Co-authored-by: Chris Truter <crisptrutski@users.noreply.github.com>

Co-authored-by: Noah Moss <32746338+noahmoss@users.noreply.github.com>

Co-authored-by: Chris Truter <crisptrutski@users.noreply.github.com>

Co-authored-by: Ngoc Khuat <qn.khuat@gmail.com>
Co-authored-by: Phoomparin Mano <poom@metabase.com>

Fixes: #50771

The new sync process failed to qualify enums outside public and
therefore failed to properly type them in database_type. This could
either result in matching the wrong type if there was a conflict or not
being typed as an enum.

Co-authored-by: Case Nelson <case@metabase.com>

Co-authored-by: metamben <103100869+metamben@users.noreply.github.com>

* Add trial banner

* Use the semantic "warning" color for background

* Simplify the "reappearance" logic

* Refine the banner logic

- Pass the bare minimum information (expiry timestamp) to the `TrialBanner`
- Take timezones into account
- Introduce utilities
- Add unit tests

* Update the remaining tests

* Clean up

* Unbreak my `defsetting` 

* Add comments

* Defer var assignment

* Fix failing tests 



* Adjust the types in order to match `master`

* Move the logic up to the parent component

* Convert the `index` component to TS

* Fix typo in the JSDoc

* Fix another typo

* Make the return statement more readable

* Remove redundant test

Co-authored-by: Nemanja Glumac <31325167+nemanjaglumac@users.noreply.github.com>

Applying the underlying records filter to a query with a join displays fields from the join

Co-authored-by: William <william@metabase.com>

Co-authored-by: Chris Truter <crisptrutski@users.noreply.github.com>

Co-authored-by: Chris Truter <crisptrutski@users.noreply.github.com>

Co-authored-by: Alexander Solovyov <alexander@solovyov.net>

Co-authored-by: Ngoc Khuat <qn.khuat@gmail.com>

Co-authored-by: Ngoc Khuat <qn.khuat@gmail.com>

Co-authored-by: Noah Moss <32746338+noahmoss@users.noreply.github.com>

Pivot Exports Can Handle `nil` in Breakout Col, and Multiple Cols of the Same Aggregation Type (#50672) (#50780)

* Pivot Exports Can Handle `nil` in Breakout Col, and Multiple Cols of the Same Aggregation Type

Fixes #50551

If a pivot table has several measures configured, they might both use the same kind of
aggregation (eg. 'sum'). Previously, this would lead to columns with the same name preventing the pivot-measures from
properly making it through the export post processing. Now, the correct aliased/deduped name is used and all columns
can be properly included in the pivot export.

Also fixes the case where a breakout column (pivot-row) contains `nil` values, which caused the pivot export to
fail. Now, all nil values are grouped and handled appropriately.

* when pivot is disabled, header should still be included

* min and max don't default to 0, add a test for the aggregations

Co-authored-by: adam-james <21064735+adam-james-v@users.noreply.github.com>

Co-authored-by: Chris Truter <crisptrutski@users.noreply.github.com>

* match against any hexadecimal bundle name

* add e2e tests skeleton

* Support display-name for ISO get-day-of-week exclusion expressions (#50606)

* Support display-name for ISO get-day-of-week exclusion expressions

Fixes #50574

* Fix light theme embeddings has different text colors (#50678)

* add cache-control tests

* support local runs

---------

Co-authored-by: Phoomparin Mano <poom@metabase.com>
Co-authored-by: metamben <103100869+metamben@users.noreply.github.com>
Co-authored-by: Mahatthana (Kelvin) Nomsawadi <me@bboykelvin.dev>

Co-authored-by: Chris Truter <crisptrutski@users.noreply.github.com>

Co-authored-by: Chris Truter <crisptrutski@users.noreply.github.com>

Co-authored-by: Chris Truter <crisptrutski@users.noreply.github.com>

This protects against us trying to read or write from an incompatible search index. In particular it prevents existing instances from being switched over to an incompatible table during an HA upgrade or downgrade.

~~It doesn't yet get rid of the needless re-initialization, but that will be a fast follow.~~ It breaks up the jobs so that we won't reindex on startup unless necessary.

