WeChat Chat Capture Tool (微信聊天记录滚动截屏工具)

Automatically scroll and capture WeChat chat windows on Windows, with optional OCR text recognition and Markdown export.

English

Introduction

A Windows desktop tool (Python + Tkinter) that captures long WeChat chat conversations by automatically scrolling the chat window and stitching the captured screenshots. It is the "screenshot" equivalent of a long-page capture for chat records.

Features

• Region selection — drag to select the exact chat area on screen
• Auto-scroll capture — scrolls the chat window automatically; configurable scroll distance (supports decimals) and capture interval
• Smart stop — stops automatically when reaching the top of the chat, or manually via ESC / stop button
• Optional OCR dedup — uses RapidOCR (ONNX Runtime) to recognize text, deduplicate overlapping screenshots, and avoid saving duplicates
• Live text preview — OCR results displayed in real time, split into left/right speaker columns
• Markdown export — each screenshot's recognized text is saved to a *.md file alongside it
• Dark UI — clean dark theme, DPI-aware, runs with low CPU usage (OCR thread-limited)
• Config persistence — region, scroll distance, interval and OCR settings are remembered across sessions

Requirements

• Windows 10/11
• Python 3.8+
• Dependencies: Pillow, pywin32, numpy, rapidocr-onnxruntime

Installation

pip install Pillow pywin32 numpy rapidocr-onnxruntime

Usage

python wechat_capture_tool.py

1. Click Select Region (框选捕获区域) and drag to cover the chat message area
2. Click Save Folder (选择保存文件夹) to pick the output directory
3. Click Start Capture (开始捕获) — a 3-second countdown lets you switch to the WeChat window
4. Capture runs automatically; press ESC or click Stop (停止) to finish

Screenshots are saved as chat_000001.png, chat_000002.png, ... plus a chat_*.md file with the recognized text for each.

Scripts

File	Description
wechat_capture_tool.py	Main GUI tool
e2e_test.py	End-to-end driver test against a real WeChat window (dev helper)
organize_chat.py	Standalone utility: OCR a batch of chat screenshots → dedupe → sort by time → output one conversation .md

License

MIT

---

中文

项目介绍

一个基于 Python + Tkinter 的 Windows 桌面工具，用于自动滚动截取微信聊天记录长图。框选聊天区域后，程序自动滚动聊天窗口并逐帧截屏，相当于聊天记录的"长截图"工具。可选内置 OCR 文本识别，自动去重并生成 Markdown 文本。

功能特性

• 框选区域 — 拖动鼠标精确框选聊天内容区域
• 自动滚动截屏 — 自动滚动聊天窗口，滚动距离（支持小数）与截屏间隔可调
• 智能停止 — 滚动到顶部自动停止，也可按 ESC 或点击"停止"手动结束
• OCR 可选去重 — 基于 RapidOCR（ONNX Runtime）识别文本，重叠内容 ≥95% 判定为无新内容，自动跳过重复截图
• 实时文本预览 — 识别结果实时显示，自动按左右分列（发送/接收）
• Markdown 导出 — 每张截图的识别文本同步保存为同目录下的 *.md 文件
• 深色界面 — 干净深色主题，DPI 感知，OCR 限线程数避免占满 CPU
• 配置记忆 — 捕获区域、滚动距离、间隔、OCR 开关等设置自动保存

环境要求

• Windows 10/11
• Python 3.8+
• 依赖：Pillow、pywin32、numpy、rapidocr-onnxruntime

安装

pip install Pillow pywin32 numpy rapidocr-onnxruntime

使用方法

python wechat_capture_tool.py

1. 点击「框选捕获区域」，拖动鼠标覆盖聊天消息区域
2. 点击「选择保存文件夹」指定输出目录
3. 点击「开始捕获」，3 秒倒计时内切换到微信窗口
4. 自动滚动截屏，按 ESC 或点击「停止」结束

截图保存为 chat_000001.png、chat_000002.png…… 每张截图同时生成对应的 chat_*.md 识别文本文件。

脚本说明

文件	说明
wechat_capture_tool.py	主程序（GUI 工具）
e2e_test.py	端到端驱动测试，在真实微信窗口上验证全流程（开发辅助）
organize_chat.py	独立工具：批量 OCR 聊天截图 → 去重 → 按时间排序 → 输出对话 md

开源协议

MIT