supervertaler 1.9.163__py3-none-any.whl

modules/pdf_rescue_Qt.py

@@ -0,0 +1,1822 @@
+"""
+PDF Rescue Module - Qt Edition
+Embeddable version of the AI-powered OCR tool for extracting text from poorly formatted PDFs
+Supports multiple AI providers: OpenAI GPT-4 Vision, Anthropic Claude Vision, Google Gemini Vision
+
+This module can be embedded in the main Supervertaler Qt application as a tab.
+Can also be run independently as a standalone application.
+"""
+
+import os
+import base64
+from pathlib import Path
+from PyQt6.QtWidgets import (
+    QWidget, QVBoxLayout, QHBoxLayout, QLabel, QPushButton, QListWidget, QTextEdit,
+    QComboBox, QCheckBox, QProgressBar, QFileDialog, QMessageBox, QSplitter,
+    QGroupBox, QFrame, QDialog, QDialogButtonBox, QPlainTextEdit, QApplication,
+    QStyleOptionButton
+)
+from PyQt6.QtCore import Qt, QTimer, QPointF, QRect
+from PyQt6.QtGui import QFont, QTextOption, QPainter, QPen, QColor, QStandardItemModel, QStandardItem
+from docx import Document
+from docx.shared import Pt
+import fitz  # PyMuPDF
+import re
+
+
+class CheckmarkCheckBox(QCheckBox):
+    """Custom checkbox with green background and white checkmark when checked - matches Supervertaler Qt style"""
+    
+    def __init__(self, text="", parent=None):
+        super().__init__(text, parent)
+        self.setStyleSheet("""
+            QCheckBox {
+                font-size: 9pt;
+                spacing: 6px;
+            }
+            QCheckBox::indicator {
+                width: 18px;
+                height: 18px;
+                border: 2px solid #999;
+                border-radius: 3px;
+                background-color: white;
+            }
+            QCheckBox::indicator:checked {
+                background-color: #4CAF50;
+                border-color: #4CAF50;
+            }
+            QCheckBox::indicator:hover {
+                border-color: #666;
+            }
+            QCheckBox::indicator:checked:hover {
+                background-color: #45a049;
+                border-color: #45a049;
+            }
+        """)
+    
+    def paintEvent(self, a0):
+        """Override paint event to draw white checkmark when checked"""
+        super().paintEvent(a0)
+        
+        if self.isChecked():
+            # Get the indicator rectangle using QStyle
+            opt = QStyleOptionButton()
+            self.initStyleOption(opt)
+            indicator_rect = self.style().subElementRect(
+                self.style().SubElement.SE_CheckBoxIndicator,
+                opt,
+                self
+            )
+            
+            if indicator_rect.isValid():
+                # Draw white checkmark
+                painter = QPainter(self)
+                painter.setRenderHint(QPainter.RenderHint.Antialiasing)
+                # Slightly thinner pen for better fit on smaller displays
+                pen_width = max(2.0, min(indicator_rect.width(), indicator_rect.height()) * 0.12)
+                painter.setPen(QPen(QColor(255, 255, 255), pen_width, Qt.PenStyle.SolidLine, Qt.PenCapStyle.RoundCap, Qt.PenJoinStyle.RoundJoin))
+                painter.setBrush(QColor(255, 255, 255))
+                
+                # Draw checkmark (✓ shape) - coordinates relative to indicator
+                # Add padding to prevent clipping on smaller displays
+                x = indicator_rect.x()
+                y = indicator_rect.y()
+                w = indicator_rect.width()
+                h = indicator_rect.height()
+                
+                # Add padding (15% on all sides) to ensure checkmark doesn't get cut off on smaller displays
+                padding = min(w, h) * 0.15
+                x += padding
+                y += padding
+                w -= padding * 2
+                h -= padding * 2
+                
+                # Checkmark path: bottom-left to middle, then middle to top-right
+                # Using proportions that create a nice checkmark shape with proper padding
+                check_x1 = x + w * 0.10  # Left point (more padding from left)
+                check_y1 = y + h * 0.50
+                check_x2 = x + w * 0.35  # Middle-bottom point
+                check_y2 = y + h * 0.70  # Bottom point (with padding from bottom)
+                check_x3 = x + w * 0.90  # Right point (more padding from right)
+                check_y3 = y + h * 0.25  # Top point (with padding from top)
+                
+                # Draw two lines forming the checkmark
+                painter.drawLine(QPointF(check_x2, check_y2), QPointF(check_x3, check_y3))
+                painter.drawLine(QPointF(check_x1, check_y1), QPointF(check_x2, check_y2))
+                
+                painter.end()
+
+
+class PDFRescueQt:
+    """
+    PDF Rescue feature - extract text from images using AI OCR
+    Can be embedded in any PyQt6 application as a tab or panel
+    """
+    
+    def __init__(self, parent_app, standalone=False):
+        """
+        Initialize PDF Rescue module
+        
+        Args:
+            parent_app: Reference to the main application (needs .load_api_keys() method or .api_keys attribute)
+            standalone: If True, running as standalone app. If False, embedded in Supervertaler
+        """
+        self.parent_app = parent_app
+        self.standalone = standalone
+        self.clients = {}  # Dictionary to store clients for different providers
+        self.image_files = []
+        self.extracted_texts = {}
+        
+        # Load API keys for all providers
+        self.api_keys = {}
+        if hasattr(parent_app, 'load_api_keys'):
+            # Supervertaler_Qt style
+            self.api_keys = parent_app.load_api_keys()
+        elif hasattr(parent_app, 'api_keys'):
+            # Direct api_keys dict
+            self.api_keys = parent_app.api_keys
+        
+        # Initialize clients for available providers
+        self._initialize_clients()
+    
+    def _initialize_clients(self):
+        """Initialize API clients for all available providers"""
+        # OpenAI
+        openai_key = self.api_keys.get('openai') or self.api_keys.get('openai_api_key')
+        if openai_key:
+            try:
+                from openai import OpenAI
+                self.clients['openai'] = OpenAI(api_key=openai_key)
+                self.log_message("✓ OpenAI client initialized")
+            except Exception as e:
+                self.log_message(f"⚠ Failed to initialize OpenAI: {e}")
+        
+        # Anthropic Claude
+        claude_key = self.api_keys.get('claude') or self.api_keys.get('anthropic')
+        if claude_key:
+            try:
+                import anthropic
+                self.clients['claude'] = anthropic.Anthropic(api_key=claude_key)
+                self.log_message("✓ Claude client initialized")
+            except ImportError:
+                self.log_message("⚠ Claude requested but 'anthropic' library not installed. Run: pip install anthropic")
+            except Exception as e:
+                self.log_message(f"⚠ Failed to initialize Claude: {e}")
+        
+        # Google Gemini
+        gemini_key = self.api_keys.get('gemini') or self.api_keys.get('google')
+        if gemini_key:
+            try:
+                import google.generativeai as genai
+                genai.configure(api_key=gemini_key)
+                self.clients['gemini'] = genai
+                self.log_message("✓ Gemini client initialized")
+            except ImportError:
+                self.log_message("⚠ Gemini requested but 'google-generativeai' library not installed. Run: pip install google-generativeai")
+            except Exception as e:
+                self.log_message(f"⚠ Failed to initialize Gemini: {e}")
+    
+    def log_message(self, message: str):
+        """Log a message to the parent app's log if available"""
+        if hasattr(self.parent_app, 'log'):
+            self.parent_app.log(f"[PDF Rescue] {message}")
+        else:
+            print(f"[PDF Rescue] {message}")
+    
+    def create_tab(self, parent):
+        """
+        Create the PDF Rescue tab UI
+        
+        Args:
+            parent: The parent widget (QWidget)
+        """
+        # Save current state before recreating UI
+        saved_files = self.image_files.copy() if hasattr(self, 'image_files') else []
+        saved_texts = self.extracted_texts.copy() if hasattr(self, 'extracted_texts') else []
+        
+        # Main layout
+        main_layout = QVBoxLayout(parent)
+        main_layout.setContentsMargins(10, 10, 10, 10)
+        main_layout.setSpacing(5)  # Reduced from 10 to 5 for tighter spacing
+        
+        # Header (matches Universal Lookup / AutoFingers style)
+        header = QLabel("🔍 PDF Rescue - AI-Powered OCR")
+        header.setStyleSheet("font-size: 16pt; font-weight: bold; color: #1976D2;")
+        main_layout.addWidget(header, 0)  # 0 = no stretch, stays compact
+        
+        # Description box (matches Universal Lookup / AutoFingers style)
+        description = QLabel(
+            "Extract text from image-based PDFs using AI vision OCR. Designed for scanned documents, screenshots, "
+            "and PDFs without accessible text. Not recommended for PDFs with selectable text - use professional tools like Adobe Acrobat instead."
