[InstituteForTheStudyOfWarBridge] Fix bridge following website update

[SqrtMinusOne]

Also increased caching time yet again because of CloudFlare.

[github-actions]

Pull request artifacts

Bridge	Context	Status
InstituteForTheStudyOfWar	1 untitled (current)	❌ Bridge returned error 0! (20394) ❌ Type: Error ❌ Message: Call to a member function find() on null
InstituteForTheStudyOfWar	1 untitled (pr)	✔️

last change: Sunday 2025-11-02 16:48:28

[Mynacol]

Thanks, this works great.
I'm still wondering if adding another toggle for fetching the full page content is necessary when it was made by default beforehand. Could it be that you experience excessive Cloudflare captchas due to your local network? I tried accessing the feed in rapid succession with the last 20 items (cache disabled) and no Cloudflare captcha/block was happening.

On the other hand, adding a limit option is standard practice and helps keeping the feed fetch time in check.

[SqrtMinusOne]

Could it be that you experience excessive Cloudflare captchas due to your local network?

Maybe. I run RSS-Bridge on a VPS I pay for with crypto; perhaps this got the network blacklisted. But it used to be fine for me as well.

I honestly have no clue how Cloudflare decides when to block requests, and it seems kinda random from site to site. It also changes over time, usually in the worse direction. E.g., on phys.org I can fetch a maximum of 3 feeds every 4 hours.

I wish they never existed :-(

[Mynacol]

I run RSS-Bridge on a VPS I pay for with crypto

That is a strong indicator that it is IP reputation based (thank your neighbors). I want to kindly ask you to remove this option again. I expect you have problems with other (Cloudflare-querying) bridges as well, and we won't add an option to each of them. There are some general "Cloudflare handling" elements in rss-bridge, e.g. curl-impersonate. But I don't want such options in specific bridges if they generally behave like others contacting Cloudflare.

I'm also confused by those HTML encoded & parts in the categories. It seems the bridge makes the right thing. But somehow they get double-encoded. @dvikan can you have a look at this?

[SqrtMinusOne]

I want to kindly ask you to remove this option again.

Ah, I've misread your original comment, sorry. The note about Cloudflare referred to a week-long caching time; the contents option was there for debugging, mostly. Removed it.

I'm also confused by those HTML encoded & parts in the categories.

No, it's the website. Adding html_entity_decode fixes the issue.

[Mynacol]

Thanks. Now this is a great PR.