Feature Request - install unaccent

[dustymc]

Is your feature request related to a problem? Please describe.

It's occasionally useful to ignore accents and diacritics and such.

Describe what you're trying to accomplish

#2874, but this could be useful for lots of searches, perhaps could even allow relaxing some rules.

Describe the solution you'd like

Install https://www.postgresql.org/docs/current/unaccent.html, build stuff (probably mostly trigram indexes) to use it

Describe alternatives you've considered

Additional context

Needs coordinated with @lkvoong re backups/copies.

Priority

This would be new functionality, I don't have time to dive in at the moment, the really cool possibilities would take some time to develop, probably relatively low.

[Jegelewicz]

Using non-ASCII characters is nice BUT in every csv download, they come out crazy in Excel (which is how a lot of us will open them). I found this solution

1. On a Windows computer, open the CSV file using Notepad.
2. Click "File > Save As".
3. . In the dialog window that appears - select "ANSI" from the "Encoding" field and "All Files (.)" from the Save as type field. Then click "Save".
4. . That's all! Open this new CSV file using Excel - your non-English characters should be displayed properly.

And I am wondering if we can get downloads in ANSI format for Excel? In other words, could we have choices in the format of downloaded data?

[dustymc]

downloads in ANSI format for Excel?

There are two possible ways in which UTF-encoded data may be represented:

1. As UTF-encoded data, or
2. Incorrectly.

Allowing a user to input "slaskie" and find data for Śląskie is potentially helpful (typing Ś isn't always trivial) and useful (possibly that's what they wanted). If that's not what they were looking for, most will understand how false positives work and there's no real harm done - they'll ignore it and move on.

Turning Śląskie into slaskie for download just sounds like a dirty trick to me. Now they've got some irreversible mess which doesn't match anything they can find anywhere and cannot be converted back to anything which does. Not a great place to find oneself!

Even if there was some defensible path to that sort of thing, ایران and 中国 do not involve accented ASCII and so are beyond the capability of the requested extension.

[Jegelewicz]

You can just say no.

[dustymc]

Tabling; might be reconsidered in context of #6524