Upload 11 files

.env.example

@@ -0,0 +1,3 @@
+GOOGLE_API_KEY=your_gemini_api_key
+EMAIL=your_email
+SECRET=your_secret

.gitignore

@@ -0,0 +1,12 @@
+# Python-generated files
+__pycache__/
+*.py[oc]
+build/
+dist/
+wheels/
+*.egg-info
+.env
+# Virtual environments
+.venv
+tests
+LLMFiles

.python-version

@@ -0,0 +1 @@
+3.12

Dockerfile

@@ -0,0 +1,33 @@
+FROM python:3.10-slim
+
+# --- System deps required by Playwright browsers ---
+RUN apt-get update && apt-get install -y \
+    wget gnupg ca-certificates curl unzip \
+    libnss3 libatk1.0-0 libatk-bridge2.0-0 libcups2 libxkbcommon0 \
+    libgtk-3-0 libgbm1 libasound2 libxcomposite1 libxdamage1 libxrandr2 \
+    libxfixes3 libpango-1.0-0 libcairo2 \
+    && rm -rf /var/lib/apt/lists/*
+
+# --- Install Playwright + Chromium ---
+RUN pip install playwright && playwright install --with-deps chromium
+
+# --- Install uv package manager ---
+RUN pip install uv
+
+# --- Copy app to container ---
+WORKDIR /app
+
+COPY . .
+
+ENV PYTHONUNBUFFERED=1
+ENV PYTHONIOENCODING=utf-8
+
+# --- Install project dependencies using uv ---
+RUN uv sync --frozen
+
+# HuggingFace Spaces exposes port 7860
+EXPOSE 7860
+
+# --- Run your FastAPI app ---
+# uvicorn must be in pyproject dependencies
+CMD ["uv", "run", "main.py"]

LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2025 Sai Vijay Ragav
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

agent.py

@@ -0,0 +1,185 @@
+from langgraph.graph import StateGraph, END, START
+from shared_store import url_time
+import time
+from langchain_core.rate_limiters import InMemoryRateLimiter
+from langgraph.prebuilt import ToolNode
+from tools import (
+    get_rendered_html, download_file, post_request,
+    run_code, add_dependencies, ocr_image_tool, transcribe_audio, encode_image_to_base64
+)
+from typing import TypedDict, Annotated, List
+from langchain_core.messages import trim_messages
+from langchain.chat_models import init_chat_model
+from langgraph.graph.message import add_messages
+import os
+from dotenv import load_dotenv
+load_dotenv()
+
+EMAIL = os.getenv("EMAIL")
+SECRET = os.getenv("SECRET")
+
+RECURSION_LIMIT = 5000
+MAX_TOKENS = 180000
+
+
+# -------------------------------------------------
+# STATE
+# -------------------------------------------------
+class AgentState(TypedDict):
+    messages: Annotated[List, add_messages]
+
+
+TOOLS = [
+    run_code, get_rendered_html, download_file,
+    post_request, add_dependencies, ocr_image_tool, transcribe_audio, encode_image_to_base64
+]
+
+
+# -------------------------------------------------
+# LLM INIT (NO SYSTEM PROMPT HERE)
+# -------------------------------------------------
+rate_limiter = InMemoryRateLimiter(
+    requests_per_second=7 / 60,
+    check_every_n_seconds=1,
+    max_bucket_size=7
+)
+
+llm = init_chat_model(
+    model_provider="google_genai",
+    model="gemini-2.5-flash",
+    rate_limiter=rate_limiter
+).bind_tools(TOOLS)
+
+
+
+# -------------------------------------------------
+# SYSTEM PROMPT (WILL BE INSERTED ONLY ONCE)
+# -------------------------------------------------
+SYSTEM_PROMPT = f"""
+You are an autonomous quiz-solving agent.
+
+Your job is to:
+1. Load each quiz page from the given URL.
+2. Extract instructions, parameters, and submit endpoint.
+3. Solve tasks exactly.
+4. Submit answers ONLY to the correct endpoint.
+5. Follow new URLs until none remain, then output END.
+
+Rules:
+- For base64 generation of an image NEVER use your own code, always use the "encode_image_to_base64" tool that's provided
+- Never hallucinate URLs or fields.
+- Never shorten endpoints.
+- Always inspect server response.
+- Never stop early.
+- Use tools for HTML, downloading, rendering, OCR, or running code.
+- Include:
+    email = {EMAIL}
+    secret = {SECRET}
+"""
+
+
+# -------------------------------------------------
+# AGENT NODE
+# -------------------------------------------------
+def agent_node(state: AgentState):
+    # time-handling
+    cur_time = time.time()
+    cur_url = os.getenv("url")
+    prev_time = url_time[cur_url]
+    offset = os.getenv("offset")
+    if prev_time is not None:
+        prev_time = float(prev_time)
+        diff = cur_time - prev_time
+
+        if diff >= 180 or (offset != "0" and (cur_time - float(offset)) > 90):
+            print("Timeout exceeded — instructing LLM to purposely submit wrong answer.", diff, "Offset=", offset)
+
+            fail_instruction = """
+            You have exceeded the time limit for this task (over 130 seconds).
+            Immediately call the `post_request` tool and submit a WRONG answer for the CURRENT quiz.
+            """
+
+            # LLM will figure out the right endpoint + JSON structure itself
+            result = llm.invoke([
+                {"role": "user", "content": fail_instruction}
+            ])
+            return {"messages": [result]}
+
+    trimmed_messages = trim_messages(
+        messages=state["messages"],
+        max_tokens=MAX_TOKENS,
+        strategy="last",
+        include_system=True,
+        start_on="human",
+        token_counter=llm,  # Use the LLM to count actual tokens, not just list length
+    )
+    
+    result = llm.invoke(trimmed_messages)
+
+    return {"messages": [result]}
+
+# -------------------------------------------------
+# ROUTE LOGIC (YOURS WITH MINOR SAFETY IMPROVES)
+# -------------------------------------------------
+def route(state):
+    last = state["messages"][-1]
+    # print("=== ROUTE DEBUG: last message type ===")
+
+    tool_calls = getattr(last, "tool_calls", None)
+
+    if tool_calls:
+        print("Route → tools")
+        return "tools"
+
+    content = getattr(last, "content", None)
+
+    if isinstance(content, str) and content.strip() == "END":
+        return END
+
+    if isinstance(content, list) and len(content) and isinstance(content[0], dict):
+        if content[0].get("text", "").strip() == "END":
+            return END
+
+    print("Route → agent")
+    return "agent"
+
+
+
+# -------------------------------------------------
+# GRAPH
+# -------------------------------------------------
+graph = StateGraph(AgentState)
+
+graph.add_node("tools", ToolNode(TOOLS))
+
+graph.add_edge(START, "agent")
+graph.add_edge("tools", "agent")
+graph.add_conditional_edges("agent", route)
+robust_retry = {
+    "initial_interval": 1,
+    "backoff_factor": 2,
+    "max_interval": 60,
+    "max_attempts": 10
+}
+
+graph.add_node("agent", agent_node, retry=robust_retry)
+app = graph.compile()
+
+
+
+# -------------------------------------------------
+# RUNNER
+# -------------------------------------------------
+def run_agent(url: str):
+    # system message is seeded ONCE here
+    initial_messages = [
+        {"role": "system", "content": SYSTEM_PROMPT},
+        {"role": "user", "content": url}
+    ]
+
+    app.invoke(
+        {"messages": initial_messages},
+        config={"recursion_limit": RECURSION_LIMIT}
+    )
+
+    print("Tasks completed successfully!")

