abstractcore 2.5.2__py3-none-any.whl → 2.5.3__py3-none-any.whl

abstractcore/compression/pil_text_renderer.py

@@ -0,0 +1,818 @@
+"""
+PIL/Pillow-based text renderer with proper bold and italic font support.
+
+This renderer directly creates images using PIL/Pillow, giving us complete control
+over font rendering and text layout.
+"""
+
+import os
+import tempfile
+from pathlib import Path
+from typing import List, Optional, Tuple
+import math
+
+from .config import GlyphConfig, RenderingConfig
+from .text_formatter import TextSegment
+from .exceptions import RenderingError
+from ..utils.structured_logging import get_logger
+
+
+class PILTextRenderer:
+    """Direct text renderer using PIL/Pillow with proper font support."""
+    
+    def __init__(self, config: GlyphConfig):
+        """
+        Initialize PIL text renderer.
+        
+        Args:
+            config: Glyph configuration
+        """
+        self.config = config
+        self.logger = get_logger(self.__class__.__name__)
+        
+        # Check dependencies
+        self._check_dependencies()
+        
+        # Track if we're using OCRB or OCRA fonts for special bold handling
+        self.using_ocrb_fonts = False
+        self.using_ocra_fonts = False
+        
+        self.logger.debug("PILTextRenderer initialized")
+    
+    def _get_effective_dimensions(self, config: RenderingConfig) -> tuple[int, int]:
+        """Get effective image dimensions (target dimensions or VLM-optimized defaults)."""
+        if config.target_width and config.target_height:
+            return (config.target_width, config.target_height)
+        else:
+            # VLM-optimized defaults: 1024x1024 works well with most vision models
+            # - Fits within Claude 3.5 Sonnet (1568x1568 max)
+            # - Efficient for GPT-4o tokenization (~1800 tokens)
+            # - Square aspect ratio (1:1) for consistent layout
+            # - Supported by all major VLM families
+            return (1024, 1024)
+    
+    def _check_dependencies(self):
+        """Check if required dependencies are available."""
+        try:
+            from PIL import Image, ImageDraw, ImageFont
+        except ImportError as e:
+            raise RenderingError(
+                f"PIL/Pillow not available: {e}. Install with: pip install pillow"
+            )
+    
+    def _estimate_text_capacity(self, config: RenderingConfig, fonts: dict) -> int:
+        """
+        Estimate how many characters can fit in the target image dimensions.
+        
+        Args:
+            config: Rendering configuration
+            fonts: Loaded fonts dictionary
+            
+        Returns:
+            Estimated character capacity
+        """
+        img_width, img_height = self._get_effective_dimensions(config)
+        
+        # Calculate available space
+        available_width = img_width - 2 * config.margin_x
+        available_height = img_height - 2 * config.margin_y
+        
+        # Account for multi-column layout
+        columns = max(1, config.columns)
+        column_gap = config.column_gap if columns > 1 else 0
+        column_width = (available_width - (columns - 1) * column_gap) / columns
+        
+        # Estimate character dimensions using regular font
+        regular_font = fonts.get('regular')
+        if regular_font:
+            # Use average character width (test with common characters)
+            test_chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789 "
+            try:
+                # Create temporary image to measure text
+                from PIL import Image, ImageDraw
+                temp_img = Image.new('RGB', (1, 1))
+                temp_draw = ImageDraw.Draw(temp_img)
+                
+                try:
+                    bbox = temp_draw.textbbox((0, 0), test_chars, font=regular_font)
+                    total_width = bbox[2] - bbox[0]
+                except AttributeError:
+                    # Fallback for older PIL versions
+                    total_width = temp_draw.textsize(test_chars, font=regular_font)[0]
+                
+                avg_char_width = total_width / len(test_chars)
+            except:
+                # Fallback estimate
+                avg_char_width = config.font_size * 0.6
+        else:
+            # Fallback estimate
+            avg_char_width = config.font_size * 0.6
+        
+        # Estimate line capacity
+        line_height = int(config.line_height * 1.3)  # Match spacing calculation
+        chars_per_line = int(column_width / avg_char_width)
+        lines_per_column = int((available_height * 1.1 + line_height) / line_height)
+        total_lines = lines_per_column * columns
+        
+        # More realistic capacity estimation
+        # The previous 30% efficiency was way too conservative
+        if min(img_width, img_height) < 600:
+            # For small images like 448x448, use more realistic estimate
+            efficiency_factor = 0.75  # Much higher to use available space
+        else:
+            efficiency_factor = 0.85
+        
+        estimated_capacity = int(chars_per_line * total_lines * efficiency_factor)
+        
+        self.logger.debug(f"Text capacity estimation: {estimated_capacity} chars "
+                         f"({chars_per_line} chars/line × {total_lines} lines × {efficiency_factor} efficiency)")
+        
+        return max(estimated_capacity, 500)  # Minimum 500 chars per page for small images
+
+    def _split_segments_into_pages(self, segments: List[TextSegment], capacity_per_page: int) -> List[List[TextSegment]]:
+        """
+        Split text segments into pages based on estimated capacity.