+        )
+        description.setWordWrap(True)
+        description.setStyleSheet("color: #666; padding: 5px; background-color: #E3F2FD; border-radius: 3px;")
+        main_layout.addWidget(description, 0)  # 0 = no stretch, stays compact
+        
+        # Split view: Files on left, Preview on right
+        splitter = QSplitter(Qt.Orientation.Horizontal)
+        
+        # LEFT: File list
+        left_widget = QWidget()
+        left_layout = QVBoxLayout(left_widget)
+        left_layout.setContentsMargins(3, 0, 3, 3)  # Zero top margin to eliminate gap
+        left_layout.setSpacing(3)
+        
+        files_label = QLabel("Images to Process")
+        files_label.setFont(QFont("Segoe UI", 9, QFont.Weight.Bold))
+        left_layout.addWidget(files_label)
+        
+        # File list
+        self.file_listbox = QListWidget()
+        self.file_listbox.setFont(QFont("Consolas", 9))
+        self.file_listbox.itemSelectionChanged.connect(self._on_file_select)
+        left_layout.addWidget(self.file_listbox)
+        
+        # Buttons
+        btn_layout = QHBoxLayout()
+        
+        pdf_btn = QPushButton("📄 Import PDF")
+        pdf_btn.clicked.connect(self._import_from_pdf)
+        pdf_btn.setStyleSheet("background-color: #9C27B0; color: white; font-weight: bold; padding: 4px 8px;")
+        pdf_btn.setToolTip("Extract all pages from a PDF file and convert them to images for OCR processing")
+        btn_layout.addWidget(pdf_btn)
+        
+        add_files_btn = QPushButton("➕ Add Image Files")
+        add_files_btn.clicked.connect(self._add_files)
+        add_files_btn.setStyleSheet("background-color: #2196F3; color: white; font-weight: bold; padding: 4px 8px;")
+        add_files_btn.setToolTip("Supported formats: .jpg, .jpeg, .png, .bmp, .gif, .tiff")
+        btn_layout.addWidget(add_files_btn)
+        
+        add_folder_btn = QPushButton("📂 Folder")
+        add_folder_btn.clicked.connect(self._add_folder)
+        add_folder_btn.setStyleSheet("background-color: #2196F3; color: white; font-weight: bold; padding: 4px 8px;")
+        add_folder_btn.setToolTip("Add all image files from a selected folder")
+        btn_layout.addWidget(add_folder_btn)
+        
+        clear_btn = QPushButton("Clear")
+        clear_btn.clicked.connect(self._clear_list)
+        clear_btn.setStyleSheet("background-color: #9E9E9E; color: white; padding: 4px 8px;")
+        clear_btn.setToolTip("Remove all files from the list")
+        btn_layout.addWidget(clear_btn)
+        
+        left_layout.addLayout(btn_layout)
+        
+        # Processing options (moved into left panel to eliminate wasted space)
+        options_group = QGroupBox("Processing Options")
+        options_layout = QVBoxLayout(options_group)
+        options_layout.setContentsMargins(8, 8, 8, 3)  # Minimal bottom margin
+        options_layout.setSpacing(3)  # Reduced spacing
+        
+        # === MODEL SELECTOR ===
+        # Model selection and formatting option
+        model_layout = QHBoxLayout()
+        
+        model_label = QLabel("AI Model:")
+        model_label.setFont(QFont("Segoe UI", 9))
+        model_layout.addWidget(model_label)
+        
+        self.model_combo = QComboBox()
+        # Organize models by provider with separators
+        self.model_combo.addItem("--- OpenAI ---")
+        self.model_combo.addItems(["gpt-4o", "gpt-4o-mini", "gpt-4-turbo", "gpt-4", "gpt-5"])
+        self.model_combo.addItem("--- Claude (Anthropic) ---")
+        self.model_combo.addItems(["claude-3-5-sonnet-20241022", "claude-3-5-haiku-20241022", "claude-3-opus-20240229"])
+        self.model_combo.addItem("--- Gemini (Google) ---")
+        self.model_combo.addItems(["gemini-2.0-flash-exp", "gemini-1.5-pro-002", "gemini-1.5-flash-002"])
+        
+        # Set default and style separator items
+        self.model_combo.setCurrentText("gpt-4o")
+        
+        # Make separator items non-selectable by disabling them
+        combo_model = self.model_combo.model()
+        if isinstance(combo_model, QStandardItemModel):
+            for i in range(self.model_combo.count()):
+                if self.model_combo.itemText(i).startswith("---"):
+                    item = combo_model.item(i)
+                    if item:
+                        item.setEnabled(False)
+                        # Make separators visually distinct (gray, centered)
+                        item.setFlags(Qt.ItemFlag.NoItemFlags)
+        
+        self.model_combo.setEditable(False)
+        self.model_combo.setToolTip("Select AI model for vision OCR processing (OpenAI, Claude, or Gemini)")
+        model_layout.addWidget(self.model_combo)
+        
+        model_layout.addSpacing(20)
+        
+        self.preserve_formatting_check = CheckmarkCheckBox("Preserve formatting (bold/italic/underline)")
+        self.preserve_formatting_check.setChecked(True)
+        self.preserve_formatting_check.setFont(QFont("Segoe UI", 9))
+        self.preserve_formatting_check.setToolTip("When enabled, the AI will use markdown to preserve text formatting (bold, italic, underline)")
+        model_layout.addWidget(self.preserve_formatting_check)
+        
+        model_layout.addStretch()
+        options_layout.addLayout(model_layout)
+        
+        # Model descriptions (prominent display in left panel)
+        model_desc_label = QLabel("Model Capabilities:")
+        model_desc_label.setFont(QFont("Segoe UI", 9, QFont.Weight.Bold))
+        options_layout.addWidget(model_desc_label)
+        
+        self.model_descriptions_text = QPlainTextEdit()
+        self.model_descriptions_text.setFont(QFont("Segoe UI", 9))
+        self.model_descriptions_text.setReadOnly(True)
+        # Size naturally to content - no fixed max height to eliminate wasted space below
+        model_descriptions = """• gpt-4o (Recommended): Fast and accurate; best balance of speed, quality, and cost; excellent for most documents including tables
+• gpt-4o-mini: Fast and economical; good for simple documents; may struggle with complex layouts or tables
+• gpt-4-turbo: Large context window (128k tokens); good for very long documents; slightly slower but handles extensive content well
+• gpt-4: Classic, reliable baseline; consistent quality; good for standard documents, though slower than gpt-4o
+• gpt-5 (Advanced Reasoning): Reasoning model; may improve table extraction and complex layouts; slower and more expensive; best for: complex tables, technical documents, structured data extraction"""
+        self.model_descriptions_text.setPlainText(model_descriptions)
+        options_layout.addWidget(self.model_descriptions_text)
+        
+        left_layout.addWidget(options_group)
+        
+        # Add stretch to push everything up and eliminate wasted space below
+        left_layout.addStretch()
+        
+        splitter.addWidget(left_widget)
+        
+        # RIGHT: Extraction Instructions and Text preview with vertical splitter
+        right_splitter = QSplitter(Qt.Orientation.Vertical)
+        
+        # Top: Extraction Instructions
+        instructions_widget = QWidget()
+        instructions_widget_layout = QVBoxLayout(instructions_widget)
+        instructions_widget_layout.setContentsMargins(0, 0, 0, 0)
+        instructions_widget_layout.setSpacing(0)
+        
+        instructions_group = QGroupBox("Extraction Instructions")
+        instructions_group_layout = QVBoxLayout(instructions_group)
+        instructions_group_layout.setContentsMargins(8, 8, 8, 8)
+        
+        instructions_header = QHBoxLayout()
+        instructions_header.setContentsMargins(0, 0, 0, 5)
+        instructions_header.addStretch()
+        
+        show_prompt_btn = QPushButton("👁️ Show Prompt")
+        show_prompt_btn.clicked.connect(self._show_full_prompt)
+        show_prompt_btn.setStyleSheet("background-color: #9C27B0; color: white; padding: 2px 8px;")
+        show_prompt_btn.setToolTip("Preview the exact prompt that will be sent to the AI model")
+        instructions_header.addWidget(show_prompt_btn)
+        
+        instructions_group_layout.addLayout(instructions_header)
+        
+        self.instructions_text = QPlainTextEdit()
+        self.instructions_text.setFont(QFont("Segoe UI", 9))
+        self.instructions_text.setMinimumHeight(60)  # Reduced minimum height
+        default_instructions = """Extract all text from this image. The image is a screenshot from a poorly formatted PDF.
+Please:
+- Extract all visible text accurately
+- Fix any obvious OCR errors or formatting issues
+- Remove extraneous line breaks within paragraphs
+- Preserve intentional paragraph breaks
+- Maintain the logical flow and structure of the content
+- For redacted/blacked-out text: insert a descriptive placeholder in square brackets in the document's language (e.g., [naam] for Dutch names, [name] for English names, [bedrag] for amounts, etc.)
+- For stamps, signatures, or images: insert a descriptive placeholder in square brackets in the document's language (e.g., [handtekening], [stempel], [signature], [stamp], etc.)
