mirror of
https://github.com/ClickHouse/ClickHouse.git
synced 2024-11-27 01:51:59 +00:00
Merge branch 'master' into rs/compile-with-cpp23
This commit is contained in:
Alexey Milovidov
GitHub
commit
f445649845
@ -89,7 +89,7 @@ CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster]
|
||||
└─────────────────────┴───────────┴──────────┴──────┘
|
||||
```
|
||||
|
||||
Первая строка отменяет предыдущее состояние объекта (пользователя). Она должен повторять все поля из ключа сортировки для отменённого состояния за исключением `Sign`.
|
||||
Первая строка отменяет предыдущее состояние объекта (пользователя). Она должна повторять все поля из ключа сортировки для отменённого состояния за исключением `Sign`.
|
||||
|
||||
Вторая строка содержит текущее состояние.
|
||||
|
||||
|
||||
@ -584,7 +584,7 @@ TTL d + INTERVAL 1 MONTH GROUP BY k1, k2 SET x = max(x), y = min(y);
|
||||
|
||||
Данные с истекшим `TTL` удаляются, когда ClickHouse мёржит куски данных.
|
||||
|
||||
Когда ClickHouse видит, что некоторые данные устарели, он выполняет внеплановые мёржи. Для управление частотой подобных мёржей, можно задать настройку `merge_with_ttl_timeout`. Если её значение слишком низкое, придется выполнять много внеплановых мёржей, которые могут начать потреблять значительную долю ресурсов сервера.
|
||||
Когда ClickHouse видит, что некоторые данные устарели, он выполняет внеплановые мёржи. Для управления частотой подобных мёржей, можно задать настройку `merge_with_ttl_timeout`. Если её значение слишком низкое, придется выполнять много внеплановых мёржей, которые могут начать потреблять значительную долю ресурсов сервера.
|
||||
|
||||
Если вы выполните запрос `SELECT` между слияниями вы можете получить устаревшие данные. Чтобы избежать этого используйте запрос [OPTIMIZE](../../../engines/table-engines/mergetree-family/mergetree.md#misc_operations-optimize) перед `SELECT`.
|
||||
|
||||
@ -679,7 +679,7 @@ TTL d + INTERVAL 1 MONTH GROUP BY k1, k2 SET x = max(x), y = min(y);
|
||||
- `policy_name_N` — название политики. Названия политик должны быть уникальны.
|
||||
- `volume_name_N` — название тома. Названия томов должны быть уникальны.
|
||||
- `disk` — диск, находящийся внутри тома.
|
||||
- `max_data_part_size_bytes` — максимальный размер куска данных, который может находится на любом из дисков этого тома. Если в результате слияния размер куска ожидается больше, чем max_data_part_size_bytes, то этот кусок будет записан в следующий том. В основном эта функция позволяет хранить новые / мелкие куски на горячем (SSD) томе и перемещать их на холодный (HDD) том, когда они достигают большого размера. Не используйте этот параметр, если политика имеет только один том.
|
||||
- `max_data_part_size_bytes` — максимальный размер куска данных, который может находиться на любом из дисков этого тома. Если в результате слияния размер куска ожидается больше, чем max_data_part_size_bytes, то этот кусок будет записан в следующий том. В основном эта функция позволяет хранить новые / мелкие куски на горячем (SSD) томе и перемещать их на холодный (HDD) том, когда они достигают большого размера. Не используйте этот параметр, если политика имеет только один том.
|
||||
- `move_factor` — доля доступного свободного места на томе, если места становится меньше, то данные начнут перемещение на следующий том, если он есть (по умолчанию 0.1). Для перемещения куски сортируются по размеру от большего к меньшему (по убыванию) и выбираются куски, совокупный размер которых достаточен для соблюдения условия `move_factor`, если совокупный размер всех партов недостаточен, будут перемещены все парты.
|
||||
- `prefer_not_to_merge` — Отключает слияние кусков данных, хранящихся на данном томе. Если данная настройка включена, то слияние данных, хранящихся на данном томе, не допускается. Это позволяет контролировать работу ClickHouse с медленными дисками.
|
||||
|
||||
@ -730,7 +730,7 @@ TTL d + INTERVAL 1 MONTH GROUP BY k1, k2 SET x = max(x), y = min(y);
|
||||
|
||||
В приведенном примере, политика `hdd_in_order` реализует прицип [round-robin](https://ru.wikipedia.org/wiki/Round-robin_(%D0%B0%D0%BB%D0%B3%D0%BE%D1%80%D0%B8%D1%82%D0%BC)). Так как в политике есть всего один том (`single`), то все записи производятся на его диски по круговому циклу. Такая политика может быть полезна при наличии в системе нескольких похожих дисков, но при этом не сконфигурирован RAID. Учтите, что каждый отдельный диск ненадёжен и чтобы не потерять важные данные это необходимо скомпенсировать за счет хранения данных в трёх копиях.
|
||||
|
||||
Если система содержит диски различных типов, то может пригодиться политика `moving_from_ssd_to_hdd`. В томе `hot` находится один SSD-диск (`fast_ssd`), а также задается ограничение на максимальный размер куска, который может храниться на этом томе (1GB). Все куски такой таблицы больше 1GB будут записываться сразу на том `cold`, в котором содержится один HDD-диск `disk1`. Также, при заполнении диска `fast_ssd` более чем на 80% данные будут переносится на диск `disk1` фоновым процессом.
|
||||
Если система содержит диски различных типов, то может пригодиться политика `moving_from_ssd_to_hdd`. В томе `hot` находится один SSD-диск (`fast_ssd`), а также задается ограничение на максимальный размер куска, который может храниться на этом томе (1GB). Все куски такой таблицы больше 1GB будут записываться сразу на том `cold`, в котором содержится один HDD-диск `disk1`. Также при заполнении диска `fast_ssd` более чем на 80% данные будут переноситься на диск `disk1` фоновым процессом.
|
||||
|
||||
Порядок томов в политиках хранения важен, при достижении условий на переполнение тома данные переносятся на следующий. Порядок дисков в томах так же важен, данные пишутся по очереди на каждый из них.
|
||||
|
||||
|
||||
@ -8,6 +8,7 @@ sidebar_label: "Клиентские библиотеки от сторонни
|
||||
|
||||
:::danger "Disclaimer"
|
||||
Яндекс не поддерживает перечисленные ниже библиотеки и не проводит тщательного тестирования для проверки их качества.
|
||||
:::
|
||||
|
||||
- Python:
|
||||
- [infi.clickhouse_orm](https://github.com/Infinidat/infi.clickhouse_orm)
|
||||
|
||||
@ -325,21 +325,21 @@ clickhouse-keeper-converter --zookeeper-logs-dir /var/lib/zookeeper/version-2 --
|
||||
Например, для кластера из 3 нод, алгоритм кворума продолжает работать при отказе не более чем одной ноды.
|
||||
|
||||
Конфигурация кластера может быть изменена динамически с некоторыми ограничениями.
|
||||
Переконфигурация также использует Raft, поэтому для добавление новой ноды кластера или исключения старой ноды из него требуется достижения кворума в рамках текущей конфигурации кластера.
|
||||
Переконфигурация также использует Raft, поэтому для добавления новой ноды кластера или исключения старой ноды требуется достижение кворума в рамках текущей конфигурации кластера.
|
||||
Если в вашем кластере произошел отказ большего числа нод, чем допускает Raft для вашей текущей конфигурации и у вас нет возможности восстановить их работоспособность, Raft перестанет работать и не позволит изменить конфигурацию стандартным механизмом.
|
||||
|
||||
Тем не менее ClickHousr Keeper имеет возможность запуститься в режиме восстановления, который позволяет переконфигурировать класте используя только одну ноду кластера.
|
||||
Тем не менее ClickHouse Keeper имеет возможность запуститься в режиме восстановления, который позволяет переконфигурировать кластер используя только одну ноду кластера.
|
||||
Этот механизм может использоваться только как крайняя мера, когда вы не можете восстановить существующие ноды кластера или запустить новый сервер с тем же идентификатором.
|
||||
|
||||
Важно:
|
||||
- Удостоверьтесь, что отказавшие ноды не смогут в дальнейшем подключиться к кластеру в будущем.
|
||||
- Не запускайте новые ноды, пока не завешите процедуру ниже.
|
||||
- Не запускайте новые ноды, пока не завершите процедуру ниже.
|
||||
|
||||
После того, как выполнили действия выше выполните следующие шаги.
|
||||
1. Выберете одну ноду Keeper, которая станет новым лидером. Учтите, что данные которые с этой ноды будут испольщзованы всем кластером, поэтому рекомендуется выбрать ноду с наиболее актуальным состоянием.
|
||||
1. Выберете одну ноду Keeper, которая станет новым лидером. Учтите, что данные с этой ноды будут использованы всем кластером, поэтому рекомендуется выбрать ноду с наиболее актуальным состоянием.
|
||||
2. Перед дальнейшими действиям сделайте резервную копию данных из директорий `log_storage_path` и `snapshot_storage_path`.
|
||||
3. Измените настройки на всех нодах кластера, которые вы собираетесь использовать.
|
||||
4. Отправьте команду `rcvr` на ноду, которую вы выбрали или остановите ее и запустите заново с аргументом `--force-recovery`. Это переведет ноду в режим восстановления.
|
||||
4. Отправьте команду `rcvr` на ноду, которую вы выбрали, или остановите ее и запустите заново с аргументом `--force-recovery`. Это переведет ноду в режим восстановления.
|
||||
5. Запускайте остальные ноды кластера по одной и проверяйте, что команда `mntr` возвращает `follower` в выводе состояния `zk_server_state` перед тем, как запустить следующую ноду.
|
||||
6. Пока нода работает в режиме восстановления, лидер будет возвращать ошибку на запрос `mntr` пока кворум не будет достигнут с помощью новых нод. Любые запросы от клиентов и постедователей будут возвращать ошибку.
|
||||
6. Пока нода работает в режиме восстановления, лидер будет возвращать ошибку на запрос `mntr` пока кворум не будет достигнут с помощью новых нод. Любые запросы от клиентов и последователей будут возвращать ошибку.
|
||||
7. После достижения кворума лидер перейдет в нормальный режим работы и станет обрабатывать все запросы через Raft. Удостоверьтесь, что запрос `mntr` возвращает `leader` в выводе состояния `zk_server_state`.
|
||||
|
||||
@ -10,6 +10,7 @@ ClickHouse поддерживает [OpenTelemetry](https://opentelemetry.io/)
|
||||
|
||||
:::danger "Предупреждение"
|
||||
Поддержка стандарта экспериментальная и будет со временем меняться.
|
||||
:::
|
||||
|
||||
## Обеспечение поддержки контекста трассировки в ClickHouse
|
||||
|
||||
|
||||
@ -26,6 +26,7 @@ ClickHouse перезагружает встроенные словари с з
|
||||
|
||||
:::danger "Внимание"
|
||||
Лучше не использовать, если вы только начали работать с ClickHouse.
|
||||
:::
|
||||
|
||||
Общий вид конфигурации:
|
||||
|
||||
@ -1064,6 +1065,7 @@ ClickHouse использует потоки из глобального пул
|
||||
|
||||
:::danger "Обратите внимание"
|
||||
Завершающий слеш обязателен.
|
||||
:::
|
||||
|
||||
**Пример**
|
||||
|
||||
@ -1330,6 +1332,7 @@ TCP порт для защищённого обмена данными с кли
|
||||
|
||||
:::danger "Обратите внимание"
|
||||
Завершающий слеш обязателен.
|
||||
:::
|
||||
|
||||
**Пример**
|
||||
|
||||
|
||||
@ -82,7 +82,7 @@ sidebar_label: "Хранение данных на внешних дисках"
|
||||
|
||||
- `type` — `encrypted`. Иначе зашифрованный диск создан не будет.
|
||||
- `disk` — тип диска для хранения данных.
|
||||
- `key` — ключ для шифрования и расшифровки. Тип: [Uint64](../sql-reference/data-types/int-uint.md). Вы можете использовать параметр `key_hex` для шифрования в шестнадцатеричной форме.
|
||||
- `key` — ключ для шифрования и расшифровки. Тип: [UInt64](../sql-reference/data-types/int-uint.md). Вы можете использовать параметр `key_hex` для шифрования в шестнадцатеричной форме.
|
||||
Вы можете указать несколько ключей, используя атрибут `id` (смотрите пример выше).
|
||||
|
||||
Необязательные параметры:
|
||||
|
||||
@ -10,6 +10,7 @@ ClickHouse поддерживает типы данных для отображ
|
||||
|
||||
:::danger "Предупреждение"
|
||||
Сейчас использование типов данных для работы с географическими структурами является экспериментальной возможностью. Чтобы использовать эти типы данных, включите настройку `allow_experimental_geo_types = 1`.
|
||||
:::
|
||||
|
||||
**См. также**
|
||||
- [Хранение географических структур данных](https://ru.wikipedia.org/wiki/GeoJSON).
|
||||
|
||||
@ -10,6 +10,7 @@ sidebar_label: Interval
|
||||
|
||||
:::danger "Внимание"
|
||||
Нельзя использовать типы данных `Interval` для хранения данных в таблице.
|
||||
:::
|
||||
|
||||
Структура:
|
||||
|
||||
|
||||
@ -34,7 +34,7 @@ SELECT tuple(1,'a') AS x, toTypeName(x)
|
||||
|
||||
## Особенности работы с типами данных {#osobennosti-raboty-s-tipami-dannykh}
|
||||
|
||||
При создании кортежа «на лету» ClickHouse автоматически определяет тип каждого аргументов как минимальный из типов, который может сохранить значение аргумента. Если аргумент — [NULL](../../sql-reference/data-types/tuple.md#null-literal), то тип элемента кортежа — [Nullable](nullable.md).
|
||||
При создании кортежа «на лету» ClickHouse автоматически определяет тип всех аргументов как минимальный из типов, который может сохранить значение аргумента. Если аргумент — [NULL](../../sql-reference/data-types/tuple.md#null-literal), то тип элемента кортежа — [Nullable](nullable.md).
|
||||
|
||||
Пример автоматического определения типа данных:
|
||||
|
||||
|
||||
@ -61,7 +61,7 @@ LAYOUT(POLYGON(STORE_POLYGON_KEY_COLUMN 1))
|
||||
- Мультиполигон. Представляет из себя массив полигонов. Каждый полигон задается двумерным массивом точек — первый элемент этого массива задает внешнюю границу полигона,
|
||||
последующие элементы могут задавать дырки, вырезаемые из него.
|
||||
|
||||
Точки могут задаваться массивом или кортежем из своих координат. В текущей реализации поддерживается только двумерные точки.
|
||||
Точки могут задаваться массивом или кортежем из своих координат. В текущей реализации поддерживаются только двумерные точки.
|
||||
|
||||
Пользователь может [загружать свои собственные данные](../../../sql-reference/dictionaries/external-dictionaries/external-dicts-dict-sources.md) во всех поддерживаемых ClickHouse форматах.
|
||||
|
||||
@ -80,7 +80,7 @@ LAYOUT(POLYGON(STORE_POLYGON_KEY_COLUMN 1))
|
||||
- `POLYGON`. Синоним к `POLYGON_INDEX_CELL`.
|
||||
|
||||
Запросы к словарю осуществляются с помощью стандартных [функций](../../../sql-reference/functions/ext-dict-functions.md) для работы со внешними словарями.
|
||||
Важным отличием является то, что здесь ключами будут являются точки, для которых хочется найти содержащий их полигон.
|
||||
Важным отличием является то, что здесь ключами являются точки, для которых хочется найти содержащий их полигон.
|
||||
|
||||
**Пример**
|
||||
|
||||
|
||||
@ -59,6 +59,7 @@ ClickHouse поддерживает следующие виды ключей:
|
||||
|
||||
:::danger "Обратите внимание"
|
||||
Ключ не надо дополнительно описывать в атрибутах.
|
||||
:::
|
||||
|
||||
### Числовой ключ {#ext_dict-numeric-key}
|
||||
|
||||
|
||||
@ -14,7 +14,7 @@ ClickHouse:
|
||||
- Периодически обновляет их и динамически подгружает отсутствующие значения.
|
||||
- Позволяет создавать внешние словари с помощью xml-файлов или [DDL-запросов](../../statements/create/dictionary.md#create-dictionary-query).
|
||||
|
||||
Конфигурация внешних словарей может находится в одном или нескольких xml-файлах. Путь к конфигурации указывается в параметре [dictionaries_config](../../../operations/server-configuration-parameters/settings.md#server_configuration_parameters-dictionaries_config).
|
||||
Конфигурация внешних словарей может находиться в одном или нескольких xml-файлах. Путь к конфигурации указывается в параметре [dictionaries_config](../../../operations/server-configuration-parameters/settings.md#server_configuration_parameters-dictionaries_config).
|
||||
|
||||
Словари могут загружаться при старте сервера или при первом использовании, в зависимости от настройки [dictionaries_lazy_load](../../../operations/server-configuration-parameters/settings.md#server_configuration_parameters-dictionaries_lazy_load).
|
||||
|
||||
|
||||
@ -22,7 +22,7 @@ sidebar_label: "Функции интроспекции"
|
||||
|
||||
ClickHouse сохраняет отчеты профилировщика в [журнал трассировки](../../operations/system-tables/trace_log.md#system_tables-trace_log) в системной таблице. Убедитесь, что таблица и профилировщик настроены правильно.
