app.py

import torch
import numpy as np
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch.nn.functional as F
import spacy
from typing import List, Dict, Tuple
import logging
import os
import gradio as gr
from fastapi.middleware.cors import CORSMiddleware
from concurrent.futures import ThreadPoolExecutor
from functools import partial
import time
from datetime import datetime
import openpyxl
from openpyxl import Workbook
from openpyxl.utils import get_column_letter
from io import BytesIO
import base64
import hashlib

# Configure logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

# Constants
MAX_LENGTH = 512
MODEL_NAME = "microsoft/deberta-v3-small"
WINDOW_SIZE = 6
WINDOW_OVERLAP = 2
CONFIDENCE_THRESHOLD = 0.65
BATCH_SIZE = 8  # Reduced batch size for CPU
MAX_WORKERS = 4  # Number of worker threads for processing

# Get password hash from environment variable (more secure)
ADMIN_PASSWORD_HASH = os.environ.get('ADMIN_PASSWORD_HASH')

if not ADMIN_PASSWORD_HASH:
    ADMIN_PASSWORD_HASH = "5e22d1ed71b273b1b2b5331f2d3e0f6cf34595236f201c6924d6bc81de27cdcb"

# Excel file path for logs
EXCEL_LOG_PATH = "/tmp/prediction_logs.xlsx"

def is_admin_password(input_text: str) -> bool:
    """
    Check if the input text matches the admin password using secure hash comparison.
    This prevents the password from being visible in the source code.
    """
    # Hash the input text
    input_hash = hashlib.sha256(input_text.strip().encode()).hexdigest()
    
    # Compare hashes (constant-time comparison to prevent timing attacks)
    return input_hash == ADMIN_PASSWORD_HASH

class TextWindowProcessor:
    def __init__(self):
        try:
            self.nlp = spacy.load("en_core_web_sm")
        except OSError:
            logger.info("Downloading spacy model...")
            spacy.cli.download("en_core_web_sm")
            self.nlp = spacy.load("en_core_web_sm")

        if 'sentencizer' not in self.nlp.pipe_names:
            self.nlp.add_pipe('sentencizer')

        disabled_pipes = [pipe for pipe in self.nlp.pipe_names if pipe != 'sentencizer']
        self.nlp.disable_pipes(*disabled_pipes)
        
        # Initialize thread pool for parallel processing
        self.executor = ThreadPoolExecutor(max_workers=MAX_WORKERS)

    def split_into_sentences(self, text: str) -> List[str]:
        doc = self.nlp(text)
        return [str(sent).strip() for sent in doc.sents]

    def create_windows(self, sentences: List[str], window_size: int, overlap: int) -> List[str]:
        if len(sentences) < window_size:
            return [" ".join(sentences)]

        windows = []
        stride = window_size - overlap
        for i in range(0, len(sentences) - window_size + 1, stride):
            window = sentences[i:i + window_size]
            windows.append(" ".join(window))
        return windows

    def create_centered_windows(self, sentences: List[str], window_size: int) -> Tuple[List[str], List[List[int]]]:
        """Create windows with better boundary handling"""
        windows = []
        window_sentence_indices = []

        for i in range(len(sentences)):
            # Calculate window boundaries centered on current sentence
            half_window = window_size // 2
            start_idx = max(0, i - half_window)
            end_idx = min(len(sentences), i + half_window + 1)

            # Create the window
            window = sentences[start_idx:end_idx]
            windows.append(" ".join(window))
            window_sentence_indices.append(list(range(start_idx, end_idx)))

        return windows, window_sentence_indices

class TextClassifier:
    def __init__(self):
        # Set thread configuration before any model loading or parallel work
        if not torch.cuda.is_available():
            torch.set_num_threads(MAX_WORKERS)
            torch.set_num_interop_threads(MAX_WORKERS)
            
        self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
        self.model_name = MODEL_NAME
        self.tokenizer = None
        self.model = None
        self.processor = TextWindowProcessor()
        self.initialize_model()

    def initialize_model(self):
        """Initialize the model and tokenizer."""
        logger.info("Initializing model and tokenizer...")
        
