
# gr.Markdown("""
# # 🤖 MasterLLM v2.0 - AI Pipeline Orchestrator
# **🏆 Bedrock Priority** with Gemini Fallback | **💾 MongoDB Sessions** | **📡 Complete REST API**
# Upload a document, describe what you want, and watch AI orchestrate the perfect pipeline!
# """)
# # State management
# session_id_state = gr.State(value=create_new_session())
# file_state = gr.State(value=None)
# with gr.Row():
# with gr.Column(scale=3):
# # Chat interface - Gradio auto-detects format from data structure
# chatbot = gr.Chatbot(label="Chat")
# # Text input
# msg = gr.Textbox(
# placeholder="💬 Type your instruction... (e.g., 'extract text from pages 1-5 and summarize')",
# label="Your Message",
# lines=2,
# max_lines=4,
# )
# with gr.Row():
# submit_btn = gr.Button("🚀 Send", variant="primary", scale=2)
# clear_btn = gr.Button("🗑️ Clear Chat", scale=1)
# with gr.Column(scale=1):
# # File upload section
# gr.Markdown("### 📁 Upload Document")
# file_upload = gr.File(
# label="PDF or Image",
# file_types=[".pdf", ".png", ".jpg", ".jpeg", ".gif", ".bmp"],
# type="filepath",
# )
# upload_status = gr.Textbox(
# label="📊 Upload Status",
# interactive=False,
# lines=10,
# max_lines=15,
# )
# # Session info
# gr.Markdown("### 🔗 Session Info")
# session_display = gr.Textbox(
# label="Session ID",
# interactive=False,
# value=lambda: session_id_state.value[:8] + "...",
# )
# # Examples
# gr.Markdown("### 💡 Example Pipelines")
# gr.Examples(
# examples=[
# "extract text from pages 1-5",
# "extract text and summarize",
# "extract text, tables, and translate to Spanish",
# "get tables from pages 2-4 and summarize",
# "text-classify-ner from entire document",
# "describe images and summarize findings",
# "extract text, detect signatures and stamps",
# ],
# inputs=msg,
# )
# # System info
# gr.Markdown("""
# ### ℹ️ System Features
# - ✅ **Bedrock** (Claude 3.5 Sonnet) priority
# - ✅ **Gemini** (gemini-2.0-flash) fallback
# - ✅ **MongoDB** session persistence
# - ✅ **Streaming** real-time updates
# - ✅ **Component-level** JSON output
# - ✅ **REST API** for integration
# ### 📊 Pipeline Flow:
# 1. **Upload** your document
# 2. **Describe** what you want
# 3. **Review** AI-generated pipeline
# 4. **Approve** to execute
# 5. **Watch** streaming updates
# 6. **Get** complete JSON results
# """)
# # Event handlers
# file_upload.upload(
# fn=handle_file_upload,
# inputs=[file_upload, session_id_state],
# outputs=[file_state, upload_status, session_id_state],
# )
# msg.submit(
# fn=chatbot_response_streaming,
# inputs=[msg, chatbot, session_id_state, file_state],
# outputs=[chatbot],
# ).then(
# lambda: "",
# outputs=msg,
# )
# submit_btn.click(
# fn=chatbot_response_streaming,
# inputs=[msg, chatbot, session_id_state, file_state],
# outputs=[chatbot],
# ).then(
# lambda: "",
# outputs=msg,
# )
# clear_btn.click(
# fn=lambda: ([], create_new_session(), None, None, "", ""),
# outputs=[chatbot, session_id_state, file_state, file_upload, msg, upload_status],
# )
# # Mount Gradio on FastAPI
# app = gr.mount_gradio_app(app, demo, path="/")
# # ========================
# # LAUNCH
# # ========================
# if __name__ == "__main__":
# import uvicorn
# port = int(os.getenv("PORT", 7860))
# print(f"""
# ╔════════════════════════════════════════════════════════════╗
# ║ ║
# ║ 🚀 MasterLLM v2.0 Starting... ║
# ║ ║
# ║ 🌐 Gradio UI: http://localhost:{port} ║
# ║ 📡 REST API: http://localhost:{port}/api/v1 ║
# ║ 📚 API Docs: http://localhost:{port}/docs ║
# ║ ║
# ║ 🏆 Bedrock: Priority (Claude 3.5 Sonnet) ║
# ║ 🔄 Gemini: Fallback (gemini-2.0-flash) ║
# ║ 💾 MongoDB: Session management ║
# ║ ║
# ╚════════════════════════════════════════════════════════════╝
# """)
# uvicorn.run(app, host="0.0.0.0", port=port)
# app.py - MasterLLM v2.0 with Bedrock Fallback System
"""
MasterLLM Pipeline Orchestrator v2.0
- Bedrock (priority) + Gemini (fallback) for pipeline generation
- Bedrock LangChain (priority) + CrewAI (fallback) for execution
- MongoDB session management
- Complete REST API
