|6 months ago|
|konfluks||9 months ago|
|.gitignore||1 year ago|
|LICENSE.md||1 year ago|
|README.md||11 months ago|
|konfluks.svg||11 months ago|
|poetry.lock||1 year ago|
|pyproject.toml||11 months ago|
|setup.py||11 months ago|
A drainage basin is a geographical feature that collects all precipitation in an area, first in to smaller streams and finally together in to the large river. Similarly, Konfluks can bring together small and dispersed streams of web content from different applications and websites together in a single large stream.
Specifically, Konfluks turns Peertube videos, iCal calendar events, other websites through their RSS and OPDS feeds and Mastodon posts under a hashtag in to Hugo page bundles. This allows one to publish from diverse sources to a single stream.
Konfluks was first made by Roel Roscam Abbing as part of lumbung.space, together with ruangrupa and Autonomic.
Konfluks tries to act as a mirror representation of the input sources. That means that whenever something remote is deleted, changed or becomes unavailable, it is also changed or deleted by Konfluks.
Konfluks tries to preserve intention. That means the above, but also by requiring explicit ways of publishing.
Konfluks works by periodically polling the remote sources, taking care not to duplicate work. It caches files, asks for last-modified headers, and skips things it has already. This makes every poll as fast and as light as possible.
Konfluks is written for clarity, not brevity nor cleverness.
Konfluks is extendable, a work in progress and a messy undertaking.
Konfluks consists of different Python scripts which each poll a particular service, say, a Peertube server, to download information and convert it in to Hugo Page Bundles
Each script part of Konfluks will essentially to the following:
Parse a source and request posts/updates/videos/a feed
- Taking care of publish ques
Create a Hugo post for each item returned, by:
- Making a folder per post in the
- Formatting post metadata as Hugo Post Frontmatter in a file called
- Grabbing local copies of media and saving them in the post folder
- Adding the post content to
- According to jinja2 templates (see
- Making a folder per post in the
The page bundles created, where possible, are given human friendly names.
Here is a typical output structure:
user@server: ~/konfluks/output: tree tv/ tv/ ├── forum-27an-mother-earth-353f93f3-5fee-49d6-b71d-8aef753f7041 │ ├── 86ccae63-3df9-443c-91f3-edce146055db.jpg │ └── index.md ├── keroncong-tugu-cafrinho-live-at-ruru-gallery-ruangrupa-jakarta-19-august-2014-e6d5bb2a-d77f-4a00-a449-992a579c8c0d │ ├── 32291aa2-a391-4219-a413-87521ff373ba.jpg │ └── index.md ├── lecture-series-1-camp-notes-on-education-8d54d3c9-0322-42af-ab6e-e954d251e076 │ ├── 0f3c835b-42c2-48a3-a2a3-a75ddac8688a.jpg │ └── index.md
curl -sSL https://raw.githubusercontent.com/python-poetry/poetry/master/get-poetry.py | python -
We use Poetry because it locks the dependencies all the way down and makes it easier to manage installation & maintenance in the long-term. Then install the dependencies & have them managed by Poetry:
Each script requires some environment variables to run, you can see the latest deployment configuration over here, look for the values under the
environment: ... stanza.
All scripts have an entrypoint described in the
pypoetry.toml which you can run via
poetry run .... For example, if you want to run the
konfluks/video.py script, you'd do:
mkdir -p testdir export OUTPUT_DIR=/testdir poetry run konfluks-vid
poetry run poetry2setup > setup.py if updating the poetry dependencies. This allows us to run
pip install . in the deployment and Pip will understand that it is just a regular Python package. If adding a new cli command, extend
pyproject.toml with a new