Version_1/main.py

import asyncio
import sys
import time
from fastapi import FastAPI, HTTPException
from pydantic import BaseModel
from urllib.parse import urlparse  # Clean URL parsing utility

# --- Keep all your existing structural imports perfectly intact ---
from models.diagnostic_matrix import DiagnosticMatrix, LayerResults
from layers.layer_1_lexical import analyze_lexical_structure
from layers.layer_2_network import analyze_network_footprint
from layers.layer_3_crypto import analyze_crypto_handshake
from layers.layer_4_sandbox import analyze_content_sandbox
from layers.layer_5_context import analyze_global_context
from fastapi.middleware.cors import CORSMiddleware

# 1. Initialize the Master Web Engine
app = FastAPI(
    title="Chimera V1 Master Security Server", 
    description="Unified API hosting Chimera V1 and future engines"
)

app.add_middleware(
    CORSMiddleware,
    allow_origins=["http://0.0.0.0:8000",
                  "http://127.0.0.1:8000"], # Update this to your frontend URL in production
    allow_credentials=False, # Must be False if mode is 'omit' in JS
    allow_methods=["*"],
    allow_headers=["*"],
    )


# 2. Define what incoming web data must look like
class URLPayload(BaseModel):
    url: str

# --- Updated Core Processing Engine with Userinfo Interception ---
async def process_payload(url: str) -> DiagnosticMatrix:
    start_time = time.perf_counter()
    circuit_breaks = []
    
    # --- SANITIZATION GATE & EXPLOIT DETECTION ---
    try:
        parsed_url = urlparse(url)
        # Check if the userinfo credential parameter is present (e.g., 'youtube' in youtube@evil-site.com)
        if parsed_url.username:
            exploit_detected = True
            # Reconstruct the true underlying network destination host target for downstream validation
            target_url = f"{parsed_url.scheme}://{parsed_url.hostname}"
            if parsed_url.path:
                target_url += parsed_url.path
            if parsed_url.query:
                target_url += f"?{parsed_url.query}"
        else:
            exploit_detected = False
            target_url = url
    except Exception:
        exploit_detected = False
        target_url = url

    async def safe_layer_4(target):
        try:
            return await asyncio.wait_for(analyze_content_sandbox(target), timeout=0.150)
        except asyncio.TimeoutError:
            circuit_breaks.append("Layer_4_Sandbox_Timeout")
            return None 

    # All analysis layers now process the TRUE, extracted network host destination target
    results = await asyncio.gather(
        analyze_lexical_structure(target_url),
        analyze_network_footprint(target_url),
        analyze_crypto_handshake(target_url),
        safe_layer_4(target_url),
        analyze_global_context(target_url),
        return_exceptions=True 
    )
    
    safe_results = []
    for res in results:
        if isinstance(res, Exception):
            # Fallback change: Under high stress/failure, assign an ambiguous 0.5 
            # penalty state instead of a completely clean 0.0 pass score
            safe_results.append(0.5)
        else:
            safe_results.append(res)
            
    l1, l2, l3, l4, l5 = safe_results
    
    # CRITICAL SECURITY PATCH APPLICATION:
    # If a userinfo redirection exploit attempt was caught at the gate, 
    # force maximum penalty points to override deceptive layer responses.
    if exploit_detected:
        l1 = max(l1 or 0.0, 1.0)  # Max out lexical abnormality risk score
        l5 = max(l5 or 0.0, 1.0)  # Max out suspicious context profile score
        circuit_breaks.append("Userinfo_Obfuscation_Intercepted")

    valid_scores = [s for s in [l1, l2, l3, l4, l5] if s is not None]
    
    if valid_scores:
        base_avg = sum(valid_scores) / len(valid_scores)
        highest_risk = max(valid_scores)
        
        # BLENDED SCORING: Combine the average with the absolute highest threat
        # to prevent pure averages from diluting a single critical vulnerability.
        composite = (base_avg * 0.5) + (highest_risk * 0.5)
    else:
        composite = 0
    
    # Determine the final system classification parameters
    if composite < 0.3:
        verdict = "PASS"
    elif composite <= 0.7:
        verdict = "ESCALATE_TO_V5"
    else:
        verdict = "QUARANTINE"
        
    execution_time = (time.perf_counter() - start_time) * 1000
    
    return DiagnosticMatrix(
        target_url=url, # Returns original string layout for security logging tracking
        composite_score=round(composite, 3),
        verdict=verdict,
        layer_scores=LayerResults(
            layer_1_lexical=round(l1, 2) if l1 is not None else 0.0,
            layer_2_network=round(l2, 2) if l2 is not None else 0.0,
            layer_3_crypto=round(l3, 2) if l3 is not None else 0.0,
            layer_4_sandbox=round(l4, 2) if l4 is not None else None,
            layer_5_context=round(l5, 2) if l5 is not None else 0.0
        ),
        execution_time_ms=round(execution_time, 2),
        circuit_breaks=circuit_breaks
    )

# 3. Create the Web Endpoint for Chimera V1 Scanner
@app.post("/predict/chimera-v1")
async def scan_url_endpoint(payload: URLPayload):
    url_to_test = payload.url.strip()
    if not url_to_test:
        raise HTTPException(status_code=400, detail="URL payload cannot be empty")
        
    if not url_to_test.startswith(('http://', 'https://')):
        url_to_test = 'http://' + url_to_test
        
    try:
        # Run your powerful parallel engine
        matrix = await process_payload(url_to_test)
        # Return the matrix directly as JSON data to the web client
        return matrix
    except Exception as e:
        raise HTTPException(status_code=500, detail=f"Internal scan failure: {str(e)}")

# Windows event loop fallback adjustments
if sys.platform == 'win32':
    asyncio.set_event_loop_policy(asyncio.WindowsSelectorEventLoopPolicy())