+- For any non-text elements that would normally appear: describe them briefly in square brackets
+- Use markdown for text formatting: **bold text**, *italic text*, __underlined text__
+- Output clean, readable text only (no commentary)"""
+        self.instructions_text.setPlainText(default_instructions)
+        instructions_group_layout.addWidget(self.instructions_text)
+        
+        instructions_widget_layout.addWidget(instructions_group)
+        right_splitter.addWidget(instructions_widget)
+        
+        # Bottom: Extracted Text Preview
+        preview_widget = QWidget()
+        preview_widget_layout = QVBoxLayout(preview_widget)
+        preview_widget_layout.setContentsMargins(0, 0, 0, 0)
+        preview_widget_layout.setSpacing(0)
+        
+        preview_label = QLabel("Extracted Text Preview")
+        preview_label.setFont(QFont("Segoe UI", 9, QFont.Weight.Bold))
+        preview_label.setContentsMargins(0, 5, 0, 5)
+        preview_widget_layout.addWidget(preview_label)
+        
+        self.preview_text = QTextEdit()
+        self.preview_text.setFont(QFont("Segoe UI", 9))
+        self.preview_text.setReadOnly(False)  # Allow editing
+        self.preview_text.setWordWrapMode(QTextOption.WrapMode.WordWrap)
+        preview_widget_layout.addWidget(self.preview_text)
+        
+        right_splitter.addWidget(preview_widget)
+        
+        # Set initial splitter sizes for right panel (30% instructions, 70% preview)
+        right_splitter.setSizes([150, 350])
+        
+        splitter.addWidget(right_splitter)
+        
+        # Set splitter sizes (1:2 ratio)
+        splitter.setSizes([300, 600])
+        
+        main_layout.addWidget(splitter, 1)  # 1 = stretch factor, expands to fill space
+        
+        # Action buttons
+        action_layout = QHBoxLayout()
+        action_layout.setSpacing(5)
+        
+        process_selected_btn = QPushButton("🔍 Process Selected")
+        process_selected_btn.clicked.connect(self._process_selected)
+        process_selected_btn.setStyleSheet("background-color: #FF9800; color: white; font-weight: bold; padding: 6px 15px;")
+        process_selected_btn.setToolTip("Process the currently selected image with AI OCR to extract text")
+        action_layout.addWidget(process_selected_btn)
+        
+        process_all_btn = QPushButton("⚡ Process ALL")
+        process_all_btn.clicked.connect(self._process_all)
+        process_all_btn.setStyleSheet("background-color: #4CAF50; color: white; font-weight: bold; padding: 6px 15px;")
+        process_all_btn.setToolTip("Process all images in the list with AI OCR. This will use API credits and may take several minutes.")
+        action_layout.addWidget(process_all_btn)
+        
+        save_docx_btn = QPushButton("💾 Export Markdown && Word")
+        save_docx_btn.clicked.connect(self._export_markdown_and_word)
+        save_docx_btn.setStyleSheet("background-color: #2196F3; color: white; font-weight: bold; padding: 6px 15px;")
+        save_docx_btn.setToolTip("Export extracted text as Markdown (.md), Word document (.docx), and session report")
+        action_layout.addWidget(save_docx_btn)
+        
+        copy_all_btn = QPushButton("📋 Copy All")
+        copy_all_btn.clicked.connect(self._copy_all_text)
+        copy_all_btn.setStyleSheet("background-color: #607D8B; color: white; font-weight: bold; padding: 6px 15px;")
+        copy_all_btn.setToolTip("Copy all extracted text from all processed images to clipboard")
+        action_layout.addWidget(copy_all_btn)
+        
+        session_report_btn = QPushButton("📊 Session Report")
+        session_report_btn.clicked.connect(self._save_session_report)
+        session_report_btn.setStyleSheet("background-color: #795548; color: white; font-weight: bold; padding: 6px 15px;")
+        session_report_btn.setToolTip("Generate and save a detailed session report in Markdown format with statistics and extracted text")
+        action_layout.addWidget(session_report_btn)
+        
+        action_layout.addStretch()
+        main_layout.addLayout(action_layout, 0)  # 0 = no stretch, stays compact
+        
+        # Status
+        self.status_label = QLabel("Ready - Add images to begin")
+        self.status_label.setFont(QFont("Segoe UI", 9))
+        self.status_label.setStyleSheet("color: #666;")
+        main_layout.addWidget(self.status_label, 0)  # 0 = no stretch, stays compact
+        
+        # Progress bar
+        self.progress = QProgressBar()
+        self.progress.setMinimum(0)
+        self.progress.setMaximum(100)
+        self.progress.setValue(0)
+        main_layout.addWidget(self.progress, 0)  # 0 = no stretch, stays compact
+        
+        # Restore state after UI creation
+        self.image_files = saved_files
+        self.extracted_texts = saved_texts
+        if self.image_files:
+            self._update_listbox()
+    
+    # === File Management Methods ===
+    
+    def _import_from_pdf(self):
+        """Import images directly from a PDF file (simple OCR-only version)"""
+        pdf_file, _ = QFileDialog.getOpenFileName(
+            parent=None,
+            caption="Select PDF File",
+            filter="PDF files (*.pdf);;All files (*.*)"
+        )
+        
+        if not pdf_file:
+            return
+        
+        try:
+            # Open PDF
+            doc = fitz.open(pdf_file)
+            total_pages = len(doc)
+            
+            if total_pages == 0:
+                QMessageBox.warning(None, "Empty PDF", "The selected PDF has no pages.")
+                return
+            
+            # Create folder for extracted images next to the PDF
+            pdf_path = Path(pdf_file)
+            pdf_name = pdf_path.stem
+            images_folder = pdf_path.parent / f"{pdf_name}_images"
+            
+            # Create folder if it doesn't exist
+            images_folder.mkdir(exist_ok=True)
+            temp_dir = str(images_folder)
+            
+            # Log start
+            self.log_message(f"Starting PDF import: {pdf_path.name}")
+            self.log_message(f"Total pages: {total_pages}")
+            
+            # Extract each page as an image
+            extracted_count = 0
+            self.status_label.setText(f"Extracting pages from PDF...")
+            QApplication.processEvents()
+            
+            for page_num in range(total_pages):
+                page = doc[page_num]
+                
+                # Render page to pixmap (image) at 2x resolution for better quality
+                zoom = 2.0
+                mat = fitz.Matrix(zoom, zoom)
+                pix = page.get_pixmap(matrix=mat)
+                
+                # Save as PNG
+                img_filename = f"{pdf_name}_page_{page_num + 1:03d}.png"
+                img_path = os.path.join(temp_dir, img_filename)
+                pix.save(img_path)
+                
+                # Add to image list
+                if img_path not in self.image_files:
+                    self.image_files.append(img_path)
+                    extracted_count += 1
+                    
+                    # Log each page
+                    self.log_message(f"  Page {page_num + 1}/{total_pages} extracted: {img_filename}")
+                
+                # Update progress
+                self.status_label.setText(
+                    f"Extracting page {page_num + 1}/{total_pages}..."
+                )
+                self.progress.setValue(page_num + 1)
+                self.progress.setMaximum(total_pages)
+                QApplication.processEvents()
+            
+            doc.close()
+            
+            # Update list
+            self._update_listbox()
+            
+            # Log completion
+            self.log_message(f"PDF import complete: {extracted_count} pages extracted to {images_folder}")
+            self.status_label.setText(f"✓ Imported {extracted_count} pages from PDF")
+            
+            QMessageBox.information(
+                None,
+                "PDF Import Complete",
+                f"Successfully extracted {extracted_count} pages from PDF!\n\n"
+                f"Images saved to:\n{images_folder}\n\n"
+                f"You can now process them with AI OCR."
