Convert a file to internal format. Args: file_path: Path to the file to extract Returns:

Convert content to JSON format. Args: specified_fields: List of specific fields to extract (uses Ollama) json

Export as markdown. Returns: The content formatted as markdown

Get the path to a specific model. Args: model_type: Type of model ('layout', 'table', 'nanonets-ocr')

Export as plain text. Returns: The content as plain text

Load and cache a document.

Extract text with layout awareness from image. Args: image_path: Path to the image file Returns:

Get output from cloud API for specific type, with caching.

Create OCR service based on provider configuration. Args: provider: OCR provider name (defaults to config)

Export as HTML. Returns: The content formatted as HTML

Check if Ollama service is available. Returns: True if Ollama is available and responding

Escape HTML special characters.

Run the MCP server.

Count tokens in text using tiktoken or fallback to character estimation.

Get Ollama client with lazy loading.

Send a styled error page.

Check if GPU is available for processing.

Export tables as CSV format. Args: table_index: Which table to export (0-based index). Default is 0 (first table).

Get cached credentials if they exist and are valid.

Get list of supported file formats. Returns: List of supported file extensions

Check if GPU is available for deep learning models. Returns: True if GPU is available, False otherwise

Run a command and handle errors.

Determine if GPU processor should be used based on GPU availability. Returns: True if GPU processor should be used, False otherwise

Clean inline markdown formatting from text.

Get OCR service instance.

Get document chunks that fit within token limit.

Parse content within a section into structured components.

Process inline markdown elements.

Stop the callback server.

Clear cached credentials.

Convert plain text to internal format. Args: text: Plain text to extract Returns: Conver

Export as plain text without GPU processing header.

Convenience function to get an authenticated access token. Args: force_reauth: Force re-authentication even if cached credential

Process the file and return a conversion result. Args: file_path: Path to the file to process Return

Process PDF file with OCR capabilities. Args: file_path: Path to the PDF file Returns: C

Clean up the extracted Word content. Args: content: Raw Word text content Returns: Clean

Clean up the extracted PowerPoint content. Args: content: Raw PowerPoint text content Returns:

Collect consecutive list items.

Convert list lines to HTML list.

Fallback to base JSON conversion method.

Convert pandas DataFrame to markdown table. Args: df: pandas DataFrame pd: pandas module reference

Extract text from numpy array region.

Join text blocks into a coherent paragraph with better text processing.

Cache credentials securely.

Download all required models. Args: force: Force re-download even if models exist progress: Show download pro

Extract text from image. Args: image_path: Path to the image file Returns: Extracted tex

Get access token, performing authentication if necessary. Args: force_reauth: Force re-authentication

Get file size in MB.

Get metadata about the file. Args: file_path: Path to the file Returns: Dictionary conta

Handle login command.

Main entry point.

Parse markdown text into structured JSON. Args: markdown_text: The markdown content to parse Returns

Run the web application.

Initialize the cloud processor. Args: api_key: API key for cloud processing (optional - uses rate-limited free tier witho

Build a hierarchical structure from flat sections list.

Check NumPy version and warn about compatibility issues.

Classify a paragraph as heading, list item, table row, or regular text.

Clean up the text content. Args: content: Raw text content Returns: Cleaned text content

Clean up the extracted web content. Args: content: Raw web text content Returns: Cleaned

Fallback to local conversion methods.

Convert PDF pages to images. Args: pdf_path: Path to the PDF file Returns: List of paths

Convert table lines to HTML table.

Convert structured table to markdown format.

Convert text blocks and tables to structured markdown.

Create hierarchical structure from flat sections list.

Detect if a URL points to a file and return file information. Args: url: URL to check Returns:

Determine heading level based on font size and position.

Determine heading level based on text characteristics.

Download model from Hugging Face using docling's logic. Args: repo_id: Hugging Face repository ID revision: G

Download model from Nanonets S3. Args: s3_url: S3 URL of the model archive local_dir: Local directory to extr

Download a specific model. Args: model_config: Model configuration dictionary force: Force re-download

Extract blockquotes from content.

Extract code blocks from content.

Extract content from API response.

Extract images from content.

Extract structured JSON using Nanonets model with specific prompt.

Extract links from content.

Extract lists from content.

Extract main content from the HTML. Args: soup: BeautifulSoup object Returns: Extracted

Extract tables directly from markdown content as fallback.

Extract paragraphs from content.

Extract tables from content.

Extract text using docling's advanced models.

Extract text from a specific region of the image.

Extract text with layout awareness using docling's neural models.

Extract text using Nanonets OCR model.

Format text as a table row.

Generate PKCE code verifier and challenge.

Get content type for file upload.

Lazily initialize the document extractor.

Calculate hash of a file for cache invalidation.

Get OCR service instance.

Get processing recommendation based on token count.

Get the appropriate processor for the file. Args: file_path: Path to the file Returns: P

Generate a summary of a section.