UniNews is a desktop application that brings news and announcements from universities into a single place.

It started as a way to avoid visiting a dozen university and department pages every day. Instead of opening several tabs, UniNews collects everything into one window, stores it locally, and lets you search and filter it.

Written in Python with PyQt6. Publishers are plain JSON files: adding a university usually means adding a few lines of configuration, not writing code.

Features

News from multiple universities in one window
RSS/Atom feeds and selector-driven HTML scraping
Local SQLite cache, so articles stay readable when a site is down
Search that ignores case and accents (υποτροφιες finds ΥΠΟΤΡΟΦΙΕΣ)
Filter by publisher or by individual source
Articles sorted by publication date, across many different date formats
Light and dark themes
Background fetching: the window stays responsive while sources load
A "Sources" view showing which sources failed on the last refresh, and why
Configurable auto-refresh, article cache size and page size

Install

Arch Linux

makepkg -si

Debian / Ubuntu

Download the .deb from the releases page, then:

sudo apt install ./uninews_0.2.0-1_all.deb

From source

git clone https://github.com/open-source-uom/UniNews.git
cd UniNews
python3 -m venv .venv
source .venv/bin/activate        # Windows: .venv\Scripts\activate
pip install -e ".[dev]"

Running

uninews # installed

python3 -m uninews # from a source checkout

Project structure

UniNews/
├── pyproject.toml            build definition and dependencies
├── PKGBUILD                  Arch package
├── debian/                   Debian/Ubuntu package
├── packaging/                desktop entry and icon
├── src/uninews/
│   ├── __main__.py           entry point
│   ├── config.py             paths (XDG) and user settings
│   ├── database.py           SQLite storage, filtering, paging
│   ├── library.py            what the window is currently showing
│   ├── article.py            the Article record
│   ├── dates.py              many date formats -> one sortable format
│   ├── text.py               whitespace and accent handling
│   ├── fetchers/             how articles are retrieved
│   │   ├── base.py           Fetcher: HTTP, cleanup, deduplication
│   │   ├── html.py           HtmlFetcher: CSS-selector scraping
│   │   ├── rss.py            RssFetcher: RSS and Atom
│   │   └── __main__.py       command-line checker
│   ├── resources/publishers/ one JSON file per publisher
│   └── ui/                   widgets; only this layer imports Qt
└── tests/

Dependencies point one way: ui -> library -> database -> fetchers. Nothing below ui/ imports Qt, and only the main window touches the database.

Adding a publisher

Create src/uninews/resources/publishers/<name>.json. Check for an RSS feed first; it is more reliable than scraping.

{
  "publisher": "MIT",
  "category": "university",
  "sources": [
    { "label": "MIT News", "type": "rss", "url": "https://news.mit.edu/rss/feed" }
  ]
}

For a site without a feed, describe its markup with CSS selectors. Open the page, inspect one article, and find the repeating container and the title link inside it.

{
  "publisher": "Example University",
  "category": "university",
  "defaults": {
    "type": "html",
    "item_selector": "article.post",
    "title_selector": "h2.entry-title a",
    "date_selector": "time, .entry-date",
    "summary_selector": ".entry-summary",
    "page_url": "{url}/page/{page}/"
  },
  "sources": [
    { "label": "Example Main News", "url": "https://example.edu/news/", "pages": 3 }
  ]
}

Key	Meaning
`type`	`rss` or `html`
`label`	Shown in the sidebar
`url`	First page of the list, or the feed
`item_selector`	Repeating element, one per article
`title_selector`	The title link inside an item
`date_selector`, `summary_selector`, `image_selector`	Optional
`pages`, `page_url`	Pagination; `{url}` and `{page}` are substituted
`enabled`	Set to `false` to park a source without deleting it

Comma-separated selectors are tried left to right, and the first match wins.

Check your work before opening the app:

python3 -m uninews.fetchers src/uninews/resources/publishers/example.json

python3 -m uninews.fetchers # or check every publisher

OK means articles were found, EMPTY means the selectors matched nothing, and FAIL prints the reason (blocked, offline, or changed markup).

Sites that need code

If selectors are not enough (a JSON API, dates only on the article page), add a module to src/uninews/fetchers/ that subclasses HtmlFetcher or Fetcher:

from uninews.fetchers.html import HtmlFetcher
 
 
class ExampleAnnouncements(HtmlFetcher):
    fetcher_type = "example_announcements"
 
    def parse_item(self, item):
        raw = super().parse_item(item)
        ...
        return raw

Import it in fetchers/__init__.py so it registers itself, then reference it in JSON with "type": "example_announcements".

Where data is stored

What	Path
Settings	`~/.config/uninews/settings.json`
Article cache	`~/.local/share/uninews/uninews.db`
Generated UI assets	`~/.cache/uninews/`
Home university (read-only, written by UniDesk)	`~/.unios/academicConfig.json`

Deleting the database makes the next refresh fetch everything again.

UniNews opens with one university selected in the sidebar, chosen in this order:

Settings → Open with in UniNews, if set to a university or "All sources".
The university in the UniOS profile (universityName in ~/.unios/academicConfig.json, set from UniDesk's Configure UniOS page).
Otherwise, "All sources" (for example, UniNews installed outside UniOS).

All universities are still fetched; the selection only filters what is shown.

Development

python3 -m pytest -q          # tests: no network, no display needed
ruff check .
ruff check . --fix

API docs are generated with Doxygen from the docstrings:

doxygen Doxyfile # output: docs/html/index.html

Tests run against saved pages in tests/fixtures/, so a failure means the code broke. To find out whether a site changed, run the fetcher checker instead.

Being a good citizen

University servers are not free to hit. UniNews fetches at most six sites in parallel and never more than one request at a time to the same site, auto-refresh is limited to once every 15 minutes, and every request identifies the application. If a site asks not to be fetched, remove it or set "enabled": false.

Roadmap

More publishers, including news sites
Managing sources from the UI
Source health history across sessions
Export and backup

Contributing

The easiest contribution is a new publisher: add a JSON file, verify it with the fetcher checker, and open a pull request.

Before opening a pull request:

python3 -m pytest -q passes.
ruff check . is clean.
Changes stay focused and easy to review.

License

GNU General Public License v3.0.