TIMESTAMP columns will now map to :type/DateTime instead of attempted
parsing them as times.

Resolves #12045

[ci sqlite]

Allows a connection URI string to be used for connecting to Mongo. When
using a connection URI you cannot provide the SSL certificate in a field on
settings page. This is a limitation of the underlying driver.

The conn-uri becomes a password field after it's been saved to avoid
displaying a password in cleartext.

Resolves #5084

[ci mongo]

When using a service account, the project-id isn't provided by the user,
but instead comes from the credentials for the service account. Handle
this case by looking for a project-id if one is provided explicitly,
otherwise defer to the credential object.

Additional change - during development, if (dev/init!) isn't called, the
event listeners are never setup, meaning that the initial syncs don't
occur.

Resolves #12648

[ci bigquery]

When the server (likely) requires SSL, you get an error message like:

    Caused by: com.mongodb.MongoTimeoutException: Timed out after 3000 ms while waiting to connect. Client view of cluster state is {type=UNKNOWN, servers=[{address=127.0.0.1:27017, type=UNKNOWN, state=CONNECTING, exception={com.mongodb.MongoSocketReadException: Prematurely reached end of stream}}]

When this is found, suggest that the user enable SSL.

Resolves #12605

[ci mongo]

* Backported fixes

* Explicitly include data.xml as a dep

* Use OffsetDateTime instead of org.h2.api.TimestampWithTimeZone [ci mysql]

* Revert JUnit tweaks -- don't seem to be working correctly

* Faster driver build strip-and-compress step by not loading Metabase core

* New cache key for FE deps

* Cache built drivers & local Maven installation of metabase-core

* Only cache the built driver JARs

* Cache built frontend

* Change driver cache key

* Try updating a specific driver [ci snowflake]

* Try updating a different driver [ci redshift] [ci snowflake]

* Need to cache driver checksums as well [ci snowflake] [ci redshift]

* Tweak what's cached for drivers (again)

This stops a failure during fingerprinting of a table from causing the
entire sync to fail.

Fingerprinting is now skipped for Google Analytics datasources, as they
do not support this (all queries against GA require a `metrics`
parameter)

Add an assert to ensure that `:metrics` is present in a GA query. The
Google library will fail with a NPE when it's not.

Resolves #12411

[ci googleanalytics]

This will allow the user to specify the server's certificate chain (the
public CA certs) or the server's private key if it's a self-signed cert
with no CA.

Resolves #3877

[ci mongo]

* Druid: don't make timestamps PKs [ci drivers]

* Fix tests [ci drivers]

* [ci drivers]

MBQL (Redshift, Vertica, Oracle): make `concat` work with any number of args

Adding SSH key support

Adding ssh key support (with passphrases)

The JSch library does not support ED25519 keys, that are the default as
of OpenSSH 7.8, but the Apache Mina SSHD library does.

Adds a end-to-end test of ssh port forwarding via a simple echo server.

[ci drivers]

Co-authored-by: Franklin Strube <thedoc8786@gmail.com>
Co-authored-by: hbc <bcxxxxxx@gmail.com>
Co-authored-by: Daniel Higginbotham <daniel@flyingmachinestudios.com>
Co-authored-by: Paul Rosenzweig <paul.a.rosenzweig@gmail.com>
Co-authored-by: Kyle Doherty <kyle.l.doherty@gmail.com>

Adds the ability to use a service account JSON key file for talking to
Google BigQuery

Resolves #4634

[ci bigquery]

Co-authored-by: Paul Rosenzweig <paul.a.rosenzweig@gmail.com>

* Druid: don't make timestamps PKs [ci drivers]

* Fix tests [ci drivers]

* [ci drivers]

MBQL (Redshift, Vertica, Oracle): make `concat` work with any number of args

* Fix #12501

* Make sure Query results_metadata comes back with inferred base type from annotate

* Support filtering against questions that use another question as their source [ci drivers]

* Test fixes 

* Lint fix [ci drivers]

* Clean up metabase.query-processor.parameters.mbql-test

* Tweak codecov requirements [ci drivers]

Newly added metadata in the query remark was blowing up for parameter
queries other than dimension / field-id queries.

