Consider sending canonical URL via HTTP header

[rgaudin]

In situations where kiwix-serve is serving on the www with ZIMs that are close to their source content, it acts like some kind of mirror.
This can have unpleasant consequences for the actual source website in that it blurs ownership to search engines and other parties.

The general solution to this is for the copy entries to indicate the source using the rel=canonical mechanism.

Should we have openzim/libzim#904, maybe via openzim/libzim#316, we would be in a good position to add this Canonical URL information.

In order to send it without altering the ZIMs, we could use the HTTP header method, by sending

Link: <https://www.source.tld/entry.html>; rel="canonical"

At the moment, we have no per-entry metatada but we frequently have a ZIM-level Source metadata.

We could send the aforementioned HTTP header with the Source value (if present obviously) on every ZIM entry. It would be abusing the purpose of the canonical URL in that it would not be the actual entry's cannonical document but at least it would indicate to search engines that kiwix-serve-served one is not.

Thoughts?

---

• https://developers.google.com/search/blog/2013/04/5-common-mistakes-with-relcanonical
• https://developers.google.com/search/docs/crawling-indexing/consolidate-duplicate-urls#rel-canonical-header-method

[benoit74]

I agree that using the Source might be a promising shortcut, to at least properly indicated to search engines and other parties that our content is a kind a mirror.

I definitely would like to have the ability to set metadata per entry (for this and also for other topics).