|
||||
|
||||
## addresssToLine {#addresstoline}
|
||||
## addressToLine {#addresstoline}
|
||||
|
||||
Преобразует адрес виртуальной памяти внутри процесса сервера ClickHouse в имя файла и номер строки в исходном коде ClickHouse.
|
||||
|
||||
|
||||
@ -9,6 +9,7 @@ slug: /ru/sql-reference/operators/exists
|
||||
|
||||
:::danger "Предупреждение"
|
||||
Ссылки на таблицы или столбцы основного запроса не поддерживаются в подзапросе.
|
||||
:::
|
||||
|
||||
**Синтаксис**
|
||||
|
||||
|
||||
@ -38,9 +38,9 @@ SELECT '1' IN (SELECT 1);
|
||||
└──────────────────────┘
|
||||
```
|
||||
|
||||
Если в качестве правой части оператора указано имя таблицы (например, `UserID IN users`), то это эквивалентно подзапросу `UserID IN (SELECT * FROM users)`. Это используется при работе с внешними данными, отправляемым вместе с запросом. Например, вместе с запросом может быть отправлено множество идентификаторов посетителей, загруженное во временную таблицу users, по которому следует выполнить фильтрацию.
|
||||
Если в качестве правой части оператора указано имя таблицы (например, `UserID IN users`), то это эквивалентно подзапросу `UserID IN (SELECT * FROM users)`. Это используется при работе с внешними данными, отправляемыми вместе с запросом. Например, вместе с запросом может быть отправлено множество идентификаторов посетителей, загруженное во временную таблицу users, по которому следует выполнить фильтрацию.
|
||||
|
||||
Если в качестве правой части оператора, указано имя таблицы, имеющий движок Set (подготовленное множество, постоянно находящееся в оперативке), то множество не будет создаваться заново при каждом запросе.
|
||||
Если в качестве правой части оператора, указано имя таблицы, имеющей движок Set (подготовленное множество, постоянно находящееся в оперативке), то множество не будет создаваться заново при каждом запросе.
|
||||
|
||||
В подзапросе может быть указано более одного столбца для фильтрации кортежей.
|
||||
Пример:
|
||||
@ -49,9 +49,9 @@ SELECT '1' IN (SELECT 1);
|
||||
SELECT (CounterID, UserID) IN (SELECT CounterID, UserID FROM ...) FROM ...
|
||||
```
|
||||
|
||||
Типы столбцов слева и справа оператора IN, должны совпадать.
|
||||
Типы столбцов слева и справа оператора IN должны совпадать.
|
||||
|
||||
Оператор IN и подзапрос могут встречаться в любой части запроса, в том числе в агрегатных и лямбда функциях.
|
||||
Оператор IN и подзапрос могут встречаться в любой части запроса, в том числе в агрегатных и лямбда-функциях.
|
||||
Пример:
|
||||
|
||||
``` sql
|
||||
@ -122,7 +122,7 @@ FROM t_null
|
||||
|
||||
Существует два варианта IN-ов с подзапросами (аналогично для JOIN-ов): обычный `IN` / `JOIN` и `GLOBAL IN` / `GLOBAL JOIN`. Они отличаются способом выполнения при распределённой обработке запроса.
|
||||
|
||||
:::note "Attention"
|
||||
:::note "Внимание"
|
||||
Помните, что алгоритмы, описанные ниже, могут работать иначе в зависимости от [настройки](../../operations/settings/settings.md) `distributed_product_mode`.
|
||||
:::
|
||||
При использовании обычного IN-а, запрос отправляется на удалённые серверы, и на каждом из них выполняются подзапросы в секциях `IN` / `JOIN`.
|
||||
@ -228,7 +228,7 @@ SELECT CounterID, count() FROM distributed_table_1 WHERE UserID IN (SELECT UserI
|
||||
SETTINGS max_parallel_replicas=3
|
||||
```
|
||||
|
||||
преобразуются на каждом сервере в
|
||||
преобразуется на каждом сервере в
|
||||
|
||||
```sql
|
||||
SELECT CounterID, count() FROM local_table_1 WHERE UserID IN (SELECT UserID FROM local_table_2 WHERE CounterID < 100)
|
||||
|
||||
@ -263,6 +263,7 @@ SELECT toDateTime('2014-10-26 00:00:00', 'Europe/Moscow') AS time, time + 60 * 6
|
||||
│ 2014-10-26 00:00:00 │ 2014-10-26 23:00:00 │ 2014-10-27 00:00:00 │
|
||||
└─────────────────────┴─────────────────────┴─────────────────────┘
|
||||
```
|
||||
:::
|
||||
|
||||
**Смотрите также**
|
||||
|
||||
|
||||
@ -6,7 +6,7 @@ sidebar_label: VIEW
|
||||
|
||||
# Выражение ALTER TABLE … MODIFY QUERY {#alter-modify-query}
|
||||
|
||||
Вы можеие изменить запрос `SELECT`, который был задан при создании [материализованного представления](../create/view.md#materialized), с помощью запроса 'ALTER TABLE … MODIFY QUERY'. Используйте его если при создании материализованного представления не использовалась секция `TO [db.]name`. Настройка `allow_experimental_alter_materialized_view_structure` должна быть включена.
|
||||
Вы можете изменить запрос `SELECT`, который был задан при создании [материализованного представления](../create/view.md#materialized), с помощью запроса 'ALTER TABLE … MODIFY QUERY'. Используйте его если при создании материализованного представления не использовалась секция `TO [db.]name`. Настройка `allow_experimental_alter_materialized_view_structure` должна быть включена.
|
||||
|
||||
Если при создании материализованного представления использовалась конструкция `TO [db.]name`, то для изменения отсоедините представление с помощью [DETACH](../detach.md), измените таблицу с помощью [ALTER TABLE](index.md), а затем снова присоедините запрос с помощью [ATTACH](../attach.md).
|
||||
|
||||
|
||||
@ -10,6 +10,7 @@ sidebar_label: OPTIMIZE
|
||||
|
||||
:::danger "Внимание"
|
||||
`OPTIMIZE` не устраняет причину появления ошибки `Too many parts`.
|
||||
:::
|
||||
|
||||
**Синтаксис**
|
||||
|
||||
|
||||
@ -2,18 +2,21 @@
|
||||
|
||||
#include <Common/SipHash.h>
|
||||
#include <Common/FieldVisitorToString.h>
|
||||
#include <DataTypes/IDataType.h>
|
||||
#include <Analyzer/ConstantNode.h>
|
||||
|
||||
#include <IO/WriteBufferFromString.h>
|
||||
#include <IO/Operators.h>
|
||||
|
||||
#include <DataTypes/IDataType.h>
|
||||
#include <DataTypes/DataTypeSet.h>
|
||||
|
||||
#include <Parsers/ASTFunction.h>
|
||||
|
||||
#include <Functions/IFunction.h>
|
||||
|
||||
#include <AggregateFunctions/IAggregateFunction.h>
|
||||
|
||||
#include <Analyzer/Utils.h>
|
||||
#include <Analyzer/ConstantNode.h>
|
||||
#include <Analyzer/IdentifierNode.h>
|
||||
|
||||
namespace DB
|
||||
@ -44,17 +47,29 @@ const DataTypes & FunctionNode::getArgumentTypes() const
|
||||
ColumnsWithTypeAndName FunctionNode::getArgumentColumns() const
|
||||
{
|
||||
const auto & arguments = getArguments().getNodes();
|
||||
size_t arguments_size = arguments.size();
|
||||
|
||||
ColumnsWithTypeAndName argument_columns;
|
||||
argument_columns.reserve(arguments.size());
|
||||
|
||||
for (const auto & arg : arguments)
|
||||
for (size_t i = 0; i < arguments_size; ++i)
|
||||
{
|
||||
ColumnWithTypeAndName argument;
|
||||
argument.type = arg->getResultType();
|
||||
if (auto * constant = arg->as<ConstantNode>())
|
||||
argument.column = argument.type->createColumnConst(1, constant->getValue());
|
||||
argument_columns.push_back(std::move(argument));
|
||||
const auto & argument = arguments[i];
|
||||
|
||||
ColumnWithTypeAndName argument_column;
|
||||
|
||||
if (isNameOfInFunction(function_name) && i == 1)
|
||||
argument_column.type = std::make_shared<DataTypeSet>();
|
||||
else
|
||||
argument_column.type = argument->getResultType();
|
||||
|
||||
auto * constant = argument->as<ConstantNode>();
|
||||
if (constant && !isNotCreatable(argument_column.type))
|
||||
argument_column.column = argument_column.type->createColumnConst(1, constant->getValue());
|
||||
|
||||
argument_columns.push_back(std::move(argument_column));
|
||||
}
|
||||
|
||||
return argument_columns;
|
||||
}
|
||||
|
||||
|
||||
@ -99,8 +99,9 @@ class InDepthQueryTreeVisitorWithContext
|
||||
public:
|
||||
using VisitQueryTreeNodeType = std::conditional_t<const_visitor, const QueryTreeNodePtr, QueryTreeNodePtr>;
|
||||
|
||||
explicit InDepthQueryTreeVisitorWithContext(ContextPtr context)
|
||||
explicit InDepthQueryTreeVisitorWithContext(ContextPtr context, size_t initial_subquery_depth = 0)
|
||||
: current_context(std::move(context))
|
||||
, subquery_depth(initial_subquery_depth)
|
||||
{}
|
||||
|
||||
/// Return true if visitor should traverse tree top to bottom, false otherwise
|
||||
@ -125,11 +126,17 @@ public:
|
||||
return current_context->getSettingsRef();
|
||||
}
|
||||
|
||||
size_t getSubqueryDepth() const
|
||||
{
|
||||
return subquery_depth;
|
||||
}
|
||||
|
||||
void visit(VisitQueryTreeNodeType & query_tree_node)
|
||||
{
|
||||
auto current_scope_context_ptr = current_context;
|
||||
SCOPE_EXIT(
|
||||
current_context = std::move(current_scope_context_ptr);
|
||||
--subquery_depth;
|
||||
);
|
||||
|
||||
if (auto * query_node = query_tree_node->template as<QueryNode>())
|
||||
@ -137,6 +144,8 @@ public:
|
||||
else if (auto * union_node = query_tree_node->template as<UnionNode>())
|
||||
current_context = union_node->getContext();
|
||||
|
||||
++subquery_depth;
|
||||
|
||||
bool traverse_top_to_bottom = getDerived().shouldTraverseTopToBottom();
|
||||
if (!traverse_top_to_bottom)
|
||||
visitChildren(query_tree_node);
|
||||
@ -145,7 +154,12 @@ public:
|
||||
|
||||
if (traverse_top_to_bottom)
|
||||
visitChildren(query_tree_node);
|
||||
|
||||
getDerived().leaveImpl(query_tree_node);
|
||||
}
|
||||
|
||||
void leaveImpl(VisitQueryTreeNodeType & node [[maybe_unused]])
|
||||
{}
|
||||
private:
|
||||
Derived & getDerived()
|
||||
{
|
||||
@ -172,6 +186,7 @@ private:
|
||||
}
|
||||
|
||||
ContextPtr current_context;
|
||||
size_t subquery_depth = 0;
|
||||
};
|
||||
|
||||
template <typename Derived>
|
||||
|
||||
@ -106,6 +106,12 @@ public:
|
||||
return locality;
|
||||
}
|
||||
|
||||
/// Set join locality
|
||||
void setLocality(JoinLocality locality_value)
|
||||
{
|
||||
locality = locality_value;
|
||||
}
|
||||
|
||||
/// Get join strictness
|
||||
JoinStrictness getStrictness() const
|
||||
{
|
||||
|
||||
@ -42,7 +42,7 @@ private:
|
||||
return;
|
||||
|
||||
const auto & storage = table_node ? table_node->getStorage() : table_function_node->getStorage();
|
||||
bool is_final_supported = storage && storage->supportsFinal() && !storage->isRemote();
|
||||
bool is_final_supported = storage && storage->supportsFinal();
|
||||
if (!is_final_supported)
|
||||
return;
|
||||
|
||||
|
||||
@ -7,8 +7,6 @@
|
||||
#include <Analyzer/ConstantNode.h>
|
||||
#include <Analyzer/HashUtils.h>
|
||||
|
||||
#include <DataTypes/DataTypeString.h>
|
||||
|
||||
namespace DB
|
||||
{
|
||||
|
||||
@ -100,6 +98,9 @@ private:
|
||||
}
|
||||
}
|
||||
|
||||
if (and_operands.size() == function_node.getArguments().getNodes().size())
|
||||
return;
|
||||
|
||||
if (and_operands.size() == 1)
|
||||
{
|
||||
/// AND operator can have UInt8 or bool as its type.
|
||||
@ -207,6 +208,9 @@ private:
|
||||
or_operands.push_back(std::move(in_function));
|
||||
}
|
||||
|
||||
if (or_operands.size() == function_node.getArguments().getNodes().size())
|
||||
return;
|
||||
|
||||
if (or_operands.size() == 1)
|
||||
{
|
||||
/// if the result type of operand is the same as the result type of OR
|
||||
@ -69,8 +69,7 @@ private:
|
||||
for (auto it = function_arguments.rbegin(); it != function_arguments.rend(); ++it)
|
||||
candidates.push_back({ *it, is_deterministic });
|
||||
|
||||
// Using DFS we traverse function tree and try to find if it uses other keys as function arguments.
|
||||
// TODO: Also process CONSTANT here. We can simplify GROUP BY x, x + 1 to GROUP BY x.
|
||||
/// Using DFS we traverse function tree and try to find if it uses other keys as function arguments.
|
||||
while (!candidates.empty())
|
||||
{
|
||||
auto [candidate, parents_are_only_deterministic] = candidates.back();
|
||||
@ -108,6 +107,7 @@ private:
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
return true;
|
||||
}
|
||||
|
||||
|
||||
@ -193,13 +193,9 @@ namespace ErrorCodes
|
||||
* lookup should not be continued, and exception must be thrown because if lookup continues identifier can be resolved from parent scope.
|
||||
*
|
||||
* TODO: Update exception messages
|
||||
* TODO: JOIN TREE subquery constant columns
|
||||
* TODO: Table identifiers with optional UUID.
|
||||
* TODO: Lookup functions arrayReduce(sum, [1, 2, 3]);
|
||||
* TODO: SELECT (compound_expression).*, (compound_expression).COLUMNS are not supported on parser level.
|
||||
* TODO: SELECT a.b.c.*, a.b.c.COLUMNS. Qualified matcher where identifier size is greater than 2 are not supported on parser level.
|
||||
* TODO: Support function identifier resolve from parent query scope, if lambda in parent scope does not capture any columns.
|
||||
* TODO: Scalar subqueries cache.
|
||||
*/
|
||||
|
||||
namespace
|
||||
@ -701,7 +697,9 @@ struct IdentifierResolveScope
|
||||
}
|
||||
|
||||
if (auto * union_node = scope_node->as<UnionNode>())
|
||||
{
|
||||
context = union_node->getContext();
|
||||
}
|
||||
else if (auto * query_node = scope_node->as<QueryNode>())
|
||||
{
|
||||
context = query_node->getContext();
|
||||
@ -1336,6 +1334,9 @@ private:
|
||||
/// Global resolve expression node to projection names map
|
||||
std::unordered_map<QueryTreeNodePtr, ProjectionNames> resolved_expressions;
|
||||
|
||||
/// Global resolve expression node to tree size
|
||||
std::unordered_map<QueryTreeNodePtr, size_t> node_to_tree_size;
|
||||
|
||||
/// Global scalar subquery to scalar value map
|
||||
std::unordered_map<QueryTreeNodePtrWithHash, Block> scalar_subquery_to_scalar_value;
|
||||
|
||||
@ -1864,7 +1865,10 @@ void QueryAnalyzer::evaluateScalarSubqueryIfNeeded(QueryTreeNodePtr & node, Iden
|
||||
|
||||
Block scalar_block;
|
||||
|
||||
QueryTreeNodePtrWithHash node_with_hash(node);
|
||||
auto node_without_alias = node->clone();
|
||||
node_without_alias->removeAlias();
|
||||
|
||||
QueryTreeNodePtrWithHash node_with_hash(node_without_alias);
|
||||
auto scalar_value_it = scalar_subquery_to_scalar_value.find(node_with_hash);
|
||||
|
||||
if (scalar_value_it != scalar_subquery_to_scalar_value.end())
|
||||
@ -1954,21 +1958,7 @@ void QueryAnalyzer::evaluateScalarSubqueryIfNeeded(QueryTreeNodePtr & node, Iden
|
||||
*
|
||||
* Example: SELECT (SELECT 2 AS x, x)
|
||||
*/
|
||||
NameSet block_column_names;
|
||||
size_t unique_column_name_counter = 1;
|
||||
|
||||
for (auto & column_with_type : block)
|
||||
{
|
||||
if (!block_column_names.contains(column_with_type.name))
|
||||
{
|
||||
block_column_names.insert(column_with_type.name);
|
||||
continue;
|
||||
}
|
||||
|
||||
column_with_type.name += '_';
|
||||
column_with_type.name += std::to_string(unique_column_name_counter);
|
||||
++unique_column_name_counter;
|
||||
}
|
||||
makeUniqueColumnNamesInBlock(block);
|
||||
|
||||
scalar_block.insert({
|
||||
ColumnTuple::create(block.getColumns()),
|
||||
@ -2348,7 +2338,13 @@ QueryTreeNodePtr QueryAnalyzer::tryResolveTableIdentifierFromDatabaseCatalog(con
|
||||
storage_id = context->resolveStorageID(storage_id);
|
||||
bool is_temporary_table = storage_id.getDatabaseName() == DatabaseCatalog::TEMPORARY_DATABASE;
|
||||
|
||||
auto storage = DatabaseCatalog::instance().tryGetTable(storage_id, context);
|
||||
StoragePtr storage;
|
||||
|
||||
if (is_temporary_table)
|
||||
storage = DatabaseCatalog::instance().getTable(storage_id, context);
|
||||
else
|
||||
storage = DatabaseCatalog::instance().tryGetTable(storage_id, context);
|
||||
|
||||
if (!storage)
|
||||
return {};
|
||||
|
||||
@ -2914,7 +2910,10 @@ QueryTreeNodePtr QueryAnalyzer::tryResolveIdentifierFromTableExpression(const Id
|
||||
break;
|
||||
|
||||
IdentifierLookup column_identifier_lookup = {qualified_identifier_with_removed_part, IdentifierLookupContext::EXPRESSION};
|
||||
if (tryBindIdentifierToAliases(column_identifier_lookup, scope) ||
|
||||
if (tryBindIdentifierToAliases(column_identifier_lookup, scope))
|
||||
break;
|
||||
|
||||
if (table_expression_data.should_qualify_columns &&
|
||||
tryBindIdentifierToTableExpressions(column_identifier_lookup, table_expression_node, scope))
|
||||
break;
|
||||
|
||||
@ -3018,11 +3017,39 @@ QueryTreeNodePtr QueryAnalyzer::tryResolveIdentifierFromJoin(const IdentifierLoo
|
||||
|
||||
resolved_identifier = std::move(result_column_node);
|
||||
}
|
||||
else if (scope.joins_count == 1 && scope.context->getSettingsRef().single_join_prefer_left_table)
|
||||
else if (left_resolved_identifier->isEqual(*right_resolved_identifier, IQueryTreeNode::CompareOptions{.compare_aliases = false}))
|
||||
{
|
||||
const auto & identifier_path_part = identifier_lookup.identifier.front();
|
||||
auto * left_resolved_identifier_column = left_resolved_identifier->as<ColumnNode>();
|
||||
auto * right_resolved_identifier_column = right_resolved_identifier->as<ColumnNode>();
|
||||
|
||||
if (left_resolved_identifier_column && right_resolved_identifier_column)
|
||||
{
|
||||
const auto & left_column_source_alias = left_resolved_identifier_column->getColumnSource()->getAlias();
|
||||
const auto & right_column_source_alias = right_resolved_identifier_column->getColumnSource()->getAlias();
|
||||
|
||||
/** If column from right table was resolved using alias, we prefer column from right table.
