reports/create_reports.py

from agents import Agent, OpenAIChatCompletionsModel, AsyncOpenAI, Runner, RunConfig
import os
from bson import ObjectId
from dotenv import load_dotenv, find_dotenv
from reports.crawl import crawl
import json
from reports.generate_report import generate_report
from reports.email_sender import send_email_with_pdf, send_ssl_alert_email
from reports.ssl_checker import check_ssl_certificate_async
import datetime
from reports.database import db
from reports.gemini_token_tracker import track_token_usage


load_dotenv(find_dotenv())

external_client = AsyncOpenAI(
    api_key=os.getenv("GEMINI_API_KEY"),
    base_url="https://generativelanguage.googleapis.com/v1beta/openai/",
)

llm_model = OpenAIChatCompletionsModel(
    model="gemini-2.5-flash",
    openai_client=external_client
)

config = RunConfig(
    model=llm_model,
    model_provider=external_client,
    tracing_disabled=True
)

instructions = """
You are a professional Web SEO Audit Agent built to analyze individual page crawl data and provide precise, actionable, and technically accurate SEO recommendations.

You will receive crawl data for a single web page, including meta tags, schema details, broken links, images, and structure. Your goal is to generate a **structured, concise, and page-specific SEO report**.

### Your Output Must Be a Valid JSON Object and Include:
1. **seo_summary**  
   - Give a concise summary of the SEO health of this specific page.
   - Do not mention total number of pages or crawl scope. Focus only on this page.

2. **detected_issues**  
   - List clear technical or SEO issues (if any).  
   - If no issues exist, say: “No major SEO issues detected.”

3. **ai_recommendations**  
   - Suggest clear, technical, and developer-friendly improvements for each issue.
   - Be specific about what to fix and how.  
   - If no issues exist, say: “No action needed. Page is well optimized.”

4. **structured_data_schema_feedback**  
   - Mention what schema types were found.
   - Suggest relevant additional schemas (if beneficial).

5. **optimization_suggestions**  
   - Include best practices related to this specific page only (if applicable).

6. **priority_level**  
   - Assign High, Medium, or Low to each recommendation (if any).

### Important Rules:
- ❌ Do NOT mention how many pages were crawled.
- ❌ Do NOT mention crawl depth, website size, or accessibility of other pages.
- ❌ Do NOT add assumptions beyond the provided crawl data.
- ✅ Focus only on the current page’s content and structure.
- ✅ Keep tone professional, precise, and data-driven.
- ✅ Output must always be a single JSON object.

### Example Output: 
{
  "seo_summary": "Page is mostly optimized but missing a canonical tag.",
  "detected_issues": ["Missing canonical tag", "2 broken links found"],
  "ai_recommendations": [
    {"action": "Add a canonical tag for this page", "priority": "High"},
    {"action": "Fix 2 broken links to improve crawlability", "priority": "High"}
  ],
  "structured_data_schema_feedback": {
    "existing": ["Organization"],
    "suggested": ["BreadcrumbList"]
  },
  "optimization_suggestions": [
    "Optimize large images for faster load times.",
    "Add descriptive alt text for all images."
  ],
  "priority_level": {
    "canonical_tag": "High",
    "broken_links": "High",
    "image_optimization": "Medium"
  }
}
"""

agent: Agent = Agent(
    name="Web SEO Audit Agent",
    instructions=instructions,
    model=llm_model
)


async def create_report(website_url: str, user_info: dict = None, user_email: str = None, plan: str = "free",report_id=None):
    
    try:
        # If user_info is provided, extract user_email and plan from it
        if user_info:
            if user_email is None:  # Only extract from user_info if not provided separately
                user_email = user_info.get("email")
            # Update plan from user_info if available
            user_plan = user_info.get("plan")
            if user_plan:
                plan = user_plan
        # If user_info is not provided but user_email is, fetch user details from database
        elif user_email:
            user = await db["users"].find_one({"email": user_email})
            if user:
                # Remove sensitive information
                user_info = {k: v for k, v in user.items() if k not in ["password", "otp_hash", "otp_expires", "reset_otp_hash", "reset_otp_expires"]}
                # Extract plan from user info if available
                user_plan = user_info.get("plan")
                if user_plan:
                    plan = user_plan

        
        crawl_result = await crawl(website_url=website_url, plan=plan)