+        Handles large segments by splitting them if needed.
+        
+        Args:
+            segments: List of TextSegment objects
+            capacity_per_page: Estimated character capacity per page
+            
+        Returns:
+            List of pages, each containing a list of segments
+        """
+        pages = []
+        current_page = []
+        current_page_chars = 0
+        
+        for segment in segments:
+            segment_length = len(segment.text)
+            
+            # If this single segment is larger than page capacity, split it
+            if segment_length > capacity_per_page:
+                # Finish current page if it has content
+                if current_page:
+                    pages.append(current_page)
+                    current_page = []
+                    current_page_chars = 0
+                
+                # Split the large segment into chunks
+                text = segment.text
+                while text:
+                    # Take a chunk that fits the capacity
+                    chunk_size = min(capacity_per_page, len(text))
+                    
+                    # Try to break at word boundaries if possible
+                    if chunk_size < len(text):
+                        # Look for a space within the last 20% of the chunk
+                        search_start = max(0, int(chunk_size * 0.8))
+                        space_pos = text.rfind(' ', search_start, chunk_size)
+                        if space_pos > search_start:
+                            chunk_size = space_pos + 1  # Include the space
+                    
+                    chunk_text = text[:chunk_size]
+                    text = text[chunk_size:]
+                    
+                    # Create new segment with same formatting
+                    chunk_segment = TextSegment(
+                        text=chunk_text,
+                        is_bold=segment.is_bold,
+                        is_italic=segment.is_italic,
+                        is_header=segment.is_header,
+                        header_level=segment.header_level
+                    )
+                    
+                    # Add as a new page
+                    pages.append([chunk_segment])
+                
+            else:
+                # Normal segment handling
+                # Check if adding this segment would exceed capacity
+                if current_page_chars + segment_length > capacity_per_page and current_page:
+                    # Start new page
+                    pages.append(current_page)
+                    current_page = [segment]
+                    current_page_chars = segment_length
+                else:
+                    # Add to current page
+                    current_page.append(segment)
+                    current_page_chars += segment_length
+        
+        # Add the last page if it has content
+        if current_page:
+            pages.append(current_page)
+        
+        # Log pagination statistics
+        if pages:
+            page_sizes = [sum(len(s.text) for s in page) for page in pages]
+            avg_size = sum(page_sizes) / len(pages)
+            min_size = min(page_sizes)
+            max_size = max(page_sizes)
+            
+            self.logger.debug(f"Split {len(segments)} segments into {len(pages)} pages")
+            self.logger.debug(f"Page sizes - avg: {avg_size:.0f}, min: {min_size}, max: {max_size} chars")
+            
+            # Warn about very small pages (less than 20% of capacity)
+            small_pages = [i+1 for i, size in enumerate(page_sizes) if size < capacity_per_page * 0.2]
+            if small_pages:
+                self.logger.warning(f"Small pages detected (< 20% capacity): {small_pages[:5]}{'...' if len(small_pages) > 5 else ''}")
+        
+        return pages
+
+    def segments_to_images(
+        self,
+        segments: List[TextSegment],
+        config: RenderingConfig,
+        output_dir: Optional[str] = None,
+        unique_id: Optional[str] = None
+    ) -> List[Path]:
+        """
+        Convert TextSegment objects to images using PIL/Pillow.