|
||||
*
|
||||
* Example: SELECT dummy FROM system.one JOIN system.one AS A ON A.dummy = system.one.dummy;
|
||||
*
|
||||
* If alias is specified for left table, and alias is not specified for right table and identifier was resolved
|
||||
* without using left table alias, we prefer column from right table.
|
||||
*
|
||||
* Example: SELECT dummy FROM system.one AS A JOIN system.one ON A.dummy = system.one.dummy;
|
||||
*
|
||||
* Otherwise we prefer column from left table.
|
||||
*/
|
||||
if (identifier_path_part == right_column_source_alias)
|
||||
return right_resolved_identifier;
|
||||
else if (!left_column_source_alias.empty() &&
|
||||
right_column_source_alias.empty() &&
|
||||
identifier_path_part != left_column_source_alias)
|
||||
return right_resolved_identifier;
|
||||
}
|
||||
|
||||
return left_resolved_identifier;
|
||||
}
|
||||
else if (left_resolved_identifier->isEqual(*right_resolved_identifier, IQueryTreeNode::CompareOptions{.compare_aliases = false}))
|
||||
else if (scope.joins_count == 1 && scope.context->getSettingsRef().single_join_prefer_left_table)
|
||||
{
|
||||
return left_resolved_identifier;
|
||||
}
|
||||
@ -4466,6 +4493,7 @@ ProjectionNames QueryAnalyzer::resolveFunction(QueryTreeNodePtr & node, Identifi
|
||||
bool is_special_function_dict_get = false;
|
||||
bool is_special_function_join_get = false;
|
||||
bool is_special_function_exists = false;
|
||||
bool is_special_function_if = false;
|
||||
|
||||
if (!lambda_expression_untyped)
|
||||
{
|
||||
@ -4473,6 +4501,7 @@ ProjectionNames QueryAnalyzer::resolveFunction(QueryTreeNodePtr & node, Identifi
|
||||
is_special_function_dict_get = functionIsDictGet(function_name);
|
||||
is_special_function_join_get = functionIsJoinGet(function_name);
|
||||
is_special_function_exists = function_name == "exists";
|
||||
is_special_function_if = function_name == "if";
|
||||
|
||||
auto function_name_lowercase = Poco::toLower(function_name);
|
||||
|
||||
@ -4571,6 +4600,60 @@ ProjectionNames QueryAnalyzer::resolveFunction(QueryTreeNodePtr & node, Identifi
|
||||
is_special_function_in = true;
|
||||
}
|
||||
|
||||
if (is_special_function_if && !function_node_ptr->getArguments().getNodes().empty())
|
||||
{
|
||||
/** Handle special case with constant If function, even if some of the arguments are invalid.
|
||||
*
|
||||
* SELECT if(hasColumnInTable('system', 'numbers', 'not_existing_column'), not_existing_column, 5) FROM system.numbers;
|
||||
*/
|
||||
auto & if_function_arguments = function_node_ptr->getArguments().getNodes();
|
||||
auto if_function_condition = if_function_arguments[0];
|
||||
resolveExpressionNode(if_function_condition, scope, false /*allow_lambda_expression*/, false /*allow_table_expression*/);
|
||||
|
||||
auto constant_condition = tryExtractConstantFromConditionNode(if_function_condition);
|
||||
|
||||
if (constant_condition.has_value() && if_function_arguments.size() == 3)
|
||||
{
|
||||
QueryTreeNodePtr constant_if_result_node;
|
||||
QueryTreeNodePtr possibly_invalid_argument_node;
|
||||
|
||||
if (*constant_condition)
|
||||
{
|
||||
possibly_invalid_argument_node = if_function_arguments[2];
|
||||
constant_if_result_node = if_function_arguments[1];
|
||||
}
|
||||
else
|
||||
{
|
||||
possibly_invalid_argument_node = if_function_arguments[1];
|
||||
constant_if_result_node = if_function_arguments[2];
|
||||
}
|
||||
|
||||
bool apply_constant_if_optimization = false;
|
||||
|
||||
try
|
||||
{
|
||||
resolveExpressionNode(possibly_invalid_argument_node,
|
||||
scope,
|
||||
false /*allow_lambda_expression*/,
|
||||
false /*allow_table_expression*/);
|
||||
}
|
||||
catch (...)
|
||||
{
|
||||
apply_constant_if_optimization = true;
|
||||
}
|
||||
|
||||
if (apply_constant_if_optimization)
|
||||
{
|
||||
auto result_projection_names = resolveExpressionNode(constant_if_result_node,
|
||||
scope,
|
||||
false /*allow_lambda_expression*/,
|
||||
false /*allow_table_expression*/);
|
||||
node = std::move(constant_if_result_node);
|
||||
return result_projection_names;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// Resolve function arguments
|
||||
|
||||
bool allow_table_expressions = is_special_function_in;
|
||||
@ -5059,7 +5142,7 @@ ProjectionNames QueryAnalyzer::resolveFunction(QueryTreeNodePtr & node, Identifi
|
||||
|
||||
/// Do not constant fold get scalar functions
|
||||
bool disable_constant_folding = function_name == "__getScalar" || function_name == "shardNum" ||
|
||||
function_name == "shardCount";
|
||||
function_name == "shardCount" || function_name == "hostName";
|
||||
|
||||
/** If function is suitable for constant folding try to convert it to constant.
|
||||
* Example: SELECT plus(1, 1);
|
||||
@ -5085,7 +5168,8 @@ ProjectionNames QueryAnalyzer::resolveFunction(QueryTreeNodePtr & node, Identifi
|
||||
/** Do not perform constant folding if there are aggregate or arrayJoin functions inside function.
|
||||
* Example: SELECT toTypeName(sum(number)) FROM numbers(10);
|
||||
*/
|
||||
if (column && isColumnConst(*column) && (!hasAggregateFunctionNodes(node) && !hasFunctionNode(node, "arrayJoin")))
|
||||
if (column && isColumnConst(*column) && !typeid_cast<const ColumnConst *>(column.get())->getDataColumn().isDummy() &&
|
||||
(!hasAggregateFunctionNodes(node) && !hasFunctionNode(node, "arrayJoin")))
|
||||
{
|
||||
/// Replace function node with result constant node
|
||||
Field column_constant_value;
|
||||
@ -5433,9 +5517,9 @@ ProjectionNames QueryAnalyzer::resolveExpressionNode(QueryTreeNodePtr & node, Id
|
||||
}
|
||||
}
|
||||
|
||||
if (node
|
||||
&& scope.nullable_group_by_keys.contains(node)
|
||||
&& !scope.expressions_in_resolve_process_stack.hasAggregateFunction())
|
||||
validateTreeSize(node, scope.context->getSettingsRef().max_expanded_ast_elements, node_to_tree_size);
|
||||
|
||||
if (scope.nullable_group_by_keys.contains(node) && !scope.expressions_in_resolve_process_stack.hasAggregateFunction())
|
||||
{
|
||||
node = node->clone();
|
||||
node->convertToNullable();
|
||||
@ -6592,6 +6676,17 @@ void QueryAnalyzer::resolveQuery(const QueryTreeNodePtr & query_node, Identifier
|
||||
|
||||
/// Resolve query node sections.
|
||||
|
||||
NamesAndTypes projection_columns;
|
||||
|
||||
if (!scope.group_by_use_nulls)
|
||||
{
|
||||
projection_columns = resolveProjectionExpressionNodeList(query_node_typed.getProjectionNode(), scope);
|
||||
if (query_node_typed.getProjection().getNodes().empty())
|
||||
throw Exception(ErrorCodes::EMPTY_LIST_OF_COLUMNS_QUERIED,
|
||||
"Empty list of columns in projection. In scope {}",
|
||||
scope.scope_node->formatASTForErrorMessage());
|
||||
}
|
||||
|
||||
if (query_node_typed.hasWith())
|
||||
resolveExpressionNodeList(query_node_typed.getWithNode(), scope, true /*allow_lambda_expression*/, false /*allow_table_expression*/);
|
||||
|
||||
@ -6686,11 +6781,14 @@ void QueryAnalyzer::resolveQuery(const QueryTreeNodePtr & query_node, Identifier
|
||||
convertLimitOffsetExpression(query_node_typed.getOffset(), "OFFSET", scope);
|
||||
}
|
||||
|
||||
auto projection_columns = resolveProjectionExpressionNodeList(query_node_typed.getProjectionNode(), scope);
|
||||
if (scope.group_by_use_nulls)
|
||||
{
|
||||
projection_columns = resolveProjectionExpressionNodeList(query_node_typed.getProjectionNode(), scope);
|
||||
if (query_node_typed.getProjection().getNodes().empty())
|
||||
throw Exception(ErrorCodes::EMPTY_LIST_OF_COLUMNS_QUERIED,
|
||||
"Empty list of columns in projection. In scope {}",
|
||||
scope.scope_node->formatASTForErrorMessage());
|
||||
}
|
||||
|
||||
/** Resolve nodes with duplicate aliases.
|
||||
* Table expressions cannot have duplicate aliases.
|
||||
@ -6757,6 +6855,15 @@ void QueryAnalyzer::resolveQuery(const QueryTreeNodePtr & query_node, Identifier
|
||||
|
||||
validateAggregates(query_node, { .group_by_use_nulls = scope.group_by_use_nulls });
|
||||
|
||||
for (const auto & column : projection_columns)
|
||||
{
|
||||
if (isNotCreatable(column.type))
|
||||
throw Exception(ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT,
|
||||
"Invalid projection column with type {}. In scope {}",
|
||||
column.type->getName(),
|
||||
scope.scope_node->formatASTForErrorMessage());
|
||||
}
|
||||
|
||||
/** WITH section can be safely removed, because WITH section only can provide aliases to query expressions
|
||||
* and CTE for other sections to use.
|
||||
*
|
||||
|
||||
@ -60,13 +60,18 @@ bool TableNode::isEqualImpl(const IQueryTreeNode & rhs) const
|
||||
}
|
||||
|
||||
void TableNode::updateTreeHashImpl(HashState & state) const
|
||||
{
|
||||
if (!temporary_table_name.empty())
|
||||
{
|
||||
state.update(temporary_table_name.size());
|
||||
state.update(temporary_table_name);
|
||||
}
|
||||
else
|
||||
{
|
||||
auto full_name = storage_id.getFullNameNotQuoted();
|
||||
state.update(full_name.size());
|
||||
state.update(full_name);
|
||||
|
||||
state.update(temporary_table_name.size());
|
||||
state.update(temporary_table_name);
|
||||
}
|
||||
|
||||
if (table_expression_modifiers)
|
||||
table_expression_modifiers->updateTreeHash(state);
|
||||
|
||||
@ -8,6 +8,7 @@
|
||||
#include <DataTypes/DataTypeString.h>
|
||||
#include <DataTypes/DataTypeTuple.h>
|
||||
#include <DataTypes/DataTypeArray.h>
|
||||
#include <DataTypes/DataTypeLowCardinality.h>
|
||||
|
||||
#include <Functions/FunctionHelpers.h>
|
||||
#include <Functions/FunctionFactory.h>
|
||||
@ -32,6 +33,7 @@ namespace DB
|
||||
namespace ErrorCodes
|
||||
{
|
||||
extern const int LOGICAL_ERROR;
|
||||
extern const int BAD_ARGUMENTS;
|
||||
}
|
||||
|
||||
bool isNodePartOfTree(const IQueryTreeNode * node, const IQueryTreeNode * root)
|
||||
@ -79,6 +81,75 @@ bool isNameOfInFunction(const std::string & function_name)
|
||||
return is_special_function_in;
|
||||
}
|
||||
|
||||
bool isNameOfLocalInFunction(const std::string & function_name)
|
||||
{
|
||||
bool is_special_function_in = function_name == "in" ||
|
||||
function_name == "notIn" ||
|
||||
function_name == "nullIn" ||
|
||||
function_name == "notNullIn" ||
|
||||
function_name == "inIgnoreSet" ||
|
||||
function_name == "notInIgnoreSet" ||
|
||||
function_name == "nullInIgnoreSet" ||
|
||||
function_name == "notNullInIgnoreSet";
|
||||
|
||||
return is_special_function_in;
|
||||
}
|
||||
|
||||
bool isNameOfGlobalInFunction(const std::string & function_name)
|
||||
{
|
||||
bool is_special_function_in = function_name == "globalIn" ||
|
||||
function_name == "globalNotIn" ||
|
||||
function_name == "globalNullIn" ||
|
||||
function_name == "globalNotNullIn" ||
|
||||
function_name == "globalInIgnoreSet" ||
|
||||
function_name == "globalNotInIgnoreSet" ||
|
||||
function_name == "globalNullInIgnoreSet" ||
|
||||
function_name == "globalNotNullInIgnoreSet";
|
||||
|
||||
return is_special_function_in;
|
||||
}
|
||||
|
||||
std::string getGlobalInFunctionNameForLocalInFunctionName(const std::string & function_name)
|
||||
{
|
||||
if (function_name == "in")
|
||||
return "globalIn";
|
||||
else if (function_name == "notIn")
|
||||
return "globalNotIn";
|
||||
else if (function_name == "nullIn")
|
||||
return "globalNullIn";
|
||||
else if (function_name == "notNullIn")
|
||||
return "globalNotNullIn";
|
||||
else if (function_name == "inIgnoreSet")
|
||||
return "globalInIgnoreSet";
|
||||
else if (function_name == "notInIgnoreSet")
|
||||
return "globalNotInIgnoreSet";
|
||||
else if (function_name == "nullInIgnoreSet")
|
||||
return "globalNullInIgnoreSet";
|
||||
else if (function_name == "notNullInIgnoreSet")
|
||||
return "globalNotNullInIgnoreSet";
|
||||
|
||||
throw Exception(ErrorCodes::BAD_ARGUMENTS, "Invalid local IN function name {}", function_name);
|
||||
}
|
||||
|
||||
void makeUniqueColumnNamesInBlock(Block & block)
|
||||
{
|
||||
NameSet block_column_names;
|
||||
size_t unique_column_name_counter = 1;
|
||||
|
||||
for (auto & column_with_type : block)
|
||||
{
|
||||
if (!block_column_names.contains(column_with_type.name))
|
||||
{
|
||||
block_column_names.insert(column_with_type.name);
|
||||
continue;
|
||||
}
|
||||
|
||||
column_with_type.name += '_';
|
||||
column_with_type.name += std::to_string(unique_column_name_counter);
|
||||
++unique_column_name_counter;
|
||||
}
|
||||
}
|
||||
|
||||
QueryTreeNodePtr buildCastFunction(const QueryTreeNodePtr & expression,
|
||||
const DataTypePtr & type,
|
||||
const ContextPtr & context,
|
||||
@ -102,6 +173,27 @@ QueryTreeNodePtr buildCastFunction(const QueryTreeNodePtr & expression,
|
||||
return cast_function_node;
|
||||
}
|
||||
|
||||
std::optional<bool> tryExtractConstantFromConditionNode(const QueryTreeNodePtr & condition_node)
|
||||
{
|
||||
const auto * constant_node = condition_node->as<ConstantNode>();
|
||||
if (!constant_node)
|
||||
return {};
|
||||
|
||||
const auto & value = constant_node->getValue();
|
||||
auto constant_type = constant_node->getResultType();
|
||||
constant_type = removeNullable(removeLowCardinality(constant_type));
|
||||
|
||||
auto which_constant_type = WhichDataType(constant_type);
|
||||
if (!which_constant_type.isUInt8() && !which_constant_type.isNothing())
|
||||
return {};
|
||||
|
||||
if (value.isNull())
|
||||
return false;
|
||||
|
||||
UInt8 predicate_value = value.safeGet<UInt8>();
|
||||
return predicate_value > 0;
|
||||
}
|
||||
|
||||
static ASTPtr convertIntoTableExpressionAST(const QueryTreeNodePtr & table_expression_node)
|
||||
{
|
||||
ASTPtr table_expression_node_ast;
|
||||
|
||||
@ -13,6 +13,18 @@ bool isNodePartOfTree(const IQueryTreeNode * node, const IQueryTreeNode * root);
|
||||
/// Returns true if function name is name of IN function or its variations, false otherwise
|
||||
bool isNameOfInFunction(const std::string & function_name);
|
||||
|
||||
/// Returns true if function name is name of local IN function or its variations, false otherwise
|
||||
bool isNameOfLocalInFunction(const std::string & function_name);
|
||||
|
||||
/// Returns true if function name is name of global IN function or its variations, false otherwise
|
||||
bool isNameOfGlobalInFunction(const std::string & function_name);
|
||||
|
||||
/// Returns global IN function name for local IN function name
|
||||
std::string getGlobalInFunctionNameForLocalInFunctionName(const std::string & function_name);
|
||||
|
||||
/// Add unique suffix to names of duplicate columns in block
|
||||
void makeUniqueColumnNamesInBlock(Block & block);
|
||||
|
||||
/** Build cast function that cast expression into type.