[ci redshift]

* Allow a driver to customize the query remarks

This allows driver-specific metadata to be added to each query. If the
driver can apply the default metadata, it should call
default-query->remark

Removed timezone test that isn't testing the driver itself, since the
driver doesn't specify a default timezone

Removed deprecated code from the redshift test

Leave the original parameters from the front-end as :user-parameters so
that they are accessible down in the drivers

Transform the original parameters into the field_values in the query
remark

* Add more math functions to MBQL [ci drivers]

* Add non-default implementations [ci drivers]

* Add non-default implementations [ci all]

* More lenient schema [ci drivers]

* Add type hints [ci drivers]

* Fix tests [ci all]

* Fix tests [ci all]

* Presto: stddev_samp -> stddev_pop [ci all]

* Fix indentation [ci all]

* Add tests [ci drivers]

* Don't support percentile in H2 [ci drivers]

* Fix tests [ci all]

* Fix mssql [ci drivers]

* FE support for new functions

* Fix mssql [ci drivers]

* Fix mssql [ci drivers]

* Fix unit test

* Implement code review suggestions [ci all]

* Fix vertica [ci drivers]

* FE: add length & rename expt -> exp [ci drivers]

* Don't i18n displayName [ci drivers]

* FE fixes

* Cleanup aggregation/filter type predicates, isStandard, isMetric/isSegment, and isCustom

* Test: alow expressions in filters

* Alow expressions in filters [ci drivers]

* Allow expressions in :between [ci drivers]

* Allow expressions in :inside [ci drivers]

* Fix log [ci drivers]

* h2: use log10 [ci drivers]

* Fix mssql & spark [ci drivers]

* fix hive [ci drivers]

* Don't reuse :percentile name [ci drivers]

* Fix filter methods

Co-authored-by: Tom Robinson <tlrobinson@gmail.com>

*  If a driver returns base type `:type/*` (i.e., unknown type) in column metadata for native query results, attempt to infer better type by scanning values sample
*  Fix NPE in H2 native query results for columns where JDBC metadata `.getColumnClassName` returned nil
*  Rename `annotate/column-info*` to `annotate/merged-column-info` (makes the purpose of the function clearer)

* Fix extracting :minute-of-hour and :hour-of-day from BigQuery TIME columns [ci bigquery]

* Fix extract with unknown temporal types [ci bigquery]

* Upgrade Chevrotain to v6.5.0

* Switch from embedded actions to visitor

* rename tokens to lexer, remove no-longer needed Chevrotain hack from webpack.config.js

* get tests passing again

* progress converting syntax parser

* got intermediate Sum(A) test passing

* more progress on parsing expressions

* get complex expression passing

* prettier

* linting

* Add TokenizedExpression snapshot test

* Add support to parser for recovering whitespace tokens

* Add case clause [ci all]

* fix test [ci all]

* Harden test [ci all]

* More tests [ci all]

* Correctly infer case return type [ci all]

* Fix test [ci all]

* Fix tests [ci all]

* Improved syntax parser

* More parser cleanup + start adding filters and functions

* Add CASE and partial filter support, cleanup tests

* Add string extracts [ci all]

* Typo [ci all]

* Make linter happy [ci all]

* Add regex-match-first [ci all]

* Fix arglist [ci all]

* Fix test [ci all]

* Add sum-where count-where and share support. Refactor aggregation tokens

* Fix tests [ci all]

* Add ns prefixes [ci all]

* Add ns qualifiers [ci all]

* Fix trim [ci all]

* Tests: ensure stable ordering of results [ci all]

* Fix trim [ci all]

* Support only 1-arg trim [ci all]

* Fix ns prefixes [ci all]

* Fix psql [ci all]

* Fix redshift [ci redshift]

* Refactor/simplify grammer to consolidate aggregation, expression, and filter functions

* Add isCase to isExpression

* Directly splice regex pattern for psql & redshift [ci all]

* Fix schema [ci all]

* Redshift: splice replace args [ci redshift]

* Add ns prefix [ci redshift]

* Typo [ci redshift]

* Filter expressions

* Add metrics and segments back in to expressions

* Disable tokenized editing

* Update tests to clojure.test [ci all]

* Add concat to sqlite [ci all]

* Add length [ci all]

* Add missing ns [ci all]

* Fix test [ci all]

* Fix messed up multimethods [ci all]

* Fix mssql [ci all]

* Fix sqlite concat [ci all]

* sqlite: correctly qoute literal strings [ci all]

* Misc improvements to syntax highlighting etc

* Cleanup tests [ci all]

* Reoder defs [ci all]

* Make linter happy [ci all]

* Mongo: add case [ci mongo]

* mongo: fix switch [ci mongo]

* typo [ci mongo]

* Improve syntax highlighter and suggestions

* Mongo: always have default fallback in case [ci mongo]

* Rearrange test to split out ones using expressions [ci mongo]

* Remove uneeded merge [ci all]

