app.py

import gradio as gr
import pandas as pd
import re
import os
import numpy as np
from datetime import datetime, timedelta
from huggingface_hub import InferenceClient, HfApi
import io
import base64
import unicodedata
import warnings
from requests.exceptions import HTTPError

# Optional: phone normalization (graceful fallback if missing) 
try:
    import phonenumbers
except Exception:
    phonenumbers = None

# Read from Space secret - check both common environment variables
HF_TOKEN = (os.getenv("HF_TOKEN", "") or os.getenv("HF_SPACES", "")).strip()

# Initialize API clients
api = HfApi(token=HF_TOKEN if HF_TOKEN else None)
client = InferenceClient(token=HF_TOKEN) if HF_TOKEN else InferenceClient()

# Your target models
DEFAULT_MODEL_ID = "tiiuae/falcon-7b-instruct"
SECONDARY_MODEL_ID = "HuggingFaceH4/zephyr-7b-beta"

def hf_inference_supported(model_id: str, task: str = "text-generation") -> bool:
    """Check if model is supported on HF's Serverless Inference API"""
    try:
        info = api.model_info(model_id, expand=["inferenceProviderMapping"])
        mapping = getattr(info, "inference_provider_mapping", None) or getattr(info, "inferenceProviderMapping", None)
        if not mapping:
            return False
        # mapping is a dict: { provider_name: { tasks: [...] , ... } }
        return any(
            prov.lower() in {"hf-inference", "hf_serverless", "hf-inference-api"} and
            (task in (details.get("tasks") or []))
            for prov, details in mapping.items()
        )
    except Exception as e:
        print(f"[HF] Provider check failed for {model_id}: {e}")
        return False

def _hf_sanity_check():
    """Enhanced sanity check with provider validation"""
    print("[HF] Token present:", bool(HF_TOKEN), "len:", len(HF_TOKEN) if HF_TOKEN else 0)
    
    if not HF_TOKEN:
        print("[HF] No token provided - API calls will fail")
        return False
    
    # Test models and their provider support
    test_models = [DEFAULT_MODEL_ID, SECONDARY_MODEL_ID]
    
    for model in test_models:
        supported = hf_inference_supported(model)
        print(f"[HF] {model}: {'✓ Supported' if supported else '✗ Not available'} on HF Serverless")
        
        if supported:
            try:
                # Quick test call
                response = client.text_generation(
                    model=model,
                    prompt="Hello",
                    max_new_tokens=5
                )
                print(f"[HF] ✓ {model} API test successful")
                return True
            except HTTPError as e:
                code = getattr(e.response, "status_code", None)
                if code == 401:
                    print(f"[HF] ✗ {model}: Invalid token")
                elif code == 403:
                    print(f"[HF] ✗ {model}: License not accepted")
                else:
                    print(f"[HF] ✗ {model}: HTTP {code}")
            except Exception as e:
                print(f"[HF] ✗ {model}: {e}")
    
    return False

# Run sanity check
_hf_sanity_check()

# --------------------------
# Utility Functions
# --------------------------

def generate_summary(prompt: str, model_id: str = DEFAULT_MODEL_ID):
    """Generate AI-powered analysis using HuggingFace models with proper error handling"""
    
    if not HF_TOKEN:
        return "⚠️ HF token missing. Set HF_TOKEN in Space Secrets and restart the Space."
    
    # Check if model is available on HF Serverless
    if not hf_inference_supported(model_id):
        return (f"⚠️ Model not available on HF's Serverless Inference API: {model_id}\n\n"
                f"This model either:\n"
                f"• Requires a Dedicated Inference Endpoint\n"
                f"• Is not supported for text-generation tasks\n"
                f"• Has restricted access\n\n"
                f"Check the HuggingFace Models Support Matrix for alternatives, "
                f"or deploy a Dedicated Inference Endpoint.")
    
    try:
        # Use the correct parameter name: prompt (not inputs)
        response = client.text_generation(
            model=model_id,
            prompt=prompt,
            max_new_tokens=500,
            temperature=0.7,
            do_sample=True,
        )
        return response
        
    except HTTPError as e:
        code = getattr(e.response, "status_code", None)
        
        if code == 401:
            return "⚠️ HF token missing/invalid. Set HF_TOKEN in Space Secrets and restart the Space."
        
        elif code == 403:
            return (f"⚠️ Access denied to {model_id}.\n\n"
                   f"Required action:\n"
                   f"• Visit: https://huggingface.co/{model_id}\n"
                   f"• Accept the model's license with the same HF account as your token\n"
                   f"• Or use a model with open access")
        
        elif code == 503:
            return (f"⚠️ Model {model_id} is currently loading or unavailable.\n"
                   f"This is common with CPU-only inference. Try again in a few minutes,\n"
                   f"or consider GPU-accelerated inference for better reliability.")
        
