Spaces:
Sleeping
Sleeping
File size: 19,896 Bytes
5ef3f54 | 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 | from fastapi import FastAPI, HTTPException, BackgroundTasks, Response
from pydantic import BaseModel, Field, ConfigDict
from typing import Optional, List, Dict
import sys
import re
import json
from huggingface_hub import repo_info
from db import db_client
from agents import create_archivist, create_librarian, run_recruiter, create_registrar, narrator, get_model, IMAGE_MODELS, AUDIO_MODELS, REASONING_MODELS
from persistence import save_to_dataset, load_from_dataset, list_saves, delete_save, persistence_manager, get_cached_media, save_cached_media
from vector import vector_model
import os
import asyncio
import httpx
app = FastAPI(title="Grim Fable: World Memory")
@app.on_event("startup")
async def startup_event():
# Verify environment on startup
token = os.getenv("HF_TOKEN")
dataset_id = os.getenv("DATASET_ID")
if not token:
print("CRITICAL: HF_TOKEN is missing!")
if not dataset_id:
print("CRITICAL: DATASET_ID is missing!")
# Check dataset accessibility
if token and dataset_id:
try:
repo_info(repo_id=dataset_id, repo_type="dataset", token=token)
print(f"Verified accessibility of dataset: {dataset_id}")
except Exception as e:
print(f"WARNING: Could not access dataset {dataset_id}: {e}")
# Bootstrap rules into the vector DB before starting
try:
print("Starting rule bootstrapping...")
await perform_bootstrap()
print("Rule bootstrapping completed.")
except Exception as e:
print(f"FAILED to bootstrap world: {e}")
import traceback
traceback.print_exc()
asyncio.create_task(inactivity_autosave_loop())
async def inactivity_autosave_loop():
while True:
try:
await asyncio.sleep(60)
if persistence_manager.should_autosave(interaction_happened=False):
print(f"Periodic autosave triggered for: {persistence_manager.current_save_name}")
await perform_save(persistence_manager.current_save_name)
except Exception as e:
print(f"Error in inactivity_autosave_loop: {e}")
await asyncio.sleep(10) # Wait a bit before retrying
class InteractRequest(BaseModel):
user_input: str
narrator_model: Optional[str] = None
reasoning_model: Optional[str] = None
image_model: Optional[str] = None
audio_model: Optional[str] = None
class ChatMessageDto(BaseModel):
text: str
is_user: bool
class ConsultGmRequest(BaseModel):
user_input: str
history: List[ChatMessageDto]
reasoning_model: Optional[str] = None
class CharacterDraft(BaseModel):
model_config = ConfigDict(populate_by_name=True)
name: str
race: str
class_name: str = Field(alias="class")
stats: Dict[str, int]
skills: List[str]
items: List[str]
starting_context: str
class WorldInitRequest(BaseModel):
prompt: Optional[str] = None
history: Optional[List[ChatMessageDto]] = None
save_name: Optional[str] = None
description: Optional[str] = None
reasoning_model: Optional[str] = None
class ConfirmRequest(BaseModel):
draft: CharacterDraft
save_name: str
description: Optional[str] = None
reasoning_model: Optional[str] = None
@app.get("/")
async def root():
return {"status": "online", "engine": "FalkorDB", "mock_mode": db_client.is_mock}
@app.get("/health")
async def health():
try:
# Simple query to verify DB connectivity
db_client.query("MATCH (n) RETURN count(n)")
return {"status": "ready", "database": "connected"}
except Exception as e:
return {"status": "degraded", "database": "disconnected", "error": str(e)}
@app.post("/consult_gm")
async def consult_gm(request: ConsultGmRequest):
try:
print(f"Received GM consultation request: {request.user_input}")
librarian = create_librarian(request.reasoning_model)
if not librarian:
raise HTTPException(status_code=500, detail="Failed to initialize Librarian for GM consultation.")
# Search for rules and context to answer the question
rules_context = await asyncio.to_thread(
librarian.run, f"Use vector_search on index 'Rule' to find information that helps answer this GM question: {request.user_input}"
)
# Get world state context if possible
world_context = ""
try:
player_data = db_client.query("MATCH (p:Player) RETURN p.name, p.hp, p.max_hp, p.strength, p.dexterity, p.constitution, p.intelligence, p.wisdom, p.charisma, p.x, p.y")
if player_data:
world_context = f"Current Player Stats: {player_data}"
except: pass
history_str = "\n".join([f"{'User' if m.is_user else 'GM'}: {m.text}" for m in request.history[-5:]]) # Last 5 messages for context
model = get_model(request.reasoning_model)
prompt = f"""
You are the Game Master (GM). A player is consulting you for information or rule clarification.
