* add useSetting hook

* example usage

* better types

* more implementations

Part of #37009.

* QB: evaluate notebook preview lazily

* fixup! QB: evaluate notebook preview lazily

* fixup! fixup! QB: evaluate notebook preview lazily

* fixup! fixup! fixup! QB: evaluate notebook preview lazily

* Move `initialCollectionId` prop to `QueryModals`

* Remove unused prop and selector

* Use mongo 5.0 for E2E tests

* Update snapshot names

* Update DB name

Fixes #38954.

* Fix applying default values for a pinned question

* fix typecheck

* Add a comment

* Add two Snowplow click tracking events to the Browse Data page, for clicks on models and tables

# Conflicts:
#	frontend/src/metabase/browse/components/BrowseModels.tsx
#	frontend/src/metabase/browse/components/TableBrowser/TableBrowser.jsx

* add backend event

* fix event tracking

* test snowplow even

* update schema type field

* more prettier

---------

Co-authored-by: Raphael Krut-Landau <raphael.kl@gmail.com>

Converted queries need to fill in types of references converted from v1
to properly pass schemas. However legacy refs still exist in
`:lib/stage-metadata` and these refs should not be touched. Checking
that `map?` is in the second position of a ref is an easy way to
determine which version ref we're looking at.

* Missing untranslated string

* Fix redundant `export`

* Misc clean up

* Remove duplicate block

* Decouple `LLMSuestionQuestionInfo` from form

* Merge loading and loaded state layouts

* Extract `getSubmittableQuestion`

* Compute `submittableQuestion` in save modal

* Fix selectors

* Fix types naming style

* Fix multiple calls to summarize endpoint

* Migrate graph.metrics to use QP metadata

* Migrate graph.metrics to use QP metadata

* Handle native columns

* Handle missing query results

* Optimize

Fixes #26202

Previously we only looked for binned field-ids in order to update
binning strategies in qp. With this change we also look for field-name
refs. This allows expressions to be binned in models.

This just bumps the metadata version of the `enable-llm-autodescription?` feature flag to 50 (`^{:added "0.50.0"}`).

* Show filters without values in fullscreen mode

* Remove `parametersWidget` from `DashboardHeader`

* Use regular filters panel in fullscreen mode

* Minor refactor

* Minor refactor

* Fix nightmode

* Fix transition

This is FE-only because this is a compatibility issue with the (legacy)
metadata coming back from the QP.

In time, the FE logic should be ported to MLv2 or made to depend on
things other than the `:name` field. MLv2 code itself (including BE
usage) should be based on the (already unique)
`:lib/desired-column-alias`.

Fixes #37517.

* Apply eslint quotes rule on e2e folder

* fixup! Apply eslint quotes rule on e2e folder

* fixup! fixup! Apply eslint quotes rule on e2e folder

* Endpoint stubs

* It's alive!!!

The summarization endpoint is live, calling OpenAI, and connected to the `POST "metabot/card/summarize"`. `POST card/` also taps the results.

* Updating prompt

* Updated response shape

* [WIP] working front-end suggest button

* Incorporating viz settings

* add suggest description to save question modal

* Updating prompt

* merge title and description suggestion into one button

* Updating prompt

* Removing dead code

* Removing unused param and fn cleanup

* Fix string comment

* Adding dashboard summary endpoint stubs

* auto-request title and description

* dashboard summary wip

* dashboard summary wip

* dashboard summary wip

* better prompt

* remove dead component

* add suggest description button in dashboard sidebar

* fix dashboard description suggestion request

* better prompt

* swap out button when generating dashboard suggestion

* adding keywords

* adding keywords

* adding keywords

* adding keywords

* polish loading question name and description suggestions

* polish UI for dashboard description suggestion

* Stubbed in usage middleware

* Beginning EE Push

Moved metabot/summarize endpoints to autodescribe/summarize in the ee codebase to push this towards being an ee feature and distinguish it from metabot features.

Added snowplow tracking.

* Isolating EE LLM Code

This change completely separates metabot code from ee llm code.

It also introduces a composable client mechanism for creating LLM clients for different tasks.

* Task-Oriented LLM Refactoring

LLM code is now organized clearly into a directory of tasks. Each ns in the tasks package provides a certain LLM task to be used in the API.

* Adding unit test

* Adding admin UI elements for EE OpenAI key

* prettier on frontend/src/metabase/services.js

* Cleaning up description text

* Updating api calls

* FE stuff -- hiding things when feature not enabled.

* Figuring out linting for snowplow file

* Figuring out linting for snowplow file

* Fixing unit test

* add token feature llm-autodscription

* change added from 0.50 to 0.49

* extract logic from SaveQuestionModal into ee plugin

* extract logic from DashboardInfoSidebar to ee plugin

* Fixing unit test

* tests

* cleaning up code for SaveQuestionModal

* gate llm question name behind openAI key setting

* remove styles from SaveQuestionModal.css

* clean up code for dashboard

* lint

* Enabling triggering based on `:llm-autodescription` feature flag

* Starting to look at using a badge for LLM results (#38797)

* Starting to look at using a badge for LLM results

* WIP - Using an Indicator to show AI "stuff"

* Prototype of LLM with badge/indicator

* Removing dashboard description AI

* Ensuring LLM icon only shows when key is present. Updating icon to newer design. TODO: Put the icon in the right place.