        from transformers import DebertaV2TokenizerFast
        
        self.tokenizer = DebertaV2TokenizerFast.from_pretrained(
            self.model_name,
            model_max_length=MAX_LENGTH,
            use_fast=True
        )
        
        self.model = AutoModelForSequenceClassification.from_pretrained(
            self.model_name,
            num_labels=2
        ).to(self.device)
            
        model_path = "model_20250209_184929_acc1.0000.pt"
        if os.path.exists(model_path):
            logger.info(f"Loading custom model from {model_path}")
            checkpoint = torch.load(model_path, map_location=self.device)
            self.model.load_state_dict(checkpoint['model_state_dict'])
        else:
            logger.warning("Custom model file not found. Using base model.")
            
        self.model.eval()

    def quick_scan(self, text: str) -> Dict:
        """Perform a quick scan using simple window analysis."""
        if not text.strip():
            return {
                'prediction': 'unknown',
                'confidence': 0.0,
                'num_windows': 0
            }

        sentences = self.processor.split_into_sentences(text)
        windows = self.processor.create_windows(sentences, WINDOW_SIZE, WINDOW_OVERLAP)

        predictions = []
        
        # Process windows in smaller batches for CPU efficiency
        for i in range(0, len(windows), BATCH_SIZE):
            batch_windows = windows[i:i + BATCH_SIZE]

            inputs = self.tokenizer(
                batch_windows,
                truncation=True,
                padding=True,
                max_length=MAX_LENGTH,
                return_tensors="pt"
            ).to(self.device)

            with torch.no_grad():
                outputs = self.model(**inputs)
                probs = F.softmax(outputs.logits, dim=-1)

                for idx, window in enumerate(batch_windows):
                    prediction = {
                        'window': window,
                        'human_prob': probs[idx][1].item(),
                        'ai_prob': probs[idx][0].item(),
                        'prediction': 'human' if probs[idx][1] > probs[idx][0] else 'ai'
                    }
                    predictions.append(prediction)

            # Clean up GPU memory if available
            del inputs, outputs, probs
            if torch.cuda.is_available():
                torch.cuda.empty_cache()

        if not predictions:
            return {
                'prediction': 'unknown',
                'confidence': 0.0,
                'num_windows': 0
            }

        avg_human_prob = sum(p['human_prob'] for p in predictions) / len(predictions)
        avg_ai_prob = sum(p['ai_prob'] for p in predictions) / len(predictions)

        return {
            'prediction': 'human' if avg_human_prob > avg_ai_prob else 'ai',
            'confidence': max(avg_human_prob, avg_ai_prob),
            'num_windows': len(predictions)
        }

    def detailed_scan(self, text: str) -> Dict:
        """Perform a detailed scan with improved sentence-level analysis."""
        # Clean up trailing whitespace
        text = text.rstrip()
        
        if not text.strip():
            return {
                'sentence_predictions': [],
                'highlighted_text': '',
                'full_text': '',
                'overall_prediction': {
                    'prediction': 'unknown',
                    'confidence': 0.0,
                    'num_sentences': 0
                }
            }

        sentences = self.processor.split_into_sentences(text)
        if not sentences:
            return {}

        # Create centered windows for each sentence
        windows, window_sentence_indices = self.processor.create_centered_windows(sentences, WINDOW_SIZE)

        # Track scores for each sentence
        sentence_appearances = {i: 0 for i in range(len(sentences))}
        sentence_scores = {i: {'human_prob': 0.0, 'ai_prob': 0.0} for i in range(len(sentences))}

        # Process windows in batches
        for i in range(0, len(windows), BATCH_SIZE):
            batch_windows = windows[i:i + BATCH_SIZE]
            batch_indices = window_sentence_indices[i:i + BATCH_SIZE]

            inputs = self.tokenizer(
                batch_windows,
                truncation=True,
                padding=True,
                max_length=MAX_LENGTH,
                return_tensors="pt"
            ).to(self.device)

            with torch.no_grad():
                outputs = self.model(**inputs)
                probs = F.softmax(outputs.logits, dim=-1)