        else:
            return f"⚠️ Inference error (HTTP {code}): {e}"
    
    except Exception as e:
        return f"⚠️ Inference error: {e}"

def create_download_link(df, filename):
    """Create downloadable CSV from DataFrame"""
    csv = df.to_csv(index=False)
    b64 = base64.b64encode(csv.encode()).decode()
    return f'<a href="data:file/csv;base64,{b64}" download="{filename}">📥 Download {filename}</a>'

# Update the Gradio interface section for the AI chatbot:

    # AI Consultant Module
    with gr.Column(visible=False) as ai_section:
        gr.Markdown("## AI-Powered Risk Consultant")
        gr.Markdown("### Chat with our AI expert about fraud detection and risk management")

        model_choice = gr.Dropdown(
            choices=[DEFAULT_MODEL_ID, SECONDARY_MODEL_ID],
            label="Choose AI Model",
            value=DEFAULT_MODEL_ID,
            info="Select the language model for analysis"
        )

        # Add model status indicator
        with gr.Row():
            gr.Markdown("""
            **Model Status**:
            - ✅ = Available on HF Serverless
            - ⚠️ = Requires license acceptance
            - ❌ = Not supported
            """)

        chatbot = gr.Chatbot(
            label="Risk Management Consultant",
            height=500
        )

        with gr.Row():
            msg = gr.Textbox(
                label="Ask about fraud detection, risk assessment, compliance...",
                placeholder="e.g., How can I improve my transaction fraud detection?",
                scale=4
            )
            submit_btn = gr.Button("Send", scale=1, variant="primary")

        msg.submit(
            chatbot_respond,
            inputs=[msg, chatbot, model_choice],
            outputs=[chatbot, msg]
        )
        submit_btn.click(
            chatbot_respond,
            inputs=[msg, chatbot, model_choice],
            outputs=[chatbot, msg]
        )

# -------------------------
# Data Validation & Normalization Helpers (Point 1)
# -------------------------

def _standardize_columns(df: pd.DataFrame) -> pd.DataFrame:
    df = df.copy()
    df.columns = [re.sub(r"\s+", "_", str(c).strip().lower()) for c in df.columns]
    return df


def _norm_str(s: str) -> str:
    if pd.isna(s):
        return ""
    s = unicodedata.normalize("NFKC", str(s))
    s = re.sub(r"\s+", " ", s).strip()
    return s


def _norm_name(name: str) -> str:
    n = _norm_str(name)
    parts = [p if len(p) <= 3 and p.isupper() else p.title() for p in n.split(" ")]
    return " ".join(parts)


def _norm_email(email: str):
    e = _norm_str(email).lower()
    if not e:
        return "", "missing_email"
    if not re.match(r"^[a-z0-9._%+\-]+@[a-z0-9.\-]+\.[a-z]{2,}$", e):
        return e, "invalid_email"
    return e, None


def _norm_phone(phone: str, default_region: str = "IN"):
    raw = _norm_str(phone)
    if not raw:
        return "", "missing_phone"
    if phonenumbers:
        try:
            num = phonenumbers.parse(raw, default_region)
            if not (phonenumbers.is_possible_number(num) and phonenumbers.is_valid_number(num)):
                return raw, "invalid_phone"
            return phonenumbers.format_number(num, phonenumbers.PhoneNumberFormat.E164), None
        except Exception:
            return raw, "invalid_phone"
    digits = re.sub(r"\D", "", raw)
    if len(digits) < 8:
        return raw, "invalid_phone"
    return "+" + digits, None


def _to_datetime(series, errors="coerce"):
    return pd.to_datetime(series, errors=errors, utc=False, infer_datetime_format=True)


def _to_numeric(series):
    return pd.to_numeric(series, errors="coerce")


def _new_issues_list():
    return []


def _add_issue(issues, row_idx, field, issue, value):
    issues.append({
        "row": int(row_idx) if pd.notna(row_idx) else None,
        "field": field,
        "issue": issue,
        "value": None if pd.isna(value) else str(value)
    })


def _issues_df(issues):
    return pd.DataFrame(issues, columns=["row", "field", "issue", "value"]) if issues else pd.DataFrame(columns=["row", "field", "issue", "value"])


