doctra 0.4.3__py3-none-any.whl → 0.5.0__py3-none-any.whl

doctra/cli/main.py

@@ -28,6 +28,7 @@ except ImportError:
 
 # Import additional modules
 from doctra.engines.layout.paddle_layout import PaddleLayoutEngine
+from doctra.cli.utils import validate_vlm_config, handle_keyboard_interrupt
 from doctra.engines.image_restoration import DocResEngine
 
 
@@ -85,7 +86,7 @@ def vlm_options(func):
     """
     func = click.option('--use-vlm/--no-vlm', default=False,
                         help='Use Vision Language Model for table/chart extraction')(func)
-    func = click.option('--vlm-provider', type=click.Choice(['gemini', 'openai']), default='gemini',
+    func = click.option('--vlm-provider', type=click.Choice(['gemini', 'openai', 'anthropic', 'openrouter', 'ollama']), default='gemini',
                         help='VLM provider to use (default: gemini)')(func)
     func = click.option('--vlm-model', type=str, default=None,
                         help='Model name to use (defaults to provider-specific defaults)')(func)
@@ -141,23 +142,6 @@ def ocr_options(func):
     return func
 
 
-def validate_vlm_config(use_vlm: bool, vlm_api_key: Optional[str]) -> None:
-    """
-    Validate VLM configuration and exit with error if invalid.
-    
-    Checks if VLM is enabled but no API key is provided, and exits
-    with an appropriate error message if the configuration is invalid.
-
-    :param use_vlm: Whether VLM processing is enabled
-    :param vlm_api_key: The VLM API key (can be None if VLM is disabled)
-    :return: None
-    :raises SystemExit: If VLM is enabled but no API key is provided
-    """
-    if use_vlm and not vlm_api_key:
-        click.echo("❌ Error: VLM API key is required when using --use-vlm", err=True)
-        click.echo("   Set the VLM_API_KEY environment variable or use --vlm-api-key", err=True)
-        click.echo("   Example: export VLM_API_KEY=your_api_key", err=True)
-        sys.exit(1)
 
 
 @cli.command()
@@ -212,7 +196,7 @@ def parse(pdf_path: Path, output_dir: Optional[Path], use_vlm: bool,
     :param verbose: Whether to enable verbose output
     :return: None
     """
-    validate_vlm_config(use_vlm, vlm_api_key)
+    validate_vlm_config(use_vlm, vlm_api_key, vlm_provider)
 
     if verbose:
         click.echo(f"🔍 Starting full PDF parsing...")
@@ -350,7 +334,7 @@ def enhance(pdf_path: Path, output_dir: Optional[Path], restoration_task: str,
     :param verbose: Whether to enable verbose output
     :return: None
     """
-    validate_vlm_config(use_vlm, vlm_api_key)
+    validate_vlm_config(use_vlm, vlm_api_key, vlm_provider)
 
     if verbose:
         click.echo(f"🔍 Starting enhanced PDF parsing with DocRes...")
@@ -488,7 +472,7 @@ def charts(pdf_path: Path, output_dir: Path, use_vlm: bool, vlm_provider: str,
     :param verbose: Whether to enable verbose output
     :return: None
     """
-    validate_vlm_config(use_vlm, vlm_api_key)
+    validate_vlm_config(use_vlm, vlm_api_key, vlm_provider)
 
     if verbose:
         click.echo(f"📊 Starting chart extraction...")
@@ -564,7 +548,7 @@ def tables(pdf_path: Path, output_dir: Path, use_vlm: bool, vlm_provider: str,
     :param verbose: Whether to enable verbose output
     :return: None
     """
-    validate_vlm_config(use_vlm, vlm_api_key)
+    validate_vlm_config(use_vlm, vlm_api_key, vlm_provider)
 
     if verbose:
         click.echo(f"📋 Starting table extraction...")
@@ -642,7 +626,7 @@ def both(pdf_path: Path, output_dir: Path, use_vlm: bool, vlm_provider: str,
     :param verbose: Whether to enable verbose output
     :return: None
     """
-    validate_vlm_config(use_vlm, vlm_api_key)
+    validate_vlm_config(use_vlm, vlm_api_key, vlm_provider)
 