                # Attribute predictions with weighted scoring
                for window_idx, indices in enumerate(batch_indices):
                    center_idx = len(indices) // 2
                    center_weight = 0.7  # Higher weight for center sentence
                    edge_weight = 0.3 / (len(indices) - 1)  # Distribute remaining weight

                    for pos, sent_idx in enumerate(indices):
                        # Apply higher weight to center sentence
                        weight = center_weight if pos == center_idx else edge_weight
                        sentence_appearances[sent_idx] += weight
                        sentence_scores[sent_idx]['human_prob'] += weight * probs[window_idx][1].item()
                        sentence_scores[sent_idx]['ai_prob'] += weight * probs[window_idx][0].item()

            # Clean up memory
            del inputs, outputs, probs
            if torch.cuda.is_available():
                torch.cuda.empty_cache()

        # Calculate final predictions with boundary smoothing
        sentence_predictions = []
        for i in range(len(sentences)):
            if sentence_appearances[i] > 0:
                human_prob = sentence_scores[i]['human_prob'] / sentence_appearances[i]
                ai_prob = sentence_scores[i]['ai_prob'] / sentence_appearances[i]

                # Apply minimal smoothing at prediction boundaries
                if i > 0 and i < len(sentences) - 1:
                    prev_human = sentence_scores[i-1]['human_prob'] / sentence_appearances[i-1]
                    prev_ai = sentence_scores[i-1]['ai_prob'] / sentence_appearances[i-1]
                    next_human = sentence_scores[i+1]['human_prob'] / sentence_appearances[i+1]
                    next_ai = sentence_scores[i+1]['ai_prob'] / sentence_appearances[i+1]

                    # Check if we're at a prediction boundary
                    current_pred = 'human' if human_prob > ai_prob else 'ai'
                    prev_pred = 'human' if prev_human > prev_ai else 'ai'
                    next_pred = 'human' if next_human > next_ai else 'ai'

                    if current_pred != prev_pred or current_pred != next_pred:
                        # Small adjustment at boundaries
                        smooth_factor = 0.1
                        human_prob = (human_prob * (1 - smooth_factor) + 
                                    (prev_human + next_human) * smooth_factor / 2)
                        ai_prob = (ai_prob * (1 - smooth_factor) + 
                                (prev_ai + next_ai) * smooth_factor / 2)

                sentence_predictions.append({
                    'sentence': sentences[i],
                    'human_prob': human_prob,
                    'ai_prob': ai_prob,
                    'prediction': 'human' if human_prob > ai_prob else 'ai',
                    'confidence': max(human_prob, ai_prob)
                })

        return {
            'sentence_predictions': sentence_predictions,
            'highlighted_text': self.format_predictions_html(sentence_predictions),
            'full_text': text,
            'overall_prediction': self.aggregate_predictions(sentence_predictions)
        }

    def format_predictions_html(self, sentence_predictions: List[Dict]) -> str:
        """Format predictions as HTML with color-coding."""
        html_parts = []
        
        for pred in sentence_predictions:
            sentence = pred['sentence']
            confidence = pred['confidence']
            
            if confidence >= CONFIDENCE_THRESHOLD:
                if pred['prediction'] == 'human':
                    color = "#90EE90"  # Light green
                else:
                    color = "#FFB6C6"  # Light red
            else:
                if pred['prediction'] == 'human':
                    color = "#E8F5E9"  # Very light green
                else:
                    color = "#FFEBEE"  # Very light red
                    
            html_parts.append(f'<span style="background-color: {color};">{sentence}</span>')
            
        return " ".join(html_parts)

    def aggregate_predictions(self, predictions: List[Dict]) -> Dict:
        """Aggregate predictions from multiple sentences into a single prediction."""
        if not predictions:
            return {
                'prediction': 'unknown',
                'confidence': 0.0,
                'num_sentences': 0
            }

        total_human_prob = sum(p['human_prob'] for p in predictions)
        total_ai_prob = sum(p['ai_prob'] for p in predictions)
        num_sentences = len(predictions)

        avg_human_prob = total_human_prob / num_sentences
        avg_ai_prob = total_ai_prob / num_sentences

        return {
            'prediction': 'human' if avg_human_prob > avg_ai_prob else 'ai',
            'confidence': max(avg_human_prob, avg_ai_prob),
            'num_sentences': num_sentences
        }

def initialize_excel_log():
    """Initialize the Excel log file if it doesn't exist."""
    if not os.path.exists(EXCEL_LOG_PATH):
        wb = Workbook()
        ws = wb.active
        ws.title = "Prediction Logs"
        