+            )
+            
+        except Exception as e:
+            QMessageBox.critical(None, "PDF Import Error", f"Failed to import PDF:\n\n{str(e)}")
+            self.log_message(f"ERROR importing PDF: {str(e)}")
+            self.status_label.setText("PDF import failed")
+    
+    def _add_files(self):
+        """Add individual image files"""
+        files, _ = QFileDialog.getOpenFileNames(
+            parent=None,
+            caption="Select Image Files",
+            filter="Image files (*.jpg *.jpeg *.png *.bmp *.gif *.tiff);;All files (*.*)"
+        )
+        
+        if files:
+            for file in files:
+                if file not in self.image_files:
+                    self.image_files.append(file)
+            self._update_listbox()
+            self.status_label.setText(f"Added {len(files)} file(s)")
+            self.log_message(f"Added {len(files)} image file(s)")
+    
+    def _add_folder(self):
+        """Add all images from a folder"""
+        folder = QFileDialog.getExistingDirectory(
+            parent=None,
+            caption="Select Folder with Images"
+        )
+        
+        if folder:
+            image_extensions = {'.jpg', '.jpeg', '.png', '.bmp', '.gif', '.tiff'}
+            files = []
+            
+            for file in sorted(os.listdir(folder)):
+                file_path = os.path.join(folder, file)
+                if os.path.isfile(file_path):
+                    ext = os.path.splitext(file)[1].lower()
+                    if ext in image_extensions and file_path not in self.image_files:
+                        files.append(file_path)
+            
+            self.image_files.extend(files)
+            self._update_listbox()
+            self.status_label.setText(f"Added {len(files)} file(s) from folder")
+            self.log_message(f"Added {len(files)} file(s) from folder: {folder}")
+    
+    def _clear_list(self):
+        """Clear all files"""
+        if self.image_files:
+            reply = QMessageBox.question(
+                None,
+                "Clear",
+                "Remove all files?",
+                QMessageBox.StandardButton.Yes | QMessageBox.StandardButton.No
+            )
+            if reply == QMessageBox.StandardButton.Yes:
+                self.image_files = []
+                self.extracted_texts = {}
+                self._update_listbox()
+                self.preview_text.clear()
+                self.status_label.setText("List cleared")
+    
+    def _update_listbox(self):
+        """Update file listbox with checkmarks for processed items"""
+        self.file_listbox.clear()
+        for i, file in enumerate(self.image_files, 1):
+            filename = os.path.basename(file)
+            status = "✓ " if file in self.extracted_texts else ""
+            self.file_listbox.addItem(f"{status}{i:2d}. {filename}")
+    
+    def _on_file_select(self):
+        """Show extracted text when file is selected"""
+        selected_items = self.file_listbox.selectedItems()
+        if not selected_items:
+            return
+        
+        idx = self.file_listbox.row(selected_items[0])
+        if idx < len(self.image_files):
+            file = self.image_files[idx]
+            if file in self.extracted_texts:
+                self.preview_text.setPlainText(self.extracted_texts[file])
+    
+    def _show_full_prompt(self):
+        """Show the exact prompt that will be sent to the AI"""
+        instructions = self.instructions_text.toPlainText().strip()
+        
+        # Apply formatting modifications like in _extract_text_from_image
+        if self.preserve_formatting_check.isChecked():
+            if "markdown for text formatting" not in instructions:
+                instructions += "\n- Use markdown for text formatting: **bold text**, *italic text*, __underlined text__"
+        else:
+            instructions = instructions.replace(
+                "\n- Use markdown for text formatting: **bold text**, *italic text*, __underlined text__", ""
+            ).replace(
+                "- Use markdown for text formatting: **bold text**, *italic text*, __underlined text__", ""
+            )
+        
+        # Create popup dialog
+        popup = QDialog()
+        popup.setWindowTitle("Full Prompt Preview")
+        popup.resize(700, 600)
+        
+        layout = QVBoxLayout(popup)
+        
+        # Title
+        title = QLabel("Exact Prompt Sent to OpenAI API")
+        title.setFont(QFont("Segoe UI", 12, QFont.Weight.Bold))
+        layout.addWidget(title)
+        
+        # Info frame
+        info_group = QGroupBox("Configuration")
+        info_layout = QVBoxLayout(info_group)
+        
+        model_label = QLabel(f"Model: {self.model_combo.currentText()}")
+        model_label.setFont(QFont("Segoe UI", 9, QFont.Weight.Bold))
+        info_layout.addWidget(model_label)
+        
+        formatting_status = "✓ Enabled" if self.preserve_formatting_check.isChecked() else "✗ Disabled"
+        formatting_label = QLabel(f"Formatting Preservation: {formatting_status}")
+        formatting_label.setFont(QFont("Segoe UI", 9))
+        info_layout.addWidget(formatting_label)
+        
+        tokens_label = QLabel("Max Tokens: 4000")
+        tokens_label.setFont(QFont("Segoe UI", 9))
+        info_layout.addWidget(tokens_label)
+        
+        layout.addWidget(info_group)
+        
+        # Prompt text
+        prompt_group = QGroupBox("Full Instructions Text")
+        prompt_layout = QVBoxLayout(prompt_group)
+        
+        prompt_text = QPlainTextEdit()
+        prompt_text.setFont(QFont("Consolas", 9))
+        prompt_text.setPlainText(instructions)
+        prompt_text.setReadOnly(True)
+        prompt_layout.addWidget(prompt_text)
+        
+        layout.addWidget(prompt_group)
+        
+        # Note
+        note = QLabel(
+            "Note: The image is sent as base64-encoded data along with these instructions."
+        )
+        note.setStyleSheet("color: #666;")
+        note.setFont(QFont("Segoe UI", 8))
+        layout.addWidget(note)
+        
+        # Close button
+        button_box = QDialogButtonBox(QDialogButtonBox.StandardButton.Close)
+        button_box.rejected.connect(popup.close)
+        layout.addWidget(button_box)
+        
+        popup.exec()
+    
+    def _save_session_report(self):
+        """Generate and save a session report in markdown format"""
+        if not self.extracted_texts:
+            QMessageBox.warning(
+                None,
+                "No Data",
+                "No OCR processing has been performed yet.\n\n"
+                "Process some images first to generate a session report."
+            )
+            return
+        
+        # Ask for save location
+        output_file, _ = QFileDialog.getSaveFileName(
+            parent=None,
+            caption="Save Session Report",
+            filter="Markdown files (*.md);;Text files (*.txt);;All files (*.*)",
+            initialFilter="Markdown files (*.md)"
+        )
+        
+        if not output_file:
+            return
+        
+        try:
+            from datetime import datetime
+            
+            # Generate report content
+            report_lines = []
+            report_lines.append("# PDF Rescue - Session Report\n")
+            report_lines.append("**Generated by [Supervertaler](https://supervertaler.com/) • by Michael Beijer**\n\n")
+            report_lines.append(f"**Date**: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}\n\n")
+            report_lines.append("---\n\n")
+            
+            # Configuration section
+            report_lines.append("## Configuration\n\n")
+            report_lines.append(f"- **Model**: {self.model_combo.currentText()}\n")
+            formatting_status = "Enabled ✓" if self.preserve_formatting_check.isChecked() else "Disabled ✗"
+            report_lines.append(f"- **Formatting Preservation**: {formatting_status}\n")
+            report_lines.append(f"- **Total Images Processed**: {len(self.extracted_texts)}\n")
+            report_lines.append(f"- **Total Images in List**: {len(self.image_files)}\n\n")
+            
+            # Instructions used
+            report_lines.append("## Extraction Instructions\n\n")
+            report_lines.append("```\n")
+            instructions = self.instructions_text.toPlainText().strip()
+            if self.preserve_formatting_check.isChecked():
+                if "markdown for text formatting" not in instructions:
+                    instructions += "\n- Use markdown for text formatting: **bold text**, *italic text*, __underlined text__"
+            report_lines.append(instructions)
+            report_lines.append("\n```\n\n")
+            
+            # Processing summary
+            report_lines.append("## Processing Summary\n\n")
+            report_lines.append("| # | Image File | Status |\n")
+            report_lines.append("|---|------------|--------|\n")
+            
+            for i, file in enumerate(self.image_files, 1):
+                filename = os.path.basename(file)
+                status = "✓ Processed" if file in self.extracted_texts else "⧗ Pending"
+                report_lines.append(f"| {i} | {filename} | {status} |\n")
+            
+            report_lines.append("\n---\n\n")
+            
+            # Extracted text for each image
+            report_lines.append("## Extracted Text\n\n")
+            
+            for i, file in enumerate(self.image_files, 1):
+                if file in self.extracted_texts:
+                    filename = os.path.basename(file)
+                    report_lines.append(f"### Page {i}: {filename}\n\n")
+                    report_lines.append("```\n")
+                    report_lines.append(self.extracted_texts[file])
+                    report_lines.append("\n```\n\n")
+                    report_lines.append("---\n\n")
+            
+            # Statistics
+            report_lines.append("## Statistics\n\n")
+            texts_list = list(self.extracted_texts.values())
+            total_chars = sum(len(text) for text in texts_list)
+            total_words = sum(len(text.split()) for text in texts_list)
+            report_lines.append(f"- **Total Characters Extracted**: {total_chars:,}\n")
+            report_lines.append(f"- **Total Words Extracted**: {total_words:,}\n")
+            report_lines.append(f"- **Average Characters per Page**: {total_chars // len(self.extracted_texts) if self.extracted_texts else 0:,}\n")
+            report_lines.append(f"- **Average Words per Page**: {total_words // len(self.extracted_texts) if self.extracted_texts else 0:,}\n\n")
+            
+            # Footer
+            report_lines.append("---\n\n")
+            report_lines.append("*Report generated by **PDF Rescue** - AI-Powered OCR Tool*\n\n")
+            report_lines.append("*Part of [**Supervertaler**](https://supervertaler.com/) • by Michael Beijer*\n")
+            
+            # Write to file
+            with open(output_file, 'w', encoding='utf-8') as f:
+                f.writelines(report_lines)
+            
+            self.log_message(f"Session report saved: {Path(output_file).name}")
+            self.status_label.setText(f"✓ Report saved to {os.path.basename(output_file)}")
+            
+            reply = QMessageBox.question(
+                None,
+                "Success",
+                f"Session report saved successfully!\n\n"
+                f"File: {Path(output_file).name}\n\n"
+                "Open the report now?",
+                QMessageBox.StandardButton.Yes | QMessageBox.StandardButton.No
+            )
+            if reply == QMessageBox.StandardButton.Yes:
+                os.startfile(output_file)
+        
+        except Exception as e:
+            QMessageBox.critical(None, "Error", f"Failed to save report:\n\n{str(e)}")
+    
+    # === OCR Processing Methods ===
+    
+    def _encode_image(self, image_path):
+        """Encode image to base64"""
+        with open(image_path, "rb") as image_file:
+            return base64.b64encode(image_file.read()).decode('utf-8')
+    
+    def _extract_text_from_image(self, image_path):
+        """Use AI Vision to extract text from image (supports OpenAI, Claude, Gemini)"""
+        try:
+            instructions = self.instructions_text.toPlainText().strip()