     if verbose:
         click.echo(f"📊📋 Starting chart and table extraction...")
@@ -972,6 +956,9 @@ def info():
     click.echo("\nVLM Providers:")
     click.echo("  • Gemini (Google) - gemini-2.5-pro, gemini-2.5-flash, gemini-2.5-flash-lite, gemini-2.0-flash")
     click.echo("  • OpenAI - gpt-5, gpt-5-mini, gpt-4.1, gpt-4.1-mini, gpt-4o")
+    click.echo("  • Anthropic - claude-opus-4-1, claude-3.5-sonnet, claude-3-haiku")
+    click.echo("  • OpenRouter - x-ai/grok-4, meta-llama/llama-3.1-405b-instruct")
+    click.echo("  • Ollama (Local) - llava:latest, gemma3:latest, llama3.2-vision:latest")
 
     # Available layout models
     click.echo("\nLayout Detection Models:")

doctra/cli/utils.py

@@ -13,20 +13,21 @@ from pathlib import Path
 from doctra.utils.progress import create_beautiful_progress_bar, create_notebook_friendly_bar
 
 
-def validate_vlm_config(use_vlm: bool, vlm_api_key: Optional[str]) -> None:
+def validate_vlm_config(use_vlm: bool, vlm_api_key: Optional[str], vlm_provider: str = "gemini") -> None:
     """
     Validate VLM configuration and exit with error if invalid.
     
-    Checks if VLM is enabled but no API key is provided, and exits
+    Checks if VLM is enabled but no API key is provided (except for Ollama), and exits
     with an appropriate error message if the configuration is invalid.
 
     :param use_vlm: Whether VLM processing is enabled
-    :param vlm_api_key: The VLM API key (can be None if VLM is disabled)
+    :param vlm_api_key: The VLM API key (can be None if VLM is disabled or using Ollama)
+    :param vlm_provider: VLM provider name (default: "gemini")
     :return: None
-    :raises SystemExit: If VLM is enabled but no API key is provided
+    :raises SystemExit: If VLM is enabled but no API key is provided (except for Ollama)
     """
-    if use_vlm and not vlm_api_key:
-        click.echo("❌ Error: VLM API key is required when using --use-vlm", err=True)
+    if use_vlm and vlm_provider != "ollama" and not vlm_api_key:
+        click.echo("❌ Error: VLM API key is required when using --use-vlm (except for Ollama)", err=True)
         click.echo("   Set the VLM_API_KEY environment variable or use --vlm-api-key", err=True)
         click.echo("   Example: export VLM_API_KEY=your_api_key", err=True)
         sys.exit(1)

