* APIv2: Make imports_skip_reason flag not appear when imports not requested
* Update lib/plausible/stats/query_result.ex
Co-authored-by: hq1 <hq@mtod.org>
---------
Co-authored-by: hq1 <hq@mtod.org>
`persist_session/1` is used in `populate_native_stats/1` test helper
to merge session data. It sometimes gets `:ok` instead of the actual
session data so that tests randomly fail with:
```
** (BadMapError) expected a map, got: :ok
code: populate_stats(site, [build(:pageview)])
stacktrace:
(elixir 1.17.1) lib/map.ex:486: Map.take/2
(plausible 0.0.1) lib/plausible/clickhouse_event_v2.ex:92: Plausible.ClickhouseEventV2.merge_session/2
(plausible 0.0.1) test/support/test_utils.ex:188: anonymous fn/2 in Plausible.TestUtils.populate_native_stats/1
(elixir 1.17.1) lib/enum.ex:2531: Enum."-reduce/3-lists^foldl/2-0-"/3
(plausible 0.0.1) test/support/test_utils.ex:184: Plausible.TestUtils.populate_native_stats/1
(plausible 0.0.1) test/support/test_utils.ex:179: Plausible.TestUtils.populate_stats/1
test/workers/send_site_setup_emails_test.exs:46: (test)
```
Offset-based pagination is used to make sure Looker integration
is able to work as efficiently as possible. To know how many
requests users need to do `include.total_rows` option was added.
* On the dashboard, allow Metrics (aka column configurations) to
* specify width they need
* specify if they are sortable
* Default sort order for breakdown endpoints is hardcoded on the dashboard (needed to show what column the report is sorted by)
* Unifies Google Keywords modal with other breakdowns
* query.date_range is now in UTC instead of user timezone
This simplifies things down the line and fixes several bugs where
query.date_range is cast to naivedatetime for ecto purposes
Many places still remain broken:
- comparison queries
- `to_date_range` calls
* Make default_for_date_range not care about time zones
* Make timezone parameter mandatory for to_date_range
* Simplify utc_date_range, update legacy query builder
* Fix more cases where query date range is needed
* query.date_range -> query.utc_time_range
* Query.date_range/1 function
* ensure_include_imported update
* Clean up send_email_report
* auto https
* changelog
* enable auto-tls only when http_port is 80
* make erlang ssl less verbose
* simplify config
* simplify config
* fix ce_dev typo
* fix another typo
---------
Co-authored-by: Adrian Gruntkowski <adrian.gruntkowski@gmail.com>
* Safeguard session queries relying on `sign` from faulty old session entries
* Comment updated metric
Co-authored-by: Karl-Aksel Puulmann <macobo@users.noreply.github.com>
* Apply safeguards to `bounce_rate` metric only
* Add note to bounce rate definition in SQL fragments as well
* Add test for graceful bounce rate handling in breakdown
* Make user_id more unique
* Add a note to the test
* Move regression test to APIv2 tests
---------
Co-authored-by: Karl-Aksel Puulmann <macobo@users.noreply.github.com>
* fix deciding whether a filter operation is freeChoice
* fix displaying loading spinner when loading combobox options
* extract lastLoadTimestamp into a separate context
* prettier
* order migrations
* migrate in 'streaks'
* cleanup
* docs
* tests
* add comment
* changelog
* continue
* run migrations tests in ci
* add pending_streams cmd
* fix ci
* add docs
* fix test
* simplify test
* only test v2+
* fix test
* return async: true
* cleanup
* add 'no pending migration'
* drop migrate/0 and pending_migrations/0
---------
Co-authored-by: Adrian Gruntkowski <adrian.gruntkowski@gmail.com>
* remove ch timeouts from ce
* just infinity
* make timeout configurable
* quadruple
* add test
* improve test
* fix test filtering in ci
* just don't compile the test in ee
* fix test in ee
* maybe this would work
* rm test
---------
Co-authored-by: Adrian Gruntkowski <adrian.gruntkowski@gmail.com>
* Rename matches/does_not_match filters internally
These have never been exposed to the frontend/user directly, only via
APIv1 filtering syntax. As such we are free to rename these without
breaking things
* Rename function arguments for consistency, simplify
* Add support for `match`/`not_match` operators for query apiv2
These match the string against a regular expression, as defined in
https://github.com/google/re2/wiki/Syntax
* not_match -> match_not
* does_not_contain -> contains_not
Note that for backwards compatibility:
- Browser handles does_not_contain in URL
- Backend will handle does_not_contain in queries for a day where we will remove it for better autocompletion
* not_matches_wildcard -> matches_wildcard_not
* prettier
* match -> matches
* Fix and test fix for matches_wildcard against prop when prop is missing
* Custom properties support for matches/matches_not
* Restore contains_not
* Test contains and contains_not behavior for custom properties
* Replace native tooltips with simplified, generic component
* Make tooltip a button?
