issue_comments

Thanks Simon, My use case is using Datasette for full text search type ahead. That was working pretty well. The _search_wildcard= option will be awesome. Thanks

Shipped in 0.36 https://datasette.readthedocs.io/en/stable/changelog.html#v0-36

https://datasette.readthedocs.io/en/stable/changelog.html#v0-36

I considered just updating the GitHub release with a link to the release notes, but I'm building other things against the GitHub releases API such that it's nicer to have the release notes live there.

I bet I could use pandoc to automate the conversion of the formats in one or other direction.

Before shipping there are a few additions I can make to the https://datasette.readthedocs.io/en/stable/ecosystem.html page:

• https://github.com/simonw/shapefile-to-sqlite
• https://github.com/simonw/datasette-mask-columns (which will itself depend on 0.36 being out)
• https://github.com/simonw/datasette-auth-existing-cookies
• https://github.com/simonw/datasette-sentry

I could also add links to associated blog entries directly to those descriptions on the ecosystem page:

• https://simonwillison.net/2020/Jan/29/weeknotes-datasette-cookies-sentry/
• https://simonwillison.net/2020/Feb/19/shapefile-to-sqlite/

Interesting feature suggestion.

My initial instinct was that this would be better handled using the layered nature of Docker - so build a Docker image with datasette package and then have a separate custom script which takes that image, copies in the extra data and outputs a new image.

But... datasette package is already meant to be more convenient than messing around with Docker by hand like this - so actually having a --copy option like you describe here feels like it's within scope of what datasette package is meant to do.

So yeah - if you're happy to design this I think it would be worth us adding.

Small design suggestion: allow --copy to be applied multiple times, so you can do something like this:

datasette package \
    --copy ~/project/templates /templates \
    --copy ~/project/README.md /README.md \
    data.db

Also since Click arguments can take multiple options I don't think you need to have the : in there - although if it better matches Docker's own UI it might be more consistent to have it.

This is a smart optimization. I'd like to add a test for it, since the logic is a tiny bit more complex.

Thanks for spotting this!

I'm afraid this is by design: supporting wildcards in that mode was more-or-less accidental and often resulted in unexpected user-facing syntax errors when users accidentally entered search text that included characters with special meaning to SQLite.

You can still execute wildcard searches but you have to do so using custom SQL statements.

I'm open to suggestions for more friendly ways to support this functionality. Would an alternative _search_wildcard= URL parameter be worth supporting I wonder? What's your specific use-case for wildcards?

Done. Updated documentation is here: https://datasette.readthedocs.io/en/latest/plugins.html#prepare-connection-conn-database-datasette

While I'm at it, the hook should accept an optional database name argument as well.