+        
+        Args:
+            segments: List of TextSegment objects with formatting
+            config: Rendering configuration
+            output_dir: Output directory for images
+            unique_id: Unique identifier for this rendering
+            
+        Returns:
+            List of paths to rendered images
+        """
+        try:
+            from PIL import Image, ImageDraw, ImageFont
+            
+            # Setup output directory
+            if output_dir is None:
+                output_dir = tempfile.mkdtemp(prefix="glyph_pil_")
+            else:
+                os.makedirs(output_dir, exist_ok=True)
+            
+            output_dir = Path(output_dir)
+            unique_id = unique_id or "render"
+            
+            # Load fonts
+            fonts = self._load_fonts(config)
+            
+            # Estimate text capacity and split into pages if needed
+            capacity_per_page = self._estimate_text_capacity(config, fonts)
+            total_chars = sum(len(segment.text) for segment in segments)
+            
+            self.logger.debug(f"Text pagination: {total_chars} total chars, "
+                             f"{capacity_per_page} chars/page capacity")
+            
+            if total_chars > capacity_per_page:
+                # Split into multiple pages
+                pages = self._split_segments_into_pages(segments, capacity_per_page)
+                self.logger.info(f"Text split into {len(pages)} pages for rendering")
+            else:
+                # Single page
+                pages = [segments]
+                self.logger.debug("Text fits in single page")
+            
+            # Render each page
+            image_paths = []
+            
+            for page_idx, page_segments in enumerate(pages):
+                self.logger.debug(f"Rendering page {page_idx + 1}/{len(pages)} "
+                               f"({len(page_segments)} segments, "
+                               f"{sum(len(s.text) for s in page_segments)} chars)")
+                
+                # Calculate text layout for this page
+                columns_data = self._layout_text(page_segments, fonts, config)
+                
+                # Get target dimensions
+                img_width, img_height = self._get_effective_dimensions(config)
+                
+                # Create image with exact dimensions
+                # Use white background for better compression
+                image = Image.new('RGB', (img_width, img_height), 'white')
+                
+                # Set DPI information on the image
+                dpi_tuple = (config.dpi, config.dpi)
+                image.info['dpi'] = dpi_tuple
+                
+                draw = ImageDraw.Draw(image)
+                
+                # Render text for this page
+                self._render_text_to_image(draw, columns_data, fonts, config)
+                
+                # Save image with page number
+                image_path = output_dir / f"{unique_id}_page_{page_idx + 1}.png"
+                image.save(image_path, 'PNG', optimize=True, dpi=dpi_tuple)
+                image_paths.append(image_path)
+                
+                self.logger.debug(f"Generated page {page_idx + 1}: {image_path} ({img_width}x{img_height})")
+            
+            self.logger.info(f"Rendered {len(pages)} pages total")
+            return image_paths
+            
+        except Exception as e:
+            self.logger.error(f"PIL text rendering failed: {e}")
+            raise RenderingError(f"Failed to render text with PIL: {e}") from e
+    
+    def _load_fonts(self, config: RenderingConfig) -> dict:
+        """Load fonts for different styles with custom font support."""
+        from PIL import ImageFont
+        import os
+        
+        fonts = {}
+        font_size = config.font_size
+        
+        self.logger.debug(f"Loading fonts with size: {font_size}")
+        
+        try:
+            # Check if custom font path is specified
+            if config.font_path and os.path.exists(config.font_path):
+                self.logger.info(f"Using custom font path: {config.font_path}")
+                try:
+                    # Try to load the custom font for all styles
+                    base_font = ImageFont.truetype(config.font_path, font_size)
+                    fonts['regular'] = base_font
+                    fonts['bold'] = base_font  # Use same font for all styles
+                    fonts['italic'] = base_font
+                    fonts['bold_italic'] = base_font
+                    
+                    self.logger.info(f"Successfully loaded custom font: {config.font_path}")
+                    return fonts
+                except Exception as e:
+                    self.logger.warning(f"Failed to load custom font {config.font_path}: {e}. Falling back to system fonts.")
