ExStruct

About

ExStruct is a robust Python library and CLI tool designed to extract comprehensive structured data from Excel workbooks. It meticulously parses cells, identifies table candidates, and captures rich elements like shapes, charts, SmartArt, merged cell ranges, print areas, and even auto page-break areas. With multiple output modes (light, standard, verbose) and tunable table detection heuristics, ExStruct provides flexible control over the extraction process, making it ideal for preparing complex Excel data for large language model (LLM) and retrieval-augmented generation (RAG) applications.

Key Features

• Extracts cells, table candidates, shapes, charts, SmartArt, merged cells, and print/page-break areas.
• Provides tunable table detection heuristics to optimize accuracy for diverse datasets.
• Gracefully falls back to basic extraction (cells + table candidates) if Excel COM is unavailable.
• Supports output in JSON, YAML, and TOON formats, with optional rendering to PDF and PNGs.
• 111 GitHub stars
• Offers 'light', 'standard', and 'verbose' output modes for varying levels of detail, including hyperlinks and formula maps.

Use Cases

• Prepare complex Excel spreadsheets for AI processing in LLM/RAG pipelines.
• Automate the conversion of Excel documents (tables, charts, flowcharts) into machine-readable JSON for AI reasoning benchmarks.
• Digitize and structure data from diverse Excel application forms for automated data entry or analysis.