# ---------- Per-module preparation ----------

def _prepare_transactions_df(df_raw: pd.DataFrame):
    issues = _new_issues_list()
    df = _standardize_columns(df_raw)

    required = {"customer_id", "amount", "timestamp"}
    missing = required - set(df.columns)
    if missing:
        return None, _issues_df([{"row": None, "field": "/".join(sorted(missing)), "issue": "missing_required_columns", "value": ""}]), f"Missing required columns: {sorted(missing)}"

    df["customer_id"] = df["customer_id"].astype(str).apply(_norm_str)
    df["amount"] = _to_numeric(df["amount"])
    df["timestamp"] = _to_datetime(df["timestamp"])

    for c in ["merchant_category", "merchant", "country", "device_id", "ip_address"]:
        if c in df.columns:
            df[c] = df[c].astype(str).apply(_norm_str)

    for idx, v in df["amount"].items():
        if pd.isna(v):
            _add_issue(issues, idx, "amount", "non_numeric_amount", df_raw.loc[idx, "amount"])

    for idx, v in df["timestamp"].items():
        if pd.isna(v):
            _add_issue(issues, idx, "timestamp", "unparseable_timestamp", df_raw.loc[idx, "timestamp"])

    clean = df.dropna(subset=["customer_id", "amount", "timestamp"]).copy()
    kept = len(clean); total = len(df_raw)
    quality = f"Validated {total} rows → kept {kept}, dropped {total - kept} due to invalid amount/timestamp/customer_id."
    return clean, _issues_df(issues), quality


def _prepare_kyc_df(df_raw: pd.DataFrame):
    issues = _new_issues_list()
    df = _standardize_columns(df_raw)

    required = {"customer_id", "name", "email", "dob"}
    missing = required - set(df.columns)
    if missing:
        return None, _issues_df([{"row": None, "field": "/".join(sorted(missing)), "issue": "missing_required_columns", "value": ""}]), f"Missing required columns: {sorted(missing)}"

    df["customer_id"] = df["customer_id"].astype(str).apply(_norm_str)
    df["name"] = df["name"].astype(str).apply(_norm_name)

    emails = []
    for idx, e in df["email"].items():
        ne, err = _norm_email(e)
        emails.append(ne)
        if err:
            _add_issue(issues, idx, "email", err, e)
    df["email"] = emails

    if "phone" in df.columns:
        phones = []
        for idx, p in df["phone"].items():
            np_, err = _norm_phone(p, default_region="IN")
            phones.append(np_)
            if err:
                _add_issue(issues, idx, "phone", err, p)
        df["phone"] = phones

    df["dob_parsed"] = _to_datetime(df["dob"])
    now = pd.Timestamp.now(tz=None)
    too_old_cutoff = now - pd.DateOffset(years=120)
    for idx, d in df["dob_parsed"].items():
        if pd.isna(d):
            _add_issue(issues, idx, "dob", "unparseable_dob", df_raw.loc[idx, "dob"])
        elif d > now:
            _add_issue(issues, idx, "dob", "future_dob", df_raw.loc[idx, "dob"])
        elif d < too_old_cutoff:
            _add_issue(issues, idx, "dob", "age_over_120", df_raw.loc[idx, "dob"])

    clean = df.dropna(subset=["customer_id", "name", "email", "dob_parsed"]).copy()
    kept = len(clean); total = len(df_raw)
    quality = f"KYC validated {total} rows → kept {kept}, dropped {total - kept} due to email/phone/DOB issues."
    return clean, _issues_df(issues), quality


def _prepare_sanctions_customer_df(df_raw: pd.DataFrame):
    issues = _new_issues_list()
    df = _standardize_columns(df_raw)

    required = {"customer_id", "name"}
    missing = required - set(df.columns)
    if missing:
        return None, _issues_df([{"row": None, "field": "/".join(sorted(missing)), "issue": "missing_required_columns", "value": ""}]), f"Missing required columns: {sorted(missing)}"

    df["customer_id"] = df["customer_id"].astype(str).apply(_norm_str)
    df["name"] = df["name"].astype(str).apply(_norm_name)

    if "dob" in df.columns:
        df["dob_parsed"] = _to_datetime(df["dob"])
        for idx, d in df["dob_parsed"].items():
            if pd.isna(d):
                _add_issue(issues, idx, "dob", "unparseable_dob", df_raw.loc[idx, "dob"])

    if "country" in df.columns:
        df["country"] = df["country"].astype(str).apply(_norm_str)

    clean = df.dropna(subset=["customer_id", "name"]).copy()
    quality = f"Sanctions input validated {len(df_raw)} rows → kept {len(clean)}, dropped {len(df_raw)-len(clean)}."
    return clean, _issues_df(issues), quality


def _prepare_sanctions_list_df(sanctions_file):
    if sanctions_file is None:
        return None, pd.DataFrame(), "Using built-in demo sanctions list."
    try:
        raw = pd.read_csv(sanctions_file.name)
        df = _standardize_columns(raw)
        if "name" not in df.columns:
            msg_df = _issues_df([{"row": None, "field": "name", "issue": "missing_required_columns", "value": ""}])
            return None, msg_df, "Uploaded sanctions list missing required 'name' column. Using demo list."
        df["name"] = df["name"].astype(str).apply(_norm_name)
        return df, pd.DataFrame(), "Using uploaded sanctions list."
    except Exception as e:
        warn = _issues_df([{"row": None, "field": "file", "issue": "read_error", "value": str(e)}])
        return None, warn, "Failed to read uploaded sanctions list. Using demo list."