doctra/engines/vlm/provider.py

@@ -1,86 +1,257 @@
-from __future__ import annotations
-
-# --- keep these imports to match your snippet style ---
-import io
-import PIL
-import openai
-import outlines
-from pydantic import BaseModel
-from google.genai import Client
-from outlines.inputs import Image
-from anthropic import Anthropic
-# ------------------------------------------------------
-
-def make_model(
-    vlm_provider: str | None = "gemini",
-    vlm_model: str | None = None,
-    *,
-    api_key: str | None = None,
-):
-    """
-    Build a callable Outlines model for VLM processing.
-    
-    Creates an Outlines model instance configured for Gemini, OpenAI, Anthropic, or OpenRouter
-    providers. Only one backend is active at a time, with Gemini as the default.
-
-    :param vlm_provider: VLM provider to use ("gemini", "openai", or "anthropic", default: "gemini")
-    :param vlm_model: Model name to use (defaults to provider-specific defaults)
-    :param api_key: API key for the VLM provider (required for all providers)
-    :return: Configured Outlines model instance
-    :raises ValueError: If provider is unsupported or API key is missing
-    """
-    vlm_provider = (vlm_provider or "gemini").lower()
-    
-    # Set default models if not provided
-    if vlm_model is None:
-        if vlm_provider == "gemini":
-            vlm_model = "gemini-2.5-pro"
-        elif vlm_provider == "openai":
-            vlm_model = "gpt-5"
-        elif vlm_provider == "anthropic":
-            vlm_model = "claude-opus-4-1"
-        elif vlm_provider == "openrouter":
-            vlm_model = "x-ai/grok-4"
-
-    if vlm_provider == "gemini":
-        if not api_key:
-            raise ValueError("Gemini provider requires api_key to be passed to make_model(...).")
-        # Create the model (exactly like your snippet)
-        return outlines.from_gemini(
-            Client(api_key=api_key),
-            vlm_model,
-        )
-
-    if vlm_provider == "openai":
-        if not api_key:
-            raise ValueError("OpenAI provider requires api_key to be passed to make_model(...).")
-        # this part is for the openai models (exactly like your snippet)
-        return outlines.from_openai(
-            openai.OpenAI(api_key=api_key),
-            vlm_model,
-        )
-
-    if vlm_provider == "anthropic":
-        if not api_key:
-            raise ValueError("Anthropic provider requires api_key to be passed to make_model(...).")
-        # Create the Anthropic client and model (exactly like your snippet)
-        client = Anthropic(api_key=api_key)
-        return outlines.from_anthropic(
-            client,
-            vlm_model,
-        )
-
-    if vlm_provider == "openrouter":
-        if not api_key:
-            raise ValueError("OpenRouter provider requires api_key to be passed to make_model(...).")
-        # Create the Anthropic client and model (exactly like your snippet)
-        client = openai.OpenAI(
-            base_url="https://openrouter.ai/api/v1",
-            api_key=api_key,
-        )
-        return outlines.from_openai(
-            client,
-            vlm_model
-        )
-
-    raise ValueError(f"Unsupported provider: {vlm_provider}. Use 'gemini', 'openai', or 'anthropic'.")
+from __future__ import annotations
+
+# --- keep these imports to match your snippet style ---
+import io
+import os
+import PIL
+import openai
+import outlines
+from pydantic import BaseModel
+from google.genai import Client
+from outlines.inputs import Image
+from anthropic import Anthropic
+import ollama
+# ------------------------------------------------------
+
+def make_model(
+    vlm_provider: str | None = "gemini",
+    vlm_model: str | None = None,
+    *,
+    api_key: str | None = None,
+):
+    """
+    Build a callable Outlines model for VLM processing.
+    
+    Creates an Outlines model instance configured for Gemini, OpenAI, Anthropic, OpenRouter, or Ollama
+    providers. Only one backend is active at a time, with Gemini as the default.
+
+    :param vlm_provider: VLM provider to use ("gemini", "openai", "anthropic", "openrouter", or "ollama", default: "gemini")
+    :param vlm_model: Model name to use (defaults to provider-specific defaults)
+    :param api_key: API key for the VLM provider (required for all providers except Ollama)
+    :return: Configured Outlines model instance
+    :raises ValueError: If provider is unsupported or API key is missing
+    """
+    vlm_provider = (vlm_provider or "gemini").lower()
+    
+    # Set default models if not provided
+    if vlm_model is None:
+        if vlm_provider == "gemini":
+            vlm_model = "gemini-2.5-pro"
+        elif vlm_provider == "openai":
+            vlm_model = "gpt-5"
+        elif vlm_provider == "anthropic":
+            vlm_model = "claude-opus-4-1"
+        elif vlm_provider == "openrouter":
+            vlm_model = "x-ai/grok-4"
+        elif vlm_provider == "ollama":
+            vlm_model = "llava:latest"
+
+    if vlm_provider == "gemini":
+        if not api_key:
+            raise ValueError("Gemini provider requires api_key to be passed to make_model(...).")
+        # Create the model (exactly like your snippet)
+        return outlines.from_gemini(
+            Client(api_key=api_key),
+            vlm_model,
+        )
+
+    if vlm_provider == "openai":
+        if not api_key:
+            raise ValueError("OpenAI provider requires api_key to be passed to make_model(...).")
+        # this part is for the openai models (exactly like your snippet)
+        return outlines.from_openai(
+            openai.OpenAI(api_key=api_key),
+            vlm_model,
+        )
+
+    if vlm_provider == "anthropic":
+        if not api_key:
+            raise ValueError("Anthropic provider requires api_key to be passed to make_model(...).")
+        # Create the Anthropic client and model (exactly like your snippet)
+        client = Anthropic(api_key=api_key)
+        return outlines.from_anthropic(
+            client,
+            vlm_model,
+        )
+
+    if vlm_provider == "openrouter":
+        if not api_key:
+            raise ValueError("OpenRouter provider requires api_key to be passed to make_model(...).")
+        # Create the Anthropic client and model (exactly like your snippet)
+        client = openai.OpenAI(
+            base_url="https://openrouter.ai/api/v1",
+            api_key=api_key,
+        )
+        return outlines.from_openai(
+            client,
+            vlm_model
+        )
+
+    if vlm_provider == "ollama":
+        # Ollama doesn't use Outlines, so we return a custom wrapper
