allo- / feedjack Goto Github PK

It seems timezone dependend dates get stored in the database. Every datetime object should be timezone aware so the dates in the db are independend of the planet's configuration.

Currentn development is only done on the plain theme and has some breaking changes. Old themes are preserved at https://github.com/allo-/feedjack/tree/master/feedjack/templates/feedjack/unsupported_themes.
Eventually they could be adapted to the new URLs and template structure and new django techniques (like {% static_url %}).

Should probably be done after the big changes are done, so they do not need to get fixed from change to change.

Use a base.html, so other django-apps, which expect it can find it.

Create a good HTML template and delegate all theming to CSS files.

Currently there is an optional feature to use lxml2 to sanatize html (close tags, where the closing tag is missing, etc.) to make sure the post page stays intact if some post contains broken html.

Add some filters to strip unsafe tags (<script>, <style>, ...) or use a whitelist of allowed tags and remove eventhandlers (onload, etc.), which could be an malicious html injection from the feed.

add a view, which filters subscribers by group

The display of the post-tags

Currently the user login only works via the admin page.

• Use the registration app for login.
• Use a base.html (https://github.com/allo-/feedjack/issues/13), so the app finds templates to generate the login page
• Provide Login/Logout links on the main page
• Add options for registering, disabling registration, requireing e-mail verification.

Set CSP headers (possibly configurable):

• Disallow scripts from posts
• Allow images (external, data:)
• Disallow/Allow Frames (often tracking pixels, etc.)

Remove all Inline Scripts to be able to completly forbid unsafe-inline (and prevent event-handlers like "onload" to inject scripts).

For the "1.0" release of the new code there should be a new name, as there is no trivial upgrade path from feedjack or any of its forks. So the package should have some distinct new name.

Hi, playing around with mk-fg/feedjack I found your repo - from what I understand your changes are based on the current master of his feedjack branch, however I see no pull request on his repo from your changes here. I do not understand the current situation regarding this code, would you like to explain?
Is this the new 'main' repo for feedjack?
Thanks for your attention!

• URLs with ?marked= do not make sense for users, who aren't logged in and are probably from bookmarks / history before the logout.

Filters like since, unread, groups, etc. should be get-parameters, so they can be combined without complex patterns in urls.py

Use jQuery's ajax function to load the next page with a selector for the post-list and append it to the current post list.

Provide a nice default theme

• Probably using some framework or even template: bootstrap / foundation / material-design / ...
• Should work without javascript, may provide more features with JS enabled
• reactive / mobile friendly

Features:

• A sidebar with feeds and feedgroups, displaying their unread-counter, textbox to filter this list (with JS)
• A top bar with ordering / filtering options (unread, read later, all, maybe "contains string")
• A nice message list, possible with features like "jump to the next article", etc.
• Possibly endless-scroll ready (inserting / removing articles without problems, clear scroll positions)

When the post list is filtered for a date range, add the range to the title. Maybe only start/end date and truncated to a coarser range depending on how many minutes/hours/days are between the posts.

Couple feedjack.models.Site with django.contrib.sites.models.Site as 1:1 relation and try to simplify site selection algorithm (get_current_site should be enough).

Add a Post<->User relationship with an intermediate model, which marks a post as

• read
• unread
• important
• favourite
• read it later
• (...)

Sidebar: with feeds list, feed/group/multiple feeds selectable.

Top: Article list with headlines
Bottom: Article view for a single article (maybe an iframe), with option to view the original in the iframe. (like classic feedreader applications or popular mail programs).

As the site will display a lot of mostly unfiltered content from feeds, it should use security headers for allowed script sources (only current domain and no inline scripts).

Update the readme with a description of the new project and a clean and easy way for installation and configuration, including settings like ´´TEMPLATES´´ with template context processors and everything what's needed to run a new instance.

Move any stuff only relevant to the old releases into another file or remove them (as they can be found in git or old feedjack releases).

move get_extra_context and page_context to context processors. Cleanup views, try to remove stuff from initview.

[ ] mark the currently selected feed in the sidebar (i.e. bold font)
[ ] add a subscribers_excluded=1,2,3 URL-Parameter
[ ] When only filters like unread are applied, allow in the sidebar to deselect single feeds, adding them to subscribers_excluded
[ ] When a subscribers=1,2,3 parameter is given, allow to select/deselect feeds via the sidebar, which add/remove them in this parameter.

Show number of unread messages in the feedlist:

• group (XX)
  • feed (YY)

d9b278a removed parsing date formats like "yesterday".
We can add the links again with precomputed dates as now() - timedelta(1), which are relative to the page (and thus permalinks) with the link instead of being always relative to now().

Meta-Bug for adding tests.

Currently there will be bigger changes, so no unit tests, yet.
Testing intended behaviour would be nice as well.

Stuff to be done:

• make sure all code can be tested (all new design changes should try to do one thing in one function)
• Add some test data (well formed feed data)
• decide on a unit test framework (nose?)
• maybe integrate some coverage framework
• write unit tests
• add a script, which tests bootstrapping a full installation

Use Dates for pagination.

• Forward: All posts after this one (ascending dates)
• Backwards: All posts after the date of 20 posts before (ascending dates)?

Add a title field to the Subscriber model and a function lambda: self.title or self.feed.title to get either a custom title for the site or the title from the feed.

• Finish all cleanup, simplification, changes to models, etc. to have a stable supportable state.
• Find a new name #21
• possibly release a "0.999" under the new or old name, with all messy migrations for migrating on your own risk.
• remove south and django migrations for a fresh start, add a hint to 0.999 for people who want to try migrating from an older version.
• Have a clean installation documentation, which covers the current state #22.

referrer "never" does not work with csrf middleware.

Try to use more ready made cache decorators and so on for the caching.