* Revert "Make tooltip a button?"
This reverts commit adacc560f9.
Previous migration took forever on prod, likely because Map lookups are linear time in complexity.
`transform/3` helps achieve the same functionality with the help of a hash table and updated
WHERE clause allows skipping most rows which dont need updating
Co-authored-by: Uku Taht <Uku.taht@gmail.com>
* Dont include imports for time:hour and time:minute dimensions
Also include more information about import warnings in query results
* Update lib/plausible/stats/query_result.ex
Co-authored-by: RobertJoonas <56999674+RobertJoonas@users.noreply.github.com>
* Revert patch
* Imported disabled graph notice (#4522)
* add explicit skip_imported_reason for unsupported interval
* stop returning information about imports from main_graph
* return warning about interval in Stats API Timeseries
* display warning bubble about interval too short for imported data
* update changelog
* improve styling of the exclamation circle icon
* return tuple from timeseries instead of map
* rename variable
* Update CHANGELOG.md
---------
Co-authored-by: RobertJoonas <56999674+RobertJoonas@users.noreply.github.com>
After refactor, it turned out that when the _legacy_ session times out,
the conn is halted but no body is sent. This results in 500 error (`Plug.Conn.NotSentError`).
This PR fixes it by redirecting to login page.
* Remove custom background on "Choose Plan" page
* Avoid `UserAuth` test failing due to bad timing
---------
Co-authored-by: Adrian Gruntkowski <adrian.gruntkowski@gmail.com>
* Turn `Plausible.Auth.UserSession` into full schema
* Implement token based sessions and use them as default
* Ignore expired user sessions during retrieval from DB
* Implement plug bumping user session last used and timeout timestamps
* Implement Oban worker removing expired user sessions with grace period
* Implement legacy session conversion on touch, when applicable
* Update `UserAuth` moduledoc
* Extend `UserAuth` tests to account for db-backed session tokens
* Update CHANGELOG
* Add tests for `UserSessionTouch` plug
* Add test for `CleanUserSessions` worker
* Add logging of legacy session retrievals
* Use single update permitting stale records when touching user session
* Don't fetch session and user for external API endpoints (/api/event too)
* Refactor `Users.with_subscription/1` and expose helper query
* Skip fetching session in legacy `SessionTimeoutPlug`
* Rely on user session assign from `AuthContext` in `SentryContext`
* Silence legacy session warnings in `UserSessionTouchTest`
* Rely on session assign from `AuthPlug` in `SuperAdminOnlyPlug`
* Change `UserAuth` to get session, user and last subscription in one go
* Avoid refetching user session in `AuthorizeSiteAccess` plug
* Fix code formatting
* Refactor `UserAuth.get_user_token/1` (h/t @aerosol)
* Remove bogus empty opts from `scope` declarations in router
* Only touch session once an hour and keep `user.last_seen` in sync
* Bring back logging of legacy token use
* Create a regression demonstration test for race condition
* Use `ConCache.isolated/1` to force sequential processing of session events
* Revise comment in regression test
* Put lock call behind cache adapter API
* Add more explicit handling of failing lock
NOTE: Apparent double execution of lock function needs to be investigated.
* Improve slow lock cases tests
* Reduce number of session cache locks and instrument them w/ telemetry
* Format
---------
Co-authored-by: Adam Rutkowski <hq@mtod.org>
