Project No Longer Maintained

I created this for collecting analytics data for study community I organized where everything happened on Slack. That was long ago and I have no use for maintaining this.

In retrospect, if I were to build a Slack scraper from scratch, I would focus on intercepting network requests and reading message data from there, instead of scraping and parsing the HTML. Much easier and more reliable. You can even make it a script or service worker you copy/paste manually into your Slack browser tab, to avoid the authentication issues people keep having on automated browsers. That's my recommmendation if someone needs this.

It might be even as simple as starting a HAR file recording using the DevTools network tab, scrolling up to whatever point in a chat/search you want to save, saving the file, filtering cookies/credentials, and prompting LLM to write script to parse the HAR JSON.

Good luck!

Slack Web Scraper

A web scraper that navigates to a Slack workspace and saves the posts and threads of a given channel or DM.

It uses Puppeteer headless browser for loading and interacting with Slack. It doesn't depend on installing an app in the Slack workspace or aquiring an API key. Instead, it logins to your Slack account and uses that to access the channel or DM.

It's helpful for saving information from a channel or DM without needing to ask a workspace administrator to export the data.

For example, if you're in the process of leaving your current company to join another, this tool is a great way to archive everything you've said and done on Slack.

How to collect Slack data?

1. Run npm install to install the dependencies.
2. Copy the .example.env file in the project root folder and rename it to .env. Then modify following environment variables in .env:

• SLACK_WORKSPACE_URL, SLACK_EMAIL and SLACK_PASSWORD are required.

  • SLACK_WORKSPACE_URL must be the URL you login to the workspace not app.slack.com. Example: SLACK_WORKSPACE_URL=cloud-native.slack.com. Note environment variables are set without quotes.
  • SLACK_EMAIL and SLACK_PASSWORD are credentials used to login into the workspace.

• You must set one of CONVERSATION_NAMES or CHANNEL_NAMES or both.

  • The collect script will scrape the list of conversations first, then the list of channels. The list must a valid JSON array: ["element1", "element2"]. The array elements are double quoted and the last element doesn't have a trailing comma. You can escape a double quote in a string in JSON like this: ["string\"hello"]
  • Set CONVERSATION_NAMES to scrape a DM or group chat. The value is the name tag of the person or group chat name as is written under "Direct Messages" in Slack. Example: CONVERSATION_NAMES=["Iuliu Pop (Core Grad)", "John Doe"].
  • Set CHANNEL_NAMES to scrape a public or private channels. It's the name you see under "channels" side tab in Slack. Example: CHANNEL_NAMES=["general", "random"].
  • The name doesn't need to be an exact match, it must only match part of the name. For example, if a name tag includes an emoji, you can only write the part of the name tag without it and it should work.
  • The channel or the conversation must be in the list of channels or DMs in the left sidebar before running the collect script.

• SCROLL_UP_TIMEOUT is optional.

  • A timeout in seconds for when to stop scrolling up the channel history and start scraping posts. Useful when scraping channels with a long history but don't need to scrape it all. For a very active channel, it could take 60 seconds to scroll up half a year then ~20min to scrape it. Example: SCROLL_UP_TIMEOUT=30

• HEADLESS_MODE is optional.

  • Set to true to scrape with the browser in headless mode. Example: HEADLESS_MODE=true.
  • Helpful for scraping long channel/conversation histories, since the browser runs with a larger vertical viewport so can scrape it larger batches at a time. I recommend you start without running headless mode with one conversation or channel since you can see clearer if the collect scraper is working or not.

• SKIP_THREADS is optional.

  • Set to true to disable scraping threads on messages that are in channels or conversations. Example: SKIP_THREADS=true.
  • Helpful if you do not want to scrape all the replies that are made on a message but just have the main message.

3. Before starting the scrape, make sure the Slack App language is set to English. You can reset it once the scrape is finished.

4. Run npm run collect. You will see the browser open and start scraping data unless you set HEADLESS_MODE to true. In headless mode you will see status updates on the scraping process in the console output.

Tip for collecting data with Windows Subsystem for Linux

You need to configure WSL to connect to a GUI even if the browser launches in headless mode. Use this guide to configure WSL to connect to an X server installed on Windows. Before running the collect script, the X server must be open and WSL correctly configured to connect to it, or Puppeteer will fail to launch the browser.

How to parse Slack data?

1. Assuming you already ran npm run collect, you can now run npm run parse. You will be prompted to select the file to parse from the slack-data/ folder. Once the parsing is complete, a slack-data/x.json file with same name as the source HTML file will be output with the parsed posts/threads.

Contributors ✨

Thanks goes to these wonderful people (emoji key):

Iuliu Pop 🤔 💻 📖 👀 💬

William Desportes 💻 🐛

NotEdwin 🐛 💻

This project follows the all-contributors specification. Contributions of any kind welcome!

Contributing

Very open to contributions to this project! If you have questions, bug reports or features you want to see, please open an issue. If you want to contribute code, open a pull request and I'll review ASAP.

License

MIT