- Gradio UI with fancy displays
"""
import os
import json
import uuid
from datetime import datetime
from typing import List, Optional
import gradio as gr
from fastapi import FastAPI
from fastapi.middleware.cors import CORSMiddleware
from contextlib import asynccontextmanager
import asyncio
# Import our new services
from services.pipeline_generator import generate_pipeline, format_pipeline_for_display
from services.pipeline_executor import execute_pipeline_streaming
from services.session_manager import session_manager
from services.intent_classifier import intent_classifier
from api_routes import router as api_router
from api_routes_v2 import router as api_router_v2
# V3 Architecture managers
from services.pipeline_manager import get_pipeline_manager
from services.workflow_manager import get_workflow_manager
# ========================
# BACKGROUND CLEANUP TASK
# ========================
async def periodic_cleanup():
"""Cleanup old sessions every hour"""
while True:
await asyncio.sleep(3600) # Run every hour
try:
removed = session_manager.cleanup_old_sessions(max_age_hours=24)
if removed > 0:
print(f"🧹 Cleaned up {removed} inactive sessions")
except Exception as e:
print(f"⚠️ Cleanup error: {e}")
@asynccontextmanager
async def lifespan(app: FastAPI):
"""Manage application lifecycle"""
# Startup
print("🚀 Starting MasterLLM v2.0...")
task = asyncio.create_task(periodic_cleanup())
yield
# Shutdown
task.cancel()
session_manager.close()
print("🛑 MasterLLM shut down gracefully")
# ========================
# FASTAPI APP
# ========================
app = FastAPI(
title="MasterLLM v2.0 - AI Pipeline Orchestrator",
description="Bedrock + Gemini fallback system with MongoDB sessions",
version="2.0.0",
lifespan=lifespan
)
# CORS Configuration
app.add_middleware(
CORSMiddleware,
allow_origins=[os.getenv("FRONTEND_ORIGIN", "http://localhost:3000")],
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
# Mount API routes
app.include_router(api_router) # V1 API (legacy)
app.include_router(api_router_v2) # V2 API (enhanced with intent classification)
# ========================
# CONVERSATION STATE
# ========================
class ConversationState:
INITIAL = "initial"
PIPELINE_PROPOSED = "pipeline_proposed"
PIPELINE_APPROVED = "pipeline_approved"
EXECUTING = "executing"
COMPLETED = "completed"
ERROR = "error"
# ========================
# GRADIO UI HANDLERS
# ========================
def create_new_session():
"""Create a new session"""
return session_manager.create_session()
def handle_file_upload(file_path, session_id):
"""
Handle file upload with V3 architecture integration
- Uploads file to S3
- Creates record in files collection
- Generates presigned URL
- Stores file metadata in MongoDB
"""
if not file_path:
return None, json.dumps({
"status": "error",
"message": "No file uploaded"
}, indent=2), session_id
if not session_id:
session_id = create_new_session()
file_name = os.path.basename(file_path)
try:
# V3: Upload to S3 and create file record
from services.s3_manager import get_s3_manager
from services.schemas import FileSchema, schema_to_dict
from datetime import datetime
s3 = get_s3_manager()
file_id = str(uuid.uuid4())
file_size = os.path.getsize(file_path) if os.path.exists(file_path) else 0
# Determine MIME type
mime_type = "application/pdf"
if file_path.lower().endswith(('.png', '.jpg', '.jpeg')):
mime_type = "image/jpeg"
elif file_path.lower().endswith('.gif'):
mime_type = "image/gif"
# Upload to S3
s3_key = f"sessions/{session_id}/files/{file_id}_{file_name}"
with open(file_path, 'rb') as f:
s3_result = s3.upload_file(
key=s3_key,
file_obj=f,
content_type=mime_type,
add_prefix=False
)
# Generate presigned URL (7-day max)
presigned = s3.generate_presigned_url(
s3_key,
expires_in=604800, # 7 days
add_prefix=False
)
# Create file record in MongoDB
mongodb_uri = os.getenv("MONGODB_URI")
mongodb_db = os.getenv("MONGODB_DB", "masterllm")
from pymongo import MongoClient
mongo_client = MongoClient(mongodb_uri)
files_collection = mongo_client[mongodb_db]["files"]