+            
+            # Add or remove formatting instruction based on checkbox
+            if self.preserve_formatting_check.isChecked():
+                if "markdown for text formatting" not in instructions:
+                    instructions += "\n- Use markdown for text formatting: **bold text**, *italic text*, __underlined text__"
+            else:
+                # Remove markdown instruction if present
+                instructions = instructions.replace(
+                    "\n- Use markdown for text formatting: **bold text**, *italic text*, __underlined text__", ""
+                ).replace(
+                    "- Use markdown for text formatting: **bold text**, *italic text*, __underlined text__", ""
+                )
+            
+            # Get selected model
+            model = self.model_combo.currentText()
+            
+            # Skip separator items
+            if model.startswith("---"):
+                return "[ERROR: Please select a valid model, not a separator]"
+            
+            # Determine provider from model name
+            provider = self._get_provider_from_model(model)
+            
+            if provider not in self.clients:
+                return f"[ERROR: {provider.title()} client not initialized. Check API key in api_keys.txt]"
+            
+            # Call appropriate provider
+            if provider == "openai":
+                return self._extract_with_openai(image_path, model, instructions)
+            elif provider == "claude":
+                return self._extract_with_claude(image_path, model, instructions)
+            elif provider == "gemini":
+                return self._extract_with_gemini(image_path, model, instructions)
+            else:
+                return f"[ERROR: Unknown provider: {provider}]"
+        
+        except Exception as e:
+            return f"[ERROR extracting text: {str(e)}]"
+    
+    def _get_provider_from_model(self, model):
+        """Determine provider from model name"""
+        if model.startswith("gpt") or model.startswith("o1") or model.startswith("o3"):
+            return "openai"
+        elif model.startswith("claude"):
+            return "claude"
+        elif model.startswith("gemini"):
+            return "gemini"
+        else:
+            return "unknown"
+    
+    def _extract_with_openai(self, image_path, model, instructions):
+        """Extract text using OpenAI Vision API"""
+        base64_image = self._encode_image(image_path)
+        
+        api_params = {
+            "model": model,
+            "messages": [
+                {
+                    "role": "user",
+                    "content": [
+                        {"type": "text", "text": instructions},
+                        {
+                            "type": "image_url",
+                            "image_url": {"url": f"data:image/jpeg;base64,{base64_image}"}
+                        }
+                    ]
+                }
+            ]
+        }
+        
+        # Use appropriate token parameter based on model
+        if model.startswith("gpt-5") or model.startswith("o1"):
+            api_params["max_completion_tokens"] = 4000
+        else:
+            api_params["max_tokens"] = 4000
+        
+        response = self.clients['openai'].chat.completions.create(**api_params)
+        return response.choices[0].message.content
+    
+    def _extract_with_claude(self, image_path, model, instructions):
+        """Extract text using Claude Vision API"""
+        import base64
+        
+        # Read image and encode to base64
+        with open(image_path, "rb") as image_file:
+            image_data = base64.standard_b64encode(image_file.read()).decode("utf-8")
+        
+        # Determine media type from file extension
+        ext = os.path.splitext(image_path)[1].lower()
+        media_type_map = {
+            '.jpg': 'image/jpeg',
+            '.jpeg': 'image/jpeg',
+            '.png': 'image/png',
+            '.gif': 'image/gif',
+            '.webp': 'image/webp'
+        }
+        media_type = media_type_map.get(ext, 'image/jpeg')
+        
+        response = self.clients['claude'].messages.create(
+            model=model,
+            max_tokens=4000,
+            messages=[
+                {
+                    "role": "user",
+                    "content": [
+                        {
+                            "type": "image",
+                            "source": {
+                                "type": "base64",
+                                "media_type": media_type,
+                                "data": image_data,
+                            },
+                        },
+                        {
+                            "type": "text",
+                            "text": instructions
+                        }
+                    ],
+                }
+            ],
+        )
+        
+        return response.content[0].text
+    
+    def _extract_with_gemini(self, image_path, model, instructions):
+        """Extract text using Gemini Vision API"""
+        from PIL import Image
+        
+        # Load image
+        img = Image.open(image_path)
+        
+        # Create model instance
+        gemini_model = self.clients['gemini'].GenerativeModel(model)
+        
+        # Generate content with image and prompt
+        response = gemini_model.generate_content([instructions, img])
+        
+        return response.text
+    
+    def _process_selected(self):
+        """Process currently selected image"""
+        selected_items = self.file_listbox.selectedItems()
+        if not selected_items:
+            QMessageBox.warning(None, "No Selection", "Please select an image to process")
+            return
+        
+        idx = self.file_listbox.row(selected_items[0])
+        if idx >= len(self.image_files):
+            return
+        
+        file = self.image_files[idx]
+        filename = os.path.basename(file)
+        
+        # Get provider info for status display
+        model = self.model_combo.currentText()
+        provider = self._get_provider_from_model(model).title()
+        
+        self.log_message(f"Processing selected image with {provider} ({model}): {filename}")
+        self.status_label.setText(f"Processing with {provider}... {filename}")
+        QApplication.processEvents()
+        
+        text = self._extract_text_from_image(file)
+        self.extracted_texts[file] = text
+        
+        self.preview_text.setPlainText(text)
+        
+        self._update_listbox()
+        self.log_message(f"Successfully processed: {filename}")
+        self.status_label.setText(f"✓ Processed {filename} with {provider}")
+    
+    def _process_all(self):
+        """Process all images in the list"""
+        if not self.image_files:
+            QMessageBox.warning(None, "No Files", "Please add images first")
+            return
+        
+        # Get provider info for confirmation
+        model = self.model_combo.currentText()
+        provider = self._get_provider_from_model(model).title()
+        
+        reply = QMessageBox.question(
+            None,
+            "Process All",
+            f"Process all {len(self.image_files)} images with {provider} ({model})?\n\n"
+            "This will use API credits and may take several minutes.",
+            QMessageBox.StandardButton.Yes | QMessageBox.StandardButton.No
+        )
+        if reply != QMessageBox.StandardButton.Yes:
+            return
+        
+        self.log_message(f"Starting batch processing with {provider}: {len(self.image_files)} images")
+        self.progress.setMaximum(len(self.image_files))
+        self.progress.setValue(0)
+        
+        for i, file in enumerate(self.image_files, 1):
+            filename = os.path.basename(file)
+            self.status_label.setText(f"[{provider}] Processing {i}/{len(self.image_files)}: {filename}...")
+            QApplication.processEvents()
+            
+            if file not in self.extracted_texts:
+                text = self._extract_text_from_image(file)
+                self.extracted_texts[file] = text
+                self.log_message(f"  [{i}/{len(self.image_files)}] Processed: {filename}")
+            else:
+                self.log_message(f"  [{i}/{len(self.image_files)}] Skipped (already processed): {filename}")
+            
+            self.progress.setValue(i)
+            self._update_listbox()
+        
+        self.log_message(f"Batch processing complete: {len(self.image_files)} images processed")
+        self.status_label.setText(f"✓ Processed all {len(self.image_files)} images with {provider}!")
+        QMessageBox.information(
+            None,
+            "Complete",
+            f"Successfully processed {len(self.image_files)} images with {provider}!\n\n"
+            "Click 'Save DOCX' to export the text."
+        )
+    
+    # === Export Methods ===
+    
+    def _add_formatted_text(self, doc, text):
+        """
+        Add text to document with markdown formatting parsed
+        Supports: **bold**, *italic*, __underline__
+        Also handles multi-column layouts with [START COLUMN X] / [END COLUMN X] markers
+        """
+        # Check if text has column markers
+        if '[START COLUMN' in text or '[COLUMN 1]' in text or '[COLUMN 2]' in text:
+            self._add_multi_column_text(doc, text)
+            return
+        
+        # Split text into paragraphs (separated by double newlines or more)
+        # This prevents treating every single line as a paragraph
+        paragraphs = re.split(r'\n\s*\n', text)
+        
+        for para_text in paragraphs:
+            if not para_text.strip():
+                continue
+            
+            # Replace single newlines within paragraph with spaces (removes extraneous line breaks)
+            para_text = para_text.replace('\n', ' ').strip()
+            
+            para = doc.add_paragraph()
+            para.paragraph_format.line_spacing = 1.15
+            para.paragraph_format.space_after = Pt(6)  # Reduced from 12 to 6 for tighter spacing
+            
+            # Parse markdown formatting using regex
+            remaining = para_text
+            
+            while remaining:
+                # Check for bold (**text**)
+                bold_match = re.match(r'\*\*(.*?)\*\*', remaining)
+                if bold_match:
+                    run = para.add_run(bold_match.group(1))
+                    run.bold = True
+                    remaining = remaining[bold_match.end():]
+                    continue
+                
+                # Check for underline (__text__)
+                underline_match = re.match(r'__(.*?)__', remaining)
+                if underline_match:
+                    run = para.add_run(underline_match.group(1))
+                    run.underline = True
+                    remaining = remaining[underline_match.end():]
+                    continue
+                
+                # Check for italic (*text*)
+                italic_match = re.match(r'\*(.*?)\*', remaining)
+                if italic_match:
+                    run = para.add_run(italic_match.group(1))
+                    run.italic = True
+                    remaining = remaining[italic_match.end():]
+                    continue
+                
+                # No formatting - add plain text until next marker or end
+                next_marker = len(remaining)
+                for marker in ['**', '*', '__']:
+                    pos = remaining.find(marker)
+                    if pos != -1 and pos < next_marker:
+                        next_marker = pos
+                
+                if next_marker == 0:
+                    # Edge case: marker at start but no match (e.g., single * or **)
+                    para.add_run(remaining[0])
+                    remaining = remaining[1:]
+                else:
+                    plain_text = remaining[:next_marker] if next_marker < len(remaining) else remaining
+                    if plain_text:
+                        para.add_run(plain_text)
+                    remaining = remaining[next_marker:]
+    
+    def _add_multi_column_text(self, doc, text):
+        """
+        Handle multi-column text layout using a Word table
+        Supports markers like [START COLUMN 1], [END COLUMN 1], etc.