        # Handle the new return format from crawl function
        if isinstance(crawl_result, dict) and "status" in crawl_result:
            if crawl_result["status"] == "error":
                print(f"X Crawl failed: {crawl_result['message']}")
                await db["reports"].update_one(
                    {"_id": ObjectId(report_id)},
                    {"$set": {
                        "status": "failed",
                        "error_message": crawl_result["message"],
                        "updated_at": datetime.datetime.utcnow()
                    }}
                )
                return {"status": "error", "message": crawl_result["message"], "data": None}
            elif crawl_result["status"] == "success":
                crawl_data = crawl_result["data"]
            else:
                print(f"X Crawl returned unexpected status: {crawl_result}")
                await db["reports"].update_one(
                    {"_id": ObjectId(report_id)},
                    {"$set": {
                        "status": "failed",
                        "error_message": crawl_result["message"],
                        "updated_at": datetime.datetime.utcnow()
                    }}
                )
                return {"status": "error", "message": "Crawl returned unexpected status.", "data": None}
        else:
            # Fallback for old format (backward compatibility)
            crawl_data = crawl_result

        if not crawl_data or not isinstance(crawl_data, list):
            print(f"X Crawl failed or returned invalid data: {crawl_data}")
            await db["reports"].update_one(
                {"_id": ObjectId(report_id)},
                {"$set": {
                    "status": "failed",
                    "error_message": "Crawl returned invalid data",
                    "updated_at": datetime.datetime.utcnow()
                }}
            )
            return {"status": "error", "message": "Crawl failed or returned invalid data.", "data": None}

        # ============================================
        # RATE LIMIT PROTECTION: Limit pages per audit
        # ============================================
        # Check remaining daily requests (20 requests/day hard limit)
        from reports.gemini_token_tracker import get_current_usage
        try:
            usage = await get_current_usage()
            requests_today = usage.get("requests_count", 0)
            remaining_requests = max(0, 20 - requests_today)
            
            if remaining_requests < len(crawl_data):
                print(f"⚠️ Low on daily requests: {remaining_requests} remaining. Limiting to {remaining_requests} pages.")
                crawl_data = crawl_data[:remaining_requests]
        except Exception as e:
            print(f"⚠️ Could not check request count: {e}")
        # ============================================

        page_reports = []
        total_tokens_used = 0
        api_error_occurred = False
        api_error_message = None

        for page_data in crawl_data:
            url = page_data.get("url") if isinstance(page_data, dict) else None

            try:
                # Agent Layer for suggestion
                ai_result = await Runner.run(
                    agent,
                    input=json.dumps({"page_data": page_data}, ensure_ascii=False, indent=2),
                    run_config=config
                )

                # Extract token usage from AI result if available
                page_tokens = 0
                if hasattr(ai_result, 'usage') and ai_result.usage:
                    page_tokens = ai_result.usage.total_tokens or 0
                    total_tokens_used += page_tokens
                    print(f"📊 Page {url}: {page_tokens} tokens used")

                #  Convert string output into JSON object
                try:
                    parsed_result = json.loads(ai_result.final_output.strip("`json\n").strip("`"))
                except json.JSONDecodeError:
                    print(f"! Warning: Could not parse AI result for {url}. Storing raw string instead.")
                    parsed_result = {"raw_output": ai_result.final_output}


                page_reports.append({
                    "url": url,
                    "ai_result": parsed_result,
                    "tokens_used": page_tokens
                })
                
            except Exception as e:
                # Track API errors (including rate limits)
                api_error_occurred = True
                api_error_message = str(e)
                print(f"❌ API error for {url}: {e}")
                
                # Check if it's a rate limit error (429)
                if "429" in str(e) or "RESOURCE_EXHAUSTED" in str(e) or "exceeded your current quota" in str(e):
                    print(f"⚠️ RATE LIMIT HIT: Gemini API quota exceeded!")
                    # Track this as a failed request for monitoring
                    await track_token_usage(
                        tokens=0,  # No tokens consumed on failed request
                        endpoint="seo_audit_failed_rate_limit",
                        user_email=user_email
                    )
                    break  # Stop processing more pages
        
        # Track total token usage for this audit (only if we had successful API calls)
        if total_tokens_used > 0:
            print(f"📊 Total tokens used for audit: {total_tokens_used}")
            await track_token_usage(
                tokens=total_tokens_used,
                endpoint="seo_audit",
                user_email=user_email
            )
        elif api_error_occurred:
            # Log that audit completed with errors
            print(f"⚠️ Audit completed with API errors. No tokens tracked.")
            