        # Set column headers
        headers = ["timestamp", "word_count", "prediction", "confidence", 
                   "execution_time_ms", "analysis_mode", "full_text"]
        
        for col_num, header in enumerate(headers, 1):
            ws.cell(row=1, column=col_num, value=header)
        
        # Adjust column widths for better readability
        ws.column_dimensions[get_column_letter(1)].width = 20  # timestamp
        ws.column_dimensions[get_column_letter(2)].width = 10  # word_count
        ws.column_dimensions[get_column_letter(3)].width = 10  # prediction
        ws.column_dimensions[get_column_letter(4)].width = 10  # confidence
        ws.column_dimensions[get_column_letter(5)].width = 15  # execution_time_ms
        ws.column_dimensions[get_column_letter(6)].width = 15  # analysis_mode
        ws.column_dimensions[get_column_letter(7)].width = 100  # full_text
            
        # Save the workbook
        wb.save(EXCEL_LOG_PATH)
        logger.info(f"Initialized Excel log file at {EXCEL_LOG_PATH}")

def log_prediction_data(input_text, word_count, prediction, confidence, execution_time, mode):
    """Log prediction data to an Excel file in the /tmp directory."""
    # Initialize the Excel file if it doesn't exist
    if not os.path.exists(EXCEL_LOG_PATH):
        initialize_excel_log()
    
    try:
        # Load the existing workbook
        wb = openpyxl.load_workbook(EXCEL_LOG_PATH)
        ws = wb.active
        
        # Get the next row number
        next_row = ws.max_row + 1
        
        # Clean up the input text for Excel storage (replace problematic characters)
        cleaned_text = input_text.replace("\n", " ")
        
        # Prepare row data
        timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
        row_data = [
            timestamp, 
            word_count, 
            prediction, 
            f"{confidence:.2f}", 
            f"{execution_time:.2f}", 
            mode, 
            cleaned_text
        ]
        
        # Add the data to the worksheet
        for col_num, value in enumerate(row_data, 1):
            ws.cell(row=next_row, column=col_num, value=value)
        
        # Save the workbook
        wb.save(EXCEL_LOG_PATH)
        logger.info(f"Successfully logged prediction data to {EXCEL_LOG_PATH}")
        return True
        
    except Exception as e:
        logger.error(f"Error logging prediction data to Excel: {str(e)}")
        return False

def get_logs_as_base64():
    """Read the Excel logs file and return as base64 for downloading."""
    if not os.path.exists(EXCEL_LOG_PATH):
        return None
    
    try:
        # Read the Excel file into memory
        with open(EXCEL_LOG_PATH, "rb") as f:
            file_data = f.read()
        
        # Encode the file as base64
        base64_data = base64.b64encode(file_data).decode('utf-8')
        return base64_data
        
    except Exception as e:
        logger.error(f"Error reading Excel logs: {str(e)}")
        return None

def analyze_text(text: str, mode: str, classifier: TextClassifier) -> tuple:
    """Analyze text using specified mode and return formatted results."""
    # Check if the input text matches the admin password using secure comparison
    if is_admin_password(text.strip()):
        # Return logs instead of analysis
        base64_data = get_logs_as_base64()
        logs_timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
        logs_filename = f"prediction_logs_{logs_timestamp}.xlsx"
        
        if base64_data:
            # Create downloadable HTML with the logs
            html_content = f"""
            <div style="background-color: #e6f7ff; padding: 15px; border-radius: 5px;">
                <h3>Admin Access Granted - Prediction Logs</h3>
                <p>Logs retrieved at: {datetime.now().strftime("%Y-%m-%d %H:%M:%S")}</p>
                <p>Excel file contains all prediction data with full text of all submissions.</p>
                <a href="data:application/vnd.openxmlformats-officedocument.spreadsheetml.sheet;base64,{base64_data}" 
                   download="{logs_filename}" 
                   style="display: inline-block; margin-top: 10px; padding: 10px 15px; 
                          background-color: #4CAF50; color: white; text-decoration: none; 
                          border-radius: 4px;">
                    Download Excel Logs
                </a>
            </div>
            """
        else:
            html_content = """
            <div style="background-color: #ffe6e6; padding: 15px; border-radius: 5px;">
                <h3>Admin Access Granted - No Logs Found</h3>
                <p>No prediction logs were found or there was an error reading the logs file.</p>
            </div>
            """
        