+            
+            # Check if custom font name is specified
+            if config.font_name:
+                self.logger.info(f"Trying to load font by name: {config.font_name}")
+                
+                # Special handling for OCRB font family
+                if config.font_name.upper() == "OCRB":
+                    return self._load_ocrb_font_family(font_size)
+                
+                # Special handling for OCRA font family
+                if config.font_name.upper() == "OCRA":
+                    return self._load_ocra_font_family(font_size)
+                
+                try:
+                    # Try to load by name (works on some systems)
+                    base_font = ImageFont.truetype(config.font_name, font_size)
+                    fonts['regular'] = base_font
+                    fonts['bold'] = base_font
+                    fonts['italic'] = base_font
+                    fonts['bold_italic'] = base_font
+                    
+                    self.logger.info(f"Successfully loaded font by name: {config.font_name}")
+                    return fonts
+                except Exception as e:
+                    self.logger.warning(f"Failed to load font by name {config.font_name}: {e}. Falling back to system fonts.")
+            
+            # Fall back to system fonts with good readability
+            self.logger.info("Using default system fonts")
+            font_paths = {
+                'regular': [
+                    '/System/Library/Fonts/Helvetica.ttc',  # macOS
+                    '/System/Library/Fonts/Arial.ttf',
+                    '/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf',  # Linux
+                    'arial.ttf'  # Windows
+                ],
+                'bold': [
+                    '/System/Library/Fonts/Helvetica.ttc',  # Contains bold variant
+                    '/System/Library/Fonts/Arial Bold.ttf',
+                    '/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf',  # Linux
+                    'arialbd.ttf'  # Windows
+                ],
+                'italic': [
+                    '/System/Library/Fonts/Helvetica.ttc',  # Contains italic variant
+                    '/System/Library/Fonts/Arial Italic.ttf',
+                    '/usr/share/fonts/truetype/dejavu/DejaVuSans-Oblique.ttf',  # Linux
+                    'ariali.ttf'  # Windows
+                ],
+                'bold_italic': [
+                    '/System/Library/Fonts/Helvetica.ttc',  # Contains bold italic variant
+                    '/System/Library/Fonts/Arial Bold Italic.ttf',
+                    '/usr/share/fonts/truetype/dejavu/DejaVuSans-BoldOblique.ttf',  # Linux
+                    'arialbi.ttf'  # Windows
+                ]
+            }
+            
+            # Load regular font
+            fonts['regular'] = self._load_font_from_paths(font_paths['regular'], font_size, 'regular')
+            
+            # Load bold font
+            fonts['bold'] = self._load_font_from_paths(font_paths['bold'], font_size, 'bold')
+            
+            # Load italic font  
+            fonts['italic'] = self._load_font_from_paths(font_paths['italic'], font_size, 'italic')
+            
+            # Load bold italic font
+            fonts['bold_italic'] = self._load_font_from_paths(font_paths['bold_italic'], font_size, 'bold_italic')
+            
+            self.logger.info(f"Loaded PIL fonts: regular={fonts['regular'] is not None}, "
+                           f"bold={fonts['bold'] is not None}, italic={fonts['italic'] is not None}")
+            
+        except Exception as e:
+            self.logger.warning(f"Failed to load system fonts, using default: {e}")
+            # Fallback to default font
+            fonts = {
+                'regular': ImageFont.load_default(),
+                'bold': ImageFont.load_default(),
+                'italic': ImageFont.load_default(),
+                'bold_italic': ImageFont.load_default()
+            }
+        
+        return fonts
+    
+    def _load_ocrb_font_family(self, font_size: int) -> dict:
+        """Load OCRB font family with proper regular and italic variants."""