|
||||
* If resolve = true, then result cast function is resolved during build, otherwise
|
||||
* result cast function is not resolved during build.
|
||||
@ -22,6 +34,9 @@ QueryTreeNodePtr buildCastFunction(const QueryTreeNodePtr & expression,
|
||||
const ContextPtr & context,
|
||||
bool resolve = true);
|
||||
|
||||
/// Try extract boolean constant from condition node
|
||||
std::optional<bool> tryExtractConstantFromConditionNode(const QueryTreeNodePtr & condition_node);
|
||||
|
||||
/** Add table expression in tables in select query children.
|
||||
* If table expression node is not of identifier node, table node, query node, table function node, join node or array join node type throws logical error exception.
|
||||
*/
|
||||
|
||||
@ -16,6 +16,7 @@ namespace ErrorCodes
|
||||
{
|
||||
extern const int NOT_AN_AGGREGATE;
|
||||
extern const int NOT_IMPLEMENTED;
|
||||
extern const int BAD_ARGUMENTS;
|
||||
}
|
||||
|
||||
class ValidateGroupByColumnsVisitor : public ConstInDepthQueryTreeVisitor<ValidateGroupByColumnsVisitor>
|
||||
@ -283,4 +284,52 @@ void assertNoFunctionNodes(const QueryTreeNodePtr & node,
|
||||
visitor.visit(node);
|
||||
}
|
||||
|
||||
void validateTreeSize(const QueryTreeNodePtr & node,
|
||||
size_t max_size,
|
||||
std::unordered_map<QueryTreeNodePtr, size_t> & node_to_tree_size)
|
||||
{
|
||||
size_t tree_size = 0;
|
||||
std::vector<std::pair<QueryTreeNodePtr, bool>> nodes_to_process;
|
||||
nodes_to_process.emplace_back(node, false);
|
||||
|
||||
while (!nodes_to_process.empty())
|
||||
{
|
||||
const auto [node_to_process, processed_children] = nodes_to_process.back();
|
||||
nodes_to_process.pop_back();
|
||||
|
||||
if (processed_children)
|
||||
{
|
||||
++tree_size;
|
||||
node_to_tree_size.emplace(node_to_process, tree_size);
|
||||
continue;
|
||||
}
|
||||
|
||||
auto node_to_size_it = node_to_tree_size.find(node_to_process);
|
||||
if (node_to_size_it != node_to_tree_size.end())
|
||||
{
|
||||
tree_size += node_to_size_it->second;
|
||||
continue;
|
||||
}
|
||||
|
||||
nodes_to_process.emplace_back(node_to_process, true);
|
||||
|
||||
for (const auto & node_to_process_child : node_to_process->getChildren())
|
||||
{
|
||||
if (!node_to_process_child)
|
||||
continue;
|
||||
|
||||
nodes_to_process.emplace_back(node_to_process_child, false);
|
||||
}
|
||||
|
||||
auto * constant_node = node_to_process->as<ConstantNode>();
|
||||
if (constant_node && constant_node->hasSourceExpression())
|
||||
nodes_to_process.emplace_back(constant_node->getSourceExpression(), false);
|
||||
}
|
||||
|
||||
if (tree_size > max_size)
|
||||
throw Exception(ErrorCodes::BAD_ARGUMENTS,
|
||||
"Query tree is too big. Maximum: {}",
|
||||
max_size);
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
@ -7,7 +7,7 @@ namespace DB
|
||||
|
||||
struct ValidationParams
|
||||
{
|
||||
bool group_by_use_nulls;
|
||||
bool group_by_use_nulls = false;
|
||||
};
|
||||
|
||||
/** Validate aggregates in query node.
|
||||
@ -31,4 +31,11 @@ void assertNoFunctionNodes(const QueryTreeNodePtr & node,
|
||||
std::string_view exception_function_name,
|
||||
std::string_view exception_place_message);
|
||||
|
||||
/** Validate tree size. If size of tree is greater than max size throws exception.
|
||||
* Additionally for each node in tree, update node to tree size map.
|
||||
*/
|
||||
void validateTreeSize(const QueryTreeNodePtr & node,
|
||||
size_t max_size,
|
||||
std::unordered_map<QueryTreeNodePtr, size_t> & node_to_tree_size);
|
||||
|
||||
}
|
||||
|
||||
@ -113,11 +113,17 @@ ASTPtr WindowNode::toASTImpl() const
|
||||
|
||||
window_definition->parent_window_name = parent_window_name;
|
||||
|
||||
if (hasPartitionBy())
|
||||
{
|
||||
window_definition->children.push_back(getPartitionByNode()->toAST());
|
||||
window_definition->partition_by = window_definition->children.back();
|
||||
}
|
||||
|
||||
if (hasOrderBy())
|
||||
{
|
||||
window_definition->children.push_back(getOrderByNode()->toAST());
|
||||
window_definition->order_by = window_definition->children.back();
|
||||
}
|
||||
|
||||
window_definition->frame_is_default = window_frame.is_default;
|
||||
window_definition->frame_type = window_frame.type;
|
||||
|
||||
@ -110,23 +110,4 @@ ThreadGroupStatusPtr CurrentThread::getGroup()
|
||||
return current_thread->getThreadGroup();
|
||||
}
|
||||
|
||||
MemoryTracker * CurrentThread::getUserMemoryTracker()
|
||||
{
|
||||
if (unlikely(!current_thread))
|
||||
return nullptr;
|
||||
|
||||
auto * tracker = current_thread->memory_tracker.getParent();
|
||||
while (tracker && tracker->level != VariableContext::User)
|
||||
tracker = tracker->getParent();
|
||||
|
||||
return tracker;
|
||||
}
|
||||
|
||||
void CurrentThread::flushUntrackedMemory()
|
||||
{
|
||||
if (unlikely(!current_thread))
|
||||
return;
|
||||
current_thread->flushUntrackedMemory();
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
@ -40,12 +40,6 @@ public:
|
||||
/// Group to which belongs current thread
|
||||
static ThreadGroupStatusPtr getGroup();
|
||||
|
||||
/// MemoryTracker for user that owns current thread if any
|
||||
static MemoryTracker * getUserMemoryTracker();
|
||||
|
||||
/// Adjust counters in MemoryTracker hierarchy if untracked_memory is not 0.
|
||||
static void flushUntrackedMemory();
|
||||
|
||||
/// A logs queue used by TCPHandler to pass logs to a client
|
||||
static void attachInternalTextLogsQueue(const std::shared_ptr<InternalTextLogsQueue> & logs_queue,
|
||||
LogsLevel client_logs_level);
|
||||
|
||||
@ -38,6 +38,8 @@ public:
|
||||
String getName() const override { return name; }
|
||||
size_t getNumberOfArguments() const override { return 0; }
|
||||
bool isSuitableForShortCircuitArgumentsExecution(const DataTypesWithConstInfo & /*arguments*/) const override { return true; }
|
||||
bool isDeterministic() const override { return false; }
|
||||
bool isDeterministicInScopeOfQuery() const override { return false; }
|
||||
|
||||
DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override
|
||||
{
|
||||
|
||||
@ -14,28 +14,33 @@ namespace ErrorCodes
|
||||
static size_t encodeURL(const char * __restrict src, size_t src_size, char * __restrict dst, bool space_as_plus)
|
||||
{
|
||||
char * dst_pos = dst;
|
||||
for (size_t i = 0; i < src_size; i++)
|
||||
for (size_t i = 0; i < src_size; ++i)
|
||||
{
|
||||
if ((src[i] >= '0' && src[i] <= '9') || (src[i] >= 'a' && src[i] <= 'z') || (src[i] >= 'A' && src[i] <= 'Z')
|
||||
|| src[i] == '-' || src[i] == '_' || src[i] == '.' || src[i] == '~')
|
||||
{
|
||||
*dst_pos++ = src[i];
|
||||
*dst_pos = src[i];
|
||||
++dst_pos;
|
||||
}
|
||||
else if (src[i] == ' ' && space_as_plus)
|
||||
{
|
||||
*dst_pos++ = '+';
|
||||
*dst_pos = '+';
|
||||
++dst_pos;
|
||||
}
|
||||
else
|
||||
{
|
||||
*dst_pos++ = '%';
|
||||
*dst_pos++ = hexDigitUppercase(src[i] >> 4);
|
||||
*dst_pos++ = hexDigitUppercase(src[i] & 0xf);
|
||||
dst_pos[0] = '%';
|
||||
++dst_pos;
|
||||
writeHexByteUppercase(src[i], dst_pos);
|
||||
dst_pos += 2;
|
||||
}
|
||||
}
|
||||
*dst_pos++ = src[src_size];
|
||||
*dst_pos = 0;
|
||||
++dst_pos;
|
||||
return dst_pos - dst;
|
||||
}
|
||||
|
||||
|
||||
/// We assume that size of the dst buf isn't less than src_size.
|
||||
static size_t decodeURL(const char * __restrict src, size_t src_size, char * __restrict dst, bool plus_as_space)
|
||||
{
|
||||
@ -120,10 +125,14 @@ struct CodeURLComponentImpl
|
||||
ColumnString::Chars & res_data, ColumnString::Offsets & res_offsets)
|
||||
{
|
||||
if (code_strategy == encode)
|
||||
//the destination(res_data) string is at most three times the length of the source string
|
||||
{
|
||||
/// the destination(res_data) string is at most three times the length of the source string
|
||||
res_data.resize(data.size() * 3);
|
||||
}
|
||||
else
|
||||
{
|
||||
res_data.resize(data.size());
|
||||
}
|
||||
|
||||
size_t size = offsets.size();
|
||||
res_offsets.resize(size);
|
||||
|
||||
@ -18,7 +18,6 @@
|
||||
#include <Parsers/ASTInsertQuery.h>
|
||||
#include <Parsers/queryToString.h>
|
||||
#include <Storages/IStorage.h>
|
||||
#include <Common/CurrentThread.h>
|
||||
#include <Common/SipHash.h>
|
||||
#include <Common/FieldVisitorHash.h>
|
||||
#include <Common/DateLUT.h>
|
||||
@ -104,10 +103,9 @@ bool AsynchronousInsertQueue::InsertQuery::operator==(const InsertQuery & other)
|
||||
return query_str == other.query_str && settings == other.settings;
|
||||
}
|
||||
|
||||
AsynchronousInsertQueue::InsertData::Entry::Entry(String && bytes_, String && query_id_, MemoryTracker * user_memory_tracker_)
|
||||
AsynchronousInsertQueue::InsertData::Entry::Entry(String && bytes_, String && query_id_)
|
||||
: bytes(std::move(bytes_))
|
||||
, query_id(std::move(query_id_))
|
||||
, user_memory_tracker(user_memory_tracker_)
|
||||
, create_time(std::chrono::system_clock::now())
|
||||
{
|
||||
}
|
||||
@ -236,7 +234,7 @@ AsynchronousInsertQueue::push(ASTPtr query, ContextPtr query_context)
|
||||
if (auto quota = query_context->getQuota())
|
||||
quota->used(QuotaType::WRITTEN_BYTES, bytes.size());
|
||||
|
||||
auto entry = std::make_shared<InsertData::Entry>(std::move(bytes), query_context->getCurrentQueryId(), CurrentThread::getUserMemoryTracker());
|
||||
auto entry = std::make_shared<InsertData::Entry>(std::move(bytes), query_context->getCurrentQueryId());
|
||||
|
||||
InsertQuery key{query, settings};
|
||||
InsertDataPtr data_to_process;
|
||||
|
||||
@ -1,7 +1,6 @@
|
||||
#pragma once
|
||||
|
||||
#include <Parsers/IAST_fwd.h>
|
||||
#include <Common/CurrentThread.h>
|
||||
#include <Common/ThreadPool.h>
|
||||
#include <Core/Settings.h>
|
||||
#include <Poco/Logger.h>
|
||||
@ -60,31 +59,6 @@ private:
|
||||
UInt128 calculateHash() const;
|
||||
};
|
||||
|
||||
struct UserMemoryTrackerSwitcher
|
||||
{
|
||||
explicit UserMemoryTrackerSwitcher(MemoryTracker * new_tracker)
|
||||
{
|
||||
auto * thread_tracker = CurrentThread::getMemoryTracker();
|
||||
prev_untracked_memory = current_thread->untracked_memory;
|
||||
prev_memory_tracker_parent = thread_tracker->getParent();
|
||||
|
||||
current_thread->untracked_memory = 0;
|
||||
thread_tracker->setParent(new_tracker);
|
||||
}
|
||||
|
||||
~UserMemoryTrackerSwitcher()
|
||||
{
|
||||
CurrentThread::flushUntrackedMemory();
|
||||
auto * thread_tracker = CurrentThread::getMemoryTracker();
|
||||
|
||||
current_thread->untracked_memory = prev_untracked_memory;
|
||||
thread_tracker->setParent(prev_memory_tracker_parent);
|
||||
}
|
||||
|
||||
MemoryTracker * prev_memory_tracker_parent;
|
||||
Int64 prev_untracked_memory;
|
||||
};
|
||||
|
||||
struct InsertData
|
||||
{
|
||||
struct Entry
|
||||
@ -92,10 +66,9 @@ private:
|
||||
public:
|
||||
const String bytes;
|
||||
const String query_id;
|
||||
MemoryTracker * const user_memory_tracker;
|
||||
const std::chrono::time_point<std::chrono::system_clock> create_time;
|
||||
|
||||
Entry(String && bytes_, String && query_id_, MemoryTracker * user_memory_tracker_);
|
||||
Entry(String && bytes_, String && query_id_);
|
||||
|
||||
void finish(std::exception_ptr exception_ = nullptr);
|
||||
std::future<void> getFuture() { return promise.get_future(); }
|
||||
@ -106,19 +79,6 @@ private:
|
||||
std::atomic_bool finished = false;
|
||||
};
|
||||
|
||||
~InsertData()
|
||||
{
|
||||
auto it = entries.begin();
|
||||
// Entries must be destroyed in context of user who runs async insert.
|
||||
// Each entry in the list may correspond to a different user,
|
||||
// so we need to switch current thread's MemoryTracker parent on each iteration.
|
||||
while (it != entries.end())
|
||||
{
|
||||
UserMemoryTrackerSwitcher switcher((*it)->user_memory_tracker);
|
||||
it = entries.erase(it);
|
||||
}
|
||||
}
|
||||
|
||||
using EntryPtr = std::shared_ptr<Entry>;
|
||||
|
||||
std::list<EntryPtr> entries;
|
||||
|
||||
@ -44,6 +44,10 @@ public:
|
||||
const auto & on_expr = table_join->getOnlyClause();
|
||||
bool support_conditions = !on_expr.on_filter_condition_left && !on_expr.on_filter_condition_right;
|
||||
|
||||
if (!on_expr.analyzer_left_filter_condition_column_name.empty() ||
|
||||
!on_expr.analyzer_right_filter_condition_column_name.empty())
|
||||
support_conditions = false;
|
||||
|
||||
/// Key column can change nullability and it's not handled on type conversion stage, so algorithm should be aware of it
|
||||
bool support_using_and_nulls = !table_join->hasUsing() || !table_join->joinUseNulls();
|
||||
|
||||
|
||||
@ -226,6 +226,12 @@ BlockIO InterpreterSelectQueryAnalyzer::execute()
|
||||
return result;
|
||||
}
|
||||
|
||||
QueryPlan & InterpreterSelectQueryAnalyzer::getQueryPlan()
|
||||
{
|
||||
planner.buildQueryPlanIfNeeded();
|
||||
return planner.getQueryPlan();
|
||||
}
|
||||
|
||||
QueryPlan && InterpreterSelectQueryAnalyzer::extractQueryPlan() &&
|
||||
{
|
||||
planner.buildQueryPlanIfNeeded();
|
||||
|
||||
@ -51,6 +51,8 @@ public:
|
||||
|
||||
BlockIO execute() override;
|
||||
|
||||
QueryPlan & getQueryPlan();
|
||||
|
||||
QueryPlan && extractQueryPlan() &&;
|
||||
|
||||
QueryPipelineBuilder buildQueryPipeline();
|
||||
|
||||
@ -50,7 +50,16 @@ void ReplaceQueryParameterVisitor::visit(ASTPtr & ast)
|
||||
void ReplaceQueryParameterVisitor::visitChildren(ASTPtr & ast)
|
||||
{
|
||||
for (auto & child : ast->children)
|
||||
{
|
||||
void * old_ptr = child.get();
|
||||
visit(child);
|
||||
void * new_ptr = child.get();
|
||||
|
||||
/// Some AST classes have naked pointers to children elements as members.
