add api to enqueue and process job analysis

README.md

@@ -127,8 +127,13 @@ python webapp.py
 
 ### Connect to Redis (Local)
 ```
-docker-compose up -d
-redis-cli -h localhost -p 6379 -a {REDIS_PASSWORD}
+docker compose up -d
+docker exec -it {container-id} bash
+redis-cli -h localhost -p 6379 -a trading-agents
+
+Run worker:
+rq worker --url redis://:{{REDIS_PASSWORD}}@{{REDIS_HOST}}:{{REDIS_PORT}}/{{REDIS_DB}} --with-scheduler
+
 ```
 
 ### Required APIs

requirements.txt

@@ -29,3 +29,4 @@ telethon
 fastapi
 uvicorn[standard]
 redis[hiredis]
+rq

service.py

@@ -0,0 +1,87 @@
+from tradingagents.external.redis.repo import redis_queue, redis_repo
+from tradingagents.domain.model import AnalysisMeta,  AnalysisStatus
+from tradingagents.domain.response import EnqueueAnalysisResponse
+from rq import get_current_job
+from tradingagents.external.redis.repo import RQ_RETRIES
+from tradingagents.graph.trading_graph import TradingAgentsGraph
+from tradingagents.dataflows.config import get_config
+
+DEFAULT_USER = "global_user"
+
+# Initialize trading agent once at startup
+def create_trading_agent():
+    """Create trading agent with fixed configuration"""
+    return TradingAgentsGraph(debug=True, config=get_config())
+
+# Create the trading agent instance once
+trading_agent = create_trading_agent()
+
+def process_job(user_id: str, symbol: str, date: str):
+    try:
+        job = get_current_job()
+        
+        attempt = job.meta.get("attempt", 1)
+        job.meta["attempt"] = attempt
+        job.save_meta()
+
+        print(f"INFO: Processing job-id {job.id} for symbol {symbol} and date {date} by user {user_id}")
+
+        # Update status to RUNNING
+        redis_repo.update_status_analysis_meta(job_id=job.id, status=AnalysisStatus.RUNNING)
+
+        final_state, decision = trading_agent.propagate(ticker=symbol, trade_date=date)
+
+        print(f"INFO: Decision for job-id {job.id}: {decision}")
+
+        # Save the final result
+        redis_repo.save_result(job_id=job.id, final_trade=final_state["final_trade_decision"])
+        # Update status to DONE
+        redis_repo.update_status_analysis_meta(job_id=job.id, status=AnalysisStatus.DONE)
+        
+        print(f"INFO: Completed job-id {job.id} for symbol {symbol}")
+    except Exception as e:
+        job.meta["attempt"] = attempt + 1
+        job.save_meta()
+        print(f"ERROR: Failed to process job-id {job.id}: {e} (Attempt {attempt} of {RQ_RETRIES})")
+        # Update status to FAILED
+        redis_repo.update_status_analysis_meta(job_id=job.id, status=AnalysisStatus.FAILED)
+
+
+def enqueue_analysis(symbol: str, date: str) -> EnqueueAnalysisResponse:
+    """
+    Enqueue a background task to analyze trading data for a given symbol and date.
+
+    Args:
+        symbol (str): The trading symbol to analyze (e.g., "BTC/USDT").
+        date (str): The date for which to perform the analysis in YYYY-MM-DD format.
+    Returns:
+        EnqueueAnalysisResponse: The response containing job_id, status, and message.
+    """
+    try:
+        # Check if the analysis is on cooldown, if cooldown return the job-id
+        job_id, ttl = redis_repo.get_cooldown(DEFAULT_USER, symbol) 
+        if job_id:
+            return EnqueueAnalysisResponse(
+                job_id=job_id,
+                status="on_cooldown",
+                message=f"Analysis for {symbol} is on cooldown. Please try again later. TTL: {ttl} seconds remaining.",
+            )
+        
+        # If not on cooldown, enqueue the task, insert cooldown key with TTL 6 hours, insert with status pending redis key for analysis analysis:job:{job_id}
+        task = redis_queue.enqueue(process_job, DEFAULT_USER, symbol, date)
+
+        redis_repo.save_cooldown(DEFAULT_USER, symbol, task.id)
+        redis_repo.create_analysis_meta(AnalysisMeta.new(job_id=task.id, user_id=DEFAULT_USER, symbol=symbol, trade_date=date))
+
+        return EnqueueAnalysisResponse(
+            job_id=task.id,
+            status="enqueued",
+            message=f"Analysis for {symbol} has been enqueued successfully."
+        )
+    except Exception as e:
+        print(f"ERROR: Failed to enqueue analysis task: {e}")
+        return EnqueueAnalysisResponse(
+            job_id=None,
+            status="error",
+            message=f"Failed to enqueue analysis task: {str(e)}"
+        )