+        """
+        # Parse columns from text
+        columns = {}
+        current_column = None
+        lines = text.split('\n')
+        
+        for line in lines:
+            # Check for column start marker
+            if '[START COLUMN' in line.upper() or '[COLUMN' in line.upper():
+                # Extract column number
+                import re
+                match = re.search(r'\[(?:START )?COLUMN[:\s]+(\d+)\]', line, re.IGNORECASE)
+                if match:
+                    current_column = int(match.group(1))
+                    if current_column not in columns:
+                        columns[current_column] = []
+                continue
+            
+            # Check for column end marker
+            if '[END COLUMN' in line.upper():
+                current_column = None
+                continue
+            
+            # Add line to current column
+            if current_column is not None:
+                columns[current_column].append(line)
+        
+        # If we found columns, create a table layout
+        if columns:
+            num_columns = max(columns.keys()) if columns else 2
+            table = doc.add_table(rows=1, cols=num_columns)
+            table.style = 'Table Grid'
+            
+            # Set column widths to be equal
+            from docx.shared import Inches
+            for col_idx in range(num_columns):
+                for cell in table.columns[col_idx].cells:
+                    cell.width = Inches(6.5 / num_columns)
+            
+            # Fill each column
+            for col_num in sorted(columns.keys()):
+                if col_num <= num_columns:
+                    cell = table.cell(0, col_num - 1)
+                    column_text = '\n'.join(columns[col_num])
+                    
+                    # Remove the cell's default paragraph and add formatted text
+                    cell.text = ''
+                    for para_text in column_text.split('\n'):
+                        if para_text.strip():
+                            para = cell.add_paragraph(para_text.strip())
+                            para.paragraph_format.line_spacing = 1.0
+                            para.paragraph_format.space_after = Pt(6)
+            
+            # Remove table borders for cleaner look
+            for row in table.rows:
+                for cell in row.cells:
+                    tcPr = cell._element.get_or_add_tcPr()
+                    tcBorders = tcPr.find('.//{http://schemas.openxmlformats.org/wordprocessingml/2006/main}tcBorders')
+                    if tcBorders is None:
+                        from docx.oxml import OxmlElement
+                        tcBorders = OxmlElement('w:tcBorders')
+                        tcPr.append(tcBorders)
+                    # Set all borders to none
+                    for border_name in ['top', 'left', 'bottom', 'right', 'insideH', 'insideV']:
+                        border = tcBorders.find('.//{http://schemas.openxmlformats.org/wordprocessingml/2006/main}' + border_name)
+                        if border is None:
+                            from docx.oxml import OxmlElement
+                            border = OxmlElement(f'w:{border_name}')
+                            tcBorders.append(border)
+                        border.set('{http://schemas.openxmlformats.org/wordprocessingml/2006/main}val', 'none')
+        else:
+            # No columns found, fall back to regular formatting
+            self._add_formatted_text(doc, text)
+    
+    def _parse_markdown_table(self, table_lines):
+        """Parse markdown table lines into rows and cells"""
+        rows = []
+        for line in table_lines:
+            line = line.strip()
+            if not line or not line.startswith('|'):
+                continue
+            # Split by pipe, remove first and last empty elements
+            cells = [cell.strip() for cell in line.split('|')[1:-1]]
+            # Skip separator row (contains only dashes, colons, and spaces)
+            is_separator = all(
+                cell.replace('-', '').replace(':', '').replace(' ', '').strip() == '' 
+                for cell in cells
+            )
+            if is_separator:
+                continue
+            rows.append(cells)
+        return rows
+    
+    def _extract_markdown_tables(self, text):
+        """Extract markdown tables from text and return (text_without_tables, tables_list)"""
+        lines = text.split('\n')
+        result_lines = []
+        tables = []
+        current_table_lines = []
+        in_table = False
+        
+        for i, line in enumerate(lines):
+            stripped = line.strip()
+            # Check if line looks like a table row (starts and ends with |)
+            is_table_row = stripped.startswith('|') and stripped.endswith('|') and '|' in stripped[1:-1]
+            
+            if is_table_row:
+                if not in_table:
+                    in_table = True
+                    current_table_lines = []
+                current_table_lines.append(line)
+            else:
+                # End of table
+                if in_table:
+                    # Parse the table
+                    parsed_table = self._parse_markdown_table(current_table_lines)
+                    if parsed_table and len(parsed_table) > 0:
+                        tables.append(parsed_table)
+                    # Add placeholder for table
+                    result_lines.append(f"[TABLE_{len(tables) - 1}]")
+                    in_table = False
+                    current_table_lines = []
+                result_lines.append(line)
+        
+        # Handle table at end of text
+        if in_table and current_table_lines:
+            parsed_table = self._parse_markdown_table(current_table_lines)
+            if parsed_table and len(parsed_table) > 0:
+                tables.append(parsed_table)
+            result_lines.append(f"[TABLE_{len(tables) - 1}]")
+        
+        text_without_tables = '\n'.join(result_lines)
+        return text_without_tables, tables
+    
+    def _add_markdown_table_to_doc(self, doc, table_rows):
+        """Add a Word table from markdown table rows"""
+        if not table_rows or len(table_rows) == 0:
+            return
+        
+        # Determine number of columns (use first row, usually header)
+        num_cols = len(table_rows[0]) if table_rows else 1
+        num_rows = len(table_rows)
+        
+        # Create Word table
+        table = doc.add_table(rows=num_rows, cols=num_cols)
+        table.style = 'Light Grid Accent 1'
+        
+        # Populate table cells
+        for row_idx, row_data in enumerate(table_rows):
+            # Ensure row has enough cells
+            while len(row_data) < num_cols:
+                row_data.append('')
+            
+            for col_idx in range(num_cols):
+                cell_text = row_data[col_idx] if col_idx < len(row_data) else ''
+                # Remove markdown formatting from table cells (keep plain text for now)
+                # Could enhance this to support formatting in cells later
+                cell_text = cell_text.replace('**', '').replace('*', '').replace('__', '')
+                table.rows[row_idx].cells[col_idx].text = cell_text
+        
+        # Add spacing after table
+        doc.add_paragraph()
+    
+    def _add_formatted_text_with_tables(self, doc, text):
+        """Add text to document with markdown formatting and tables parsed"""
+        # Extract tables first
+        text_without_tables, tables = self._extract_markdown_tables(text)
+        
+        # Split text into paragraphs (now with table placeholders)
+        parts = text_without_tables.split('\n')
+        
+        current_paragraph_text = []
+        
+        for part in parts:
+            # Check if this is a table placeholder
+            if part.strip().startswith('[TABLE_') and part.strip().endswith(']'):
+                # Add any accumulated paragraph text
+                if current_paragraph_text:
+                    para_text = '\n'.join(current_paragraph_text).strip()
+                    if para_text:
+                        self._add_formatted_text(doc, para_text)
+                    current_paragraph_text = []
+                
+                # Extract table index
+                try:
+                    table_idx = int(part.strip()[7:-1])  # Extract number from [TABLE_0]
+                    if 0 <= table_idx < len(tables):
+                        self._add_markdown_table_to_doc(doc, tables[table_idx])
+                except (ValueError, IndexError):
+                    pass
+            else:
+                current_paragraph_text.append(part)
+        
+        # Add any remaining paragraph text
+        if current_paragraph_text:
+            para_text = '\n'.join(current_paragraph_text).strip()
+            if para_text:
+                self._add_formatted_text(doc, para_text)
+    
+    def _generate_markdown_file(self, output_file):
+        """Generate markdown file from extracted texts"""
+        with open(output_file, 'w', encoding='utf-8') as f:
+            # Header
+            f.write("# Extracted Text from Images\n\n")
+            f.write("*Generated by PDF Rescue (a Supervertaler module)*\n\n")
+            f.write("---\n\n")
+            
+            # Add extracted text in order
+            for i, file in enumerate(self.image_files, 1):
+                if file in self.extracted_texts:
+                    # Page header
+                    f.write(f"## Page {i}: {os.path.basename(file)}\n\n")
+                    
+                    # Text content (already in markdown format)
+                    text = self.extracted_texts[file]
+                    f.write(text)
+                    f.write("\n\n")
+                    f.write("---\n\n")
+    
+    def _markdown_to_docx(self, md_file, docx_file):
+        """Convert markdown file to DOCX with proper table handling"""
+        # Read markdown file
+        with open(md_file, 'r', encoding='utf-8') as f:
+            md_content = f.read()
+        
+        # Create Word document
+        doc = Document()
+        
+        # Parse markdown content line by line
+        lines = md_content.split('\n')
+        i = 0
+        
+        while i < len(lines):
+            line = lines[i].rstrip()  # Keep left whitespace, remove right