+        from PIL import ImageFont
+        import os
+        
+        fonts = {}
+        
+        # Define paths to OCRB font files relative to the package
+        try:
+            # Get the path to the assets directory
+            assets_dir = Path(__file__).parent.parent / "assets"
+            ocrb_regular_path = assets_dir / "OCRB.ttf"
+            ocrb_italic_path = assets_dir / "OCRBL.ttf"
+            
+            self.logger.info(f"Loading OCRB font family from assets directory")
+            
+            # Load regular font
+            if ocrb_regular_path.exists():
+                fonts['regular'] = ImageFont.truetype(str(ocrb_regular_path), font_size)
+                fonts['bold'] = ImageFont.truetype(str(ocrb_regular_path), font_size)  # Use regular for bold
+                self.logger.info(f"Loaded OCRB regular font: {ocrb_regular_path}")
+            else:
+                self.logger.warning(f"OCRB regular font not found at: {ocrb_regular_path}")
+                fonts['regular'] = ImageFont.load_default()
+                fonts['bold'] = ImageFont.load_default()
+            
+            # Load italic font (OCRBL)
+            if ocrb_italic_path.exists():
+                fonts['italic'] = ImageFont.truetype(str(ocrb_italic_path), font_size)
+                fonts['bold_italic'] = ImageFont.truetype(str(ocrb_italic_path), font_size)  # Use italic for bold-italic
+                self.logger.info(f"Loaded OCRB italic font: {ocrb_italic_path}")
+            else:
+                self.logger.warning(f"OCRB italic font not found at: {ocrb_italic_path}")
+                fonts['italic'] = fonts['regular']  # Fall back to regular
+                fonts['bold_italic'] = fonts['regular']
+            
+            self.logger.info("Successfully loaded OCRB font family with proper italic support")
+            self.using_ocrb_fonts = True  # Set flag for special bold handling
+            return fonts
+            
+        except Exception as e:
+            self.logger.error(f"Failed to load OCRB font family: {e}")
+            # Fall back to default fonts
+            return {
+                'regular': ImageFont.load_default(),
+                'bold': ImageFont.load_default(),
+                'italic': ImageFont.load_default(),
+                'bold_italic': ImageFont.load_default()
+            }
+    
+    def _load_ocra_font_family(self, font_size: int) -> dict:
+        """Load OCRA font family with special handling (no italic variant available)."""
+        from PIL import ImageFont
+        import os
+        
+        fonts = {}
+        
+        # Define path to OCRA font file relative to the package
+        try:
+            # Get the path to the assets directory
+            assets_dir = Path(__file__).parent.parent / "assets"
+            ocra_regular_path = assets_dir / "OCRA.ttf"
+            
+            self.logger.info(f"Loading OCRA font family from assets directory")
+            
+            # Load regular font (OCRA only has one variant)
+            if ocra_regular_path.exists():
+                fonts['regular'] = ImageFont.truetype(str(ocra_regular_path), font_size)
+                fonts['bold'] = ImageFont.truetype(str(ocra_regular_path), font_size)  # Use regular for bold
+                fonts['italic'] = ImageFont.truetype(str(ocra_regular_path), font_size)  # Use regular for italic
+                fonts['bold_italic'] = ImageFont.truetype(str(ocra_regular_path), font_size)  # Use regular for bold-italic
+                self.logger.info(f"Loaded OCRA regular font: {ocra_regular_path}")
+            else:
+                self.logger.warning(f"OCRA regular font not found at: {ocra_regular_path}")
+                fonts['regular'] = ImageFont.load_default()
+                fonts['bold'] = ImageFont.load_default()
+                fonts['italic'] = ImageFont.load_default()
+                fonts['bold_italic'] = ImageFont.load_default()
+            
+            self.logger.info("Successfully loaded OCRA font family (using regular font for all styles)")
+            self.using_ocra_fonts = True  # Set flag for special bold handling
+            return fonts
+            
+        except Exception as e:
+            self.logger.error(f"Failed to load OCRA font family: {e}")
+            # Fall back to default fonts
+            return {
+                'regular': ImageFont.load_default(),
+                'bold': ImageFont.load_default(),
+                'italic': ImageFont.load_default(),
+                'bold_italic': ImageFont.load_default()
+            }
+    
+    def _load_font_from_paths(self, paths: List[str], size: int, style: str = 'regular'):
+        """Try to load font from a list of possible paths."""