tradingagents/domain/model.py

@@ -0,0 +1,34 @@
+from dataclasses import dataclass
+from enum import Enum
+from typing import Optional
+import time
+
+
+class AnalysisStatus(str, Enum):
+    PENDING = "pending"
+    RUNNING = "running"
+    DONE = "done"
+    FAILED = "failed"
+
+
+@dataclass
+class AnalysisMeta:
+    job_id: str
+    user_id: str
+    symbol: str
+    status: AnalysisStatus
+    trade_date: str # "trade_date": final_state["trade_date"],
+    updated_at: float
+    created_at: float = time.time()
+
+    @staticmethod
+    def new(job_id: str, user_id: str, symbol: str, trade_date: str) -> "AnalysisMeta":
+        return AnalysisMeta(
+            job_id=job_id,
+            user_id=user_id,
+            symbol=symbol,
+            status=AnalysisStatus.PENDING,
+            trade_date=trade_date,
+            updated_at=time.time(),
+            created_at=time.time(),
+        )

tradingagents/domain/response.py

@@ -0,0 +1,8 @@
+from dataclasses import dataclass
+from typing import Optional
+
+@dataclass
+class EnqueueAnalysisResponse:
+    job_id: Optional[str]
+    status: str
+    message: str

tradingagents/external/redis/client.py

@@ -11,6 +11,7 @@ def get_redis_client() -> Redis:
     if _client is None:
         try:
             config = get_config()
+            print(f"INFO: Creating Redis connection pool config {config}")
             retry = Retry(ExponentialBackoff(), retries=5)
 
             pool = ConnectionPool(

tradingagents/external/redis/repo.py

@@ -1,18 +1,105 @@
+import time
 from tradingagents.external.redis.client import get_redis_client
+from tradingagents.domain.model import AnalysisMeta, AnalysisStatus
+from rq import Queue, Retry
+from redis import Redis
 
-redis = get_redis_client()
+# TODO: Move to config
+RQ_RETRIES = 3
+RQ_INTERVAL = [30, 60, 120]
+
+ANALYSIS_META_KEY = "analysis:meta:{job_id}"
+ANALYSIS_RESULT_KEY = "analysis:result:{job_id}"
+ANALYSIS_COOLDOWN_KEY = "tradingagents-analysis-cooldown-{user_id}:{symbol}"
 
 class RedisRepo:
-    def get(self, key: str):
-        return redis.get(key)
-
-    def set(self, key: str, value: str, ex: int | None = None):
-        return redis.set(key, value, ex=ex)
-
-    def delete(self, key: str):
-        return redis.delete(key)
+    def __init__(self, redis: Redis):
+        self.redis = redis
 
     def exists(self, key: str) -> bool:
-        return redis.exists(key) == 1
+        return self.redis.exists(key) == 1
+    
+    def create_cooldown_key(self, user_id: str, symbol: str) -> str:
+        return ANALYSIS_COOLDOWN_KEY.format(user_id=user_id, symbol=symbol)
+    
+    def save_cooldown(self, user_id: str, symbol: str, job_id: str, ttl: int = 6 * 3600):
+        key = self.create_cooldown_key(user_id, symbol)
+        self.redis.set(key, job_id, ex=ttl)
 
