This project provides an automated solution for generating technical proposals, specifically tailored for Chinese requirements. It assists engineers and technical writers by transforming product documentation and requirements matrices into structured, point-by-point proposal responses, significantly reducing manual effort and improving consistency.

How It Works

The system processes a product manual (in .docx format) and a requirements table (in .xlsx format). It extracts relevant content from the manual, matching features to requirements. If a match is found, it copies and formats the manual's content; otherwise, it leverages a large language model (like ERNIE-Speed-8K or ChatGPT) to generate the necessary text. The output is a formatted proposal document (.docx) and a technical requirements deviation table, complete with automatically generated chapter numbers.

Quick Start & Requirements

• Install dependencies: pip install -r requirements.txt
• Requires a large language model API key (OpenAI or Baidu Qianfan).
• Input files: Template.docx (product manual) and requirements_table.xlsx.
• Official documentation: Not explicitly linked, but usage is detailed in the README.

Highlighted Details

• Automatically generates structured proposals with H1, H2, H3 headings and body text, images, and bullet points.
• Supports copying and rewriting product manual content, including images and tables.
• Generates technical requirements deviation tables with "Answer: Fully supported, {LLM-generated text}" format.
• Breaks down product manuals into reusable documents for efficiency.

Maintenance & Community

The project is associated with WhaleOps Open Source, the team behind Apache DolphinScheduler and Apache SeaTunnel. Contact information for commercial inquiries is provided.

Licensing & Compatibility

The README does not explicitly state a license. Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

The system relies on specific Word styling ("Body Text," "Heading 1," "Heading 2," "Heading 3") for correct formatting. Image resizing is handled up to a specified MAX_WIDTH_CM. The project appears to be primarily focused on Chinese language proposals.