def _prepare_credit_df(df_raw: pd.DataFrame):
    issues = _new_issues_list()
    df = _standardize_columns(df_raw)

    required = {"customer_id"}
    missing = required - set(df.columns)
    if missing:
        return None, _issues_df([{"row": None, "field": "/".join(sorted(missing)), "issue": "missing_required_columns", "value": ""}]), f"Missing required columns: {sorted(missing)}"

    df["customer_id"] = df["customer_id"].astype(str).apply(_norm_str)

    numeric_cols = ["credit_score", "utilization_rate", "debt_to_income", "income", "recent_defaults"]
    for c in numeric_cols:
        if c in df.columns:
            df[c] = _to_numeric(df[c])

    if "credit_score" in df.columns:
        for idx, v in df["credit_score"].items():
            if pd.isna(v):
                _add_issue(issues, idx, "credit_score", "non_numeric", df_raw.loc[idx, "credit_score"])
            elif not (250 <= v <= 950):
                _add_issue(issues, idx, "credit_score", "out_of_range_250_950", v)

    if "utilization_rate" in df.columns:
        for idx, v in df["utilization_rate"].items():
            if pd.isna(v):
                _add_issue(issues, idx, "utilization_rate", "non_numeric", df_raw.loc[idx, "utilization_rate"])
            elif not (0 <= v <= 1.5):
                _add_issue(issues, idx, "utilization_rate", "out_of_range_0_1.5", v)

    if "debt_to_income" in df.columns:
        for idx, v in df["debt_to_income"].items():
            if pd.isna(v):
                _add_issue(issues, idx, "debt_to_income", "non_numeric", df_raw.loc[idx, "debt_to_income"])
            elif not (0 <= v <= 2):
                _add_issue(issues, idx, "debt_to_income", "out_of_range_0_2", v)

    if "income" in df.columns:
        for idx, v in df["income"].items():
            if pd.isna(v):
                _add_issue(issues, idx, "income", "non_numeric", df_raw.loc[idx, "income"])
            elif v < 0:
                _add_issue(issues, idx, "income", "negative_income", v)

    if "recent_defaults" in df.columns:
        for idx, v in df["recent_defaults"].items():
            if pd.isna(v):
                _add_issue(issues, idx, "recent_defaults", "non_numeric", df_raw.loc[idx, "recent_defaults"])
            elif v < 0:
                _add_issue(issues, idx, "recent_defaults", "negative_count", v)

    clean = df.dropna(subset=["customer_id"]).copy()
    kept = len(clean); total = len(df_raw)
    quality = f"Credit input validated {total} rows → kept {kept}, dropped {total - kept} (non-critical fields coerced with issues recorded)."
    return clean, _issues_df(issues), quality


# -------------------------
# 1. Transaction Fraud (Enhanced + Validation)
# -------------------------

def process_transaction_file(file):
    """Process transaction data for fraud detection"""
    try:
        df_raw = pd.read_csv(file.name)
        df, dq_issues, quality = _prepare_transactions_df(df_raw)
        if df is None:
            return pd.DataFrame(), quality, "", dq_issues

        # Enhanced fraud detection rules (on clean df)
        high_risk_mc = (df["merchant_category"] == "HIGH_RISK") if "merchant_category" in df.columns else False
        suspicious_conditions = (
            (df['amount'] > 10000) |
            (df['amount'] < 0) |
            (high_risk_mc) |
            (df.groupby('customer_id')['amount'].transform('sum') > 50000)
        )

        suspicious = df[suspicious_conditions].copy()
        def _reason(x):
            if x['amount'] > 10000: return 'Large Amount'
            if x['amount'] < 0: return 'Negative Amount'
            if ('merchant_category' in df.columns and x.get('merchant_category') == 'HIGH_RISK'): return 'High Risk Merchant'
            return 'Daily Limit Exceeded'
        if not suspicious.empty:
            suspicious['risk_reason'] = suspicious.apply(_reason, axis=1)

        prompt = f"""You are a financial fraud analyst. Analyze these suspicious transactions:
Sample:
{df.head(10).to_string()}
Suspicious Found: {len(suspicious)}
Data Quality: {quality}
Provide a risk assessment and recommended actions."""
        summary = generate_summary(prompt)

        stats = f"{quality}\nFound {len(suspicious)} suspicious transactions out of {len(df)} usable rows."
        return suspicious, summary, stats, dq_issues

    except Exception as e:
        return pd.DataFrame(), f"Error processing file: {str(e)}", "", pd.DataFrame()


