docx-md

Word DOCX (OOXML) – docx-md Overview

Three entry points: Read – 输出 compact Markdown (default, 令牌-efficient) or full JSON; Modify – 应用ly AI-returned edits to the docx; Finalize – accept all revisions and 移除 all comments. Implemented via OOXML (ZIP + XML). No commercial Word libraries required.

工作流 Goal Action 获取 document for AI Read: 运行 read script → Markdown (default) or JSON. Markdown includes block索引 markers for edit tar获取ing. 应用ly AI edits to docx Modify: 运行 应用ly script with docx + edits JSON → new docx with 追踪 changes and comments. Deliver final version Finalize: 运行 finalize script → new docx with no revisions/comments. LLM-oriented 流水线 Read – 解析 docx; 输出 Markdown (default) or JSON. Markdown uses prefix per block; revisions: {+inserted+} {-删除d-}; comments: [comment: text]. 发送 the 输出 + task prompt to the 模型; require the 模型 to 输出 only the edit JSON: block索引, originalContent, content, basis . Modify – Script infers op from block索引, originalContent, content, basis; converts to OOXML (w:ins / w:del / comment anchors), then write back to Word. Finalize – When the user confirms, 运行 finalize to accept all revisions and 移除 all comments.

See references/llm-流水线.md for the Markdown 格式化, JSON 模式, and edit 格式化.

• Read

解析 word/document.xml (w:body only) and word/comments.xml. 输出 Markdown (default) or JSON. Markdown is compact and 令牌-efficient.

Script: scripts/read_docx.py

# Default: Markdown 输出 (令牌-efficient) python3 技能s/docx-md/scripts/read_docx.py document.docx python3 技能s/docx-md/scripts/read_docx.py document.docx -o 结果.md

# JSON 输出 (full structure) python3 技能s/docx-md/scripts/read_docx.py document.docx -f json -o 结果.json

Options:

-o, --输出 – 输出 path (default: stdout) -f, --格式化 – md (default) or json

• Modify

输入: docx path + edit JSON { modifications: [{ block索引, originalContent, content, basis }] } (same block索引 as read 输出). Flow: Convert JSON to OOXML (w:ins / w:del / comments), then write back to Word.

Script: scripts/应用ly_edits_docx.py. Use - as edits file to read JSON from stdin.

python3 技能s/docx-md/scripts/应用ly_edits_docx.py document.docx edits.json -o 输出.docx python3 技能s/docx-md/scripts/应用ly_edits_docx.py document.docx - -o 输出.docx # stdin

Options: --author (default: "Review")

• Finalize

Accept all revisions (flatten to final text), 移除 all comments. Save as new docx. Uses docx-revisions to accept revisions (preserves encoding), then 移除s comment markup via regex on raw bytes.

Script: scripts/finalize_docx.py

Requires: pip 安装 docx-revisions (see requirements.txt)

python3 技能s/docx-md/scripts/finalize_docx.py 输入.docx -o 输出.docx

Resources scripts/ read_docx.py – Read: python3 scripts/read_docx.py document.docx [-o out.md] [-f md|json] 应用ly_edits_docx.py – Modify: python3 scripts/应用ly_edits_docx.py document.docx edits.json -o 输出.docx finalize_docx.py – Finalize: python3 scripts/finalize_docx.py 输入.docx -o 输出.docx references/ ooxml.md – OOXML layout (document.xml, comments.xml, revisions, comments) llm-流水线.md – 流水线: read → Markdown/JSON → 模型 edits → modify; defines Markdown 格式化, JSON shape (block索引, originalContent, content, basis)