|
||||
/// We have to replace them if the child was replaced.
|
||||
if (new_ptr != old_ptr)
|
||||
ast->updatePointerToChild(old_ptr, new_ptr);
|
||||
}
|
||||
}
|
||||
|
||||
const String & ReplaceQueryParameterVisitor::getParamValue(const String & name)
|
||||
@ -89,6 +98,7 @@ void ReplaceQueryParameterVisitor::visitQueryParameter(ASTPtr & ast)
|
||||
literal = value;
|
||||
else
|
||||
literal = temp_column[0];
|
||||
|
||||
ast = addTypeConversionToAST(std::make_shared<ASTLiteral>(literal), type_name);
|
||||
|
||||
/// Keep the original alias.
|
||||
|
||||
@ -256,6 +256,11 @@ protected:
|
||||
void formatQueryImpl(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const override;
|
||||
|
||||
bool isOneCommandTypeOnly(const ASTAlterCommand::Type & type) const;
|
||||
|
||||
void forEachPointerToChild(std::function<void(void**)> f) override
|
||||
{
|
||||
f(reinterpret_cast<void **>(&command_list));
|
||||
}
|
||||
};
|
||||
|
||||
}
|
||||
|
||||
@ -94,5 +94,12 @@ public:
|
||||
void formatImpl(const FormatSettings & format, FormatState &, FormatStateStacked) const override;
|
||||
ASTPtr getRewrittenASTWithoutOnCluster(const WithoutOnClusterASTRewriteParams &) const override;
|
||||
QueryKind getQueryKind() const override;
|
||||
};
|
||||
|
||||
void forEachPointerToChild(std::function<void(void**)> f) override
|
||||
{
|
||||
f(reinterpret_cast<void **>(&backup_name));
|
||||
f(reinterpret_cast<void **>(&base_backup_name));
|
||||
}
|
||||
};
|
||||
|
||||
}
|
||||
|
||||
@ -25,5 +25,11 @@ public:
|
||||
ASTPtr clone() const override;
|
||||
|
||||
void formatImpl(const FormatSettings & s, FormatState & state, FormatStateStacked frame) const override;
|
||||
};
|
||||
|
||||
void forEachPointerToChild(std::function<void(void**)> f) override
|
||||
{
|
||||
f(reinterpret_cast<void **>(&expr));
|
||||
}
|
||||
};
|
||||
|
||||
}
|
||||
|
||||
@ -91,6 +91,11 @@ public:
|
||||
ASTPtr clone() const override;
|
||||
|
||||
void formatImpl(const FormatSettings & s, FormatState & state, FormatStateStacked frame) const override;
|
||||
|
||||
void forEachPointerToChild(std::function<void(void**)> f) override
|
||||
{
|
||||
f(reinterpret_cast<void **>(&elem));
|
||||
}
|
||||
};
|
||||
|
||||
ASTPtr ASTColumnsElement::clone() const
|
||||
|
||||
@ -32,6 +32,17 @@ public:
|
||||
void formatImpl(const FormatSettings & s, FormatState & state, FormatStateStacked frame) const override;
|
||||
|
||||
bool isExtendedStorageDefinition() const;
|
||||
|
||||
void forEachPointerToChild(std::function<void(void**)> f) override
|
||||
{
|
||||
f(reinterpret_cast<void **>(&engine));
|
||||
f(reinterpret_cast<void **>(&partition_by));
|
||||
f(reinterpret_cast<void **>(&primary_key));
|
||||
f(reinterpret_cast<void **>(&order_by));
|
||||
f(reinterpret_cast<void **>(&sample_by));
|
||||
f(reinterpret_cast<void **>(&ttl_table));
|
||||
f(reinterpret_cast<void **>(&settings));
|
||||
}
|
||||
};
|
||||
|
||||
|
||||
@ -57,6 +68,16 @@ public:
|
||||
return (!columns || columns->children.empty()) && (!indices || indices->children.empty()) && (!constraints || constraints->children.empty())
|
||||
&& (!projections || projections->children.empty());
|
||||
}
|
||||
|
||||
void forEachPointerToChild(std::function<void(void**)> f) override
|
||||
{
|
||||
f(reinterpret_cast<void **>(&columns));
|
||||
f(reinterpret_cast<void **>(&indices));
|
||||
f(reinterpret_cast<void **>(&primary_key));
|
||||
f(reinterpret_cast<void **>(&constraints));
|
||||
f(reinterpret_cast<void **>(&projections));
|
||||
f(reinterpret_cast<void **>(&primary_key));
|
||||
}
|
||||
};
|
||||
|
||||
|
||||
@ -126,6 +147,19 @@ public:
|
||||
|
||||
protected:
|
||||
void formatQueryImpl(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const override;
|
||||
|
||||
void forEachPointerToChild(std::function<void(void**)> f) override
|
||||
{
|
||||
f(reinterpret_cast<void **>(&columns_list));
|
||||
f(reinterpret_cast<void **>(&inner_storage));
|
||||
f(reinterpret_cast<void **>(&storage));
|
||||
f(reinterpret_cast<void **>(&as_table_function));
|
||||
f(reinterpret_cast<void **>(&select));
|
||||
f(reinterpret_cast<void **>(&comment));
|
||||
f(reinterpret_cast<void **>(&table_overrides));
|
||||
f(reinterpret_cast<void **>(&dictionary_attributes_list));
|
||||
f(reinterpret_cast<void **>(&dictionary));
|
||||
}
|
||||
};
|
||||
|
||||
}
|
||||
|
||||
@ -47,6 +47,11 @@ public:
|
||||
ASTPtr clone() const override;
|
||||
|
||||
void formatImpl(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const override;
|
||||
|
||||
void forEachPointerToChild(std::function<void(void**)> f) override
|
||||
{
|
||||
f(reinterpret_cast<void **>(¶meters));
|
||||
}
|
||||
};
|
||||
|
||||
|
||||
|
||||
@ -41,6 +41,11 @@ public:
|
||||
}
|
||||
|
||||
QueryKind getQueryKind() const override { return QueryKind::ExternalDDL; }
|
||||
|
||||
void forEachPointerToChild(std::function<void(void**)> f) override
|
||||
{
|
||||
f(reinterpret_cast<void **>(&from));
|
||||
}
|
||||
};
|
||||
|
||||
}
|
||||
|
||||
@ -33,6 +33,11 @@ public:
|
||||
bool hasSecretParts() const override;
|
||||
|
||||
void updateTreeHashImpl(SipHash & hash_state) const override;
|
||||
|
||||
void forEachPointerToChild(std::function<void(void**)> f) override
|
||||
{
|
||||
f(reinterpret_cast<void **>(&second));
|
||||
}
|
||||
};
|
||||
|
||||
|
||||
|
||||
@ -23,6 +23,12 @@ public:
|
||||
|
||||
ASTPtr clone() const override;
|
||||
void formatImpl(const FormatSettings & s, FormatState & state, FormatStateStacked frame) const override;
|
||||
|
||||
void forEachPointerToChild(std::function<void(void**)> f) override
|
||||
{
|
||||
f(reinterpret_cast<void **>(&expr));
|
||||
f(reinterpret_cast<void **>(&type));
|
||||
}
|
||||
};
|
||||
|
||||
}
|
||||
|
||||
@ -18,6 +18,11 @@ public:
|
||||
|
||||
ASTPtr clone() const override;
|
||||
void formatImpl(const FormatSettings & s, FormatState & state, FormatStateStacked frame) const override;
|
||||
|
||||
void forEachPointerToChild(std::function<void(void**)> f) override
|
||||
{
|
||||
f(reinterpret_cast<void **>(&query));
|
||||
}
|
||||
};
|
||||
|
||||
}
|
||||
|
||||
@ -27,6 +27,12 @@ public:
|
||||
String getID(char) const override { return "TableOverride " + table_name; }
|
||||
ASTPtr clone() const override;
|
||||
void formatImpl(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const override;
|
||||
|
||||
void forEachPointerToChild(std::function<void(void**)> f) override
|
||||
{
|
||||
f(reinterpret_cast<void **>(&columns));
|
||||
f(reinterpret_cast<void **>(&storage));
|
||||
}
|
||||
};
|
||||
|
||||
/// List of table overrides, for example:
|
||||
|
||||
@ -175,6 +175,16 @@ public:
|
||||
field = nullptr;
|
||||
}
|
||||
|
||||
/// After changing one of `children` elements, update the corresponding member pointer if needed.
|
||||
void updatePointerToChild(void * old_ptr, void * new_ptr)
|
||||
{
|
||||
forEachPointerToChild([old_ptr, new_ptr](void ** ptr) mutable
|
||||
{
|
||||
if (*ptr == old_ptr)
|
||||
*ptr = new_ptr;
|
||||
});
|
||||
}
|
||||
|
||||
/// Convert to a string.
|
||||
|
||||
/// Format settings.
|
||||
@ -295,6 +305,10 @@ public:
|
||||
protected:
|
||||
bool childrenHaveSecretParts() const;
|
||||
|
||||
/// Some AST classes have naked pointers to children elements as members.
|
||||
/// This method allows to iterate over them.
|
||||
virtual void forEachPointerToChild(std::function<void(void**)>) {}
|
||||
|
||||
private:
|
||||
size_t checkDepthImpl(size_t max_depth) const;
|
||||
|
||||
|
||||
@ -80,6 +80,15 @@ protected:
|
||||
{
|
||||
throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Method formatImpl is not supported by MySQLParser::ASTAlterCommand.");
|
||||
}
|
||||
|
||||
void forEachPointerToChild(std::function<void(void**)> f) override
|
||||
{
|
||||
f(reinterpret_cast<void **>(&index_decl));
|
||||
f(reinterpret_cast<void **>(&default_expression));
|
||||
f(reinterpret_cast<void **>(&additional_columns));
|
||||
f(reinterpret_cast<void **>(&order_by_columns));
|
||||
f(reinterpret_cast<void **>(&properties));
|
||||
}
|
||||
};
|
||||
|
||||
class ParserAlterCommand : public IParserBase
|
||||
|
||||
@ -31,6 +31,13 @@ protected:
|
||||
{
|
||||
throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Method formatImpl is not supported by MySQLParser::ASTCreateDefines.");
|
||||
}
|
||||
|
||||
void forEachPointerToChild(std::function<void(void**)> f) override
|
||||
{
|
||||
f(reinterpret_cast<void **>(&columns));
|
||||
f(reinterpret_cast<void **>(&indices));
|
||||
f(reinterpret_cast<void **>(&constraints));
|
||||
}
|
||||
};
|
||||
|
||||
class ParserCreateDefines : public IParserBase
|
||||
@ -44,4 +51,3 @@ protected:
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
|
||||
@ -214,9 +214,14 @@ public:
|
||||
{
|
||||
/// Constness of limit is validated during query analysis stage
|
||||
limit_length = query_node.getLimit()->as<ConstantNode &>().getValue().safeGet<UInt64>();
|
||||
}
|
||||
|
||||
if (query_node.hasOffset())
|
||||
if (query_node.hasOffset() && limit_length)
|
||||
{
|
||||
/// Constness of offset is validated during query analysis stage
|
||||
limit_offset = query_node.getOffset()->as<ConstantNode &>().getValue().safeGet<UInt64>();
|
||||
}
|
||||
}
|
||||
else if (query_node.hasOffset())
|
||||
{
|
||||
/// Constness of offset is validated during query analysis stage
|
||||
limit_offset = query_node.getOffset()->as<ConstantNode &>().getValue().safeGet<UInt64>();
|
||||
|
||||
@ -45,7 +45,7 @@ bool GlobalPlannerContext::hasColumnIdentifier(const ColumnIdentifier & column_i
|
||||
return column_identifiers.contains(column_identifier);
|
||||
}
|
||||
|
||||
PlannerContext::PlannerContext(ContextPtr query_context_, GlobalPlannerContextPtr global_planner_context_)
|
||||
PlannerContext::PlannerContext(ContextMutablePtr query_context_, GlobalPlannerContextPtr global_planner_context_)
|
||||
: query_context(std::move(query_context_))
|
||||
, global_planner_context(std::move(global_planner_context_))
|
||||
{}
|
||||
|
||||
@ -88,16 +88,22 @@ class PlannerContext
|
||||
{
|
||||
public:
|
||||
/// Create planner context with query context and global planner context
|
||||
PlannerContext(ContextPtr query_context_, GlobalPlannerContextPtr global_planner_context_);
|
||||
PlannerContext(ContextMutablePtr query_context_, GlobalPlannerContextPtr global_planner_context_);
|
||||
|
||||
/// Get planner context query context
|
||||
const ContextPtr & getQueryContext() const
|
||||
ContextPtr getQueryContext() const
|
||||
{
|
||||
return query_context;
|
||||
}
|
||||
|
||||
/// Get planner context query context
|
||||
ContextPtr & getQueryContext()
|
||||
/// Get planner context mutable query context
|
||||
const ContextMutablePtr & getMutableQueryContext() const
|
||||
{
|
||||
return query_context;
|
||||
}
|
||||
|
||||
/// Get planner context mutable query context
|
||||
ContextMutablePtr & getMutableQueryContext()
|
||||
{
|
||||
return query_context;
|
||||
}
|
||||
@ -137,12 +143,18 @@ public:
|
||||
*/
|
||||
TableExpressionData * getTableExpressionDataOrNull(const QueryTreeNodePtr & table_expression_node);
|
||||
|
||||
/// Get table expression node to data read only map
|
||||
/// Get table expression node to data map
|
||||
const std::unordered_map<QueryTreeNodePtr, TableExpressionData> & getTableExpressionNodeToData() const
|
||||
{
|
||||
return table_expression_node_to_data;
|
||||
}
|
||||
|
||||
/// Get table expression node to data map
|
||||
std::unordered_map<QueryTreeNodePtr, TableExpressionData> & getTableExpressionNodeToData()
|
||||
{
|
||||
return table_expression_node_to_data;
|
||||
}
|
||||
|
||||
/** Get column node identifier.
|
||||
* For column node source check if table expression data is registered.
|
||||
* If table expression data is not registered exception is thrown.
|
||||
@ -184,7 +196,7 @@ public:
|
||||
|
||||
private:
|
||||
/// Query context
|
||||
ContextPtr query_context;
|
||||
ContextMutablePtr query_context;
|
||||
|
||||
/// Global planner context
|
||||
GlobalPlannerContextPtr global_planner_context;
|
||||
|
||||
@ -34,15 +34,13 @@ namespace
|
||||
* It is client responsibility to update filter analysis result if filter column must be removed after chain is finalized.
|
||||
*/
|
||||
FilterAnalysisResult analyzeFilter(const QueryTreeNodePtr & filter_expression_node,
|
||||
const ColumnsWithTypeAndName & current_output_columns,
|
||||
const ColumnsWithTypeAndName & input_columns,
|
||||
const PlannerContextPtr & planner_context,
|
||||
ActionsChain & actions_chain)
|
||||
{
|
||||
const auto & filter_input = current_output_columns;
|
||||
|
||||
FilterAnalysisResult result;
|
||||
|
||||
result.filter_actions = buildActionsDAGFromExpressionNode(filter_expression_node, filter_input, planner_context);
|
||||
result.filter_actions = buildActionsDAGFromExpressionNode(filter_expression_node, input_columns, planner_context);
|
||||
result.filter_column_name = result.filter_actions->getOutputs().at(0)->result_name;
|
||||
actions_chain.addStep(std::make_unique<ActionsChainStep>(result.filter_actions));
|
||||
|
||||
@ -52,8 +50,8 @@ FilterAnalysisResult analyzeFilter(const QueryTreeNodePtr & filter_expression_no
|
||||
/** Construct aggregation analysis result if query tree has GROUP BY or aggregates.
|
||||
* Actions before aggregation are added into actions chain, if result is not null optional.