# -------------------------
# 2. KYC Fraud Analysis (Enhanced + Validation)
# -------------------------

def process_kyc_file(file):
    """Process KYC data for identity fraud detection"""
    try:
        df_raw = pd.read_csv(file.name)
        df, dq_issues, quality = _prepare_kyc_df(df_raw)
        if df is None:
            return pd.DataFrame(), quality, "", dq_issues

        flagged_records = []

        dup_email = df[df.duplicated('email', keep=False)]
        if not dup_email.empty:
            dup_email = dup_email.copy(); dup_email['flag_reason'] = 'Duplicate Email'
            flagged_records.append(dup_email)

        if 'phone' in df.columns:
            dup_phone = df[df.duplicated('phone', keep=False)]
            if not dup_phone.empty:
                dup_phone = dup_phone.copy(); dup_phone['flag_reason'] = 'Duplicate Phone'
                flagged_records.append(dup_phone)

        if 'name' in df.columns:
            suspicious_names = df[
                df['name'].str.contains(r'^[A-Z]+$', na=False) |
                df['name'].str.contains(r'\d', na=False) |
                (df['name'].str.len() < 3)
            ].copy()
            if not suspicious_names.empty:
                suspicious_names['flag_reason'] = 'Suspicious Name Pattern'
                flagged_records.append(suspicious_names)

        flagged_df = pd.concat(flagged_records, ignore_index=True).drop_duplicates() if flagged_records else pd.DataFrame()

        prompt = f"""You are a KYC fraud analyst. Review identity records for potential fraud.
Total: {len(df)}
Flagged: {len(flagged_df)}
Data Quality: {quality}
Flag reasons distribution:
{flagged_df['flag_reason'].value_counts().to_string() if not flagged_df.empty else 'None'}
Recommend verification steps."""
        summary = generate_summary(prompt)

        stats = f"{quality}\nFlagged {len(flagged_df)} suspicious KYC records out of {len(df)} usable rows."
        return flagged_df, summary, stats, dq_issues

    except Exception as e:
        return pd.DataFrame(), f"Error processing KYC file: {str(e)}", "", pd.DataFrame()


# -------------------------
# 3. Sanctions Check (Enhanced + Validation)
# -------------------------

def process_sanctions_file(file, sanctions_file=None):
    """Process customer data against sanctions/PEP lists"""
    try:
        customer_raw = pd.read_csv(file.name)
        customers, cust_issues, cust_quality = _prepare_sanctions_customer_df(customer_raw)

        sanc_df, sanc_issues, sanc_quality = _prepare_sanctions_list_df(sanctions_file)
        if sanc_df is None:
            default_sanctions = [
                "John Doe", "Jane Smith", "Muhammad Ali", "Vladimir Putin",
                "Kim Jong Un", "Alexander Petrov", "Maria Gonzalez"
            ]
            sanc_df = pd.DataFrame({"name": [_norm_name(x) for x in default_sanctions]})

        dq_issues = pd.concat([cust_issues, sanc_issues], ignore_index=True) if not cust_issues.empty or not sanc_issues.empty else pd.DataFrame()

        if customers is None:
            msg = f"{cust_quality} | {sanc_quality}"
            return pd.DataFrame(), msg, "", dq_issues

        # Exact matches
        exact = customers[customers["name"].isin(set(sanc_df["name"]))].copy()
        if not exact.empty:
            exact["match_type"] = "Exact Match"

        # Simple fuzzy: shared >=2 tokens
        sanc_tokens = [set(n.lower().split()) for n in sanc_df["name"]]
        fuzzy_idx = []
        for idx, row in customers.iterrows():
            tokens = set(str(row["name"]).lower().split())
            for st in sanc_tokens:
                if len(tokens & st) >= 2:
                    fuzzy_idx.append(idx); break
        fuzzy = customers.loc[sorted(set(fuzzy_idx))].copy() if fuzzy_idx else pd.DataFrame()
        if not fuzzy.empty:
            fuzzy["match_type"] = "Fuzzy Match"

        flagged = pd.concat([exact, fuzzy]).drop_duplicates()

        prompt = f"""You are a compliance officer conducting sanctions screening.
Customers screened: {len(customers)}
Potential matches: {len(flagged)}
Customer sample:
{customers.head(5).to_string()}
Data Quality: {cust_quality} | {sanc_quality}
Assess risk and recommend EDD steps for any matches."""
        summary = generate_summary(prompt)

        stats = f"{cust_quality} | {sanc_quality}\nFound {len(flagged)} potential matches out of {len(customers)} customers."
        return flagged, summary, stats, dq_issues

    except Exception as e:
        return pd.DataFrame(), f"Error processing sanctions check: {str(e)}", "", pd.DataFrame()