+        from PIL import ImageFont
+        
+        for path in paths:
+            try:
+                if Path(path).exists():
+                    # For .ttc files (TrueType Collections), try different indices for different styles
+                    if path.endswith('.ttc'):
+                        # Map styles to font indices in Helvetica.ttc
+                        style_indices = {
+                            'regular': 0,      # Helvetica Regular
+                            'bold': 1,         # Helvetica Bold  
+                            'italic': 2,       # Helvetica Oblique
+                            'bold_italic': 3   # Helvetica Bold Oblique
+                        }
+                        index = style_indices.get(style, 0)
+                        try:
+                            font = ImageFont.truetype(path, size, index=index)
+                            self.logger.debug(f"Loaded {style} font from {path} index {index}")
+                            return font
+                        except Exception as e:
+                            self.logger.warning(f"Failed to load {style} from {path} index {index}: {e}")
+                            continue
+                    else:
+                        font = ImageFont.truetype(path, size)
+                        self.logger.debug(f"Loaded {style} font from {path}")
+                        return font
+            except Exception as e:
+                self.logger.debug(f"Failed to load font from {path}: {e}")
+                continue
+        
+        # Fallback to default
+        self.logger.warning(f"Using default font for {style}")
+        return ImageFont.load_default()
+    
+    def _layout_text(self, segments: List[TextSegment], fonts: dict, config: RenderingConfig) -> List[List[List[dict]]]:
+        """
+        Layout text segments into columns and lines with word wrapping.
+        
+        Returns:
+            List of columns, where each column is a list of lines,
+            and each line is a list of text chunks with formatting info
+        """
+        lines = []
+        current_line = []
+        current_line_width = 0
+        
+        # Calculate available width using effective dimensions
+        img_width, img_height = self._get_effective_dimensions(config)
+        margin_x = config.margin_x
+        available_width = img_width - 2 * margin_x
+        
+        # Handle multi-column layout properly
+        columns = max(1, config.columns)
+        column_gap = config.column_gap if columns > 1 else 0
+        column_width = (available_width - (columns - 1) * column_gap) / columns
+        
+        self.logger.debug(f"Layout: img_width={img_width}, available_width={available_width}, "
+                         f"columns={columns}, column_width={column_width}")
+        
+        for segment in segments:
+            if segment.text == "\n":
+                # Force line break
+                if current_line:
+                    lines.append(current_line)
+                    current_line = []
+                    current_line_width = 0
+                continue
+            
+            # Get appropriate font
+            font = self._get_font_for_segment(segment, fonts)
+            
+            # Handle space-only segments (like "   " from single newlines)
+            if segment.text.strip() == "":
+                # This is a space-only segment - preserve it
+                space_width = self._get_text_width(segment.text, font, segment)
+                if current_line_width + space_width <= column_width:
+                    # Add spaces to current line
+                    current_line.append({
+                        'text': segment.text,
+                        'font': font,
+                        'segment': segment
+                    })
+                    current_line_width += space_width
+                else:
+                    # Start new line with spaces
+                    if current_line:
+                        lines.append(current_line)
+                    current_line = [{
+                        'text': segment.text,
+                        'font': font,
+                        'segment': segment
+                    }]
+                    current_line_width = space_width
+                continue
+            
+            # Split segment text into words but preserve spaces more carefully
+            # Use a simple approach: split on spaces but keep track of them