+            
+            # Skip empty lines (add spacing as paragraph)
+            if not line.strip():
+                doc.add_paragraph()
+                i += 1
+                continue
+            
+            # Handle headers
+            if line.startswith('# '):
+                title = doc.add_heading(line[2:].strip(), 0)
+                title.runs[0].font.size = Pt(16)
+                i += 1
+                continue
+            elif line.startswith('## '):
+                heading = doc.add_heading(line[3:].strip(), level=2)
+                heading.runs[0].font.size = Pt(12)
+                i += 1
+                continue
+            elif line.startswith('### '):
+                heading = doc.add_heading(line[4:].strip(), level=3)
+                i += 1
+                continue
+            
+            # Handle horizontal rule
+            if line.strip() == '---' or line.strip() == '***':
+                doc.add_paragraph()
+                i += 1
+                continue
+            
+            # Handle italics line (branding) - simple single-line italic
+            stripped = line.strip()
+            if stripped.startswith('*') and stripped.endswith('*') and stripped.count('*') == 2 and len(stripped) > 2:
+                para = doc.add_paragraph()
+                run = para.add_run(stripped[1:-1])
+                run.italic = True
+                run.font.size = Pt(9)
+                i += 1
+                continue
+            
+            # Check if this might be a table
+            if line.strip().startswith('|') and line.strip().endswith('|') and '|' in line.strip()[1:-1]:
+                # Collect all table lines
+                table_lines = [line]
+                i += 1
+                while i < len(lines):
+                    next_line = lines[i].rstrip()
+                    # Continue if it's a table row or separator row
+                    if next_line.strip().startswith('|') and next_line.strip().endswith('|'):
+                        table_lines.append(next_line)
+                        i += 1
+                    else:
+                        break
+                
+                # Parse and add table
+                parsed_table = self._parse_markdown_table(table_lines)
+                if parsed_table and len(parsed_table) > 0:
+                    self._add_markdown_table_to_doc(doc, parsed_table)
+                continue
+            
+            # Regular paragraph text - collect until empty line or special marker
+            para_lines = [line]
+            i += 1
+            while i < len(lines):
+                next_line = lines[i].rstrip()
+                # Stop at empty line or special markers
+                if not next_line.strip():
+                    break
+                if next_line.strip().startswith('#') or next_line.strip() in ['---', '***']:
+                    break
+                if next_line.strip().startswith('|') and next_line.strip().endswith('|'):
+                    break
+                para_lines.append(next_line)
+                i += 1
+            
+            # Process paragraph with tables
+            para_text = '\n'.join(para_lines)
+            if para_text.strip():
+                self._add_formatted_text_with_tables(doc, para_text)
+        
+        # Save document
+        doc.save(docx_file)
+    
+    def _export_markdown_and_word(self):
+        """Export extracted text as markdown, convert to DOCX, and generate session report"""
+        if not self.extracted_texts:
+            QMessageBox.warning(
+                None,
+                "No Text",
+                "No extracted text to export.\n\n"
+                "Process images first."
+            )
+            return
+        
+        # Ask for base file name (will generate .md, .docx, and _report.md)
+        output_file, _ = QFileDialog.getSaveFileName(
+            parent=None,
+            caption="Export Markdown & Word Documents",
+            filter="Markdown files (*.md);;All files (*.*)",
+            initialFilter="Markdown files (*.md)"
+        )
+        
+        if not output_file:
+            return
+        
+        # Generate file names
+        base_path = Path(output_file)
+        if base_path.suffix != '.md':
+            base_path = base_path.with_suffix('.md')
+        
+        md_file = str(base_path)
+        docx_file = str(base_path.with_suffix('.docx'))
+        report_file = str(base_path.parent / f"{base_path.stem}_report.md")
+        
+        self.log_message(f"Exporting documents: {base_path.name}")
+        
+        try:
+            # Step 1: Generate markdown file
+            self._generate_markdown_file(md_file)
+            self.log_message(f"✓ Markdown file created: {base_path.name}")
+            
+            # Step 2: Convert markdown to DOCX
+            self._markdown_to_docx(md_file, docx_file)
+            self.log_message(f"✓ Word document created: {base_path.stem}.docx")
+            
+            # Step 3: Generate session report
+            self._save_session_report_to_file(report_file)
+            self.log_message(f"✓ Session report created: {base_path.stem}_report.md")
+            
+            # Success message
+            files_created = f"✓ {base_path.name}\n✓ {base_path.stem}.docx\n✓ {base_path.stem}_report.md"
+            self.status_label.setText(f"✓ Exported 3 files successfully")
+            
+            reply = QMessageBox.question(
+                None,
+                "Export Complete",
+                f"Successfully exported all documents!\n\n"
+                f"Files created:\n{files_created}\n\n"
+                "Open the files now?",
+                QMessageBox.StandardButton.Yes | QMessageBox.StandardButton.No
+            )
+            if reply == QMessageBox.StandardButton.Yes:
+                os.startfile(md_file)
+                os.startfile(docx_file)
+                os.startfile(report_file)
+        
+        except Exception as e:
+            QMessageBox.critical(None, "Error", f"Failed to export documents:\n\n{str(e)}")
+            self.log_message(f"✗ Export error: {str(e)}")
+    
+    def _save_session_report_to_file(self, output_file):
+        """Generate and save session report to specified file (internal method)"""
+        from datetime import datetime
+        
+        # Generate report content
+        report_lines = []
+        report_lines.append("# PDF Rescue - Session Report\n")
+        report_lines.append("**Generated by [Supervertaler](https://supervertaler.com/) • by Michael Beijer**\n\n")
+        report_lines.append(f"**Date**: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}\n\n")
+        report_lines.append("---\n\n")
+        
+        # Configuration section
+        report_lines.append("## Configuration\n\n")
+        report_lines.append(f"- **Model**: {self.model_combo.currentText()}\n")
+        formatting_status = "Enabled ✓" if self.preserve_formatting_check.isChecked() else "Disabled ✗"
+        report_lines.append(f"- **Formatting Preservation**: {formatting_status}\n")
+        report_lines.append(f"- **Total Images Processed**: {len(self.extracted_texts)}\n")
+        report_lines.append(f"- **Total Images in List**: {len(self.image_files)}\n\n")
+        
+        # Instructions used
+        report_lines.append("## Extraction Instructions\n\n")
+        report_lines.append("```\n")
+        instructions = self.instructions_text.toPlainText().strip()
+        if self.preserve_formatting_check.isChecked():
+            if "markdown for text formatting" not in instructions:
+                instructions += "\n- Use markdown for text formatting: **bold text**, *italic text*, __underlined text__"
+        report_lines.append(instructions)
+        report_lines.append("\n```\n\n")
+        
+        # Processing summary
+        report_lines.append("## Processing Summary\n\n")
+        report_lines.append("| # | Image File | Status |\n")
+        report_lines.append("|---|------------|--------|\n")
+        
+        for i, file in enumerate(self.image_files, 1):
+            filename = os.path.basename(file)
+            status = "✓ Processed" if file in self.extracted_texts else "⧗ Pending"
+            report_lines.append(f"| {i} | {filename} | {status} |\n")
+        
+        report_lines.append("\n---\n\n")
+        
+        # Extracted text for each image
+        report_lines.append("## Extracted Text\n\n")
+        
+        for i, file in enumerate(self.image_files, 1):
+            if file in self.extracted_texts:
+                filename = os.path.basename(file)
+                report_lines.append(f"### Page {i}: {filename}\n\n")
+                report_lines.append("```\n")
+                report_lines.append(self.extracted_texts[file])
+                report_lines.append("\n```\n\n")
+                report_lines.append("---\n\n")
+        
+        # Statistics
+        report_lines.append("## Statistics\n\n")
+        texts_list = list(self.extracted_texts.values())
+        total_chars = sum(len(text) for text in texts_list)
+        total_words = sum(len(text.split()) for text in texts_list)
+        report_lines.append(f"- **Total Characters Extracted**: {total_chars:,}\n")
+        report_lines.append(f"- **Total Words Extracted**: {total_words:,}\n")
+        report_lines.append(f"- **Average Characters per Page**: {total_chars // len(self.extracted_texts) if self.extracted_texts else 0:,}\n")
+        report_lines.append(f"- **Average Words per Page**: {total_words // len(self.extracted_texts) if self.extracted_texts else 0:,}\n\n")
+        
+        # Footer
+        report_lines.append("---\n\n")
+        report_lines.append("*Report generated by **PDF Rescue** - AI-Powered OCR Tool*\n\n")
+        report_lines.append("*Part of [**Supervertaler**](https://supervertaler.com/) • by Michael Beijer*\n")
+        
+        # Write to file
+        with open(output_file, 'w', encoding='utf-8') as f:
+            f.writelines(report_lines)
+    
+    def _save_to_docx(self):
+        """Save all extracted text to a Word document"""
+        if not self.extracted_texts:
+            QMessageBox.warning(
+                None,
+                "No Text",
+                "No extracted text to save.\n\n"
+                "Process images first."