file_record = FileSchema(
file_id=file_id,
session_id=session_id,
uploaded_at=datetime.utcnow(),
file_name=file_name,
file_size=file_size,
mime_type=mime_type,
s3_bucket=s3.bucket_name,
s3_key=s3_key,
presigned_url=presigned["presigned_url"],
presigned_expires_at=presigned["presigned_expires_at"]
)
files_collection.insert_one(schema_to_dict(file_record))
print(f"✅ File uploaded to S3 and recorded: {file_id}")
# Update session with file reference
session_manager.update_session(session_id, {
"current_file": file_path, # Keep local path for backward compatibility
"current_file_id": file_id, # V3: file_id reference
"state": ConversationState.INITIAL
})
# Add system message with file info
session_manager.add_message(
session_id,
"system",
f"File uploaded: {file_name}",
file_data={"file_id": file_id, "file_name": file_name}
)
status = {
"status": "success",
"message": f"File '{file_name}' uploaded successfully",
"file_info": {
"file_id": file_id,
"name": file_name,
"size_bytes": file_size,
"s3_key": s3_key,
"presigned_url": presigned["presigned_url"],
"expires_at": presigned["presigned_expires_at"]
},
"next_action": "💬 Now tell me what you'd like to do with this document"
}
return file_path, json.dumps(status, indent=2), session_id
except Exception as e:
# Fallback to old behavior if V3 integration fails
print(f"⚠️ File upload V3 integration failed: {e}")
print(f" Falling back to local file path")
# Update session with local path only
session_manager.update_session(session_id, {
"current_file": file_path,
"state": ConversationState.INITIAL
})
# Add system message
session_manager.add_message(
session_id,
"system",
f"File uploaded: {file_name}"
)
status = {
"status": "partial_success",
"message": f"File '{file_name}' uploaded locally (S3 upload failed)",
"file_info": {
"name": file_name,
"path": file_path,
"size_bytes": os.path.getsize(file_path) if os.path.exists(file_path) else 0
},
"warning": str(e),
"next_action": "💬 Now tell me what you'd like to do with this document"
}
return file_path, json.dumps(status, indent=2), session_id
def format_chat_history(history, new_user_msg, new_assistant_msg):
"""
Convert chat history to new Gradio format (list of dicts with role/content)
Handles both old format (tuples) and new format (dicts)
"""
messages = []
# Handle existing history - could be in old or new format
if history:
# Check if already in new format (list of dicts with 'role' and 'content')
if isinstance(history[0], dict) and 'role' in history[0]:
# Already in new format, just copy it
messages = list(history)
else:
# Old format (list of tuples), convert it
for item in history:
if isinstance(item, (list, tuple)) and len(item) == 2:
user_msg, bot_msg = item
messages.append({"role": "user", "content": user_msg})
messages.append({"role": "assistant", "content": bot_msg})
# Add new messages
messages.append({"role": "user", "content": new_user_msg})
messages.append({"role": "assistant", "content": new_assistant_msg})
return messages
def chatbot_response_streaming(message: str, history: List, session_id: str, file_path: str = None):
"""
Handle chat messages with streaming updates
Uses intent classification + Bedrock (priority) → Gemini (fallback) for both generation and execution
"""
# Get or create session
session = session_manager.get_session(session_id)
if not session:
session_id = create_new_session()
session = session_manager.get_session(session_id)
# Update file path if provided
if file_path:
session_manager.update_session(session_id, {"current_file": file_path})
session = session_manager.get_session(session_id)
# Add user message to session
session_manager.add_message(session_id, "user", message)
current_state = session.get("state", ConversationState.INITIAL)
# ========================
# CLASSIFY USER INTENT
# ========================
intent_data = intent_classifier.classify_intent(message)
# ========================
# HANDLE CASUAL CHAT & QUESTIONS
# ========================
if intent_data["intent"] in ["casual_chat", "question"] and current_state == ConversationState.INITIAL:
friendly_response = intent_classifier.get_friendly_response(intent_data["intent"], message)
session_manager.add_message(session_id, "assistant", friendly_response)
yield format_chat_history(history, message, friendly_response)
return
# ========================
# HANDLE UNCLEAR INTENT
# ========================
if intent_data["intent"] == "unclear" and current_state == ConversationState.INITIAL:
friendly_response = intent_classifier.get_friendly_response("unclear", message)
session_manager.add_message(session_id, "assistant", friendly_response)
yield format_chat_history(history, message, friendly_response)
return
# ========================
# HANDLE WORKFLOW SAVE REQUEST (V3)
# ========================
pending_workflow = session.get("pending_workflow_save")
if pending_workflow and any(phrase in message.lower() for phrase in ["save", "yes", "sure", "ok"]):
try:
from services.s3_manager import get_s3_manager
pipeline_mgr = get_pipeline_manager()
workflow_mgr = get_workflow_manager()
s3 = get_s3_manager()
# V3: Get full pipeline document from S3 using pipeline_id
pipeline_id = pending_workflow.get("pipeline_id")
pipeline_doc = pipeline_mgr.get_full_pipeline_document(pipeline_id)
if pipeline_doc and pipeline_doc.get("definition"):
# Extract definition from pipeline document
pipeline_def = pipeline_doc["definition"]
# Save as workflow with source tracking
workflow_id = workflow_mgr.save_workflow(
session_id=session_id,
pipeline_definition=pipeline_def,
user_message=message,
source_pipeline_id=pipeline_id,
pipeline_status=pipeline_doc.get("status", "unknown")
)
# Clear pending
session_manager.update_session(session_id, {"pending_workflow_save": None})
response = f"✅ **Workflow Saved!**\n\nWorkflow ID: `{workflow_id}`\nName: {pending_workflow['pipeline_name']}\nSource Pipeline: `{pipeline_id[:8]}...`\n\nYou can now reuse this workflow anytime!\n\nWhat else can I help you with?"
session_manager.add_message(session_id, "assistant", response)
yield format_chat_history(history, message, response)
return
else:
# Pipeline document not found
session_manager.update_session(session_id, {"pending_workflow_save": None})
response = "⚠️ Sorry, I couldn't find the pipeline to save. The workflow save request has expired.\n\nWhat else can I help you with?"
session_manager.add_message(session_id, "assistant", response)
yield format_chat_history(history, message, response)
return
except Exception as e:
session_manager.update_session(session_id, {"pending_workflow_save": None})
response = f"❌ Failed to save workflow: {str(e)}\n\nWhat would you like to do next?"
session_manager.add_message(session_id, "assistant", response)
yield format_chat_history(history, message, response)
return
# ========================
# STATE: INITIAL - Generate Pipeline ONLY if intent requires it
# ========================
if current_state == ConversationState.INITIAL:
# Only generate pipeline if user explicitly requested it
if not intent_data.get("requires_pipeline", False):
# Not a pipeline request - give friendly response
friendly_response = "I'm here to help process documents! Please tell me what you'd like to do with your document.\n\nFor example:\n- 'extract text and summarize'\n- 'get tables from pages 2-5'\n- 'translate to Spanish'\n\nType 'help' to see all capabilities!"
session_manager.add_message(session_id, "assistant", friendly_response)
yield format_chat_history(history, message, friendly_response)
return
# Check if file is uploaded
if not session.get("current_file"):
response_text = "📁 Please upload a document first before I can process it!\n\nClick the 'Upload Document' button to get started."
session_manager.add_message(session_id, "assistant", response_text)
yield format_chat_history(history, message, response_text)
return
try:
# Generate pipeline using Bedrock → Gemini fallback
yield format_chat_history(history, message, "🤖 Analyzing your request and creating a pipeline...\n⏳ This will take just a moment...")