|
||||
*/
|
||||
std::pair<std::optional<AggregationAnalysisResult>, std::optional<ColumnsWithTypeAndName>> analyzeAggregation(const QueryTreeNodePtr & query_tree,
|
||||
const ColumnsWithTypeAndName & current_output_columns,
|
||||
std::optional<AggregationAnalysisResult> analyzeAggregation(const QueryTreeNodePtr & query_tree,
|
||||
const ColumnsWithTypeAndName & input_columns,
|
||||
const PlannerContextPtr & planner_context,
|
||||
ActionsChain & actions_chain)
|
||||
{
|
||||
@ -69,9 +67,7 @@ std::pair<std::optional<AggregationAnalysisResult>, std::optional<ColumnsWithTyp
|
||||
|
||||
Names aggregation_keys;
|
||||
|
||||
const auto & group_by_input = current_output_columns;
|
||||
|
||||
ActionsDAGPtr before_aggregation_actions = std::make_shared<ActionsDAG>(group_by_input);
|
||||
ActionsDAGPtr before_aggregation_actions = std::make_shared<ActionsDAG>(input_columns);
|
||||
before_aggregation_actions->getOutputs().clear();
|
||||
|
||||
std::unordered_set<std::string_view> before_aggregation_actions_output_node_names;
|
||||
@ -203,14 +199,14 @@ std::pair<std::optional<AggregationAnalysisResult>, std::optional<ColumnsWithTyp
|
||||
aggregation_analysis_result.grouping_sets_parameters_list = std::move(grouping_sets_parameters_list);
|
||||
aggregation_analysis_result.group_by_with_constant_keys = group_by_with_constant_keys;
|
||||
|
||||
return { aggregation_analysis_result, available_columns_after_aggregation };
|
||||
return aggregation_analysis_result;
|
||||
}
|
||||
|
||||
/** Construct window analysis result if query tree has window functions.
|
||||
* Actions before window functions are added into actions chain, if result is not null optional.
|
||||
*/
|
||||
std::optional<WindowAnalysisResult> analyzeWindow(const QueryTreeNodePtr & query_tree,
|
||||
const ColumnsWithTypeAndName & current_output_columns,
|
||||
const ColumnsWithTypeAndName & input_columns,
|
||||
const PlannerContextPtr & planner_context,
|
||||
ActionsChain & actions_chain)
|
||||
{
|
||||
@ -220,11 +216,9 @@ std::optional<WindowAnalysisResult> analyzeWindow(const QueryTreeNodePtr & query
|
||||
|
||||
auto window_descriptions = extractWindowDescriptions(window_function_nodes, *planner_context);
|
||||
|
||||
const auto & window_input = current_output_columns;
|
||||
|
||||
PlannerActionsVisitor actions_visitor(planner_context);
|
||||
|
||||
ActionsDAGPtr before_window_actions = std::make_shared<ActionsDAG>(window_input);
|
||||
ActionsDAGPtr before_window_actions = std::make_shared<ActionsDAG>(input_columns);
|
||||
before_window_actions->getOutputs().clear();
|
||||
|
||||
std::unordered_set<std::string_view> before_window_actions_output_node_names;
|
||||
@ -299,12 +293,11 @@ std::optional<WindowAnalysisResult> analyzeWindow(const QueryTreeNodePtr & query
|
||||
* It is client responsibility to update projection analysis result with project names actions after chain is finalized.
|
||||
*/
|
||||
ProjectionAnalysisResult analyzeProjection(const QueryNode & query_node,
|
||||
const ColumnsWithTypeAndName & current_output_columns,
|
||||
const ColumnsWithTypeAndName & input_columns,
|
||||
const PlannerContextPtr & planner_context,
|
||||
ActionsChain & actions_chain)
|
||||
{
|
||||
const auto & projection_input = current_output_columns;
|
||||
auto projection_actions = buildActionsDAGFromExpressionNode(query_node.getProjectionNode(), projection_input, planner_context);
|
||||
auto projection_actions = buildActionsDAGFromExpressionNode(query_node.getProjectionNode(), input_columns, planner_context);
|
||||
|
||||
auto projection_columns = query_node.getProjectionColumns();
|
||||
size_t projection_columns_size = projection_columns.size();
|
||||
@ -347,13 +340,11 @@ ProjectionAnalysisResult analyzeProjection(const QueryNode & query_node,
|
||||
* Actions before sort are added into actions chain.
|
||||
*/
|
||||
SortAnalysisResult analyzeSort(const QueryNode & query_node,
|
||||
const ColumnsWithTypeAndName & current_output_columns,
|
||||
const ColumnsWithTypeAndName & input_columns,
|
||||
const PlannerContextPtr & planner_context,
|
||||
ActionsChain & actions_chain)
|
||||
{
|
||||
const auto & order_by_input = current_output_columns;
|
||||
|
||||
ActionsDAGPtr before_sort_actions = std::make_shared<ActionsDAG>(order_by_input);
|
||||
ActionsDAGPtr before_sort_actions = std::make_shared<ActionsDAG>(input_columns);
|
||||
auto & before_sort_actions_outputs = before_sort_actions->getOutputs();
|
||||
before_sort_actions_outputs.clear();
|
||||
|
||||
@ -436,13 +427,12 @@ SortAnalysisResult analyzeSort(const QueryNode & query_node,
|
||||
* Actions before limit by are added into actions chain.
|
||||
*/
|
||||
LimitByAnalysisResult analyzeLimitBy(const QueryNode & query_node,
|
||||
const ColumnsWithTypeAndName & current_output_columns,
|
||||
const ColumnsWithTypeAndName & input_columns,
|
||||
const PlannerContextPtr & planner_context,
|
||||
const NameSet & required_output_nodes_names,
|
||||
ActionsChain & actions_chain)
|
||||
{
|
||||
const auto & limit_by_input = current_output_columns;
|
||||
auto before_limit_by_actions = buildActionsDAGFromExpressionNode(query_node.getLimitByNode(), limit_by_input, planner_context);
|
||||
auto before_limit_by_actions = buildActionsDAGFromExpressionNode(query_node.getLimitByNode(), input_columns, planner_context);
|
||||
|
||||
NameSet limit_by_column_names_set;
|
||||
Names limit_by_column_names;
|
||||
@ -480,8 +470,7 @@ PlannerExpressionsAnalysisResult buildExpressionAnalysisResult(const QueryTreeNo
|
||||
std::optional<FilterAnalysisResult> where_analysis_result_optional;
|
||||
std::optional<size_t> where_action_step_index_optional;
|
||||
|
||||
const auto * input_columns = actions_chain.getLastStepAvailableOutputColumnsOrNull();
|
||||
ColumnsWithTypeAndName current_output_columns = input_columns ? *input_columns : join_tree_input_columns;
|
||||
ColumnsWithTypeAndName current_output_columns = join_tree_input_columns;
|
||||
|
||||
if (query_node.hasWhere())
|
||||
{
|
||||
@ -490,9 +479,9 @@ PlannerExpressionsAnalysisResult buildExpressionAnalysisResult(const QueryTreeNo
|
||||
current_output_columns = actions_chain.getLastStepAvailableOutputColumns();
|
||||
}
|
||||
|
||||
auto [aggregation_analysis_result_optional, aggregated_columns_optional] = analyzeAggregation(query_tree, current_output_columns, planner_context, actions_chain);
|
||||
if (aggregated_columns_optional)
|
||||
current_output_columns = std::move(*aggregated_columns_optional);
|
||||
auto aggregation_analysis_result_optional = analyzeAggregation(query_tree, current_output_columns, planner_context, actions_chain);
|
||||
if (aggregation_analysis_result_optional)
|
||||
current_output_columns = actions_chain.getLastStepAvailableOutputColumns();
|
||||
|
||||
std::optional<FilterAnalysisResult> having_analysis_result_optional;
|
||||
std::optional<size_t> having_action_step_index_optional;
|
||||
|
||||
@ -246,17 +246,87 @@ bool applyTrivialCountIfPossible(
|
||||
return true;
|
||||
}
|
||||
|
||||
JoinTreeQueryPlan buildQueryPlanForTableExpression(const QueryTreeNodePtr & table_expression,
|
||||
const SelectQueryInfo & select_query_info,
|
||||
const SelectQueryOptions & select_query_options,
|
||||
PlannerContextPtr & planner_context,
|
||||
bool is_single_table_expression)
|
||||
void prepareBuildQueryPlanForTableExpression(const QueryTreeNodePtr & table_expression, PlannerContextPtr & planner_context)
|
||||
{
|
||||
const auto & query_context = planner_context->getQueryContext();
|
||||
const auto & settings = query_context->getSettingsRef();
|
||||
|
||||
auto & table_expression_data = planner_context->getTableExpressionDataOrThrow(table_expression);
|
||||
auto columns_names = table_expression_data.getColumnNames();
|
||||
|
||||
auto * table_node = table_expression->as<TableNode>();
|
||||
auto * table_function_node = table_expression->as<TableFunctionNode>();
|
||||
auto * query_node = table_expression->as<QueryNode>();
|
||||
auto * union_node = table_expression->as<UnionNode>();
|
||||
|
||||
/** The current user must have the SELECT privilege.
|
||||
* We do not check access rights for table functions because they have been already checked in ITableFunction::execute().
|
||||
*/
|
||||
if (table_node)
|
||||
{
|
||||
auto column_names_with_aliases = columns_names;
|
||||
const auto & alias_columns_names = table_expression_data.getAliasColumnsNames();
|
||||
column_names_with_aliases.insert(column_names_with_aliases.end(), alias_columns_names.begin(), alias_columns_names.end());
|
||||
checkAccessRights(*table_node, column_names_with_aliases, query_context);
|
||||
}
|
||||
|
||||
if (columns_names.empty())
|
||||
{
|
||||
NameAndTypePair additional_column_to_read;
|
||||
|
||||
if (table_node || table_function_node)
|
||||
{
|
||||
const auto & storage = table_node ? table_node->getStorage() : table_function_node->getStorage();
|
||||
const auto & storage_snapshot = table_node ? table_node->getStorageSnapshot() : table_function_node->getStorageSnapshot();
|
||||
additional_column_to_read = chooseSmallestColumnToReadFromStorage(storage, storage_snapshot);
|
||||
|
||||
}
|
||||
else if (query_node || union_node)
|
||||
{
|
||||
const auto & projection_columns = query_node ? query_node->getProjectionColumns() : union_node->computeProjectionColumns();
|
||||
NamesAndTypesList projection_columns_list(projection_columns.begin(), projection_columns.end());
|
||||
additional_column_to_read = ExpressionActions::getSmallestColumn(projection_columns_list);
|
||||
}
|
||||
else
|
||||
{
|
||||
throw Exception(ErrorCodes::LOGICAL_ERROR, "Expected table, table function, query or union. Actual {}",
|
||||
table_expression->formatASTForErrorMessage());
|
||||
}
|
||||
|
||||
auto & global_planner_context = planner_context->getGlobalPlannerContext();
|
||||
const auto & column_identifier = global_planner_context->createColumnIdentifier(additional_column_to_read, table_expression);
|
||||
columns_names.push_back(additional_column_to_read.name);
|
||||
table_expression_data.addColumn(additional_column_to_read, column_identifier);
|
||||
}
|
||||
|
||||
/// Limitation on the number of columns to read
|
||||
if (settings.max_columns_to_read && columns_names.size() > settings.max_columns_to_read)
|
||||
throw Exception(ErrorCodes::TOO_MANY_COLUMNS,
|
||||
"Limit for number of columns to read exceeded. Requested: {}, maximum: {}",
|
||||
columns_names.size(),
|
||||
settings.max_columns_to_read);
|
||||
}
|
||||
|
||||
JoinTreeQueryPlan buildQueryPlanForTableExpression(QueryTreeNodePtr table_expression,
|
||||
const SelectQueryInfo & select_query_info,
|
||||
const SelectQueryOptions & select_query_options,
|
||||
PlannerContextPtr & planner_context,
|
||||
bool is_single_table_expression,
|
||||
bool wrap_read_columns_in_subquery)
|
||||
{
|
||||
const auto & query_context = planner_context->getQueryContext();
|
||||
const auto & settings = query_context->getSettingsRef();
|
||||
|
||||
auto & table_expression_data = planner_context->getTableExpressionDataOrThrow(table_expression);
|
||||
|
||||
QueryProcessingStage::Enum from_stage = QueryProcessingStage::Enum::FetchColumns;
|
||||
|
||||
if (wrap_read_columns_in_subquery)
|
||||
{
|
||||
auto columns = table_expression_data.getColumns();
|
||||
table_expression = buildSubqueryToReadColumnsFromTableExpression(columns, table_expression, query_context);
|
||||
}
|
||||
|
||||
auto * table_node = table_expression->as<TableNode>();
|
||||
auto * table_function_node = table_expression->as<TableFunctionNode>();
|
||||
auto * query_node = table_expression->as<QueryNode>();
|
||||
@ -264,8 +334,6 @@ JoinTreeQueryPlan buildQueryPlanForTableExpression(const QueryTreeNodePtr & tabl
|
||||
|
||||
QueryPlan query_plan;
|
||||
|
||||
auto & table_expression_data = planner_context->getTableExpressionDataOrThrow(table_expression);
|
||||
|
||||
if (table_node || table_function_node)
|
||||
{
|
||||
const auto & storage = table_node ? table_node->getStorage() : table_function_node->getStorage();
|
||||
@ -362,32 +430,6 @@ JoinTreeQueryPlan buildQueryPlanForTableExpression(const QueryTreeNodePtr & tabl
|
||||
|
||||
auto columns_names = table_expression_data.getColumnNames();
|
||||
|
||||
/** The current user must have the SELECT privilege.
|
||||
* We do not check access rights for table functions because they have been already checked in ITableFunction::execute().
|
||||
*/
|
||||
if (table_node)
|
||||
{
|
||||
auto column_names_with_aliases = columns_names;
|
||||
const auto & alias_columns_names = table_expression_data.getAliasColumnsNames();
|
||||
column_names_with_aliases.insert(column_names_with_aliases.end(), alias_columns_names.begin(), alias_columns_names.end());
|
||||
checkAccessRights(*table_node, column_names_with_aliases, planner_context->getQueryContext());
|
||||
}
|
||||
|
||||
/// Limitation on the number of columns to read
|
||||
if (settings.max_columns_to_read && columns_names.size() > settings.max_columns_to_read)
|
||||
throw Exception(ErrorCodes::TOO_MANY_COLUMNS,
|
||||
"Limit for number of columns to read exceeded. Requested: {}, maximum: {}",
|
||||
columns_names.size(),
|
||||
settings.max_columns_to_read);
|
||||
|
||||
if (columns_names.empty())
|
||||
{
|
||||
auto additional_column_to_read = chooseSmallestColumnToReadFromStorage(storage, storage_snapshot);
|
||||
const auto & column_identifier = planner_context->getGlobalPlannerContext()->createColumnIdentifier(additional_column_to_read, table_expression);
|
||||
columns_names.push_back(additional_column_to_read.name);
|
||||
table_expression_data.addColumn(additional_column_to_read, column_identifier);
|
||||
}
|
||||
|
||||
bool need_rewrite_query_with_final = storage->needRewriteQueryWithFinal(columns_names);
|
||||
if (need_rewrite_query_with_final)
|
||||
{
|
||||
@ -423,6 +465,17 @@ JoinTreeQueryPlan buildQueryPlanForTableExpression(const QueryTreeNodePtr & tabl
|
||||
{
|
||||
from_stage = storage->getQueryProcessingStage(query_context, select_query_options.to_stage, storage_snapshot, table_expression_query_info);
|
||||
storage->read(query_plan, columns_names, storage_snapshot, table_expression_query_info, query_context, from_stage, max_block_size, max_streams);
|
||||
|
||||
if (query_context->hasQueryContext() && !select_query_options.is_internal)
|
||||
{
|
||||
auto local_storage_id = storage->getStorageID();
|
||||
query_context->getQueryContext()->addQueryAccessInfo(
|
||||
backQuoteIfNeed(local_storage_id.getDatabaseName()),
|
||||
local_storage_id.getFullTableName(),
|
||||
columns_names,
|
||||
{},
|
||||
{});
|
||||
}
|
||||
}
|
||||
|
||||
if (query_plan.isInitialized())
|
||||
@ -464,16 +517,6 @@ JoinTreeQueryPlan buildQueryPlanForTableExpression(const QueryTreeNodePtr & tabl
|
||||
}
|
||||
else
|
||||
{
|
||||
if (table_expression_data.getColumnNames().empty())
|
||||
{
|
||||
const auto & projection_columns = query_node ? query_node->getProjectionColumns() : union_node->computeProjectionColumns();
|
||||
NamesAndTypesList projection_columns_list(projection_columns.begin(), projection_columns.end());
|
||||
auto additional_column_to_read = ExpressionActions::getSmallestColumn(projection_columns_list);
|
||||
|
||||
const auto & column_identifier = planner_context->getGlobalPlannerContext()->createColumnIdentifier(additional_column_to_read, table_expression);
|
||||
table_expression_data.addColumn(additional_column_to_read, column_identifier);
|
||||
}
|
||||
|
||||
auto subquery_options = select_query_options.subquery();
|
||||
Planner subquery_planner(table_expression, subquery_options, planner_context->getGlobalPlannerContext());
|
||||
/// Propagate storage limits to subquery
|
||||
@ -516,10 +559,11 @@ JoinTreeQueryPlan buildQueryPlanForTableExpression(const QueryTreeNodePtr & tabl
|
||||
planner.buildQueryPlanIfNeeded();
|
||||
|
||||
auto expected_header = planner.getQueryPlan().getCurrentDataStream().header;
|
||||
materializeBlockInplace(expected_header);
|
||||
|
||||
if (!blocksHaveEqualStructure(query_plan.getCurrentDataStream().header, expected_header))
|
||||
{
|
||||
materializeBlockInplace(expected_header);
|
||||
|
||||
auto rename_actions_dag = ActionsDAG::makeConvertingActions(
|
||||
query_plan.getCurrentDataStream().header.getColumnsWithTypeAndName(),
|
||||
expected_header.getColumnsWithTypeAndName(),
|
||||
@ -1059,14 +1103,40 @@ JoinTreeQueryPlan buildJoinTreeQueryPlan(const QueryTreeNodePtr & query_node,
|
||||
const ColumnIdentifierSet & outer_scope_columns,
|
||||
PlannerContextPtr & planner_context)
|
||||
{
|
||||
const auto & query_node_typed = query_node->as<QueryNode &>();
|
||||
auto table_expressions_stack = buildTableExpressionsStack(query_node_typed.getJoinTree());
|
||||
auto table_expressions_stack = buildTableExpressionsStack(query_node->as<QueryNode &>().getJoinTree());
|
||||
size_t table_expressions_stack_size = table_expressions_stack.size();
|
||||
bool is_single_table_expression = table_expressions_stack_size == 1;
|
||||
|
||||
std::vector<ColumnIdentifierSet> table_expressions_outer_scope_columns(table_expressions_stack_size);
|
||||
ColumnIdentifierSet current_outer_scope_columns = outer_scope_columns;
|
||||
|
||||
/// For each table, table function, query, union table expressions prepare before query plan build
|
||||
for (size_t i = 0; i < table_expressions_stack_size; ++i)
|
||||
{
|
||||
const auto & table_expression = table_expressions_stack[i];
|
||||
auto table_expression_type = table_expression->getNodeType();
|
||||
if (table_expression_type == QueryTreeNodeType::JOIN ||
|
||||
table_expression_type == QueryTreeNodeType::ARRAY_JOIN)
|
||||
continue;
|
||||
|
||||
prepareBuildQueryPlanForTableExpression(table_expression, planner_context);
|
||||
}
|
||||
|
||||
/** If left most table expression query plan is planned to stage that is not equal to fetch columns,
|
||||
* then left most table expression is responsible for providing valid JOIN TREE part of final query plan.