* Cleanup -- still need to figure out how to position the indicator and not collapse the radio when selected.

* Updating save question form with LLM values does not reset form

* Cleanup

* Correct icon placement

* Adding MB branding to star

* add ai icon, tweak display

* Fix z-index, tweak padding

* cleanup

* remove inactive code

* Cleaning llm description task cobwebs out

* fix collection creation

* Cleaning lint and merge errors

* Adding "save-question-modal" test id back in

* Fixing jest test

* Touching a file to unstick CI

* Untouching a file to unstick CI

* Fixing some includes from the rebase.

* eslint fixes

* eslint fixes

* Fixing types

* Using `PLUGIN_IS_EE_BUILD.isEEBuild()` instead of `isEnterprise()`

* Removing dead styling

* Adding test ids and fixing cypress tests

* Fixing e2e tests

* Incorporating feedback

* Update frontend/src/metabase-types/api/settings.ts

Co-authored-by: Jesse Devaney <22608765+JesseSDevaney@users.noreply.github.com>

* Update frontend/src/metabase/core/components/EditableText/EditableText.tsx

Co-authored-by: Jesse Devaney <22608765+JesseSDevaney@users.noreply.github.com>

* Update frontend/src/metabase/core/components/Input/Input.tsx

Co-authored-by: Jesse Devaney <22608765+JesseSDevaney@users.noreply.github.com>

* Update frontend/src/metabase/core/components/TextArea/TextArea.styled.tsx

Co-authored-by: Jesse Devaney <22608765+JesseSDevaney@users.noreply.github.com>

* Update frontend/src/metabase/core/components/TextArea/TextArea.styled.tsx

Co-authored-by: Jesse Devaney <22608765+JesseSDevaney@users.noreply.github.com>

* Update frontend/src/metabase/dashboard/components/DashboardInfoSidebar/DashboardInfoSidebar.tsx

Co-authored-by: Jesse Devaney <22608765+JesseSDevaney@users.noreply.github.com>

* Fixing import

* Rebasing on master to fix conflicts

* Fixing types

* Revert "Update frontend/src/metabase/core/components/EditableText/EditableText.tsx"

This reverts commit fe11817086bff50f90336f30d33a1431cae662c5.

* Revert "Update frontend/src/metabase/core/components/Input/Input.tsx"

This reverts commit eef21407ee32e636a8b4e2594d502d5bbaa36870.

* Revert "Update frontend/src/metabase/core/components/TextArea/TextArea.styled.tsx"

This reverts commit dd1c22b993a6576d92547749d2eb93cb68411585.

* Revert "Update frontend/src/metabase/core/components/TextArea/TextArea.styled.tsx"

This reverts commit b169d05364e92c032ae5b6c972888a86079921e2.

* Fixing e2e tests

* address feedback

* Fixing e2e tests

* Fixing modal for models

---------

Co-authored-by: Jesse Devaney <22608765+JesseSDevaney@users.noreply.github.com>
Co-authored-by: Emmad Usmani <emmadusmani@berkeley.edu>
Co-authored-by: Kyle Doherty <5248953+kdoh@users.noreply.github.com>
Co-authored-by: Anton Kulyk <kuliks.anton@gmail.com>
Co-authored-by: Ryan Laurie <iethree@gmail.com>

* Show query execution time in Query Builder footer

* Show execution time for all types of queries

* Revert "Show execution time for all types of queries"

This reverts commit 429cb124c59a7a3314719663ebe1a3cbefa2b95b.

* Update assertions

* Remove redundant condition

* Add tests for query execution time

* Format code

* Do not use FunctionComponent

* Hide dashboard actions on public/static embed dashboards

This is not what we want. See
https://metaboat.slack.com/archives/C057T1QTB3L/p1696532502602629?thread_ts=1696519212.704419&cid=C057T1QTB3L

* Add E2E tests

* Review: Use TS type guard

* Fix the logic in the previous commit

Remove note

* WIP flow :semantic-type through QP for aggregations

Fixes: #38022

WIP need to add and fix some tests and confirm if this is the most correct place to fix the problem. Seems like it's a
good spot to put things though.

* Add test asserting that :semantic_type does flow through QP

Also fix a couple tests that now have a :semantic_type key.

* Add different aggregation types to semantic-type test

Working on adding tests and fixing various aggregation semantic types

* Test :semantic-type for all aggregations and adjust some metadata-methods

* Better use of existing methods via derive

* Properly set hierarchy between ::count-aggregation and ::quantity-aggregation