pipeline = generate_pipeline(
user_input=message,
file_path=session.get("current_file"),
prefer_bedrock=True
)
# V3: Create pipeline_id and upload to S3
pipeline_id = str(uuid.uuid4())
# Build initial pipeline document
pipeline_doc = {
"pipeline_id": pipeline_id,
"session_id": session_id,
"pipeline_name": pipeline.get("pipeline_name"),
"status": "proposed",
"created_at": datetime.utcnow().isoformat() + "Z",
"created_by_message": message,
"definition": pipeline,
"execution": None,
"results": None
}
# Upload to S3
from services.s3_manager import get_s3_manager
s3 = get_s3_manager()
pipeline_s3_key = f"sessions/{session_id}/pipelines/{pipeline_id}.json"
s3.upload_json(pipeline_s3_key, pipeline_doc, add_prefix=False)
# Create metadata in MongoDB
from services.pipeline_manager import get_pipeline_manager
pipeline_mgr = get_pipeline_manager()
pipeline_mgr.create_pipeline_metadata(
pipeline_id=pipeline_id,
session_id=session_id,
pipeline_name=pipeline.get("pipeline_name"),
s3_key=pipeline_s3_key,
status="proposed",
created_by_message=message
)
# Update session with reference only (not full pipeline)
session_manager.update_session(session_id, {
"current_pipeline_id": pipeline_id,
"current_pipeline_s3_key": pipeline_s3_key,
"state": ConversationState.PIPELINE_PROPOSED
})
# Create user-friendly display
pipeline_name = pipeline.get("pipeline_name", "Document Processing")
steps_list = pipeline.get("pipeline_steps", [])
steps_summary = "\n".join([f" {i+1}. **{step.get('tool', 'Unknown')}**" for i, step in enumerate(steps_list)])
friendly_display = f"""🎯 **Pipeline Created: {pipeline_name}**
Here's what I'll do:
{steps_summary}
**Ready to proceed?**
- Type **'approve'** or **'yes'** to execute
- Type **'reject'** or **'no'** to cancel
- Describe changes to modify the plan"""
# Add technical details in collapsible format
response_text = friendly_display + f"\n\n<details>\n<summary>📊 Technical Details (for developers)</summary>\n\n```json\n{json.dumps(pipeline, indent=2)}\n```\n</details>"
session_manager.add_message(session_id, "assistant", response_text)
yield format_chat_history(history, message, response_text)
return
except Exception as e:
error_response = f"❌ **Oops!** I encountered an error while creating the pipeline:\n\n{str(e)}\n\nPlease try rephrasing your request or type 'help' for examples."
session_manager.add_message(session_id, "assistant", error_response)
yield format_chat_history(history, message, error_response)
return
# ========================
# STATE: PIPELINE_PROPOSED - Handle Approval/Rejection
# ========================
elif current_state == ConversationState.PIPELINE_PROPOSED:
user_input = message.lower().strip()
# APPROVE - Execute the pipeline
if "approve" in user_input or "yes" in user_input:
session_manager.update_session(session_id, {"state": ConversationState.EXECUTING})
# V3: Get pipeline references from session
pipeline_s3_key = session.get("current_pipeline_s3_key")
pipeline_id = session.get("current_pipeline_id")
# Download pipeline from S3
from services.s3_manager import get_s3_manager
s3 = get_s3_manager()
from services.pipeline_manager import get_pipeline_manager
pipeline_mgr = get_pipeline_manager()
if pipeline_s3_key:
try:
pipeline_doc = s3.download_json(pipeline_s3_key, add_prefix=False)
plan = pipeline_doc["definition"]
except Exception as e:
print(f"⚠️ Failed to download pipeline from S3: {e}")
plan = {}
pipeline_id = session_id
else:
# Fallback for old sessions without S3 storage
plan = session.get("proposed_pipeline", {})
pipeline_id = session_id
# Update pipeline status to executing
if pipeline_s3_key and pipeline_doc:
pipeline_doc["status"] = "executing"
pipeline_doc["execution"] = {
"started_at": datetime.utcnow().isoformat() + "Z",
"executor": "unknown",
"components_status": []
}
s3.upload_json(pipeline_s3_key, pipeline_doc, add_prefix=False)
pipeline_mgr.update_pipeline_status(pipeline_id, "executing")
execution_id = pipeline_id
# Initial status - User-friendly
initial_message = f"✅ **Approved!** Starting execution of: **{plan.get('pipeline_name', 'pipeline')}**\n\n🚀 Processing... please wait...\n_(Using {plan.get('_generator', 'AI')} - {plan.get('_model', 'model')})_"
yield format_chat_history(history, message, initial_message)
steps_completed = []
final_payload = None
executor_used = "unknown"
progress_messages = []
try:
# Execute pipeline with Bedrock → CrewAI fallback
for event in execute_pipeline_streaming(
pipeline=plan,
file_path=session.get("current_file"),
session_id=session_id,
prefer_bedrock=True
):
event_type = event.get("type")
