Skip to content

A Python script that aggregates recent popular posts from your Mastodon timeline

License

Notifications You must be signed in to change notification settings

hodgesmr/mastodon_digest

Repository files navigation

Mastodon Digest

⚠️ Update, 2023-05-21: Mastodon Digest is currently in maintenance-only mode. I do not have the capacity to continue fielding feature requests or PRs that fundamentally alter the way the tool works. If you have an idea for a new feature, forks are encouraged. I will continue to push updates for bug or security fixes. If you're looking for options that are more actively maintained, I suggest: Fediview or this fork. I'm heartened by how much positive attention this got among many Mastodon communities. For me this was always a toy. I'm glad I could push some open source code that gave people room to experiment.

This is a Python project that generates a digest of popular Mastodon posts from your home timeline. The digest is generated locally. The digests present two lists: posts from users you follow, and boosts from your followers. Each list is constructed by respecting your server-side content filters and identifying content that you haven't yet interacted with. Digests are automatically opened locally in your web browser. You can adjust the digest algorithm to suit your liking (see Command arguments). The digest will not contain posts from users who include #nobot or #noindex in their bio.

Mastodon Digest

Run

You can run in Docker or in a local python environment. But first, set up your environment:

Before you can run the tool locally, you need to copy .env.example to .env (which is ignored by git) and fill in the relevant environment variables:

cp .env.example .env
  • MASTODON_TOKEN : This is your access token. You can generate one on your home instance under Preferences > Development. Your token only needs Read permissions.
  • MASTODON_BASE_URL : This is the protocol-aware URL of your Mastodon home instance. For example, if you are @Gargron@mastodon.social, then you would set https://mastodon.social.

Both the Docker container and the python script will construct the environment from the .env file. This is usually sufficient and you can stop here. However, you may optionally construct your environment manually. This is may be useful for deployed environments.

Docker

First, build the image:

make build

Then you can generate and open a digest:

make run

You can also pass command arguments:

make run FLAGS="-n 8 -s ExtendedSimpleWeighted -t lax"

Local

Mastodon Digest has been tested to work on Python 3.9 and above.

With Make

If your system Python meets that, you can:

make local

You can also pass command arguments:

make local FLAGS="-n 8 -s ExtendedSimpleWeighted -t lax"

Manually

Althernatively if you have a different Python 3.9 environment, you can:

pip install -r requirements.txt

Then generate a Mastodon Digest with:

python run.py

Through either method, the digest is written to render/index.html by default. You can then view it with the browser of your choice.

Command arguments

A number of command arguments are available to adjust the algorithm. You can see the command arguments by passing the -h flag:

python run.py -h
usage: mastodon_digest [-h] [-f TIMELINE] [-n {1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24}]
                       [-s {ExtendedSimple,ExtendedSimpleWeighted,Simple,SimpleWeighted}] [-t {lax,normal,strict}]
                       [-o OUTPUT_DIR] [--theme {light,default}]

options:
  -h, --help            show this help message and exit
  -f TIMELINE           The timeline to summarize: Expects 'home', 'local' or 'federated', or 'list:id', 'hashtag:tag' (default:
                        home)
  -n {1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24}
                        The number of hours to include in the Mastodon Digest (default: 12)
  -s {ExtendedSimple,ExtendedSimpleWeighted,Simple,SimpleWeighted}
                        Which post scoring criteria to use. Simple scorers take a geometric mean of boosts and favs. Extended
                        scorers include reply counts in the geometric mean. Weighted scorers multiply the score by an inverse
                        square root of the author's followers, to reduce the influence of large accounts. (default:
                        SimpleWeighted)
  -t {lax,normal,strict}
                        Which post threshold criteria to use. lax = 90th percentile, normal = 95th percentile, strict = 98th
                        percentile (default: normal)
  -o OUTPUT_DIR         Output directory for the rendered digest (default: ./render/)
  --theme {light,default}
                        Named template theme with which to render the digest (default: default)

If you are running with Docker and make, you can pass flags as:

make run FLAGS="-n 8 -s ExtendedSimpleWeighted -t lax"

Algorithm Options

  • -f : Timeline feed to source from. home is the default.
    • home : Your home timeline.
    • local : The local timeline for your instance; all the posts from users in an instance. This is more useful on small/medium-sized instances. Consider using a much smaller value for -n to limit the number of posts analysed.
    • federated : The federated public timeline on your instance; all posts that your instance has seen. This is useful for discovering posts on very small or personal instances.
    • hashtag:HashTagName : The timeline for the specified #hashtag. (Do not include the # in the name.)
    • list:3 : A list timeline. Lists are given numeric IDs (as in their URL, e.g. https://example.social/lists/2), which you must use for input here, not the list name.
  • -n : Number of hours to look back when building your digest. This can be an integer from 1 to 24. Defaults to 12. I've found that 12 works well in the morning and 8 works well in the evening.
  • -s : Scoring method to use. SimpleWeighted is the default.
    • Simple : Each post is scored with a modified geometric mean of its number of boosts and its number of favorites.
    • SimpleWeighted : The same as Simple, but every score is multiplied by the inverse of the square root of the author's follower count. Therefore, authors with very large audiences will need to meet higher boost and favorite numbers. This is the default scorer.
    • ExtendedSimple : Each post is scored with a modified geometric mean of its number of boosts, its number of favorites, and its number of replies.
    • ExtendedSimpleWeighted : The same as ExtendedSimple, but every score is multiplied by the inverse of the square root of the author's follower count. Therefore, authors with very large audiences will need to meet higher boost, favorite, and reply numbers.
  • -t : Threshold for scores to include. normal is the default
    • lax : Posts must achieve a score within the 90th percentile.
    • normal : Posts must achieve a score within the 95th percentile. This is the default threshold.
    • strict : Posts must achive a score within the 98th percentile.

I'm still experimenting with these, so it's possible that I change the defaults in the future.

Theme Options

Specify a render template theme with the --theme <theme-name> argument.

Two basic templates for the digest are provided, default and light. You can create new templates by adding a directory to templates/themes/my-theme/. You must create index.html.jinja as the root template.

Template fragments placed inside themes/common/ can be re-used by any template, which is helpful to try and keep things DRY-er (for example, include scripts.html.jinja for the current version of the Mastodon iframe embed JavaScript.)

The available view variables are:

  • posts : Array of posts to display
  • boosts : Array of boosts to display
  • hours : Hours rendered
  • mastodon_base_url : The base URL for this mastodon instance, as defined in env.
  • rendered_at : The time the digest was generated
  • timeline_name : The timeline used to generated the digest (e.g. home, local, hashtag:introductions)
  • threshold : The threshold for scores included
  • scorer : The scoring method used

Each post and boost is a ScoredPost object:

  • url : The canonical URL of the post.
  • get_home_url(mastodon_base_url): The URL of the post, translated to the mastodon_base_url instance provided.
  • info : The full underlying status dict for the post, documented by mastodon.py here.

When developing themes, you can run the digest in development mode, which uses theme files from the local filesystem rather than rebuilding the docker image every time you make a change:

make dev FLAGS="--theme my-theme"

A Matt Hodges project

This project is maintained by @MattHodges.

Please use it for good, not evil.

About

A Python script that aggregates recent popular posts from your Mastodon timeline

Resources

License

Stars

Watchers

Forks

Packages

No packages published