+        return OllamaModelWrapper(vlm_model)
+
+    raise ValueError(f"Unsupported provider: {vlm_provider}. Use 'gemini', 'openai', 'anthropic', 'openrouter', or 'ollama'.")
+
+
+class OllamaModelWrapper:
+    """
+    Wrapper class to make Ollama compatible with the Outlines interface.
+    
+    This class provides a callable interface that matches the Outlines model
+    signature, allowing Ollama to be used as a drop-in replacement for other
+    VLM providers in the Doctra framework.
+    """
+    
+    def __init__(self, model_name: str):
+        """
+        Initialize the Ollama model wrapper.
+        
+        :param model_name: Name of the Ollama model to use (e.g., "llava:latest", "gemma3:latest")
+        """
+        self.model_name = model_name
+    
+    def __call__(self, prompt, schema):
+        """
+        Call the Ollama model with the given prompt and schema.
+        
+        :param prompt: List containing [text_prompt, Image] - the text prompt and PIL Image
+        :param schema: Pydantic model class for structured output
+        :return: Structured data object matching the provided schema
+        """
+        if not isinstance(prompt, list) or len(prompt) != 2:
+            raise ValueError("Prompt must be a list with [text, image] format")
+        
+        text_prompt, image = prompt
+        
+        # Convert Image object to bytes for Ollama
+        # The Image object from Outlines might be a PIL Image or a different type
+        try:
+            # Try to get the PIL Image from the Outlines Image object
+            if hasattr(image, 'image'):
+                pil_image = image.image
+            elif hasattr(image, '_image'):
+                pil_image = image._image
+            else:
+                pil_image = image
+            
+            # Convert to bytes
+            img_buffer = io.BytesIO()
+            pil_image.save(img_buffer, format='JPEG')
+            img_bytes = img_buffer.getvalue()
+        except Exception as e:
+            # Try alternative approach - save the image directly to a file
+            import tempfile
+            with tempfile.NamedTemporaryFile(suffix='.jpg', delete=False) as tmp_file:
+                try:
+                    if hasattr(image, 'image'):
+                        image.image.save(tmp_file.name, format='JPEG')
+                    else:
+                        image.save(tmp_file.name, format='JPEG')
+                    with open(tmp_file.name, 'rb') as f:
+                        img_bytes = f.read()
+                    os.unlink(tmp_file.name)
+                except Exception as e2:
+                    raise
+        
+        # Save image to temporary file for Ollama
+        import tempfile
+        with tempfile.NamedTemporaryFile(suffix='.jpg', delete=False) as tmp_file:
+            tmp_file.write(img_bytes)
+            tmp_path = tmp_file.name
+        
+        try:
+            # Call Ollama with the image and prompt
+            response = ollama.chat(
+                messages=[{
+                    "role": "user",
+                    "content": text_prompt,
+                    "images": [tmp_path],
+                }],
+                model=self.model_name,
+                format=schema.model_json_schema(),  # Use Pydantic schema for structured output
+            )
+            
+            # Handle different response formats
+            if 'message' in response and 'content' in response['message']:
+                content = response['message']['content']
+            elif 'response' in response:
+                content = response['response']
+            else:
+                content = str(response)
+            
+            # Try to parse as JSON
+            try:
+                result = schema.model_validate_json(content)
+                return result
+            except Exception as json_error:
+                # Try to extract data manually from text response
+                return self._extract_from_text_response(content, schema)
+            
+        except Exception as e:
+            # Return a default structure to prevent crashes
+            return schema(
+                title="Extraction Failed",
+                description="Failed to extract data from image",
+                headers=["Error"],
+                rows=[["Could not process image"]]
+            )
+        finally:
+            # Clean up temporary file
+            import os
+            try:
+                os.unlink(tmp_path)
+            except:
+                pass
+    
+    def _extract_from_text_response(self, content: str, schema):
+        """
+        Extract structured data from text response when JSON parsing fails.
+        
+        :param content: Text response from Ollama
+        :param schema: Pydantic schema class
+        :return: Structured data object
+        """
+        try:
+            # Try to find JSON in the response
+            import re
+            import json
+            
+            # Look for JSON-like content
+            json_match = re.search(r'\{.*\}', content, re.DOTALL)
+            if json_match:
+                json_str = json_match.group()
+                return schema.model_validate_json(json_str)
+            
+            # If no JSON found, create a basic structure
+            lines = content.split('\n')
+            title = "Extracted Data"
+            description = content[:300] if len(content) > 300 else content
+            
+            # Try to extract headers and rows from text
+            headers = ["Column 1", "Column 2"]  # Default headers
+            rows = [["Data 1", "Data 2"]]  # Default row
+            
+            # Look for table-like patterns
+            for line in lines:
+                if '|' in line and len(line.split('|')) > 2:
+                    # This looks like a table row
+                    cells = [cell.strip() for cell in line.split('|') if cell.strip()]
+                    if len(cells) > 1:
+                        rows.append(cells)
+            
+            return schema(
+                title=title,
+                description=description,
+                headers=headers,
+                rows=rows
+            )
+            
+        except Exception as e:
+            # Return minimal structure
+            return schema(
+                title="Text Extraction",
+                description=content[:300] if len(content) > 300 else content,
+                headers=["Content"],
+                rows=[[content[:100]]]
+            )