# -------------------------
# 4. Credit Risk Analysis (Enhanced + Validation)
# -------------------------

def process_credit_file(file):
    """Process credit data for risk assessment"""
    try:
        df_raw = pd.read_csv(file.name)
        df, dq_issues, quality = _prepare_credit_df(df_raw)
        if df is None:
            return pd.DataFrame(), quality, "", dq_issues

        masks = []
        if 'credit_score' in df.columns:
            masks.append(df['credit_score'] < 600)
        if 'utilization_rate' in df.columns:
            masks.append(df['utilization_rate'] > 0.8)
        if 'debt_to_income' in df.columns:
            masks.append(df['debt_to_income'] > 0.4)
        if 'recent_defaults' in df.columns:
            masks.append(df['recent_defaults'] > 0)
        if 'income' in df.columns:
            masks.append(df['income'] < 30000)

        if masks:
            risk_mask = pd.concat(masks, axis=1).any(axis=1)
            risky_customers = df[risk_mask].copy()
            risky_customers['risk_score'] = 0
            for m in masks:
                risky_customers.loc[m, 'risk_score'] += 1
            risky_customers['risk_level'] = risky_customers['risk_score'].apply(
                lambda x: 'High' if x >= 3 else 'Medium' if x >= 2 else 'Low'
            )
        else:
            risky_customers = pd.DataFrame()

        prompt = f"""You are a credit risk analyst. Assess these customer credit profiles:
Total Customers: {len(df)}
High-Risk Customers: {len(risky_customers)}
Risk Distribution:
{risky_customers['risk_level'].value_counts().to_string() if not risky_customers.empty else 'No high-risk customers identified'}
Data Quality: {quality}
Provide risk assessment insights and recommend credit policies or monitoring actions."""
        summary = generate_summary(prompt)

        stats = f"{quality}\nIdentified {len(risky_customers)} high-risk customers out of {len(df)} usable rows."
        return risky_customers, summary, stats, dq_issues

    except Exception as e:
        return pd.DataFrame(), f"Error processing credit risk file: {str(e)}", "", pd.DataFrame()


# -------------------------
# 5. Chatbot (Enhanced)
# -------------------------

def chatbot_respond(message, history, model_choice):
    """Enhanced chatbot for fraud and risk analysis queries"""
    conversation = ""
    for msg, response in history:
        conversation += f"User: {msg}\nAssistant: {response}\n\n"

    prompt = f"""You are an expert fraud analyst and risk management consultant. Help users with:
- Transaction fraud detection
- KYC/Identity verification
- Sanctions screening
- Credit risk assessment
- Regulatory compliance
- Financial crime prevention
Previous conversation:
{conversation}
User: {message}
Assistant:"""

    try:
        response = generate_summary(prompt, model_id=model_choice)
        history.append((message, response))
        return history, ""
    except Exception as e:
        error_response = f"I apologize, but I encountered an error: {str(e)}"
        history.append((message, error_response))
        return history, ""


# -------------------------
# Navigation Functions (Updated)
# -------------------------

def update_section(section):
    """Updated navigation function to properly show/hide sections"""
    # Set all sections to hidden first
    visibility = {
        "overview": False, 
        "transaction": False, 
        "kyc": False, 
        "sanctions": False, 
        "credit": False, 
        "ai": False
    }
    
    # Show only the requested section
    visibility[section] = True
    
    return (
        gr.update(visible=visibility["overview"]),
        gr.update(visible=visibility["transaction"]),
        gr.update(visible=visibility["kyc"]),
        gr.update(visible=visibility["sanctions"]),
        gr.update(visible=visibility["credit"]),
        gr.update(visible=visibility["ai"]),
        gr.update(visible=(section != "overview"))  # Show back button only when not on overview
    )


# -------------------------
# Gradio Interface
# -------------------------
with gr.Blocks(theme=gr.themes.Soft(), title="🛡️ Fraud Detector Analyst") as demo:

    # Overview/Front Page
    with gr.Column(visible=True) as overview_section:
        gr.Markdown(
            """
        # 🛡️ Fraud Detector Analyst
        ## Multi-Module Risk Intelligence Platform