|
||||
*
|
||||
* Examples: Distributed, LiveView, Merge storages.
|
||||
*/
|
||||
auto left_table_expression = table_expressions_stack.front();
|
||||
auto left_table_expression_query_plan = buildQueryPlanForTableExpression(left_table_expression,
|
||||
select_query_info,
|
||||
select_query_options,
|
||||
planner_context,
|
||||
is_single_table_expression,
|
||||
false /*wrap_read_columns_in_subquery*/);
|
||||
if (left_table_expression_query_plan.from_stage != QueryProcessingStage::FetchColumns)
|
||||
return left_table_expression_query_plan;
|
||||
|
||||
for (Int64 i = static_cast<Int64>(table_expressions_stack_size) - 1; i >= 0; --i)
|
||||
{
|
||||
table_expressions_outer_scope_columns[i] = current_outer_scope_columns;
|
||||
@ -1120,19 +1190,23 @@ JoinTreeQueryPlan buildJoinTreeQueryPlan(const QueryTreeNodePtr & query_node,
|
||||
}
|
||||
else
|
||||
{
|
||||
const auto & table_expression_data = planner_context->getTableExpressionDataOrThrow(table_expression);
|
||||
if (table_expression_data.isRemote() && i != 0)
|
||||
throw Exception(ErrorCodes::UNSUPPORTED_METHOD,
|
||||
"JOIN with multiple remote storages is unsupported");
|
||||
if (table_expression == left_table_expression)
|
||||
{
|
||||
query_plans_stack.push_back(std::move(left_table_expression_query_plan)); /// NOLINT
|
||||
left_table_expression = {};
|
||||
continue;
|
||||
}
|
||||
|
||||
/** If table expression is remote and it is not left most table expression, we wrap read columns from such
|
||||
* table expression in subquery.
|
||||
*/
|
||||
bool is_remote = planner_context->getTableExpressionDataOrThrow(table_expression).isRemote();
|
||||
query_plans_stack.push_back(buildQueryPlanForTableExpression(table_expression,
|
||||
select_query_info,
|
||||
select_query_options,
|
||||
planner_context,
|
||||
is_single_table_expression));
|
||||
|
||||
if (query_plans_stack.back().from_stage != QueryProcessingStage::FetchColumns)
|
||||
break;
|
||||
is_single_table_expression,
|
||||
is_remote /*wrap_read_columns_in_subquery*/));
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
@ -18,6 +18,7 @@
|
||||
#include <Functions/IFunction.h>
|
||||
#include <Functions/FunctionFactory.h>
|
||||
|
||||
#include <Analyzer/Utils.h>
|
||||
#include <Analyzer/FunctionNode.h>
|
||||
#include <Analyzer/ConstantNode.h>
|
||||
#include <Analyzer/TableNode.h>
|
||||
@ -61,6 +62,8 @@ void JoinClause::dump(WriteBuffer & buffer) const
|
||||
for (const auto & dag_node : dag_nodes)
|
||||
{
|
||||
dag_nodes_dump += dag_node->result_name;
|
||||
dag_nodes_dump += " ";
|
||||
dag_nodes_dump += dag_node->result_type->getName();
|
||||
dag_nodes_dump += ", ";
|
||||
}
|
||||
|
||||
|
||||
@ -101,6 +101,17 @@ public:
|
||||
return column_names;
|
||||
}
|
||||
|
||||
NamesAndTypes getColumns() const
|
||||
{
|
||||
NamesAndTypes result;
|
||||
result.reserve(column_names.size());
|
||||
|
||||
for (const auto & column_name : column_names)
|
||||
result.push_back(column_name_to_column.at(column_name));
|
||||
|
||||
return result;
|
||||
}
|
||||
|
||||
ColumnIdentifiers getColumnIdentifiers() const
|
||||
{
|
||||
ColumnIdentifiers result;
|
||||
|
||||
@ -4,6 +4,7 @@
|
||||
#include <Parsers/ASTSelectQuery.h>
|
||||
#include <Parsers/ASTSubquery.h>
|
||||
|
||||
#include <DataTypes/DataTypesNumber.h>
|
||||
#include <DataTypes/DataTypeLowCardinality.h>
|
||||
#include <DataTypes/DataTypeNullable.h>
|
||||
|
||||
@ -19,6 +20,7 @@
|
||||
|
||||
#include <Analyzer/Utils.h>
|
||||
#include <Analyzer/ConstantNode.h>
|
||||
#include <Analyzer/ColumnNode.h>
|
||||
#include <Analyzer/FunctionNode.h>
|
||||
#include <Analyzer/QueryNode.h>
|
||||
#include <Analyzer/UnionNode.h>
|
||||
@ -341,27 +343,6 @@ QueryTreeNodePtr mergeConditionNodes(const QueryTreeNodes & condition_nodes, con
|
||||
return function_node;
|
||||
}
|
||||
|
||||
std::optional<bool> tryExtractConstantFromConditionNode(const QueryTreeNodePtr & condition_node)
|
||||
{
|
||||
const auto * constant_node = condition_node->as<ConstantNode>();
|
||||
if (!constant_node)
|
||||
return {};
|
||||
|
||||
const auto & value = constant_node->getValue();
|
||||
auto constant_type = constant_node->getResultType();
|
||||
constant_type = removeNullable(removeLowCardinality(constant_type));
|
||||
|
||||
auto which_constant_type = WhichDataType(constant_type);
|
||||
if (!which_constant_type.isUInt8() && !which_constant_type.isNothing())
|
||||
return {};
|
||||
|
||||
if (value.isNull())
|
||||
return false;
|
||||
|
||||
UInt8 predicate_value = value.safeGet<UInt8>();
|
||||
return predicate_value > 0;
|
||||
}
|
||||
|
||||
QueryTreeNodePtr replaceTablesAndTableFunctionsWithDummyTables(const QueryTreeNodePtr & query_node,
|
||||
const ContextPtr & context,
|
||||
ResultReplacementMap * result_replacement_map)
|
||||
@ -391,4 +372,36 @@ QueryTreeNodePtr replaceTablesAndTableFunctionsWithDummyTables(const QueryTreeNo
|
||||
return query_node->cloneAndReplace(replacement_map);
|
||||
}
|
||||
|
||||
QueryTreeNodePtr buildSubqueryToReadColumnsFromTableExpression(const NamesAndTypes & columns,
|
||||
const QueryTreeNodePtr & table_expression,
|
||||
const ContextPtr & context)
|
||||
{
|
||||
auto projection_columns = columns;
|
||||
|
||||
QueryTreeNodes subquery_projection_nodes;
|
||||
subquery_projection_nodes.reserve(projection_columns.size());
|
||||
|
||||
for (const auto & column : projection_columns)
|
||||
subquery_projection_nodes.push_back(std::make_shared<ColumnNode>(column, table_expression));
|
||||
|
||||
if (subquery_projection_nodes.empty())
|
||||
{
|
||||
auto constant_data_type = std::make_shared<DataTypeUInt64>();
|
||||
subquery_projection_nodes.push_back(std::make_shared<ConstantNode>(1UL, constant_data_type));
|
||||
projection_columns.push_back({"1", std::move(constant_data_type)});
|
||||
}
|
||||
|
||||
auto context_copy = Context::createCopy(context);
|
||||
updateContextForSubqueryExecution(context_copy);
|
||||
|
||||
auto query_node = std::make_shared<QueryNode>(std::move(context_copy));
|
||||
|
||||
query_node->resolveProjectionColumns(projection_columns);
|
||||
query_node->getProjection().getNodes() = std::move(subquery_projection_nodes);
|
||||
query_node->getJoinTree() = table_expression;
|
||||
query_node->setIsSubquery(true);
|
||||
|
||||
return query_node;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
@ -63,13 +63,15 @@ bool queryHasWithTotalsInAnySubqueryInJoinTree(const QueryTreeNodePtr & query_no
|
||||
/// Returns `and` function node that has condition nodes as its arguments
|
||||
QueryTreeNodePtr mergeConditionNodes(const QueryTreeNodes & condition_nodes, const ContextPtr & context);
|
||||
|
||||
/// Try extract boolean constant from condition node
|
||||
std::optional<bool> tryExtractConstantFromConditionNode(const QueryTreeNodePtr & condition_node);
|
||||
|
||||
/// Replace tables nodes and table function nodes with dummy table nodes
|
||||
using ResultReplacementMap = std::unordered_map<QueryTreeNodePtr, QueryTreeNodePtr>;
|
||||
QueryTreeNodePtr replaceTablesAndTableFunctionsWithDummyTables(const QueryTreeNodePtr & query_node,
|
||||
const ContextPtr & context,
|
||||
ResultReplacementMap * result_replacement_map = nullptr);
|
||||
|
||||
/// Build subquery to read specified columns from table expression
|
||||
QueryTreeNodePtr buildSubqueryToReadColumnsFromTableExpression(const NamesAndTypes & columns,
|
||||
const QueryTreeNodePtr & table_expression,
|
||||
const ContextPtr & context);
|
||||
|
||||
}
|
||||
|
||||
@ -12,10 +12,19 @@ ISourceStep::ISourceStep(DataStream output_stream_)
|
||||
QueryPipelineBuilderPtr ISourceStep::updatePipeline(QueryPipelineBuilders, const BuildQueryPipelineSettings & settings)
|
||||
{
|
||||
auto pipeline = std::make_unique<QueryPipelineBuilder>();
|
||||
QueryPipelineProcessorsCollector collector(*pipeline, this);
|
||||
|
||||
/// For `Source` step, since it's not add new Processors to `pipeline->pipe`
|
||||
/// in `initializePipeline`, but make an assign with new created Pipe.
|
||||
/// And Processors for the Step is added here. So we do not need to use
|
||||
/// `QueryPipelineProcessorsCollector` to collect Processors.
|
||||
initializePipeline(*pipeline, settings);
|
||||
auto added_processors = collector.detachProcessors();
|
||||
processors.insert(processors.end(), added_processors.begin(), added_processors.end());
|
||||
|
||||
/// But we need to set QueryPlanStep manually for the Processors, which
|
||||
/// will be used in `EXPLAIN PIPELINE`
|
||||
for (auto & processor : processors)
|
||||
{
|
||||
processor->setQueryPlanStep(this);
|
||||
}
|
||||
return pipeline;
|
||||
}
|
||||
|
||||
|
||||
@ -519,8 +519,9 @@ AggregationInputOrder buildInputOrderInfo(
|
||||
|
||||
enreachFixedColumns(sorting_key_dag, fixed_key_columns);
|
||||
|
||||
for (auto it = matches.cbegin(); it != matches.cend(); ++it)
|
||||
for (const auto * output : dag->getOutputs())
|
||||
{
|
||||
auto it = matches.find(output);
|
||||
const MatchedTrees::Match * match = &it->second;
|
||||
if (match->node)
|
||||
{
|
||||
|
||||
@ -10,7 +10,6 @@ namespace DB
|
||||
* You can render it with:
|
||||
* dot -T png < pipeline.dot > pipeline.png
|
||||
*/
|
||||
|
||||
template <typename Processors, typename Statuses>
|
||||
void printPipeline(const Processors & processors, const Statuses & statuses, WriteBuffer & out)
|
||||
{
|
||||
@ -70,5 +69,4 @@ void printPipeline(const Processors & processors, WriteBuffer & out)
|
||||
/// If QueryPlanStep wasn't set for processor, representation may be not correct.
|
||||
/// If with_header is set, prints block header for each edge.
