Hacky Python scripts for downloading your Twitter likes and converting to HTML. Includes support for downloading user avatars and image media in tweets. Scrapes tweets using the GraphQL API powering Twitter.com - the equivalent of you scrolling through all your likes in your web browser, only saved locally forever!

Note only currently supports grabbing the liked tweet. So if it was a quote tweet, does not download the RT'd tweet. If it's a reply or part of a thread, does not download the other tweets.

Meant to jumpstart you to at least getting your tweets offline from Twitter, if not building something much better!

Intended for use with Python 3.7. First install requirements:

pip install -r requirements.txt

(It's just requests). Next you'll need to populate the config.json file (create it by copying config_template.json) with credentials needed to match how your web browser gets your likes from Twitter.com:

1. Open up your web browser and ensure the Network web debugging tab is open so you can inspect network requests (in Chrome, it is under Open Chrome Developer Console > Network)
2. Navigate to https://twitter.com/<your_user_handle>/likes
3. Look for a network request to an api.twitter.com domain path ending in /Likes (you can type /likes in the filter box at the top left of the debug console). You need to find userId, authorization, cookie and x-csrf-token values from the network request and fill the corresponding fields in config.json.
4. From the request headers, copy paste the following details (you can right click on the request and select Copy > Copy as cURL and paste into an editor window for the ease of finding these details): a. Copy the Authorization value (find it as authorization: Bearer xxx) and save as HEADER_AUTHORIZATION in config.json b. Copy the Cookies value (find it as cookie: xxx) and save as HEADER_COOKIES in config.json. Note that while pasting cookie value, you would need to escape any existing double quotes by prefixing them with a backslash (\). c. Copy the x-csrf-token (find it as x-csrf-token: xxx) value and save as HEADER_CSRF in config.json
5. Find your Twitter user ID (available in the /Likes request params, or elsewhere) and save as USER_ID in config.json. If you click on Payload tab, you may find something like "userId": "xxx", where xxx is your numeric user ID.

By default, liked tweets will be saved to a file liked_tweets.json in this repo's folder path. If you'd like to override this, set new path as OUTPUT_JSON_FILE_PATH in config.json.

Run as follows:

python download_tweets.py

Should provide output like the following:

Starting retrieval of likes for Twitter user 1234...
Fetching likes page: 1...
Fetching likes page: 2...
Fetching likes page: 3...
Done. Likes JSON saved to: liked_tweets.json

The output JSON will be a list of dictionaries like the following:

[
   {
      "tweet_id": "780770946428829696",
      "user_id": "265447323",
      "user_handle": "LeahTiscione",
      "user_name": "Leah Tiscione",
      "user_avatar_url": "https://pbs.twimg.com/profile_images/1563330281838284805/aUtIY2vj_normal.jpg",
      "tweet_content":"What are you hiding in your locked instagram? sandwiches? Sunsets???? let us see your nephew!!!!",
      "tweet_media_urls": [],
      "tweet_created_at": "Sun Mar 13 15:16:45 +0000 2011"
   }
]

Every time the download_tweets.py is successfully run, the last Tweet ID is saved to config.json and in a subsequent run gets used as the "Stop ID", which means retrieval stops when this ID is found in the feed. Assuming the tweet was never unliked, this works like incremental retrieval such that you can run periodically and not download those tweets that have already been downloaded before. These incremental downloads are also saved as separate files with a numeric suffix (e.g., Liked_tweets.2.json, Liked_tweets.3.json etc.) and avoid overwriting an existing file. To enable downloading all liked tweets again, edit config.json and set the TWEET_STOP_ID to null before retriggering the download.

If you want your tweets as a local HTML file, you can run the second script to convert the output JSON file from the above step.

NOTE: This will attempt to download all media images and tweet author avatars locally by default to avoid relying on Twitter hosting. You can override this by changing the DOWNLOAD_IMAGES boolean in config.json to false.

1. Be sure the OUTPUT_JSON_FILE_PATH value in config.json is pointing to the output JSON file of your tweets.
2. Run:

python parse_tweets_json_to_html.py

This will download all images (if specified; saved to tweet_likes_html/images) and construct an HTML file at tweet_likes_html/index.html containing all liked tweets, as well as individual HTML files within tweet_likes_html/tweets/.