# Info events (fallback notifications, etc.)
if event_type == "info":
info_message = f"ℹ️ {event.get('message')}\n_(Executor: {event.get('executor', 'unknown')})_"
progress_messages.append(info_message)
accumulated_response = initial_message + "\n\n" + "\n".join(progress_messages)
yield format_chat_history(history, message, accumulated_response)
# Component status updates (V3)
elif event_type == "component_status":
step_info = {
"step": event.get("step", 0),
"component": event.get("component", "processing"),
"status": event.get("status", "running"),
"message": event.get("message", ""),
"executor": event.get("executor", "unknown")
}
if "observation" in event:
step_info["observation"] = event.get("observation")
steps_completed.append(step_info)
executor_used = event.get("executor", executor_used)
# User-friendly status
status_emoji = "✅" if event.get('status') == 'completed' else "⏳"
component_msg = event.get('message', '')
status_line = f"{status_emoji} Step {event.get('step', 0)}: {event.get('component', 'processing')}"
if component_msg:
status_line += f" - {component_msg}"
progress_messages.append(status_line)
current_progress = "\n".join(progress_messages[-5:]) # Last 5 steps
accumulated = initial_message + "\n\n" + current_progress
yield format_chat_history(history, message, accumulated)
# Legacy step updates (backwards compatibility)
elif event_type == "step":
step_info = {
"step": event.get("step", 0),
"tool": event.get("tool", "processing"),
"status": event.get("status", "running"),
"executor": event.get("executor", "unknown")
}
# Add observation if available (tool output)
if "observation" in event:
step_info["observation"] = event.get("observation")
# Add tool input if available
if "input" in event:
step_info["input"] = event.get("input")
steps_completed.append(step_info)
executor_used = event.get("executor", executor_used)
# Create user-friendly progress message
step_num = event.get('step', 0)
tool_name = event.get('tool', 'processing')
if event.get('status') == 'completed' and 'observation' in event:
obs_preview = str(event.get('observation'))[:80]
step_message = f"✅ **Step {step_num}:** {tool_name} - Completed!\n _Preview: {obs_preview}..._"
elif event.get('status') == 'executing':
step_message = f"⏳ **Step {step_num}:** {tool_name} - Processing..."
else:
step_message = f"📍 **Step {step_num}:** {tool_name}"
progress_messages.append(step_message)
accumulated_response = initial_message + "\n\n" + "\n\n".join(progress_messages)
yield format_chat_history(history, message, accumulated_response)
# Final result
elif event_type == "final":
final_payload = event.get("data")
executor_used = event.get("executor", executor_used)
# Error
elif event_type == "error":
error_msg = event.get("error", "Unknown error")
friendly_error = f"❌ **Pipeline Failed**\n\nError: {error_msg}\n\nCompleted {len(steps_completed)} step(s) before failure.\n\nWhat would you like to do next?"
session_manager.update_session(session_id, {"state": ConversationState.INITIAL})
session_manager.add_message(session_id, "assistant", friendly_error)
yield format_chat_history(history, message, friendly_error)
return
# Process final result
# V3: Update pipeline document with results in S3
if pipeline_s3_key:
try:
pipeline_doc = s3.download_json(pipeline_s3_key, add_prefix=False)
pipeline_doc["status"] = "completed"
if pipeline_doc.get("execution"):
pipeline_doc["execution"]["completed_at"] = datetime.utcnow().isoformat() + "Z"
pipeline_doc["execution"]["executor"] = executor_used
pipeline_doc["results"] = {
"final_output_url": final_payload.get("final_output_url"),
"final_output_expires_at": final_payload.get("final_output_expires_at"),
"components_executed": final_payload.get("components_executed"),
"last_node_output": final_payload.get("last_node_output"),
"workflow_status": "completed"
}
s3.upload_json(pipeline_s3_key, pipeline_doc, add_prefix=False)
# Update MongoDB metadata
pipeline_mgr.update_pipeline_status(
pipeline_id,
"completed",
final_output_url=final_payload.get("final_output_url"),
final_output_expires_at=final_payload.get("final_output_expires_at")
)
except Exception as e:
print(f"⚠️ Failed to update pipeline document: {e}")
# Update session state
session_manager.update_session(session_id, {
"state": ConversationState.INITIAL
})
# V3: Store pending workflow save info with pipeline_id
session_manager.update_session(session_id, {
"pending_workflow_save": {
"pipeline_id": pipeline_id,
"pipeline_name": plan.get("pipeline_name", "Untitled")
}
})
# REMOVED: Deprecated save_pipeline_execution - now handled by pipeline_manager
# Create user-friendly final response with workflow save prompt
success_count = len([s for s in steps_completed if s.get("status") == "completed"])
# V3: Check if we have S3 final output URL
final_output_url = final_payload.get("final_output_url", "")
final_output_expires = final_payload.get("final_output_expires_at", "")
# Build URL section
url_section = ""
if final_output_url:
url_section = f"\n📊 **Results Available:**\n- [Download All Outputs]({final_output_url})\n- Expires: {final_output_expires}\n"
friendly_final = f"""🎉 **Pipeline Completed Successfully!**
**Summary:**
- Pipeline: {plan.get('pipeline_name', 'Document Processing')}
- Total Steps: {len(steps_completed)}
- Successful: {success_count}
- Executor: {executor_used}
{url_section}
💾 **Would you like to save this workflow for future use?**
Type **'save workflow'** or **'yes'** to save it.