        Welcome to the comprehensive fraud detection and risk management platform powered by AI. 
        Choose a module below to get started with your risk analysis.
        """
        )

        with gr.Row():
            with gr.Column():
                gr.Markdown("### 🔍 Detection Modules")

                transaction_btn = gr.Button(
                    "📊 Transaction Fraud Detection",
                    size="lg",
                    variant="primary"
                )
                gr.Markdown("*Detect suspicious transaction patterns and anomalies*")

                kyc_btn = gr.Button(
                    "🆔 KYC Identity Fraud Analysis",
                    size="lg",
                    variant="primary"
                )
                gr.Markdown("*Identify duplicate accounts and synthetic identities*")

                sanctions_btn = gr.Button(
                    "🌍 Sanctions & PEP Screening",
                    size="lg",
                    variant="primary"
                )
                gr.Markdown("*Screen customers against global sanctions lists*")

            with gr.Column():
                gr.Markdown("### 🎯 Assessment Tools")

                credit_btn = gr.Button(
                    "💳 Credit Risk Assessment",
                    size="lg",
                    variant="primary"
                )
                gr.Markdown("*Evaluate credit risk and default probability*")

                ai_btn = gr.Button(
                    "💬 AI-Powered Risk Consultant",
                    size="lg",
                    variant="secondary"
                )
                gr.Markdown("*Get expert guidance on fraud and risk management*")

        gr.Markdown(
            """
        ---
        ### 🚀 Quick Start Guide
        1. **Choose a module** from the buttons above based on your analysis needs
        2. **Upload your CSV data** following the format requirements  
        3. **Review AI-powered insights** and flagged records
        4. **Download results** for further investigation

        ### 📋 Supported Data Formats
        - **Transactions**: `customer_id, amount, merchant_category, timestamp`
        - **KYC Records**: `customer_id, name, email, phone, dob, address`
        - **Customer Lists**: `customer_id, name, dob, country`
        - **Credit Profiles**: `customer_id, credit_score, utilization_rate, income`
        """
        )

    # Navigation Back Button (for all modules)
    back_btn = gr.Button("← Back to Main Menu", visible=False, variant="secondary")

    # Transaction Fraud Module
    with gr.Column(visible=False) as transaction_section:
        gr.Markdown("## 📊 Transaction Fraud Detection")
        gr.Markdown("### Upload transaction data to detect fraudulent patterns")
        with gr.Row():
            trans_file = gr.File(
                label="Upload Transaction CSV",
                file_types=[".csv"],
                type="filepath"
            )

        with gr.Row():
            with gr.Column():
                trans_summary = gr.Textbox(
                    label="AI Analysis Summary",
                    lines=8,
                    interactive=False
                )
            with gr.Column():
                trans_stats = gr.Textbox(
                    label="Detection Statistics",
                    lines=3,
                    interactive=False
                )

        trans_results = gr.Dataframe(
            label="Suspicious Transactions",
            interactive=False
        )
        trans_issues = gr.Dataframe(
            label="Data Quality Issues",
            interactive=False
        )

        trans_file.upload(
            process_transaction_file,
            inputs=[trans_file],
            outputs=[trans_results, trans_summary, trans_stats, trans_issues]
        )

    # KYC Fraud Module
    with gr.Column(visible=False) as kyc_section:
        gr.Markdown("## 🆔 KYC Identity Fraud Analysis")
        gr.Markdown("### Detect identity fraud and synthetic accounts in customer onboarding data")
        with gr.Row():
            kyc_file = gr.File(
                label="Upload KYC Customer Data CSV",
                file_types=[".csv"],
                type="filepath"
            )

        with gr.Row():
            with gr.Column():
                kyc_summary = gr.Textbox(
                    label="KYC Fraud Analysis",
                    lines=8,
                    interactive=False
                )
            with gr.Column():
                kyc_stats = gr.Textbox(
                    label="KYC Statistics",
                    lines=3,
                    interactive=False
                )

        kyc_results = gr.Dataframe(
            label="Flagged KYC Records",
            interactive=False
        )
        kyc_issues = gr.Dataframe(
            label="Data Quality Issues",
            interactive=False
        )

        kyc_file.upload(
            process_kyc_file,
            inputs=[kyc_file],
            outputs=[kyc_results, kyc_summary, kyc_stats, kyc_issues]
        )