Relevant Rules: {rules_context}
World Context: {world_context}
Recent Consultation History:
{history_str}
Current Question: {request.user_input}
Provide a helpful, concise, and in-character response based on the rules and world state. Do NOT change any data in the database.
"""
# model is an HfApiModel, it has a generate method
response = await asyncio.to_thread(model.generate, messages=[{"role": "user", "content": prompt}])
response_text = response.content if hasattr(response, 'content') else str(response)
return {"response": response_text}
except Exception as e:
print(f"Error in /consult_gm: {e}")
import traceback
traceback.print_exc()
raise HTTPException(status_code=500, detail=str(e))
@app.post("/interact")
async def interact(request: InteractRequest, background_tasks: BackgroundTasks):
try:
print(f"Received interaction request: {request.user_input}")
persistence_manager.update_interaction()
should_save = persistence_manager.should_autosave(interaction_happened=True)
librarian = create_librarian(request.reasoning_model)
archivist = create_archivist(request.reasoning_model)
if not librarian or not archivist:
raise HTTPException(status_code=500, detail="Failed to initialize Librarian or Archivist agents.")
rules_context = await asyncio.to_thread(
librarian.run, f"Use vector_search on index 'Rule' to find relevant game mechanics for: {request.user_input}"
)
memories_context = await asyncio.to_thread(
librarian.run, f"Use vector_search on index 'Memory' to find relevant past events for: {request.user_input}"
)
context = f"Rules:\n{rules_context}\n\nMemories:\n{memories_context}"
# Archivist agent execution is blocking, run in thread
changes = await asyncio.to_thread(
archivist.run, f"Context: {context}\nUser Action: {request.user_input}\nUpdate the world state in FalkorDB. Use the smart_validator to ensure the Cypher respects game rules and user intent. Coordinate tracking (x,y) is mandatory. Summarize the changes."
)
response = await asyncio.to_thread(
narrator.run, context=context, user_input=request.user_input, changes=changes, model_id=request.narrator_model
)
background_tasks.add_task(record_memory, request.user_input, response)
if should_save:
background_tasks.add_task(perform_save, persistence_manager.current_save_name)
return {"response": response}
except Exception as e:
print(f"Error in /interact: {e}")
import traceback
traceback.print_exc()
if isinstance(e, HTTPException): raise e
raise HTTPException(status_code=500, detail=str(e))
async def record_memory(user_input: str, response: str):
try:
model = get_model()
summary_prompt = f"Summarize this game exchange for long-term memory:\nUser: {user_input}\nNarrator: {response}"
# Model inference is blocking, run in thread
summary = await asyncio.to_thread(model.generate, messages=[{"role": "user", "content": summary_prompt}])
summary_text = summary.content if hasattr(summary, 'content') else str(summary)
content = f"User: {user_input} | Narrator: {response}"
embedding = vector_model.encode(summary_text)
query = "CREATE (m:Memory {content: $content, summary: $summary, embedding: vecf32($embedding), timestamp: timestamp()})"
await asyncio.to_thread(db_client.query, query, {
"content": content,
"summary": summary_text,
"embedding": embedding
})
except Exception as e:
print(f"Error recording memory: {e}")
async def perform_save(save_name: str, description: str = None):
try:
db_path = "/data/world.db"
await asyncio.to_thread(db_client.save_db, db_path)
if await asyncio.to_thread(save_to_dataset, save_name, db_path, description=description):
persistence_manager.needs_save = False
persistence_manager.last_save_time = os.path.getmtime(db_path) if os.path.exists(db_path) else persistence_manager.last_save_time
except Exception as e:
print(f"Error performing save: {e}")
@app.post("/world/save")
async def save_world(save_name: str, background_tasks: BackgroundTasks, description: Optional[str] = None):
persistence_manager.current_save_name = save_name
background_tasks.add_task(perform_save, save_name, description)
return {"status": "save_initiated", "save_name": save_name}
@app.delete("/world/save/{save_name}")