main.py

@@ -0,0 +1,61 @@
+from fastapi import FastAPI, Request, BackgroundTasks
+from fastapi.responses import JSONResponse
+from fastapi.exceptions import HTTPException
+from fastapi.middleware.cors import CORSMiddleware
+from agent import run_agent
+from dotenv import load_dotenv
+import uvicorn
+import os
+from shared_store import url_time, BASE64_STORE
+import time
+
+load_dotenv()
+
+EMAIL = os.getenv("EMAIL") 
+SECRET = os.getenv("SECRET")
+
+app = FastAPI()
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],  # or specific domains
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+START_TIME = time.time()
+@app.get("/healthz")
+def healthz():
+    """Simple liveness check."""
+    return {
+        "status": "ok",
+        "uptime_seconds": int(time.time() - START_TIME)
+    }
+
+@app.post("/solve")
+async def solve(request: Request, background_tasks: BackgroundTasks):
+    try:
+        data = await request.json()
+    except Exception:
+        raise HTTPException(status_code=400, detail="Invalid JSON")
+    if not data:
+        raise HTTPException(status_code=400, detail="Invalid JSON")
+    url = data.get("url")
+    secret = data.get("secret")
+    if not url or not secret:
+        raise HTTPException(status_code=400, detail="Invalid JSON")
+    
+    if secret != SECRET:
+        raise HTTPException(status_code=403, detail="Invalid secret")
+    url_time.clear() 
+    BASE64_STORE.clear()  
+    print("Verified starting the task...")
+    os.environ["url"] = url
+    os.environ["offset"] = "0"
+    url_time[url] = time.time()
+    background_tasks.add_task(run_agent, url)
+
+    return JSONResponse(status_code=200, content={"status": "ok"})
+
+
+if __name__ == "__main__":
+    uvicorn.run(app, host="0.0.0.0", port=7860)

pyproject.toml

@@ -0,0 +1,38 @@
+[project]
+name = "tdsproject2"
+version = "0.1.0"
+description = "Add your description here"
+readme = "README.md"
+requires-python = ">=3.12"
+dependencies = [
+    "playwright>=1.56.0",
+    "beautifulsoup4>=4.14.2",
+    "langgraph>=1.0.3",
+    "langchain>=0.2.0",
+    "langchain-community>=0.2.0",
+    "langchain-google-genai>=1.0.0",
+    "google-genai>=0.17.0",
+    "jsonpatch>=1.33",
+    "python-dotenv>=1.2.1",
+    "pandas>=2.3.3",
+    "fastapi>=0.121.3",
+    "uvicorn>=0.38.0",
+    "requests>=2.32.5",
+    "pillow>=12.0.0",
+    "pytesseract>=0.3.13",
+    "speechrecognition>=3.14.4",
+    "pydub>=0.25.1",
+    "geopy>=2.4.1",
+    "scikit-learn>=1.7.2",
+    "matplotlib>=3.10.7",
+    "pypdf2>=3.0.1",
+    "ffmpeg-python>=0.2.0",
+    "numpy>=2.3.5",
+    "networkx>=3.6",
+    "fuzzywuzzy>=0.18.0",
+    "python-levenshtein>=0.27.3",
+    "duckdb>=1.4.2",
+    "pypdf>=6.4.0",
+    "scipy>=1.16.3",
+    "haversine>=2.9.0",
+]

shared_store.py

@@ -0,0 +1,2 @@
+BASE64_STORE = {}
+url_time = {}