    # Sanctions Check Module
    with gr.Column(visible=False) as sanctions_section:
        gr.Markdown("## 🌍 Sanctions & PEP Screening")
        gr.Markdown("### Screen customers against sanctions lists and PEP databases")
        with gr.Row():
            sanctions_customer_file = gr.File(
                label="Upload Customer List CSV",
                file_types=[".csv"],
                type="filepath"
            )
            sanctions_list_file = gr.File(
                label="Upload Sanctions List CSV (Optional)",
                file_types=[".csv"],
                type="filepath"
            )

        with gr.Row():
            with gr.Column():
                sanctions_summary = gr.Textbox(
                    label="Sanctions Screening Results",
                    lines=8,
                    interactive=False
                )
            with gr.Column():
                sanctions_stats = gr.Textbox(
                    label="Screening Statistics",
                    lines=3,
                    interactive=False
                )

        sanctions_results = gr.Dataframe(
            label="Flagged Customers",
            interactive=False
        )
        sanctions_issues = gr.Dataframe(
            label="Data Quality Issues",
            interactive=False
        )

        sanctions_customer_file.upload(
            lambda f1, f2: process_sanctions_file(f1, f2),
            inputs=[sanctions_customer_file, sanctions_list_file],
            outputs=[sanctions_results, sanctions_summary, sanctions_stats, sanctions_issues]
        )

    # Credit Risk Module
    with gr.Column(visible=False) as credit_section:
        gr.Markdown("## 💳 Credit Risk Assessment")
        gr.Markdown("### Assess credit risk and default probability for loan applicants")
        with gr.Row():
            credit_file = gr.File(
                label="Upload Credit Profile CSV",
                file_types=[".csv"],
                type="filepath"
            )

        with gr.Row():
            with gr.Column():
                credit_summary = gr.Textbox(
                    label="Credit Risk Analysis",
                    lines=8,
                    interactive=False
                )
            with gr.Column():
                credit_stats = gr.Textbox(
                    label="Risk Statistics",
                    lines=3,
                    interactive=False
                )

        credit_results = gr.Dataframe(
            label="High-Risk Customers",
            interactive=False
        )
        credit_issues = gr.Dataframe(
            label="Data Quality Issues",
            interactive=False
        )

        credit_file.upload(
            process_credit_file,
            inputs=[credit_file],
            outputs=[credit_results, credit_summary, credit_stats, credit_issues]
        )

    # AI Consultant Module
    with gr.Column(visible=False) as ai_section:
        gr.Markdown("## 💬 AI-Powered Risk Consultant")
        gr.Markdown("### Chat with our AI expert about fraud detection and risk management")

        model_choice = gr.Dropdown(
            choices=[DEFAULT_MODEL_ID, SECONDARY_MODEL_ID],
            label="Choose AI Model",
            value=DEFAULT_MODEL_ID,
            info="Select the language model for analysis"
        )

        chatbot = gr.Chatbot(
            label="Risk Management Consultant",
            height=500
        )

        with gr.Row():
            msg = gr.Textbox(
                label="Ask about fraud detection, risk assessment, compliance...",
                placeholder="e.g., How can I improve my transaction fraud detection?",
                scale=4
            )
            submit_btn = gr.Button("Send", scale=1, variant="primary")

        msg.submit(
            chatbot_respond,
            inputs=[msg, chatbot, model_choice],
            outputs=[chatbot, msg]
        )
        submit_btn.click(
            chatbot_respond,
            inputs=[msg, chatbot, model_choice],
            outputs=[chatbot, msg]
        )

    # Button Event Handlers - Fixed to properly navigate between sections
    transaction_btn.click(
        fn=lambda: update_section("transaction"),
        inputs=[],
        outputs=[overview_section, transaction_section, kyc_section, sanctions_section, credit_section, ai_section, back_btn]
    )

    kyc_btn.click(
        fn=lambda: update_section("kyc"),
        inputs=[],
        outputs=[overview_section, transaction_section, kyc_section, sanctions_section, credit_section, ai_section, back_btn]
    )

    sanctions_btn.click(
        fn=lambda: update_section("sanctions"),
        inputs=[],
        outputs=[overview_section, transaction_section, kyc_section, sanctions_section, credit_section, ai_section, back_btn]
    )

    credit_btn.click(
        fn=lambda: update_section("credit"),
        inputs=[],
        outputs=[overview_section, transaction_section, kyc_section, sanctions_section, credit_section, ai_section, back_btn]
    )

    ai_btn.click(
        fn=lambda: update_section("ai"),
        inputs=[],
        outputs=[overview_section, transaction_section, kyc_section, sanctions_section, credit_section, ai_section, back_btn]
    )

    # Back button handler - Returns to overview page
    back_btn.click(
        fn=lambda: update_section("overview"),
        inputs=[],
        outputs=[overview_section, transaction_section, kyc_section, sanctions_section, credit_section, ai_section, back_btn]
    )

    # Footer (visible on all pages)
    with gr.Row():
        gr.Markdown(
            """
        ---
        **⚠️ Disclaimer:** This tool is for demonstration purposes. Always validate results with domain experts and comply with relevant regulations.
        """
        )

if __name__ == "__main__":
    demo.launch(
        share=True,
        server_name="0.0.0.0",
        server_port=7860
    )