✅ All done! What else can I help you with?
<details>
<summary>📊 Detailed Results (for developers)</summary>
```json
{json.dumps({"status": "completed", "executor": executor_used, "pipeline": plan.get("pipeline_name"), "result": final_payload, "steps": steps_completed}, indent=2)}
```
</details>"""
final_response = friendly_final
else:
final_response = f"✅ **Pipeline Completed!**\n\nExecuted {len(steps_completed)} steps using {executor_used}.\n\nReady for your next task!"
session_manager.update_session(session_id, {"state": ConversationState.INITIAL})
session_manager.add_message(session_id, "assistant", final_response)
yield format_chat_history(history, message, final_response)
return
except Exception as e:
friendly_error = f"❌ **Pipeline Execution Failed**\n\nError: {str(e)}\n\nCompleted {len(steps_completed)} step(s) before failure.\n\n<details>\n<summary>📋 Error Details</summary>\n\n```\n{str(e)}\n```\n</details>\n\nWould you like to try again with a different approach?"
session_manager.update_session(session_id, {"state": ConversationState.INITIAL})
session_manager.add_message(session_id, "assistant", friendly_error)
yield format_chat_history(history, message, friendly_error)
return
# REJECT - Cancel the pipeline
elif "reject" in user_input or "no" in user_input:
session_manager.update_session(session_id, {
"state": ConversationState.INITIAL,
"proposed_pipeline": None
})
friendly_rejection = "👍 No problem! Pipeline cancelled.\n\nWhat else would you like me to help you with?"
session_manager.add_message(session_id, "assistant", friendly_rejection)
yield format_chat_history(history, message, friendly_rejection)
return
# EDIT - Request modifications
elif "edit" in user_input or "modify" in user_input:
current_pipeline = session.get("proposed_pipeline", {})
friendly_edit_help = f"""📝 **Edit Mode**
Current pipeline: **{current_pipeline.get('pipeline_name', 'Unknown')}**
Describe what you'd like to change. For example:
- "Add summarization at the end"
- "Remove table extraction"
- "Only process pages 1-3"
- "Translate to French instead of Spanish"
Or type 'approve' to run the current plan as-is."""
session_manager.add_message(session_id, "assistant", friendly_edit_help)
yield format_chat_history(history, message, friendly_edit_help)
return
# Try to modify pipeline based on user input
else:
if len(message.strip()) > 5:
try:
original_plan = session.get("proposed_pipeline", {})
edit_context = f"Original: {original_plan.get('pipeline_name')}. User wants: {message}"
# Generate new pipeline with modification
new_pipeline = generate_pipeline(
user_input=edit_context,
file_path=session.get("current_file"),
prefer_bedrock=True
)
session_manager.update_session(session_id, {
"proposed_pipeline": new_pipeline,
"state": ConversationState.PIPELINE_PROPOSED
})
formatted = format_pipeline_for_display(new_pipeline)
response = formatted + f"\n\n```json\n{json.dumps(new_pipeline, indent=2)}\n```"
session_manager.add_message(session_id, "assistant", response)
yield format_chat_history(history, message, response)
return
except Exception as e:
error_response = {
"status": "edit_failed",
"error": str(e),
"message": "Could not modify the plan",
"action": "Try 'approve' to run as-is, or 'reject' to start over"
}
response = f"```json\n{json.dumps(error_response, indent=2)}\n```"
session_manager.add_message(session_id, "assistant", response)
yield format_chat_history(history, message, response)
return
# Default waiting message
response_data = {
"status": "waiting_for_confirmation",
"message": "Please type 'approve', 'reject', or describe changes",
"hint": "You can also say 'edit' for modification hints"
}
response = f"```json\n{json.dumps(response_data, indent=2)}\n```"
session_manager.add_message(session_id, "assistant", response)
yield format_chat_history(history, message, response)
return
# Default fallback
response = json.dumps({"status": "ready", "message": "Ready for your next instruction"}, indent=2)
session_manager.add_message(session_id, "assistant", response)
yield format_chat_history(history, message, response)
# ========================
# GRADIO UI
# ========================
# Simple Blocks initialization for HF Spaces compatibility (older Gradio version)
with gr.Blocks(title="MasterLLM v2.0 - AI Pipeline Orchestrator") as demo:
gr.Markdown("""
# 🤖 MasterLLM v2.0 - AI Pipeline Orchestrator
**🏆 Bedrock Priority** with Gemini Fallback | **💾 MongoDB Sessions** | **📡 Complete REST API**
Upload a document, describe what you want, and watch AI orchestrate the perfect pipeline!
