anulman/docx-sax

docx-sax is a developer tool for reading the internal structure of Word docx files. Instead of giving you a finished document object with paragraphs and headings, it streams out low level events as it walks through the file's OpenXML package: when a part starts and ends, what relationships exist between files inside the package, when an XML element opens or closes, and when text is found. It keeps details like namespaces, attribute values, file paths, and event order intact, which makes it useful for people who need precise control over how a docx file is parsed rather than a simplified summary. The core is a .NET 8 library built on Microsoft's Open XML SDK. Around that core, the project offers several ways to use the parser depending on your platform. There is a command line tool that outputs one JSON event per line, a Node.js wrapper that reads docx files from disk, and a browser wrapper built on WebAssembly that reads file bytes directly, useful for parsing uploaded documents in a web page without a server round trip. All three interfaces share the same event structure, so code written against one behaves consistently across the others. The project is explicit about what it will not do. It has no interest in offering paragraph or heading level concepts, no document editing or authoring features, and no JSON first design in its core library. It is meant purely as a low level walker over the file format, leaving any higher level interpretation to whoever uses it. The repository includes a Next.js demo app that shows the browser version parsing an uploaded docx file live. Development uses standard dotnet and npm tooling, with a continuous integration setup that builds and tests across Linux, Windows, and macOS. The project is still early stage, versioned in the 0.x range while its event format continues to change, and it recommends prerelease version numbers until the API is considered stable.