zulip: Search doesn't match words in URL, only whole hostname or path
Details on what matches and doesn’t match: https://github.com/zulip/zulip/issues/9165#issuecomment-855061552
Links with additional reports of the issue and discussion:
- 2018 user report and some product discussion of desired behavior: https://chat.zulip.org/#narrow/stream/137-feedback/topic/indexing.20urls/near/655441
- 2020 explanation below of how this can make a big difference in the effectiveness of search: https://github.com/zulip/zulip/issues/9165#issuecomment-596857254
- 2020
#backend
discussion of implementation strategies: https://chat.zulip.org/#narrow/stream/3-backend/topic/Searching.20parts.20of.20url.20.239165/near/996827 - 2021 user report, discussion of the underlying subsystem that provides the current behavior, and product discussion of desired behavior: https://chat.zulip.org/#narrow/stream/137-feedback/topic/How.20to.20get.20search.20to.20work.20for.20this/near/1171632
- 2021 “An organization administrator I reached out to mentioned this as their top issue”: https://github.com/zulip/zulip/issues/9165#issuecomment-983039342, below
- 2023 user report: https://chat.zulip.org/#narrow/stream/137-feedback/topic/link.20terms.20do.20not.20show.20up.20in.20search.20results/near/1580761
Original report follows.
The search function right now will only find URLs if they’re entered fully. I.e. I will only find https://www.blog.google if I enter www.blog.google
(it won’t find it if I type blog.google
or .google
). An example of this can be seen here: https://youtu.be/j6e_l6-3APQ
Similarly if I want to find something that has a SEO URL format (title of the post or whatever in the URL), I cannot find them easily. This is also shown in the video with me searching for zulip (with the URL to this repo being visible in the stream).
This is on Zulip 1.8.0. ( Also, is there a way to see the server version from within the app itself? I can find it out easily by checking the running version on the server, but I can’t seem to find a way for other users or administrators to see the currently running server version. Event the /stats
site doesn’t contain it. )
About this issue
- Original URL
- State: open
- Created 6 years ago
- Reactions: 8
- Comments: 28 (14 by maintainers)
Commits related to this issue
- search: Allow searching of messages with link by typing parts of url. This commit adds a postgres function that extract urls from rendered_content of message using REGEXP_MATCH in sql and then replac... — committed to thedeveloperr/zulip by thedeveloperr 4 years ago
- search: Allow searching of messages with link by typing parts of url. This commit adds a postgres function that extract urls from rendered_content of message using REGEXP_MATCH in sql and then replac... — committed to thedeveloperr/zulip by thedeveloperr 4 years ago
- search: Allow searching of messages with link by typing parts of url. This commit adds a postgres function that extract urls from rendered_content of message using REGEXP_MATCH in sql and then replac... — committed to thedeveloperr/zulip by thedeveloperr 4 years ago
- search: add unittest for pgroogna parts-of-URL search. Fixes #9165. — committed to asah/zulip by asah 2 years ago
- search: add unittest for pgroogna parts-of-URL search. Fixes #9165. — committed to asah/zulip by asah 2 years ago
- search: add unittest for pgroogna parts-of-URL search. Fixes #9165. — committed to asah/zulip by asah 2 years ago
@timabbott Can I work on this ?
For us, one of the benefits of zulip is that your conversations can become a searchable repository of knowledge - if you’ve already explained something once you can search for the conversation and send a link to the conversation rather than repeating yourself.
Unfortunately, this bug makes that workflow a lot more difficult as often the uniquely identifying information was in a url. We try and include a blurb with all urls but, sometimes we forget and it can then be a very frustrating / time-consuming experience trying to find the right conversation.
It would be very useful if searching for words contained in urls Just Worked.
An organization administrator I reached out to mentioned this as their top issue.