async def delete_world_save(save_name: str):
if delete_save(save_name):
return {"status": "deleted", "save_name": save_name}
raise HTTPException(status_code=404, detail="Save not found or could not be deleted")
@app.post("/world/load")
async def load_world(save_name: str, background_tasks: BackgroundTasks):
try:
db_path = "/data/world.db"
if await asyncio.to_thread(load_from_dataset, save_name, db_path):
await asyncio.to_thread(db_client.load_db, db_path)
persistence_manager.current_save_name = save_name
# Restart after sending response to apply new RDB file
background_tasks.add_task(restart_server)
return {"status": "loaded", "save_name": save_name, "notice": "Server restarting to apply world state"}
raise HTTPException(status_code=404, detail="Save not found")
except Exception as e:
print(f"Error in /world/load: {e}")
import traceback
traceback.print_exc()
if isinstance(e, HTTPException): raise e
raise HTTPException(status_code=500, detail=str(e))
@app.get("/world/saves")
async def get_saves():
try:
return {"saves": list_saves()}
except Exception as e:
print(f"Error in /world/saves: {e}")
import traceback
traceback.print_exc()
raise HTTPException(status_code=500, detail=str(e))
@app.get("/world/media")
async def get_media(entity_id: str, media_type: str, prompt: str, model_id: Optional[str] = None):
"""Fetches cached media or generates new media for an entity asynchronously with fallback logic."""
if model_id and model_id.lower() == "disabled":
raise HTTPException(status_code=400, detail="Media generation is disabled")
save_name = persistence_manager.current_save_name
cached_content = get_cached_media(save_name, entity_id, media_type)
if cached_content:
mime = "image/webp" if media_type == "image" else "audio/mpeg"
return Response(content=cached_content, media_type=mime)
# Preparation for generation
models_to_try = []
if model_id:
models_to_try.append(model_id)
base_list = IMAGE_MODELS if media_type == "image" else AUDIO_MODELS
for m in base_list:
if m not in models_to_try:
models_to_try.append(m)
headers = {"Authorization": f"Bearer {os.getenv('HF_TOKEN')}"}
async with httpx.AsyncClient(timeout=60.0) as client:
last_error = "Unknown error"
for mid in models_to_try:
try:
api_url = f"https://api-inference.huggingface.co/models/{mid}"
response = await client.post(api_url, headers=headers, json={"inputs": prompt})
if response.status_code == 200:
save_cached_media(save_name, entity_id, media_type, response.content)
mime = "image/webp" if media_type == "image" else "audio/mpeg"
return Response(content=response.content, media_type=mime)
else:
last_error = f"Model {mid} failed: {response.status_code} {response.text}"
continue
except Exception as e:
last_error = str(e)
continue
raise HTTPException(status_code=500, detail=f"All generation attempts failed. Last error: {last_error}")
@app.get("/world/npc/{npc_id}")
async def get_npc(npc_id: str):
# Support retrieving both Players and NPCs, and match by id or name for legacy compatibility
# RETURN properties(n) ensures we get a dictionary in both mock and live DB
query = "MATCH (n) WHERE (n:NPC OR n:Player) AND (n.id = $id OR n.name = $id) RETURN properties(n)"
result = db_client.query(query, {"id": npc_id})
if not result or len(result) == 0:
raise HTTPException(status_code=404, detail="Entity not found")
# result[0][0] is the node properties dictionary
entity = result[0][0]
return {"npc": entity}
@app.get("/world/map")
async def get_map():
# Only return Player and Location nodes. NPCs are tracked but not displayed.
query = """
MATCH (n)
WHERE n:Player OR n:Location
RETURN labels(n)[0], n.id, n.name, n.x, n.y
"""
print(f"Executing map query: {query}")
result = db_client.query(query)
print(f"Map query result: {result}")
entities = []
if result:
for row in result:
entities.append({
"type": row[0],
"id": row[1] or row[2], # Fallback to name if id is missing
"name": row[2],
"x": int(row[3]) if row[3] is not None else 0,
"y": int(row[4]) if row[4] is not None else 0
})
return {"entities": entities}
def robust_json_extract(text: str):
"""Extracts JSON and returns (draft_dict, clean_text)."""