Co-authored-by: Chris Truter <crisptrutski@users.noreply.github.com>

Co-authored-by: Chris Truter <crisptrutski@users.noreply.github.com>

See https://www.notion.so/metabase/Search-Index-Maintenance-14b69354c9018047b670cbba1c3be50d



This ensures that when one instance starts or finishes reindexing, the other instances pick up and switch to the same tables.

Note: this does not stop blowing away the currently active table on deployment yet, we need to have a notion of version for that first.

Co-authored-by: Chris Truter <crisptrutski@users.noreply.github.com>

This also avoids a brittle song and dance with non-authoritative atoms in the background. Now we juggle authoritative atoms all the time 



Co-authored-by: Chris Truter <crisptrutski@users.noreply.github.com>

* Enable underlying-records drill in table viz for multi-stage queries

* Fix e2e test

* Add tests for underyling-records drill for multi-stage queries

* Review feedback: filterv not-empty etc

* Review feedback: add a breakout-only query to drill-thru canned-queries

Co-authored-by: appleby <86076+appleby@users.noreply.github.com>

Fixes https://github.com/metabase/metabase/issues/50501



Co-authored-by: Chris Truter <crisptrutski@users.noreply.github.com>

* Table Column Ordering Should Not Prevent Pivot Exports

Prior, when viz-settings existed for table column ordering, or if some columns were hidden in a regular table viz, the
pivot export could fail.

Now, if a pivot table is being exported, the table column sorting is properly ignored and the pivot export should work.

* address feedback

Co-authored-by: adam-james <21064735+adam-james-v@users.noreply.github.com>

Co-authored-by: Chris Truter <crisptrutski@users.noreply.github.com>

Co-authored-by: Chris Truter <crisptrutski@users.noreply.github.com>

Co-authored-by: Chris Truter <crisptrutski@users.noreply.github.com>

* Adding .png version of no_results.svg to fix #49012

* Adjusting text color to  to match existing color scheme

* Removing old unused no results assets; adding new png version of no_results.svg at matching resolution to pulse_no_results@2x.png

* Fixing namespace import cycle

Co-authored-by: Tyler Plude <tsplude@gmail.com>

* remove pivot-grouping row and columns in tandem (#50476)

* remove pivot-grouping row and columns in tandem

* add pivot card scalar test for csvs and xlsx.

* improve test feedback

* add a 'todo' for pivoted json exports

* indent

* tap tap tap

---------

Co-authored-by: bryan <bryan.maass@gmail.com>
Co-authored-by: Adam James <adam.vermeer2@gmail.com>

* Pivot Export Aggregations Improved and Won't Error (#50380)

* Pivot Export Aggregations Improved and Won't Error

Fixes #50207

WIP

- Totals now have appropriate aggregation implementation
- Totals for different row and column partial paths could collide before (eg. if a pivot-col value was equal to the
- measure idx), now the paths are unique and will not collide, causing errors

* add a test

* multiple measures need to have the correct order in each row

* clean up, add some comments explaining totals paths

* fix tests

* add test to show add-rows working correctly after :rows-part :cols-part added

* tap tap tap

---------

Co-authored-by: adam-james <21064735+adam-james-v@users.noreply.github.com>
Co-authored-by: Adam James <adam.vermeer2@gmail.com>

* Add stage-number parameter to NativeParameterDimensionTarget

* Update unit test

* Recognize native query dimension parameters with stage-number

* Update test setup

---------

Co-authored-by: Tamás Benkő <tamas@metabase.com>

* Disable `?locale` on static embeds/public links

* Fix E2E tests

* Add a new line for consistency

* Don't dissoc `:id` in `before-update`

There seems to be a nasty footgun in toucan2 here.

Say you're executing an `update!` command on a set of IDs, e.g.:

```
(t2/update! :model/Card :id [:in 1 2] {:view_count 1})
```

This works if:

- both cards 1 and 2 have `view_count=1` already

- both cards 1 and 2 have `view_count!=1`

However.

If one of the two cards has `view_count=1` and another has a different
view_count, then (if the `before-update` method doesn't have the primary
key attached) Toucan emits a call to update *every card in the
database*, without a where clause at all.