doctra/parsers/structured_pdf_parser.py

@@ -88,11 +88,14 @@ class StructuredPDFParser:
         self.use_vlm = use_vlm
         self.vlm = None
         if self.use_vlm:
-            self.vlm = VLMStructuredExtractor(
-                vlm_provider=vlm_provider,
-                vlm_model=vlm_model,
-                api_key=vlm_api_key,
-            )
+            try:
+                self.vlm = VLMStructuredExtractor(
+                    vlm_provider=vlm_provider,
+                    vlm_model=vlm_model,
+                    api_key=vlm_api_key,
+                )
+            except Exception as e:
+                self.vlm = None
 
     def parse(self, pdf_path: str) -> None:
         """

doctra/ui/enhanced_parser_ui.py

@@ -65,7 +65,7 @@ def run_enhanced_parse(
 
     # Validate VLM configuration if VLM is enabled
     if use_vlm:
-        vlm_error = validate_vlm_config(use_vlm, vlm_api_key)
+        vlm_error = validate_vlm_config(use_vlm, vlm_api_key, vlm_provider)
         if vlm_error:
             return (vlm_error, None, [], "", None, None, "")
 
@@ -358,7 +358,7 @@ def create_enhanced_parser_tab() -> Tuple[gr.Tab, dict]:
         # VLM settings
         with gr.Row():
             use_vlm_enhanced = gr.Checkbox(label="Use VLM (optional)", value=False)
-            vlm_provider_enhanced = gr.Dropdown(["gemini", "openai", "anthropic", "openrouter"], value="gemini", label="VLM Provider")
+            vlm_provider_enhanced = gr.Dropdown(["gemini", "openai", "anthropic", "openrouter", "ollama"], value="gemini", label="VLM Provider")
             vlm_api_key_enhanced = gr.Textbox(type="password", label="VLM API Key", placeholder="Optional if VLM disabled")
 
         # Advanced settings accordion

doctra/ui/full_parse_ui.py

@@ -60,7 +60,7 @@ def run_full_parse(
         return ("No file provided.", None, [], [], "")
 
     # Validate VLM configuration
-    vlm_error = validate_vlm_config(use_vlm, vlm_api_key)
+    vlm_error = validate_vlm_config(use_vlm, vlm_api_key, vlm_provider)
     if vlm_error:
         return (vlm_error, None, [], [], "")
 
@@ -429,7 +429,7 @@ def create_full_parse_tab() -> Tuple[gr.Tab, dict]:
         with gr.Row():
             pdf = gr.File(file_types=[".pdf"], label="PDF")
             use_vlm = gr.Checkbox(label="Use VLM (optional)", value=False)
-            vlm_provider = gr.Dropdown(["gemini", "openai", "anthropic", "openrouter"], value="gemini", label="VLM Provider")
+            vlm_provider = gr.Dropdown(["gemini", "openai", "anthropic", "openrouter", "ollama"], value="gemini", label="VLM Provider")
             vlm_api_key = gr.Textbox(type="password", label="VLM API Key", placeholder="Optional if VLM disabled")
 