        # print("AI Recommendation: ",page_reports)
        

        # Now generate final structured report
        # First, check SSL certificate
        ssl_info = await check_ssl_certificate_async(website_url)
        
        report_result = await generate_report(
            website=website_url,
            crawl_data=crawl_data,
            ai_recommendation=page_reports,
            user_info=user_info,
            plan_type=plan,
            ssl_info=ssl_info  # Pass SSL info to PDF generator
        )

        # Handle the new return format from generate_report function
        if isinstance(report_result, dict) and "status" in report_result:
            if report_result["status"] == "error":
                print(f"X Report generation failed: {report_result['message']}")
                await db["reports"].update_one(
                    {"_id": ObjectId(report_id)},
                    {"$set": {
                        "status": "failed",
                        "error_message": report_result["message"],
                        "updated_at": datetime.datetime.utcnow()
                    }}
                )
                return {"status": "error", "message": report_result["message"], "data": None}
            elif report_result["status"] == "success":
                report = report_result
            else:
                print(f"X Report generation returned unexpected status: {report_result}")
                return {"status": "error", "message": "Report generation returned unexpected status.", "data": None}
        else:
            # Fallback for old format (backward compatibility)
            report = {"data": report_result}
            
        user_id = user_info.get("_id") if user_info else None    

        # print("Storing report in MongoDB...")
        report_data = report["data"]

        # Attach user reference
        report_data["user_id"] = ObjectId(user_id)

        # Add useful metadata
        report_data["website"] = website_url
        report_data["created_at"] = datetime.datetime.utcnow()
        report_data["status"] = "completed"
        
        # Update the existing "in-progress" report
        await db["reports"].update_one(
            {"_id": ObjectId(report_id)},  # pass report_id from /audit
            {"$set": {
                "status": "completed",
                "pdf": report_data.get("pdf"),    # PDF bytes
                "score": report_data.get("score"),
                "issues": report_data.get("issues"),
                "ai_result": report_data.get("ai_result"),
                "updated_at": datetime.datetime.utcnow()
            }}
        )

        pdf_data = report["data"]["pdf"]

        # Create a meaningful filename
        safe_site = website_url.replace("https://", "").replace("http://", "").replace("/", "_")
        timestamp = datetime.datetime.now().strftime("%Y-%m-%d")
        pdf_filename = f"SEO_Report_{safe_site}_{timestamp}.pdf"

        # Generate professional HTML email template
        from reports.email_sender import get_email_html_template

        # Extract username from user_info
        username = user_info.get("username") if user_info else None
        
        # Get backend URL from environment variable for email logo
        backend_url = os.getenv("BACKEND_URL", "http://localhost:8000")

        email_html = get_email_html_template(
            website_url=website_url,
            username=username,
            backend_url=backend_url
        )

        # Send the PDF report to the user's email
        email_result = send_email_with_pdf(
            to_email=user_email,
            subject=f"Your SEO Audit Report for {website_url}",
            message=email_html,
            pdf_data=pdf_data,
            pdf_filename=pdf_filename
        )

        # Update the existing "in-progress" report with email delivery status
        await db["reports"].update_one(
            {"_id": ObjectId(report_id)},
            {"$set": {
                "status": "completed",
                "pdf": report_data.get("pdf"),    # PDF bytes
                "score": report_data.get("score"),
                "issues": report_data.get("issues"),
                "ai_result": report_data.get("ai_result"),
                "email_sent": email_result["success"],
                "email_delivery_status": email_result["status"],
                "email_sent_at": datetime.datetime.utcnow() if email_result["success"] else None,
                "email_error": email_result.get("error"),
                "ssl_info": ssl_info,  # Store SSL info in report for reference
                "updated_at": datetime.datetime.utcnow()
            }}
        )
        
