Support for the google reader protocol

[andreyorst]

I'd like to propose an additional protocol.

I use FreshRSS, and while it works via Fever in elfeed-protocol, inability to sync the read state is bugging me quite a bit, because I usually read from my phone, so when I finally open my feeds in elfeed, I have a ton of old unread articles, that were actually read.

FreshRSS has a page about GoogleReader API:

https://freshrss.github.io/FreshRSS/en/developers/06_GoogleReader_API.html

As far as I can tell, GReader works better than Fever.

[fasheng]

You need a two-way synchronization protocol like ownCloud News although few people use it.

For Google Reader API, it's really special. It doesn't even have public documentation, everyone has different compatibility with it. After some verification work, looks that the two-way sync feature that you care about is not provided by Google Reader API, instead the clients did some workaround for it.

So this issue will be mark as "feature" and will continue to see if there is a need to implement it. However at least I will provide some builtin workaround code to sync unread stat like other clients did in the future.

Here is a simple code for fever:

(defun elfeed-protocol-fever-sync-unread-stat (host-url)
  "Set all entries in search view to read and fetch latest unread entries.
HOST-URL is the host name of Fever server with user field authentication info,
for example \"https://user@myhost.com\"."
  (mark-whole-buffer)
  (cl-loop for entry in (elfeed-search-selected)
           do (elfeed-untag-1 entry 'unread))
  (elfeed-protocol-fever--do-update host-url 'update-unread))

Reference:
https://ash7.io/blog/hello-read-you-welcome-to-google-reader-api-en-us/
miniflux/v2#2129
https://github.com/FreshRSS/FreshRSS/blob/edge/p/api/greader.php
https://github.com/miniflux/v2/blob/main/internal/googlereader/handler.go

[andreyorst]

A few more notes on Google Reader API: https://feedhq.readthedocs.io/en/latest/api/index.html

[andreyorst]

(defun elfeed-protocol-fever-sync-unread-stat (host-url)
  "Set all entries in search view to read and fetch latest unread entries.
HOST-URL is the host name of Fever server with user field authentication info,
for example \"https://user@myhost.com\"."
  (interactive
   (list (completing-read
          "feed: "
          (mapcar (lambda (fd)
                    (string-trim-left (car fd) "[^+]*\\+"))
                  elfeed-protocol-feeds))))
  (save-mark-and-excursion
    (mark-whole-buffer)
    (cl-loop for entry in (elfeed-search-selected)
             do (elfeed-untag-1 entry 'unread))
    (elfeed-protocol-fever--do-update host-url 'update-unread)))

Made an interactive prompt for choosing a specific feed from protocol-feeds

[andreyorst]

You need a two-way synchronization protocol like ownCloud News although few people use it.

After some verification work, looks that the two-way sync feature that you care about is not provided by Google Reader API, instead the clients did some workaround for it.

Yeah, the client of my choice has this feature, so I assumed it was part of GReader protocol.

For Google Reader API, it's really special. It doesn't even have public documentation, everyone has different compatibility with it

I thought the API was thorley reverse-engineered, and well documented though unofficially. AFAICS it's one of the most widespread protocols, provided by a lot of services and used in a lot of clients.

[bricka]

It's still very much beta, but I've built a package for The Old Reader, which also uses some sort of Google Reader API: https://gitlab.com/bricka/elfeed-protocol-theoldreader

It's based on their API, which they claim is based on Google Reader, but I'm not close enough to the Google Reader API itself to say for sure

[fasheng]

@bricka Great job!

[einsiedlerspiel]

Hi, I took @bricka s package and made it work with freshrss by generalizing/de-hardcoding the greader api: https://codeberg.org/lou/elfeed-protocol-freshrss

In principle this should also work with other greader api implementations but I haven't tested that.

[s0ph0s-dog]

Hi, I took @bricka s package and made it work with freshrss by generalizing/de-hardcoding the greader api: https://codeberg.org/lou/elfeed-protocol-freshrss

In principle this should also work with other greader api implementations but I haven't tested that.

Thank you so much! I was just looking through the elfeed-protocol code to see if I could ask the FreshRSS Fever API for more unread items (either a larger number or the next page) and ended up looking here after discovering that it's not possible. I look forward to trying your code once Codeberg's data center has power again 🙃

[einsiedlerspiel]

@s0ph0s-dog: I'm also mirroring the package here: https://git.repetitions.de/elfeed-protocol-freshrss/

re: pulling more unread items -- I've currently hardcoded it to download the 1000 most recent items regardless of read state, for maximum effect you can sync individual feeds (this will again download the 1000 most recent but per feed).

Making that a bit more customizable and implementing a full db sync is on my TODO.