+            )
+            return
+        
+        output_file, _ = QFileDialog.getSaveFileName(
+            parent=None,
+            caption="Save Extracted Text",
+            filter="Word Document (*.docx);;All files (*.*)",
+            initialFilter="Word Document (*.docx)"
+        )
+        
+        if not output_file:
+            return
+        
+        self.log_message(f"Saving extracted text to DOCX: {Path(output_file).name}")
+        
+        try:
+            doc = Document()
+            
+            # Add title
+            title = doc.add_heading('Extracted Text from Images', 0)
+            title.runs[0].font.size = Pt(16)
+            
+            # Add branding with hyperlink to Supervertaler
+            from docx.oxml import OxmlElement
+            from docx.oxml.ns import qn
+            
+            def add_hyperlink(paragraph, text, url):
+                """Add a hyperlink to a paragraph"""
+                # Get the paragraph element
+                part = paragraph.part
+                r_id = part.relate_to(url, 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/hyperlink', is_external=True)
+                
+                # Create the hyperlink element
+                hyperlink = OxmlElement('w:hyperlink')
+                hyperlink.set(qn('r:id'), r_id)
+                
+                # Create a new run element
+                new_run = OxmlElement('w:r')
+                rPr = OxmlElement('w:rPr')
+                
+                # Add hyperlink style
+                rStyle = OxmlElement('w:rStyle')
+                rStyle.set(qn('w:val'), 'Hyperlink')
+                rPr.append(rStyle)
+                new_run.append(rPr)
+                
+                # Add the text
+                new_run.text = text
+                hyperlink.append(new_run)
+                
+                # Add hyperlink to paragraph
+                paragraph._p.append(hyperlink)
+                return hyperlink
+            
+            branding_para = doc.add_paragraph()
+            
+            # Add text before hyperlink
+            run1 = branding_para.add_run('Generated by PDF Rescue (a ')
+            run1.font.size = Pt(9)
+            run1.italic = True
+            
+            # Add hyperlink
+            add_hyperlink(branding_para, 'Supervertaler', 'https://supervertaler.com/')
+            
+            # Add text after hyperlink
+            run2 = branding_para.add_run(' module)')
+            run2.font.size = Pt(9)
+            run2.italic = True
+            
+            # Add spacing
+            doc.add_paragraph()
+            
+            # Add extracted text in order
+            for i, file in enumerate(self.image_files, 1):
+                if file in self.extracted_texts:
+                    # Page header
+                    heading = doc.add_heading(f'Page {i}: {os.path.basename(file)}', level=2)
+                    heading.runs[0].font.size = Pt(12)
+                    
+                    # Text content with formatting
+                    text = self.extracted_texts[file]
+                    if self.preserve_formatting_check.isChecked():
+                        self._add_formatted_text(doc, text)
+                    else:
+                        # Split by double newlines to preserve paragraph breaks
+                        # Replace single newlines with spaces to remove extraneous line breaks
+                        paragraphs = re.split(r'\n\s*\n', text)
+                        for para_text in paragraphs:
+                            if para_text.strip():
+                                # Replace single newlines within paragraph with spaces
+                                para_text = para_text.replace('\n', ' ').strip()
+                                para = doc.add_paragraph(para_text)
+                                para.paragraph_format.line_spacing = 1.15
+                                para.paragraph_format.space_after = Pt(6)
+                    
+                    # Page break except for last
+                    if i < len(self.image_files):
+                        doc.add_page_break()
+            
+            doc.save(output_file)
+            
+            self.log_message(f"Successfully saved {len(self.extracted_texts)} pages to: {Path(output_file).name}")
+            self.status_label.setText(f"✓ Saved to {os.path.basename(output_file)}")
+            
+            reply = QMessageBox.question(
+                None,
+                "Success",
+                f"Document saved successfully!\n\n"
+                f"{len(self.extracted_texts)} pages of text extracted\n\n"
+                "Open the document now?",
+                QMessageBox.StandardButton.Yes | QMessageBox.StandardButton.No
+            )
+            if reply == QMessageBox.StandardButton.Yes:
+                os.startfile(output_file)
+        
+        except Exception as e:
+            QMessageBox.critical(None, "Error", f"Failed to save document:\n\n{str(e)}")
+    
+    def _copy_all_text(self):
+        """Copy all extracted text to clipboard"""
+        if not self.extracted_texts:
+            QMessageBox.warning(None, "No Text", "No extracted text to copy")
+            return
+        
+        all_text = []
+        for i, file in enumerate(self.image_files, 1):
+            if file in self.extracted_texts:
+                all_text.append(f"=== Page {i}: {os.path.basename(file)} ===\n")
+                all_text.append(self.extracted_texts[file])
+                all_text.append("\n\n")
+        
+        combined = "".join(all_text)
+        
+        # Use QApplication clipboard
+        from PyQt6.QtGui import QClipboard
+        app = QApplication.instance()
+        if app:
+            clipboard = app.clipboard()
+            clipboard.setText(combined)
+        
+        self.status_label.setText(f"✓ Copied {len(self.extracted_texts)} pages to clipboard")
+        QMessageBox.information(None, "Copied", f"Copied text from {len(self.extracted_texts)} pages to clipboard!")
+
+
+# === Standalone Application ===
+
+if __name__ == "__main__":
+    """Run PDF Rescue as a standalone application"""
+    import sys
+    from PyQt6.QtWidgets import QApplication, QMainWindow, QTextEdit, QLabel
+    from PyQt6.QtCore import Qt
+    
+    class StandaloneApp(QMainWindow):
+        """Minimal parent app for standalone mode"""
+        def __init__(self):
+            super().__init__()
+            self.setWindowTitle("PDF Rescue - AI-Powered OCR Tool")
+            self.setGeometry(100, 100, 1200, 800)
+            
+            # Central widget
+            central_widget = QWidget()
+            self.setCentralWidget(central_widget)
+            
+            # Main layout
+            main_layout = QVBoxLayout(central_widget)
+            main_layout.setContentsMargins(10, 10, 10, 10)
+            
+            # Title
+            title = QLabel("PDF Rescue - AI-Powered OCR Tool")
+            title.setFont(QFont("Segoe UI", 14, QFont.Weight.Bold))
+            title.setAlignment(Qt.AlignmentFlag.AlignCenter)
+            main_layout.addWidget(title)
+            
+            # Load API keys
+            self.api_keys = {}
+            api_file = Path("api_keys.txt")
+            if not api_file.exists():
+                # Try user_data folder
+                api_file = Path("user_data_private" if os.path.exists(".supervertaler.local") else "user_data") / "api_keys.txt"
+            
+            if api_file.exists():
+                with open(api_file, 'r', encoding='utf-8') as f:
+                    for line in f:
+                        line = line.strip()
+                        if line and not line.startswith('#') and '=' in line:
+                            key, value = line.split('=', 1)
+                            if 'openai' in key.lower():
+                                self.api_keys['openai'] = value.strip()
+            
+            if not self.api_keys.get('openai'):
+                QMessageBox.critical(
+                    self,
+                    "API Key Missing",
+                    "Could not find OpenAI API key in api_keys.txt\n\n"
+                    "Please add a line like:\nOPENAI_API_KEY=your-key-here\n\n"
+                    "Or place api_keys.txt in the user_data folder."
+                )
+                # Still create UI but warn user
+            
+            # Create PDF Rescue instance (standalone mode)
+            self.pdf_rescue = PDFRescueQt(self, standalone=True)
+            pdf_rescue_widget = QWidget()
+            self.pdf_rescue.create_tab(pdf_rescue_widget)
+            main_layout.addWidget(pdf_rescue_widget)
+            
+            # Add log at bottom
+            log_group = QGroupBox("Activity Log")
+            log_layout = QVBoxLayout(log_group)
+            
+            self.log_text = QTextEdit()
+            self.log_text.setReadOnly(True)
+            self.log_text.setMaximumHeight(150)
+            self.log_text.setFont(QFont("Consolas", 9))
+            log_layout.addWidget(self.log_text)
+            
+            main_layout.addWidget(log_group)
+        
+        def log(self, message: str):
+            """Add message to log"""
+            from datetime import datetime
+            timestamp = datetime.now().strftime("%H:%M:%S")
+            formatted_message = f"[{timestamp}] {message}"
+            self.log_text.append(formatted_message)
+        
+        def load_api_keys(self):
+            """Load API keys for compatibility"""
+            return self.api_keys
+    
+    # Create and run standalone app
+    app = QApplication(sys.argv)
+    window = StandaloneApp()
+    window.show()
+    sys.exit(app.exec())
+