        # ============ SSL ALERT EMAIL ============
        # Send SSL alert email if certificate is expiring within 7 days or already expired
        username = user_info.get("username") if user_info else None
        backend_url = os.getenv("BACKEND_URL", "http://localhost:8000")
        
        if ssl_info:
            is_expired = ssl_info.get("is_expired", False)
            days_until_expiry = ssl_info.get("days_until_expiry")
            
            # Send alert if expired OR expiring within 7 days
            should_send_alert = (
                (is_expired) or 
                (days_until_expiry is not None and 0 <= days_until_expiry <= 7)
            )
            
            if should_send_alert:
                print(f"🔒 SSL Alert: Sending SSL alert email for {website_url} (expired: {is_expired}, days remaining: {days_until_expiry})")
                
                ssl_alert_result = send_ssl_alert_email(
                    to_email=user_email,
                    website_url=website_url,
                    username=username,
                    ssl_info=ssl_info,
                    backend_url=backend_url
                )
                
                # Update report with SSL alert status
                await db["reports"].update_one(
                    {"_id": ObjectId(report_id)},
                    {"$set": {
                        "ssl_alert_sent": ssl_alert_result["success"],
                        "ssl_alert_sent_at": datetime.datetime.utcnow() if ssl_alert_result["success"] else None,
                        "ssl_alert_error": ssl_alert_result.get("error"),
                        "ssl_status": {
                            "has_ssl": ssl_info.get("has_ssl"),
                            "ssl_valid": ssl_info.get("ssl_valid"),
                            "is_expired": ssl_info.get("is_expired"),
                            "days_until_expiry": ssl_info.get("days_until_expiry"),
                            "expiry_date": ssl_info.get("expiry_date"),
                            "issuer": ssl_info.get("issuer")
                        }
                    }}
                )
                
                if ssl_alert_result["success"]:
                    print("✅ SSL alert email sent successfully")
                else:
                    print(f"❌ Failed to send SSL alert email: {ssl_alert_result.get('error')}")
            else:
                # SSL is healthy, just store the info
                await db["reports"].update_one(
                    {"_id": ObjectId(report_id)},
                    {"$set": {
                        "ssl_status": {
                            "has_ssl": ssl_info.get("has_ssl"),
                            "ssl_valid": ssl_info.get("ssl_valid"),
                            "is_expired": ssl_info.get("is_expired"),
                            "days_until_expiry": ssl_info.get("days_until_expiry"),
                            "expiry_date": ssl_info.get("expiry_date"),
                            "issuer": ssl_info.get("issuer")
                        }
                    }}
                )
                print(f"✓ SSL certificate healthy ({days_until_expiry} days remaining)")

        # ============ BUILD RESPONSE WITH SSL STATUS ============
        ssl_status_response = None
        if ssl_info:
            ssl_status_response = {
                "has_ssl": ssl_info.get("has_ssl", False),
                "ssl_valid": ssl_info.get("ssl_valid", False),
                "is_expired": ssl_info.get("is_expired", False),
                "days_until_expiry": ssl_info.get("days_until_expiry"),
                "expiry_date": ssl_info.get("expiry_date").isoformat() if ssl_info.get("expiry_date") else None,
                "issuer": ssl_info.get("issuer", "Unknown"),
                "alert_sent": ssl_info.get("is_expired", False) or (ssl_info.get("days_until_expiry") is not None and 0 <= ssl_info.get("days_until_expiry") <= 7)
            }

        print("Report stored successfully with ID:")
        return {
            "status": "success",
            "data": {
                "message": "Report generated and stored successfully",
                "ssl_status": ssl_status_response
            }
        }

    except Exception as e:
        print("Unexpected error:", str(e))

        await db["reports"].update_one(
            {"_id": ObjectId(report_id)},
            {"$set": {
                "status": "failed",
                "error_message": str(e),
                "updated_at": datetime.datetime.utcnow()
            }}
        )
        
        return {
            "status": "error",
            "message": f"Report generation failed: {str(e)}",
            "data": None
        }


# ---------------- Run Locally ----------------
if __name__ == "__main__":
    import asyncio
    asyncio.run(create_report(website_url="https://blueverse.ae", user_email="subhancontact2@gmail.com", plan="free"))