+            import re
+            
+            # Split while preserving spaces - use regex to capture spaces
+            parts = re.split(r'(\s+)', segment.text)
+            
+            for part in parts:
+                if not part:  # Skip empty parts
+                    continue
+                    
+                part_width = self._get_text_width(part, font, segment)
+                
+                # Check if part fits on current line
+                if current_line_width + part_width <= column_width or not current_line:
+                    # Add to current line
+                    current_line.append({
+                        'text': part,
+                        'font': font,
+                        'segment': segment
+                    })
+                    current_line_width += part_width
+                else:
+                    # Start new line
+                    if current_line:
+                        lines.append(current_line)
+                    current_line = [{
+                        'text': part,
+                        'font': font,
+                        'segment': segment
+                    }]
+                    current_line_width = part_width
+        
+        # Add final line
+        if current_line:
+            lines.append(current_line)
+        
+        # Now distribute lines among columns
+        if columns == 1:
+            return [lines]  # Single column
+        
+        # Multi-column: distribute lines evenly
+        column_data = [[] for _ in range(columns)]
+        lines_per_column = len(lines) // columns
+        extra_lines = len(lines) % columns
+        
+        line_idx = 0
+        for col in range(columns):
+            lines_in_this_column = lines_per_column + (1 if col < extra_lines else 0)
+            column_data[col] = lines[line_idx:line_idx + lines_in_this_column]
+            line_idx += lines_in_this_column
+        
+        return column_data
+    
+    def _get_font_for_segment(self, segment: TextSegment, fonts: dict):
+        """Get the appropriate font for a text segment."""
+        if segment.is_bold and segment.is_italic:
+            return fonts['bold_italic']
+        elif segment.is_header or segment.is_bold:
+            # Headers should ONLY be bold, never italic
+            return fonts['bold']
+        elif segment.is_italic:
+            return fonts['italic']
+        else:
+            return fonts['regular']
+    
+    def _get_text_width(self, text: str, font, segment=None) -> int:
+        """Get the width of text in pixels, accounting for stroke effects."""
+        from PIL import Image, ImageDraw
+        
+        if not text:
+            return 0
+            
+        # Create temporary image to measure text
+        temp_img = Image.new('RGB', (1, 1))
+        temp_draw = ImageDraw.Draw(temp_img)
+        
+        try:
+            bbox = temp_draw.textbbox((0, 0), text, font=font)
+            base_width = bbox[2] - bbox[0]
+        except AttributeError:
+            # Fallback for older PIL versions
+            try:
+                base_width = temp_draw.textsize(text, font=font)[0]
+            except AttributeError:
+                # Ultimate fallback - estimate based on font size
+                base_width = len(text) * config.font_size * 0.6
+        
+        # Add extra width for OCRB and OCRA bold overlay effect
+        if ((self.using_ocrb_fonts or self.using_ocra_fonts) and segment and 
+            (segment.is_bold or segment.is_header) and not segment.is_italic):
+            # Add width for enhanced horizontal overlays (0.6 pixel max offset)
+            return int(base_width + 0.6)
+        
+        return base_width
+    
+    def _get_text_height(self, font) -> int:
+        """Get the height of text in pixels."""
+        from PIL import Image, ImageDraw
+        
+        # Create temporary image to measure text
+        temp_img = Image.new('RGB', (1, 1))
+        temp_draw = ImageDraw.Draw(temp_img)
+        
+        try:
+            bbox = temp_draw.textbbox((0, 0), "Ag", font=font)
+            return bbox[3] - bbox[1]
+        except:
+            # Fallback for older PIL versions
+            return temp_draw.textsize("Ag", font=font)[1]
+    
+    def _calculate_image_size(self, columns_data: List[List[List[dict]]], fonts: dict, config: RenderingConfig) -> Tuple[int, int]:
+        """Calculate the required image size for multi-column layout with DPI scaling."""