        # Return special admin output instead of normal analysis
        return (
            html_content,
            f"Admin access granted. Logs retrieved at {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}",
            f"ADMIN MODE\nLogs available for download\nFile: {EXCEL_LOG_PATH}"
        )
        
    # Start timing for normal analysis
    start_time = time.time()
    
    # Count words in the text
    word_count = len(text.split())
    
    # If text is less than 200 words and detailed mode is selected, switch to quick mode
    original_mode = mode
    if word_count < 200 and mode == "detailed":
        mode = "quick"
    
    if mode == "quick":
        result = classifier.quick_scan(text)
        
        quick_analysis = f"""
        PREDICTION: {result['prediction'].upper()}
        Confidence: {result['confidence']*100:.1f}%
        Windows analyzed: {result['num_windows']}
        """
        
        # Add note if mode was switched
        if original_mode == "detailed":
            quick_analysis += f"\n\nNote: Switched to quick mode because text contains only {word_count} words. Minimum 200 words required for detailed analysis."
        
        # Calculate execution time in milliseconds
        execution_time = (time.time() - start_time) * 1000
        
        # Log the prediction data
        log_prediction_data(
            input_text=text, 
            word_count=word_count, 
            prediction=result['prediction'], 
            confidence=result['confidence'],
            execution_time=execution_time,
            mode=original_mode
        )
        
        return (
            text,  # No highlighting in quick mode
            "Quick scan mode - no sentence-level analysis available",
            quick_analysis
        )
    else:
        analysis = classifier.detailed_scan(text)
        
        detailed_analysis = []
        for pred in analysis['sentence_predictions']:
            confidence = pred['confidence'] * 100
            detailed_analysis.append(f"Sentence: {pred['sentence']}")
            detailed_analysis.append(f"Prediction: {pred['prediction'].upper()}")
            detailed_analysis.append(f"Confidence: {confidence:.1f}%")
            detailed_analysis.append("-" * 50)
        
        final_pred = analysis['overall_prediction']
        overall_result = f"""
        FINAL PREDICTION: {final_pred['prediction'].upper()}
        Overall confidence: {final_pred['confidence']*100:.1f}%
        Number of sentences analyzed: {final_pred['num_sentences']}
        """
        
        # Calculate execution time in milliseconds
        execution_time = (time.time() - start_time) * 1000
        
        # Log the prediction data
        log_prediction_data(
            input_text=text, 
            word_count=word_count, 
            prediction=final_pred['prediction'],
            confidence=final_pred['confidence'],
            execution_time=execution_time,
            mode=original_mode
        )
        
        return (
            analysis['highlighted_text'],
            "\n".join(detailed_analysis),
            overall_result
        )

# Initialize the classifier globally
classifier = TextClassifier()

# Create Gradio interface
demo = gr.Interface(
    fn=lambda text, mode: analyze_text(text, mode, classifier),
    inputs=[
        gr.Textbox(
            lines=8,
            placeholder="Enter text to analyze...",
            label="Input Text"
        ),
        gr.Radio(
            choices=["quick", "detailed"],
            value="quick",
            label="Analysis Mode",
            info="Quick mode for faster analysis, Detailed mode for sentence-level analysis"
        )
    ],
    outputs=[
        gr.HTML(label="Highlighted Analysis"),
        gr.Textbox(label="Sentence-by-Sentence Analysis", lines=10),
        gr.Textbox(label="Overall Result", lines=4)
    ],
    title="AI Text Detector",
    description="Analyze text to detect if it was written by a human or AI. Choose between quick scan and detailed sentence-level analysis. 200+ words suggested for accurate predictions.",
    api_name="predict",
    flagging_mode="never"
)

# Get the FastAPI app from Gradio
app = demo.app

# Add CORS middleware
app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],  # For development
    allow_credentials=True,
    allow_methods=["GET", "POST", "OPTIONS"],
    allow_headers=["*"],
)

# Ensure CORS is applied before launching
if __name__ == "__main__":
    demo.queue()
    demo.launch(
        server_name="0.0.0.0",
        server_port=7860,
        share=True
    )