         # Advanced settings accordion

doctra/ui/tables_charts_ui.py

@@ -48,7 +48,7 @@ def run_extract(
         return ("No file provided.", "", [], [], "")
     
     # Validate VLM configuration
-    vlm_error = validate_vlm_config(use_vlm, vlm_api_key)
+    vlm_error = validate_vlm_config(use_vlm, vlm_api_key, vlm_provider)
     if vlm_error:
         return (vlm_error, "", [], [], "")
 
@@ -334,7 +334,7 @@ def create_tables_charts_tab() -> Tuple[gr.Tab, dict]:
             pdf_e = gr.File(file_types=[".pdf"], label="PDF")
             target = gr.Dropdown(["tables", "charts", "both"], value="both", label="Target")
             use_vlm_e = gr.Checkbox(label="Use VLM (optional)", value=False)
-            vlm_provider_e = gr.Dropdown(["gemini", "openai", "anthropic", "openrouter"], value="gemini", label="VLM Provider")
+            vlm_provider_e = gr.Dropdown(["gemini", "openai", "anthropic", "openrouter", "ollama"], value="gemini", label="VLM Provider")
             vlm_api_key_e = gr.Textbox(type="password", label="VLM API Key", placeholder="Optional if VLM disabled")
         
         # Advanced settings accordion

doctra/ui/ui_helpers.py

@@ -261,21 +261,22 @@ def parse_markdown_by_pages(md_content: str) -> List[Dict[str, Any]]:
     return pages
 
 
-def validate_vlm_config(use_vlm: bool, vlm_api_key: str) -> Optional[str]:
+def validate_vlm_config(use_vlm: bool, vlm_api_key: str, vlm_provider: str = "gemini") -> Optional[str]:
     """
     Validate VLM configuration parameters.
     
     Args:
         use_vlm: Whether VLM is enabled
         vlm_api_key: API key for VLM provider
+        vlm_provider: VLM provider name (default: "gemini")
         
     Returns:
         Error message if validation fails, None if valid
     """
-    if use_vlm and not vlm_api_key:
-        return "❌ Error: VLM API key is required when using VLM"
+    if use_vlm and vlm_provider != "ollama" and not vlm_api_key:
+        return "❌ Error: VLM API key is required when using VLM (except for Ollama)"
     
-    if use_vlm and vlm_api_key:
+    if use_vlm and vlm_api_key and vlm_provider != "ollama":
         # Basic API key validation
         if len(vlm_api_key.strip()) < 10:
             return "❌ Error: VLM API key appears to be too short or invalid"

doctra/version.py

@@ -1,2 +1,2 @@
 """Version information for Doctra."""
-__version__ = '0.4.3'
+__version__ = '0.5.0'

{doctra-0.4.3.dist-info → doctra-0.5.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: doctra
-Version: 0.4.3
+Version: 0.5.0
 Summary: Parse, extract, and analyze documents with ease
 Home-page: https://github.com/AdemBoukhris457/Doctra
 Author: Adem Boukhris
@@ -267,6 +267,7 @@ Dynamic: requires-python
 [![stars](https://img.shields.io/github/stars/AdemBoukhris457/Doctra.svg)](https://github.com/AdemBoukhris457/Doctra)
 [![forks](https://img.shields.io/github/forks/AdemBoukhris457/Doctra.svg)](https://github.com/AdemBoukhris457/Doctra)
 [![PyPI version](https://img.shields.io/pypi/v/doctra)](https://pypi.org/project/doctra/)
+[![Documentation](https://img.shields.io/badge/documentation-available-success)](https://ademboukhris457.github.io/Doctra/index.html)
 </div>
 