* misc

* Much improved syntax highlighting using recovery mode and partial CST

* Improved suggestions

* try monospace font for expressions

* Better typing in parser, various other fixes

* Avoid parsing multiple times

* Fix tests

* Fix tests

* More test fixes

* Special case for UnderlyingRecords drill with sum-where/count-where/share

* switch back to square brackets for identifiers

* change extract to regexextract

* Lift expressions to subselect [ci all]

* Commit on enter

* rename extract to substitute, fix tests

* Refactor [ci all]

* Fix normalization [ci all]

* Correctly handle joins [ci all]

Co-authored-by: Daniel Higginbotham <daniel@flyingmachinestudios.com>
Co-authored-by: Simon Belak <simon@metabase.com>
Co-authored-by: Maz Ameli <maz@metabase.com>

* Replace util fns with ones from (updated) medley

* Clean ns

* Remove QP context cancelf and metadataf functions which aren't needed/used

* Test fixes 

* Port tests from expectations to clojure.test

* Fix docstring

* Added :question template tag type with lookup

* s/question/card/

* Fix typo: s/substitue/substitute/g

* Wrap card query parameter data in a record

Just like `FieldFilter`s.

* Unrelated: Improve test layout

* Add substitution of card query parameters in queries

This implementation is incomplete.

* Finish substitution of card query in parent query

* Basic FE support for question template tags

* Update card ID on tag name update

Also, don't use `\d` in regexes.

* yarn prettier

* Add support for MBQL queries in template tag lookup

* Return query strings directly as tag value

* Fix query type lookup

* Initialize QP store for MBQL→native conversion

* Fix MBQL to native query conversion

* Port tests from expectations to clojure.test

* Add middleware to resolve tables/fields from referenced cards

This needs to happen before `substitute-parameters` in the QP pipeline,
so that `substitute-parameters` can convert any MBQL queries (from
referenced cards) into native queries. Maybe it should be consolidated
with the other resolution steps.

* Use mbql-query util instead of manual query def

* Add docstring to now public function

* Remove unused import

* Fix docstring positioning 



* Clean ns declaration

* Convert tests to clojure.test

* Be more specific about errors we're looking expecting

Invalid queries caused unexpected exceptions to be thrown when checking
permissions.

* Recursively check permissions of referenced queries

* Add incomplete test [WIP]

* Finish permission check test for referenced MBQL queries

* Add tests for referenced native queries perms check

* Clean ns

* Use existing mbql-to-native middleware

Rather than calling driver-specific `mbql->native` directly.

* Fix formatting

* Use more comprehensive qp func for converting query to native

* front end updates

* Check that all referenced queries are from the same db

* Add test for checking db id of referenced MBQL query

* Clean ns

* add metadata and better error states to UI

* move error message beneath the question picker

* Combine `vals`+`filter`+`map` into a single `keep`

* Rename `CardQuery`→`ReferencedCardQuery`

* Cosmetic: fix typo and formatting

* Test substitution of multiple sub-queries

* Test CTE syntax substitution of multiple sub-queries

* Clean ns

* Clean ns

* Simplify referenced card substitution tests

* Test recursive sub-query substitution

* Cosmetic: typos

* Fix parameter names in comments

* Check that queries don't include circular sub-queries

* Clean ns

* Give users a more user-friendly error for circular referencing sub-queries

* Test referencing queries with parameters

* Update src/metabase/query_processor/middleware/resolve_referenced.clj

Co-Authored-By: Maz Ameli <maz@metabase.com>

* Fix error message and combine card name queries

* Wrap sub-query error in a more user friendly explanation

* Cosmetic: function argument alignment

* Test error handling of referenced sub-queries

* Remove unused 1-arity version of `card-subquery-graph`

* Cosmetic: indentation

* Normalize l10n of error messages

* Reuse functions for looking up referenced cards from query

* implement FE PR feedback

* card -> card_id backend changes

* fix tests

* fix another test

* disallow space between # and number

* Align new middleware with newer QP architecture

* Align sub-query error handling with the new QP

It seems like exceptions are handled more consistently in the new QP, so
we can be less hacky.

* (Re)wrap sub-query errors in user-friendly expalanation

* Rename `xformf`→`rff` to align with QP changes

* force cypress click on select placeholder

* move popover attachement and cap its height

* icon spacing

* customize highlighting for template tags

Co-authored-by: Paul Rosenzweig <paulrosenzweig@users.noreply.github.com>
Co-authored-by: Maz Ameli <maz@metabase.com>

[ci drivers]

* Fix test failures [ci drivers]

* Test fixes [ci drivers]