-redis_repo = RedisRepo()
+    def get_cooldown(self, user_id: str, symbol: str) -> tuple[str | None, int | None]:
+        key = self.create_cooldown_key(user_id, symbol)
+
+        job_id = self.redis.get(key)
+        if job_id is None:
+            return None, None
+
+        ttl = self.redis.ttl(key)
+
+        # Redis TTL semantics:
+        # -2 → key does not exist
+        # -1 → key exists but has no expiry
+        # >=0 → seconds remaining
+
+        if ttl < 0:
+            print(f"WARNING: Cooldown key {key} has invalid TTL {ttl}.")
+            ttl = None
+
+        return job_id, ttl
+
+
+    def _meta_key(self, job_id: str) -> str:
+        return ANALYSIS_META_KEY.format(job_id=job_id)
+
+    def _result_key(self, job_id: str) -> str:
+        return ANALYSIS_RESULT_KEY.format(job_id=job_id)
+    
+    def create_analysis_meta(self, meta: AnalysisMeta, ttl: int = 7 * 24 * 3600):
+        self.redis.hset(
+            self._meta_key(meta.job_id),
+            mapping={
+                "job_id": meta.job_id,
+                "trade_date": meta.trade_date,
+                "user_id": meta.user_id,
+                "symbol": meta.symbol,
+                "status": meta.status.value,
+                "updated_at": meta.updated_at,
+            },
+        )
+        self.redis.expire(self._meta_key(meta.job_id), ttl)
+
+    def update_status_analysis_meta(self, job_id: str, status: AnalysisStatus):
+        self.redis.hset(
+            self._meta_key(job_id),
+            mapping={
+                "status": status.value,
+                "updated_at": time.time(),
+            },
+        )
+
+    def get_analysis_meta(self, job_id: str) -> AnalysisMeta | None:
+        data = self.redis.hgetall(self._meta_key(job_id))
+        if not data:
+            return None
+
+        return AnalysisMeta(
+            job_id=data["job_id"],
+            user_id=data["user_id"],
+            symbol=data["symbol"],
+            trade_date=data["trade_date"],
+            status=AnalysisStatus(data["status"]),
+            updated_at=float(data["updated_at"]),
+            created_at=float(data.get("created_at")),
+        )
+
+    def save_result(self, job_id: str, final_trade: str, ttl: int = 7 * 24 * 3600):
+        '''
+        Save the final trading decision result to Redis. No expiration by default.
+        '''
+        self.redis.set(self._result_key(job_id), final_trade, ex=ttl)
+
+    def get_result(self, job_id: str) -> str | None:
+        return self.redis.get(self._result_key(job_id))
+
+
+redis_repo = RedisRepo(get_redis_client())
+redis_queue = Queue(connection=get_redis_client(), retry=Retry(max=RQ_RETRIES, interval=RQ_INTERVAL))

webapp.py

@@ -1,11 +1,10 @@
-from fastapi import FastAPI, HTTPException
+from fastapi import FastAPI, HTTPException, status
 from pydantic import BaseModel
 import uvicorn
 from datetime import datetime
-import asyncio
 
 # Import your trading agents
-from tradingagents.graph.trading_graph import TradingAgentsGraph
+from service import enqueue_analysis
 from tradingagents.dataflows.config import get_config
 from dotenv import load_dotenv
 
@@ -21,26 +20,17 @@ app = FastAPI(
     version="0.1.0"
 )
 
-# Pydantic models for request/response
-class TradingRequest(BaseModel):
+class TradingAnalyzeRequest(BaseModel):
     symbol: str
     date: str
 
-class TradingResponse(BaseModel):
+class TradingAnalyzeResponse(BaseModel):
     symbol: str
     date: str
-    decision: dict
+    job_id: str
     timestamp: str
     status: str
 
-# Initialize trading agent once at startup
-def create_trading_agent():
-    """Create trading agent with fixed configuration"""
-    return TradingAgentsGraph(debug=True, config=config)
-
-# Create the trading agent instance once
-trading_agent = create_trading_agent()
-
 @app.get("/")
 async def root():
     """Root endpoint"""
@@ -60,38 +50,31 @@ async def health_check():
         "service": "tradingagents-api"
     }
 
-@app.post("/trading/analyze", response_model=TradingResponse)
-async def analyze_trading_decision(request: TradingRequest):
+@app.post("/v1/trading/analyze", response_model=TradingAnalyzeResponse, status_code=status.HTTP_202_ACCEPTED,)
+async def analyze_trading_decision(request: TradingAnalyzeRequest):
     """
     Analyze trading decision for a given symbol and date
     
     Example usage:
     POST /trading/analyze
     {
-        "symbol": "NVDA",
+        "symbol": "BTC/USDT",
         "date": "2024-05-10"
     }
     """
-    try:
-        # Run the analysis (this might take a while, so we run it in a thread pool)
-        def run_analysis():
-            _, decision = trading_agent.propagate(request.symbol, request.date)
-            return decision
-        
-        # Run in thread pool to avoid blocking
-        loop = asyncio.get_event_loop()
-        decision = await loop.run_in_executor(None, run_analysis)
-        
-        return TradingResponse(
-            symbol=request.symbol,
-            date=request.date,
-            decision=decision,
-            timestamp=datetime.now().isoformat(),
-            status="success"
-        )
-        
-    except Exception as e:
-        raise HTTPException(status_code=500, detail=f"Trading analysis failed: {str(e)}")
+    response = enqueue_analysis(request.symbol, request.date)
+    print(f"INFO: Enqueue response: {response}")
+
+    if response.status == "error":
+        raise HTTPException(status_code=500, detail=response.message)
+    
+    return TradingAnalyzeResponse(
+        symbol=request.symbol,
+        date=request.date,
+        job_id=response.job_id,
+        timestamp=datetime.now().isoformat(),
+        status=response.status
+    )
 
 
 if __name__ == "__main__":