 ## 📋 Table of Contents

{doctra-0.4.3.dist-info → doctra-0.5.0.dist-info}/RECORD

@@ -1,8 +1,8 @@
 doctra/__init__.py,sha256=rNLCyODOpaPb_TTP6qmQnuWZJW9JPXrxg1IfKnvb1No,773
-doctra/version.py,sha256=UtaT-N7wXotEga348278k_4dwz6xpN5W57ulX1lo5vU,62
+doctra/version.py,sha256=A8O6Kr44VM50_wm5hKwwcjBUZJFPAk64i5o7DatMRlQ,62
 doctra/cli/__init__.py,sha256=4PTujjYRShOOUlZ7PwuWckShPWLC4v4CYIhJpzgyv1k,911
-doctra/cli/main.py,sha256=_gvG8bm-Mn1tIEw6eJUgqz9dYEo9klXGiJDJzjqgPyo,43503
-doctra/cli/utils.py,sha256=w3Bxyzczcbl_cs1Cea8C3ehv7dkGl_wecprYZXrcGhk,11772
+doctra/cli/main.py,sha256=UhWTatY3qIeutZzVo9syLG2srbs8MZuGaLo5tk9xC_M,43108
+doctra/cli/utils.py,sha256=GKSSGi-JjNXufNekqCysSev7St1t32caYMduy0Tq96s,11971
 doctra/engines/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 doctra/engines/image_restoration/__init__.py,sha256=vzcN6Rw7_U-5jIK2pdo2NlgqdLdXDShigrOGM7QLNEE,263
 doctra/engines/image_restoration/docres_engine.py,sha256=wbo-FWEb6_Twq5KqzjPgGQwcAuFD98uBAiQBEY8vN2A,21592
@@ -15,7 +15,7 @@ doctra/engines/ocr/path_resolver.py,sha256=2_7Nsekt3dCDU3oVsgdr62iMrlAhbGNfYwgh4
 doctra/engines/ocr/pytesseract_engine.py,sha256=Imz2uwju6himkBiS8CH7DLxBRe-LtmMYZiOdb_6PoQw,2911
 doctra/engines/vlm/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 doctra/engines/vlm/outlines_types.py,sha256=fQK6ru7XiXHaa8JPpaTTBaTk_zQ93ZyhFp4SyAnUdVU,1337
-doctra/engines/vlm/provider.py,sha256=aE8Eo1U-8XqAimakNlT0-T4etIyCV8rZ3DwxdqbFeTc,3131
+doctra/engines/vlm/provider.py,sha256=lXQJNxDTxBHSxuEMbF37PjETEokR9o7rc2jLWEH9RnU,9943
 doctra/engines/vlm/service.py,sha256=nygxMe7uTq6Bv70ycBPL59F2a0ESp1Hix4j833p6rUM,4343
 doctra/exporters/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 doctra/exporters/excel_writer.py,sha256=rwyqlH73P7z413BELovQY_pS6IMkkqHEho6mbPrJ2Sk,11857
@@ -26,7 +26,7 @@ doctra/exporters/markdown_writer.py,sha256=L7EjF2MB8jYX7XkZ3a3NeeEC8gnb0qzRPTzIN
 doctra/parsers/__init__.py,sha256=8M6LVzcWGpuTIK_1SMXML3ll7zK1CTHXGI5qXvqdm-A,206
 doctra/parsers/enhanced_pdf_parser.py,sha256=TG4uM_dK80-69y1C99HhSoVInHGwTb-sGJtmHBpZuMY,23756
 doctra/parsers/layout_order.py,sha256=W6b-T11H907RZ2FaZwNvnYhmvH11rpUzxC5yLkdf28k,640
-doctra/parsers/structured_pdf_parser.py,sha256=RSduGt7L5HcoB7JE7zbAjlkvEMk2XQnQhHHD8p7QjQ4,22284
+doctra/parsers/structured_pdf_parser.py,sha256=3jPulhR0agnhP1r9j48WvH53-NZVMhePAmNLzy-_fes,22391
 doctra/parsers/table_chart_extractor.py,sha256=ZD0l2V_8HBdHOAIhMIujfnd5ai3gXsSLL67VMVu3F8A,13905
 doctra/third_party/docres/inference.py,sha256=krD5EQDiqki-5uTMqqHYivhL38sfSOhYgaihI751070,13576
 doctra/third_party/docres/utils.py,sha256=N0ZVmOTB3wsinFlYu5hT84C4_MhWGdc98T8LTG-S9dA,14566