draft = None
clean_text = text
# Try markdown block first
json_match = re.search(r"(```(?:json)?\s*(.*?)\s*```)", text, re.DOTALL)
if json_match:
full_block = json_match.group(1)
content = json_match.group(2).strip()
clean_text = text.replace(full_block, "").strip()
try:
draft = json.loads(content)
except:
try:
content = re.sub(r",\s*([\]}])", r"\1", content)
draft = json.loads(content)
except: pass
if draft: return draft, clean_text
# Try searching for any { ... } block
brace_match = re.search(r"({.*})", text, re.DOTALL)
if brace_match:
full_match = brace_match.group(1)
try:
draft = json.loads(full_match.strip())
clean_text = text.replace(full_match, "").strip()
return draft, clean_text
except: pass
return None, text
@app.post("/world/init")
async def init_world(request: WorldInitRequest):
try:
print(f"Character creation step for: {request.prompt}")
user_input = request.prompt or "Hello"
history = request.history or []
response_text = await asyncio.to_thread(run_recruiter, user_input, history, request.reasoning_model)
draft, clean_text = robust_json_extract(response_text)
return {
"response": clean_text,
"draft": draft
}
except Exception as e:
print(f"Error in /world/init: {e}")
if "BadRequestError" in str(type(e)):
try:
# Attempt to log response content if it exists
if hasattr(e, 'response') and hasattr(e.response, 'text'):
print(f"Bad Request Response Body: {e.response.text}")
except: pass
import traceback
traceback.print_exc()
if isinstance(e, HTTPException): raise e
raise HTTPException(status_code=500, detail=str(e))
@app.post("/world/confirm")
async def confirm_world(request: ConfirmRequest, background_tasks: BackgroundTasks):
try:
print(f"Confirming character: {request.draft.name}")
registrar = create_registrar(request.reasoning_model)
if not registrar:
raise HTTPException(status_code=500, detail="Failed to initialize Registrar agent.")
# Registrar handles DB creation
# Use by_alias=True to ensure 'class' is used in the JSON passed to the registrar
await asyncio.to_thread(registrar.run, f"Validated Character Draft: {request.draft.model_dump_json(by_alias=True)}")
# Use character name as the save slot name
save_name = request.draft.name
persistence_manager.current_save_name = save_name
# Character confirmed - immediate save
await perform_save(save_name, request.description)
return {"status": "initialized", "save_name": save_name}
except Exception as e:
print(f"Error in /world/confirm: {e}")
import traceback
traceback.print_exc()
raise HTTPException(status_code=500, detail=str(e))
@app.post("/world/bootstrap")
async def bootstrap_world(background_tasks: BackgroundTasks):
background_tasks.add_task(perform_bootstrap)
return {"status": "bootstrap_initiated"}
async def perform_bootstrap():
import json
rules_path = "dnd_srd_rules.json"
if not os.path.exists(rules_path):
print(f"Rules file not found at {rules_path}")
return
with open(rules_path, "r") as f:
rules = json.load(f)
# Attempt to create indices, ignore if they already exist
# Using the new syntax for FalkorDB 4.x+
try:
db_client.query("CREATE VECTOR INDEX FOR (r:Rule) ON (r.embedding) OPTIONS {dimension: 384, similarityFunction: 'cosine'}")
except Exception as e:
if "already exists" not in str(e).lower(): print(f"Notice creating Rule index: {e}")
try:
db_client.query("CREATE VECTOR INDEX FOR (m:Memory) ON (m.embedding) OPTIONS {dimension: 384, similarityFunction: 'cosine'}")
except Exception as e:
if "already exists" not in str(e).lower(): print(f"Notice creating Memory index: {e}")
# Fetch all existing rule titles once to optimize bootstrapping
existing_rules = db_client.query("MATCH (r:Rule) RETURN r.title")
existing_titles = set()
if existing_rules:
for row in existing_rules:
if isinstance(row, list) and len(row) > 0:
existing_titles.add(row[0])
for rule in rules:
title = rule.get("title")
if title in existing_titles:
continue
content = rule.get("content")
tags = rule.get("tags", [])
embedding = vector_model.encode(content)
query = """
CREATE (r:Rule {title: $title, content: $content, tags: $tags, embedding: vecf32($embedding)})
"""
db_client.query(query, {
"title": title,
"content": content,
"tags": tags,
"embedding": embedding
})
print(f"Successfully bootstrapped {len(rules)} rules.")
async def restart_server():
print("Initiating server restart in 2 seconds...")
await asyncio.sleep(2)
sys.exit(0)
|