+        if not columns_data or not any(columns_data):
+            return (config.page_width, self._scale_dimension(100, config))  # Minimum size
+        
+        # Calculate width (use page width - no scaling needed as it's already in points)
+        width = config.page_width
+        
+        # Calculate height based on the tallest column with DPI scaling
+        scaled_line_height = self._scale_dimension(config.line_height, config)
+        scaled_line_spacing = int(scaled_line_height * 1.3)  # 30% more space between lines
+        scaled_margin_y = self._scale_dimension(config.margin_y, config)
+        scaled_padding = self._scale_dimension(20, config)
+        
+        max_lines_in_column = max(len(column) for column in columns_data)
+        total_height = max_lines_in_column * scaled_line_spacing + 2 * scaled_margin_y + scaled_padding
+        
+        # Add some padding
+        min_height = self._scale_dimension(100, config)
+        height = max(total_height, min_height)
+        
+        self.logger.debug(f"Image size calculation: columns={len(columns_data)}, "
+                         f"max_lines_in_column={max_lines_in_column}, "
+                         f"scaled_line_height={scaled_line_height}, scaled_line_spacing={scaled_line_spacing}, height={height}")
+        
+        return (int(width), int(height))
+    
+    def _render_text_to_image(self, draw, columns_data: List[List[List[dict]]], fonts: dict, config: RenderingConfig):
+        """Render text columns to the image with proper multi-column support."""
+        line_height = config.line_height
+        line_spacing = int(line_height * 1.3)  # Match the spacing calculation
+        
+        # Calculate column layout using effective dimensions
+        columns = len(columns_data)
+        img_width, img_height = self._get_effective_dimensions(config)
+        available_width = img_width - 2 * config.margin_x
+        column_gap = config.column_gap if columns > 1 else 0
+        column_width = (available_width - (columns - 1) * column_gap) / columns
+        
+        # Render each column
+        for col_idx, column_lines in enumerate(columns_data):
+            # Calculate column x position
+            column_x = config.margin_x + col_idx * (column_width + column_gap)
+            
+            # Render this column
+            y = config.margin_y
+            for line in column_lines:
+                x = column_x
+                
+                for chunk in line:
+                    text = chunk['text']
+                    font = chunk['font']
+                    segment = chunk.get('segment')
+                    
+                    # Draw text with special handling for OCRB bold
+                    self._draw_text_with_effects(draw, (x, y), text, font, segment)
+                    
+                    # Move x position
+                    x += self._get_text_width(text, font, segment)
+                
+                # Move to next line
+                y += line_spacing
+    
+    def _draw_text_with_effects(self, draw, position, text, font, segment):
+        """Draw text with special effects for OCRB and OCRA bold text."""
+        x, y = position
+        
+        # Check if this should be bold and we're using OCRB or OCRA fonts
+        if ((self.using_ocrb_fonts or self.using_ocra_fonts) and segment and 
+            (segment.is_bold or segment.is_header) and not segment.is_italic):
+            
+            # Use improved bold effect for OCRB and OCRA text
+            # Method: Enhanced multiple overlays for more visible bold effect
+            try:
+                # Draw the base text
+                draw.text((x, y), text, font=font, fill='black')
+                
+                # Add horizontal overlays for width (increased for 10% more visibility)
+                draw.text((x + 0.2, y), text, font=font, fill='black')
+                draw.text((x + 0.4, y), text, font=font, fill='black')
+                draw.text((x + 0.6, y), text, font=font, fill='black')  # Additional overlay
+                
+                # Add vertical overlays for height (enhanced for better visibility)
+                draw.text((x, y - 0.1), text, font=font, fill='black')
+                draw.text((x, y + 0.1), text, font=font, fill='black')  # Bottom overlay
+                
+                # Add diagonal overlays for smoother appearance
+                draw.text((x + 0.1, y - 0.05), text, font=font, fill='black')
+                
+            except Exception as e:
+                # Fallback: just draw regular text if anything fails
+                font_type = "OCRB" if self.using_ocrb_fonts else "OCRA"
+                self.logger.debug(f"{font_type} bold effect failed, using regular text: {e}")
+                draw.text((x, y), text, font=font, fill='black')
+        else:
+            # Regular text drawing
+            draw.text((x, y), text, font=font, fill='black')