@@ -52,10 +52,10 @@ doctra/ui/__init__.py,sha256=XzOOKeGSBnUREuDQiCIWds1asFSa2nypFQTJXwclROA,85
 doctra/ui/app.py,sha256=I9pX-U3VASGs4kfL6Tv3nDH2tlU4kSv5WrnsNDfYTbQ,2305
 doctra/ui/docres_ui.py,sha256=QMTsNUdw2NGlHK-mYwB-j5i2QXEndYv8Zvc8213jXVA,13034
 doctra/ui/docres_wrapper.py,sha256=BjcY5Xik9UBFPzPL-ONT2GIpTeRrYUXXzuDEq1QE28Q,4498
-doctra/ui/enhanced_parser_ui.py,sha256=OVPwv9yErjg1lL-dEVH5KWrc7YqEP7QmFa80WPhaCX0,20754
-doctra/ui/full_parse_ui.py,sha256=19EsprqeegZAj24KhAWKvyR1hW8HC3nE_f4UFpY-dfQ,18597
-doctra/ui/tables_charts_ui.py,sha256=x0YmERDyfkUruAbHqQ-Kc0_cDOuqf64l_fjBvVOULOI,16534
-doctra/ui/ui_helpers.py,sha256=LthpitCrZOpjXcQvpctyNaDz3T26V06TpAy3r_ChLhY,15584
+doctra/ui/enhanced_parser_ui.py,sha256=oImlFfpjLGs3CpOIUIx_o-1fK7ddUhUCOYW4NUiuJrA,20778
+doctra/ui/full_parse_ui.py,sha256=h-bckQq9FRbVA00l4VQXnzdLgNIrIeAtVVdHkihTPjE,18621
+doctra/ui/tables_charts_ui.py,sha256=ZcRhTbi4iB0tBi3JC-Z3w6AN6dgUOWt9sV_-iJCkaFE,16558
+doctra/ui/ui_helpers.py,sha256=Wx36d5rbUdRXQg98w45DIxH0Hib0mTMEmv2cH3ejyGI,15753
 doctra/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 doctra/utils/bbox.py,sha256=R2-95p0KiWvet3TH27TQVvCar7WJg6z0u3L21iEDF-A,674
 doctra/utils/constants.py,sha256=ZWOvNDrvETbQ_pxHiX7vUW4J5Oj8_qnov0QacUOBizI,189
@@ -66,9 +66,9 @@ doctra/utils/pdf_io.py,sha256=c8EY47Z1iqVtlLFHS_n0qGuXJ5ERFaMUd84ivXV0b9E,706
 doctra/utils/progress.py,sha256=BD9YZqYLZw6yohQnyUV3w9QsQuiIrXM_EqByOSSJsDU,11912
 doctra/utils/quiet.py,sha256=5XPS-1CtJ0sVk6qgSQctdhr_wR8mP1xoJLoUbmkXROA,387
 doctra/utils/structured_utils.py,sha256=vU84dsD8wIlTyMsA9hitorGH-eroQiVuWEpBTQBUT24,1478
-doctra-0.4.3.dist-info/licenses/LICENSE,sha256=HrhfyXIkWY2tGFK11kg7vPCqhgh5DcxleloqdhrpyMY,11558
-doctra-0.4.3.dist-info/METADATA,sha256=YoaPW5G3wdM9zNCb1M_FTM5JmDnUM4MqgS-aVMOBO-M,37033
-doctra-0.4.3.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-doctra-0.4.3.dist-info/entry_points.txt,sha256=4G2RHamA0llCiIXaQQm8EDkVK9JNGKbI7uDnXVFgIaY,47
-doctra-0.4.3.dist-info/top_level.txt,sha256=jI7E8jHci2gP9y0GYaWxlg9jG0O5n3FjHJJPLXDXMds,7
-doctra-0.4.3.dist-info/RECORD,,
+doctra-0.5.0.dist-info/licenses/LICENSE,sha256=HrhfyXIkWY2tGFK11kg7vPCqhgh5DcxleloqdhrpyMY,11558
+doctra-0.5.0.dist-info/METADATA,sha256=tall4Spu8hFtNARaVVCNl9QedT-4VUubsV4oqrMLxoc,37168
+doctra-0.5.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+doctra-0.5.0.dist-info/entry_points.txt,sha256=4G2RHamA0llCiIXaQQm8EDkVK9JNGKbI7uDnXVFgIaY,47
+doctra-0.5.0.dist-info/top_level.txt,sha256=jI7E8jHci2gP9y0GYaWxlg9jG0O5n3FjHJJPLXDXMds,7
+doctra-0.5.0.dist-info/RECORD,,