""")
# State management
session_id_state = gr.State(value=create_new_session())
file_state = gr.State(value=None)
with gr.Row():
with gr.Column(scale=3):
# Chat interface - Gradio auto-detects format from data structure
chatbot = gr.Chatbot(label="Chat")
# Text input
msg = gr.Textbox(
placeholder="💬 Type your instruction... (e.g., 'extract text from pages 1-5 and summarize')",
label="Your Message",
lines=2,
max_lines=4,
)
with gr.Row():
submit_btn = gr.Button("🚀 Send", variant="primary", scale=2)
clear_btn = gr.Button("🗑️ Clear Chat", scale=1)
with gr.Column(scale=1):
# File upload section
gr.Markdown("### 📁 Upload Document")
file_upload = gr.File(
label="PDF or Image",
file_types=[".pdf", ".png", ".jpg", ".jpeg", ".gif", ".bmp"],
type="filepath",
)
upload_status = gr.Textbox(
label="📊 Upload Status",
interactive=False,
lines=10,
max_lines=15,
)
# Session info
gr.Markdown("### 🔗 Session Info")
session_display = gr.Textbox(
label="Session ID",
interactive=False,
value=lambda: session_id_state.value[:8] + "...",
)
# Examples
gr.Markdown("### 💡 Example Pipelines")
gr.Examples(
examples=[
"extract text from pages 1-5",
"extract text and summarize",
"extract text, tables, and translate to Spanish",
"get tables from pages 2-4 and summarize",
"text-classify-ner from entire document",
"describe images and summarize findings",
"extract text, detect signatures and stamps",
],
inputs=msg,
)
# System info
gr.Markdown("""
### ℹ️ System Features
- ✅ **Bedrock** (Claude 3.5 Sonnet) priority
- ✅ **Gemini** (gemini-2.0-flash) fallback
- ✅ **MongoDB** session persistence
- ✅ **Streaming** real-time updates
- ✅ **Component-level** JSON output
- ✅ **REST API** for integration
### 📊 Pipeline Flow:
1. **Upload** your document
2. **Describe** what you want
3. **Review** AI-generated pipeline
4. **Approve** to execute
5. **Watch** streaming updates
6. **Get** complete JSON results
""")
# Event handlers
file_upload.upload(
fn=handle_file_upload,
inputs=[file_upload, session_id_state],
outputs=[file_state, upload_status, session_id_state],
)
msg.submit(
fn=chatbot_response_streaming,
inputs=[msg, chatbot, session_id_state, file_state],
outputs=[chatbot],
).then(
lambda: "",
outputs=msg,
)
submit_btn.click(
fn=chatbot_response_streaming,
inputs=[msg, chatbot, session_id_state, file_state],
outputs=[chatbot],
).then(
lambda: "",
outputs=msg,
)
clear_btn.click(
fn=lambda: ([], create_new_session(), None, None, "", ""),
outputs=[chatbot, session_id_state, file_state, file_upload, msg, upload_status],
)
# Mount Gradio on FastAPI
app = gr.mount_gradio_app(app, demo, path="/")
# ========================
# LAUNCH
# ========================
if __name__ == "__main__":
import uvicorn
port = int(os.getenv("PORT", 7860))
print(f"""
╔════════════════════════════════════════════════════════════╗
║ ║
║ 🚀 MasterLLM v2.0 Starting... ║
║ ║
║ 🌐 Gradio UI: http://localhost:{port} ║
║ 📡 REST API: http://localhost:{port}/api/v1 ║
║ 📚 API Docs: http://localhost:{port}/docs ║
║ ║
║ 🏆 Bedrock: Priority (Claude 3.5 Sonnet) ║
║ 🔄 Gemini: Fallback (gemini-2.0-flash) ║
║ 💾 MongoDB: Session management ║
║ ║
╚════════════════════════════════════════════════════════════╝
""")
uvicorn.run(app, host="0.0.0.0", port=port) |