|
||||
void printPipelineCompact(const Processors & processors, WriteBuffer & out, bool with_header);
|
||||
|
||||
}
|
||||
|
||||
@ -13,6 +13,7 @@
|
||||
#include <Interpreters/getHeaderForProcessingStage.h>
|
||||
#include <Interpreters/SelectQueryOptions.h>
|
||||
#include <Interpreters/InterpreterSelectQuery.h>
|
||||
#include <Interpreters/InterpreterSelectQueryAnalyzer.h>
|
||||
#include <QueryPipeline/narrowPipe.h>
|
||||
#include <QueryPipeline/Pipe.h>
|
||||
#include <QueryPipeline/RemoteQueryExecutor.h>
|
||||
@ -83,8 +84,12 @@ Pipe StorageHDFSCluster::read(
|
||||
auto extension = getTaskIteratorExtension(query_info.query, context);
|
||||
|
||||
/// Calculate the header. This is significant, because some columns could be thrown away in some cases like query with count(*)
|
||||
Block header =
|
||||
InterpreterSelectQuery(query_info.query, context, SelectQueryOptions(processed_stage).analyze()).getSampleBlock();
|
||||
Block header;
|
||||
|
||||
if (context->getSettingsRef().allow_experimental_analyzer)
|
||||
header = InterpreterSelectQueryAnalyzer::getSampleBlock(query_info.query, context, SelectQueryOptions(processed_stage).analyze());
|
||||
else
|
||||
header = InterpreterSelectQuery(query_info.query, context, SelectQueryOptions(processed_stage).analyze()).getSampleBlock();
|
||||
|
||||
const Scalars & scalars = context->hasQueryContext() ? context->getQueryContext()->getScalars() : Scalars{};
|
||||
|
||||
|
||||
@ -4121,9 +4121,9 @@ void MergeTreeData::delayInsertOrThrowIfNeeded(Poco::Event * until, const Contex
|
||||
ProfileEvents::increment(ProfileEvents::RejectedInserts);
|
||||
throw Exception(
|
||||
ErrorCodes::TOO_MANY_PARTS,
|
||||
"Too many parts ({}) in all partitions in total. This indicates wrong choice of partition key. The threshold can be modified "
|
||||
"Too many parts ({}) in all partitions in total in table '{}'. This indicates wrong choice of partition key. The threshold can be modified "
|
||||
"with 'max_parts_in_total' setting in <merge_tree> element in config.xml or with per-table setting.",
|
||||
parts_count_in_total);
|
||||
parts_count_in_total, getLogName());
|
||||
}
|
||||
|
||||
size_t outdated_parts_over_threshold = 0;
|
||||
@ -4137,8 +4137,8 @@ void MergeTreeData::delayInsertOrThrowIfNeeded(Poco::Event * until, const Contex
|
||||
ProfileEvents::increment(ProfileEvents::RejectedInserts);
|
||||
throw Exception(
|
||||
ErrorCodes::TOO_MANY_PARTS,
|
||||
"Too many inactive parts ({}). Parts cleaning are processing significantly slower than inserts",
|
||||
outdated_parts_count_in_partition);
|
||||
"Too many inactive parts ({}) in table '{}'. Parts cleaning are processing significantly slower than inserts",
|
||||
outdated_parts_count_in_partition, getLogName());
|
||||
}
|
||||
if (settings->inactive_parts_to_delay_insert > 0 && outdated_parts_count_in_partition >= settings->inactive_parts_to_delay_insert)
|
||||
outdated_parts_over_threshold = outdated_parts_count_in_partition - settings->inactive_parts_to_delay_insert + 1;
|
||||
@ -4151,6 +4151,7 @@ void MergeTreeData::delayInsertOrThrowIfNeeded(Poco::Event * until, const Contex
|
||||
const auto active_parts_to_throw_insert
|
||||
= query_settings.parts_to_throw_insert ? query_settings.parts_to_throw_insert : settings->parts_to_throw_insert;
|
||||
size_t active_parts_over_threshold = 0;
|
||||
|
||||
{
|
||||
bool parts_are_large_enough_in_average
|
||||
= settings->max_avg_part_size_for_too_many_parts && average_part_size > settings->max_avg_part_size_for_too_many_parts;
|
||||
@ -4160,9 +4161,10 @@ void MergeTreeData::delayInsertOrThrowIfNeeded(Poco::Event * until, const Contex
|
||||
ProfileEvents::increment(ProfileEvents::RejectedInserts);
|
||||
throw Exception(
|
||||
ErrorCodes::TOO_MANY_PARTS,
|
||||
"Too many parts ({} with average size of {}). Merges are processing significantly slower than inserts",
|
||||
"Too many parts ({} with average size of {}) in table '{}'. Merges are processing significantly slower than inserts",
|
||||
parts_count_in_partition,
|
||||
ReadableSize(average_part_size));
|
||||
ReadableSize(average_part_size),
|
||||
getLogName());
|
||||
}
|
||||
if (active_parts_to_delay_insert > 0 && parts_count_in_partition >= active_parts_to_delay_insert
|
||||
&& !parts_are_large_enough_in_average)
|
||||
|
||||
@ -201,6 +201,7 @@ MergeTreeConditionInverted::MergeTreeConditionInverted(
|
||||
rpn.push_back(RPNElement::FUNCTION_UNKNOWN);
|
||||
return;
|
||||
}
|
||||
|
||||
rpn = std::move(
|
||||
RPNBuilder<RPNElement>(
|
||||
query_info.filter_actions_dag->getOutputs().at(0), context_,
|
||||
@ -208,10 +209,10 @@ MergeTreeConditionInverted::MergeTreeConditionInverted(
|
||||
{
|
||||
return this->traverseAtomAST(node, out);
|
||||
}).extractRPN());
|
||||
return;
|
||||
}
|
||||
|
||||
ASTPtr filter_node = buildFilterNode(query_info.query);
|
||||
|
||||
if (!filter_node)
|
||||
{
|
||||
rpn.push_back(RPNElement::FUNCTION_UNKNOWN);
|
||||
@ -226,7 +227,6 @@ MergeTreeConditionInverted::MergeTreeConditionInverted(
|
||||
query_info.prepared_sets,
|
||||
[&](const RPNBuilderTreeNode & node, RPNElement & out) { return traverseAtomAST(node, out); });
|
||||
rpn = std::move(builder).extractRPN();
|
||||
|
||||
}
|
||||
|
||||
/// Keep in-sync with MergeTreeConditionFullText::alwaysUnknownOrTrue
|
||||
|
||||
@ -59,7 +59,7 @@ void appendColumnNameWithoutAlias(const ActionsDAG::Node & node, WriteBuffer & o
|
||||
{
|
||||
auto name = node.function_base->getName();
|
||||
if (legacy && name == "modulo")
|
||||
writeCString("moduleLegacy", out);
|
||||
writeCString("moduloLegacy", out);
|
||||
else
|
||||
writeString(name, out);
|
||||
|
||||
|
||||
@ -60,11 +60,11 @@ void ReplicatedMergeTreeAttachThread::run()
|
||||
|
||||
if (needs_retry)
|
||||
{
|
||||
LOG_ERROR(log, "Initialization failed. Error: {}", e.message());
|
||||
LOG_ERROR(log, "Initialization failed. Error: {}", getCurrentExceptionMessage(/* with_stacktrace */ true));
|
||||
}
|
||||
else
|
||||
{
|
||||
LOG_ERROR(log, "Initialization failed, table will remain readonly. Error: {}", e.message());
|
||||
LOG_ERROR(log, "Initialization failed, table will remain readonly. Error: {}", getCurrentExceptionMessage(/* with_stacktrace */ true));
|
||||
storage.initialization_done = true;
|
||||
}
|
||||
}
|
||||
|
||||
@ -39,11 +39,16 @@
|
||||
#include <Parsers/parseQuery.h>
|
||||
#include <Parsers/IAST.h>
|
||||
|
||||
#include <Analyzer/Utils.h>
|
||||
#include <Analyzer/ColumnNode.h>
|
||||
#include <Analyzer/FunctionNode.h>
|
||||
#include <Analyzer/TableNode.h>
|
||||
#include <Analyzer/TableFunctionNode.h>
|
||||
#include <Analyzer/QueryNode.h>
|
||||
#include <Analyzer/JoinNode.h>
|
||||
#include <Analyzer/QueryTreeBuilder.h>
|
||||
#include <Analyzer/Passes/QueryAnalysisPass.h>
|
||||
#include <Analyzer/InDepthQueryTreeVisitor.h>
|
||||
|
||||
#include <Planner/Planner.h>
|
||||
#include <Planner/Utils.h>
|
||||
@ -55,6 +60,7 @@
|
||||
#include <Interpreters/ExpressionAnalyzer.h>
|
||||
#include <Interpreters/InterpreterDescribeQuery.h>
|
||||
#include <Interpreters/InterpreterSelectQuery.h>
|
||||
#include <Interpreters/InterpreterSelectQueryAnalyzer.h>
|
||||
#include <Interpreters/InterpreterInsertQuery.h>
|
||||
#include <Interpreters/JoinedTables.h>
|
||||
#include <Interpreters/TranslateQualifiedNamesVisitor.h>
|
||||
@ -69,12 +75,14 @@
|
||||
#include <Interpreters/getCustomKeyFilterForParallelReplicas.h>
|
||||
|
||||
#include <Functions/IFunction.h>
|
||||
#include <Functions/FunctionFactory.h>
|
||||
#include <TableFunctions/TableFunctionView.h>
|
||||
#include <TableFunctions/TableFunctionFactory.h>
|
||||
|
||||
#include <Storages/IStorageCluster.h>
|
||||
|
||||
#include <Processors/Executors/PushingPipelineExecutor.h>
|
||||
#include <Processors/Executors/CompletedPipelineExecutor.h>
|
||||
#include <Processors/QueryPlan/QueryPlan.h>
|
||||
#include <Processors/QueryPlan/BuildQueryPipelineSettings.h>
|
||||
#include <Processors/QueryPlan/Optimizations/QueryPlanOptimizationSettings.h>
|
||||
@ -138,6 +146,7 @@ namespace ErrorCodes
|
||||
extern const int DISTRIBUTED_TOO_MANY_PENDING_BYTES;
|
||||
extern const int ARGUMENT_OUT_OF_BOUND;
|
||||
extern const int TOO_LARGE_DISTRIBUTED_DEPTH;
|
||||
extern const int DISTRIBUTED_IN_JOIN_SUBQUERY_DENIED;
|
||||
}
|
||||
|
||||
namespace ActionLocks
|
||||
@ -634,12 +643,278 @@ StorageSnapshotPtr StorageDistributed::getStorageSnapshotForQuery(
|
||||
namespace
|
||||
{
|
||||
|
||||
QueryTreeNodePtr buildQueryTreeDistributedTableReplacedWithLocalTable(const SelectQueryInfo & query_info,
|
||||
/// Visitor that collect column source to columns mapping from query and all subqueries
|
||||
class CollectColumnSourceToColumnsVisitor : public InDepthQueryTreeVisitor<CollectColumnSourceToColumnsVisitor>
|
||||
{
|
||||
public:
|
||||
struct Columns
|
||||
{
|
||||
NameSet column_names;
|
||||
NamesAndTypes columns;
|
||||
|
||||
void addColumn(NameAndTypePair column)
|
||||
{
|
||||
if (column_names.contains(column.name))
|
||||
return;
|
||||
|
||||
column_names.insert(column.name);
|
||||
columns.push_back(std::move(column));
|
||||
}
|
||||
};
|
||||
|
||||
const std::unordered_map<QueryTreeNodePtr, Columns> & getColumnSourceToColumns() const
|
||||
{
|
||||
return column_source_to_columns;
|
||||
}
|
||||
|
||||
void visitImpl(QueryTreeNodePtr & node)
|
||||
{
|
||||
auto * column_node = node->as<ColumnNode>();
|
||||
if (!column_node)
|
||||
return;
|
||||
|
||||
auto column_source = column_node->getColumnSourceOrNull();
|
||||
if (!column_source)
|
||||
return;
|
||||
|
||||
auto it = column_source_to_columns.find(column_source);
|
||||
if (it == column_source_to_columns.end())
|
||||
{
|
||||
auto [insert_it, _] = column_source_to_columns.emplace(column_source, Columns());
|
||||
it = insert_it;
|
||||
}
|
||||
|
||||
it->second.addColumn(column_node->getColumn());
|
||||
}
|
||||
|
||||
private:
|
||||
std::unordered_map<QueryTreeNodePtr, Columns> column_source_to_columns;
|
||||
};
|
||||
|
||||
/** Visitor that rewrites IN and JOINs in query and all subqueries according to distributed_product_mode and
|
||||
* prefer_global_in_and_join settings.
|
||||
*
|
||||
* Additionally collects GLOBAL JOIN and GLOBAL IN query nodes.
|
||||
*
|
||||
* If distributed_product_mode = deny, then visitor throws exception if there are multiple distributed tables.
|
||||
* If distributed_product_mode = local, then visitor collects replacement map for tables that must be replaced
|
||||
* with local tables.
|
||||
* If distributed_product_mode = global or prefer_global_in_and_join setting is true, then visitor rewrites JOINs and IN functions that
|
||||
* contain distributed tables to GLOBAL JOINs and GLOBAL IN functions.
|
||||
* If distributed_product_mode = allow, then visitor does not rewrite query if there are multiple distributed tables.
|
||||
*/
|
||||
class DistributedProductModeRewriteInJoinVisitor : public InDepthQueryTreeVisitorWithContext<DistributedProductModeRewriteInJoinVisitor>
|
||||
{
|
||||
public:
|
||||
using Base = InDepthQueryTreeVisitorWithContext<DistributedProductModeRewriteInJoinVisitor>;
|
||||
using Base::Base;
|
||||
|
||||
explicit DistributedProductModeRewriteInJoinVisitor(const ContextPtr & context_)
|
||||
: Base(context_)
|
||||
{}
|
||||
|
||||
struct InFunctionOrJoin
|
||||
{
|
||||
QueryTreeNodePtr query_node;
|
||||
size_t subquery_depth = 0;
|
||||
};
|
||||
|
||||
const std::unordered_map<const IQueryTreeNode *, QueryTreeNodePtr> & getReplacementMap() const
|
||||
{
|
||||
return replacement_map;
|
||||
}
|
||||
|
||||
const std::vector<InFunctionOrJoin> & getGlobalInOrJoinNodes() const
|
||||
{
|
||||
return global_in_or_join_nodes;
|
||||
}
|
||||
|
||||
static bool needChildVisit(QueryTreeNodePtr & parent, QueryTreeNodePtr & child)
|
||||
{
|
||||
auto * function_node = parent->as<FunctionNode>();
|
||||
if (function_node && isNameOfGlobalInFunction(function_node->getFunctionName()))
|
||||
return false;
|
||||
|
||||
auto * join_node = parent->as<JoinNode>();
|
||||
if (join_node && join_node->getLocality() == JoinLocality::Global && join_node->getRightTableExpression() == child)
|
||||
return false;
|
||||
|
||||
return true;
|
||||
}
|
||||
|
||||
void visitImpl(QueryTreeNodePtr & node)
|
||||
{
|
||||
auto * function_node = node->as<FunctionNode>();
|
||||
auto * join_node = node->as<JoinNode>();
|
||||
|
||||
if ((function_node && isNameOfGlobalInFunction(function_node->getFunctionName())) ||
|
||||
(join_node && join_node->getLocality() == JoinLocality::Global))
|
||||
{
|
||||
InFunctionOrJoin in_function_or_join_entry;
|
||||
in_function_or_join_entry.query_node = node;
|
||||
in_function_or_join_entry.subquery_depth = getSubqueryDepth();
|
||||
global_in_or_join_nodes.push_back(std::move(in_function_or_join_entry));
|
||||
return;
|
||||
}
|
||||
|
||||
if ((function_node && isNameOfLocalInFunction(function_node->getFunctionName())) ||
|
||||
(join_node && join_node->getLocality() != JoinLocality::Global))
|
||||
{
|
||||
InFunctionOrJoin in_function_or_join_entry;
|
||||
in_function_or_join_entry.query_node = node;
|
||||
in_function_or_join_entry.subquery_depth = getSubqueryDepth();
|
||||
in_function_or_join_stack.push_back(in_function_or_join_entry);
|
||||
return;
|
||||
}
|
||||
|
||||
if (node->getNodeType() == QueryTreeNodeType::TABLE)
|
||||
tryRewriteTableNodeIfNeeded(node);
|
||||
}
|
||||
|
||||
void leaveImpl(QueryTreeNodePtr & node)
|
||||
{
|
||||
if (!in_function_or_join_stack.empty() && node.get() == in_function_or_join_stack.back().query_node.get())
|
||||
in_function_or_join_stack.pop_back();
|
||||
}
|
||||
|
||||
private:
|
||||
void tryRewriteTableNodeIfNeeded(const QueryTreeNodePtr & table_node)
|
||||
{
|
||||
const auto & table_node_typed = table_node->as<TableNode &>();
|
||||
const auto * distributed_storage = typeid_cast<const StorageDistributed *>(table_node_typed.getStorage().get());
|
||||
if (!distributed_storage)
|
||||
return;
|
||||
|
||||
bool distributed_valid_for_rewrite = distributed_storage->getShardCount() >= 2;
|
||||
if (!distributed_valid_for_rewrite)
|
||||
return;
|
||||
|
||||
auto distributed_product_mode = getSettings().distributed_product_mode;
|
||||
|
||||
if (distributed_product_mode == DistributedProductMode::LOCAL)
|
||||
{
|
||||
StorageID remote_storage_id = StorageID{distributed_storage->getRemoteDatabaseName(),
|
||||
distributed_storage->getRemoteTableName()};
|
||||
auto resolved_remote_storage_id = getContext()->resolveStorageID(remote_storage_id);
|
||||
const auto & distributed_storage_columns = table_node_typed.getStorageSnapshot()->metadata->getColumns();
|
||||
auto storage = std::make_shared<StorageDummy>(resolved_remote_storage_id, distributed_storage_columns);
|
||||
auto replacement_table_expression = std::make_shared<TableNode>(std::move(storage), getContext());
|
||||
replacement_map.emplace(table_node.get(), std::move(replacement_table_expression));
|
||||
}
|
||||
else if ((distributed_product_mode == DistributedProductMode::GLOBAL || getSettings().prefer_global_in_and_join) &&
|
||||
!in_function_or_join_stack.empty())
|
||||
{
|
||||
auto * in_or_join_node_to_modify = in_function_or_join_stack.back().query_node.get();
|
||||
|
||||
if (auto * in_function_to_modify = in_or_join_node_to_modify->as<FunctionNode>())
|
||||
{
|
||||
auto global_in_function_name = getGlobalInFunctionNameForLocalInFunctionName(in_function_to_modify->getFunctionName());
|
||||
auto global_in_function_resolver = FunctionFactory::instance().get(global_in_function_name, getContext());
|
||||
in_function_to_modify->resolveAsFunction(global_in_function_resolver->build(in_function_to_modify->getArgumentColumns()));
|
||||
}
|
||||
else if (auto * join_node_to_modify = in_or_join_node_to_modify->as<JoinNode>())
|
||||
{
|
||||
join_node_to_modify->setLocality(JoinLocality::Global);
|
||||
}
|
||||
|
||||
global_in_or_join_nodes.push_back(in_function_or_join_stack.back());
|
||||
}
|
||||
else if (distributed_product_mode == DistributedProductMode::ALLOW)
|
||||
{
|
||||
return;
|
||||
}
|
||||
else if (distributed_product_mode == DistributedProductMode::DENY)
|
||||
{
|
||||
throw Exception(ErrorCodes::DISTRIBUTED_IN_JOIN_SUBQUERY_DENIED,
|
||||
"Double-distributed IN/JOIN subqueries is denied (distributed_product_mode = 'deny'). "
|
||||
"You may rewrite query to use local tables "
|
||||
"in subqueries, or use GLOBAL keyword, or set distributed_product_mode to suitable value.");
|
||||
}
|
||||
}
|
||||
|
||||
std::vector<InFunctionOrJoin> in_function_or_join_stack;
|
||||
std::unordered_map<const IQueryTreeNode *, QueryTreeNodePtr> replacement_map;
|
||||
std::vector<InFunctionOrJoin> global_in_or_join_nodes;
|
||||
};
|
||||
|
||||
/** Execute subquery node and put result in mutable context temporary table.
|
||||
* Returns table node that is initialized with temporary table storage.
|
||||
*/
|
||||
QueryTreeNodePtr executeSubqueryNode(const QueryTreeNodePtr & subquery_node,
|
||||
ContextMutablePtr & mutable_context,
|
||||
size_t subquery_depth)
|
||||
{
|
||||
auto subquery_hash = subquery_node->getTreeHash();
|
||||
String temporary_table_name = fmt::format("_data_{}_{}", subquery_hash.first, subquery_hash.second);
|
||||
|
||||
const auto & external_tables = mutable_context->getExternalTables();
|
||||
auto external_table_it = external_tables.find(temporary_table_name);
|
||||
if (external_table_it != external_tables.end())
|
||||
{
|
||||
auto temporary_table_expression_node = std::make_shared<TableNode>(external_table_it->second, mutable_context);
|
||||
temporary_table_expression_node->setTemporaryTableName(temporary_table_name);
|
||||
return temporary_table_expression_node;
|
||||
}
|
||||
|
||||
auto subquery_options = SelectQueryOptions(QueryProcessingStage::Complete, subquery_depth, true /*is_subquery*/);
|
||||
auto context_copy = Context::createCopy(mutable_context);
|
||||
updateContextForSubqueryExecution(context_copy);
|
||||
|
||||
InterpreterSelectQueryAnalyzer interpreter(subquery_node, context_copy, subquery_options);
|
||||
auto & query_plan = interpreter.getQueryPlan();
|
||||
|
||||
auto sample_block_with_unique_names = query_plan.getCurrentDataStream().header;
|
||||
makeUniqueColumnNamesInBlock(sample_block_with_unique_names);
|
||||
|
||||
if (!blocksHaveEqualStructure(sample_block_with_unique_names, query_plan.getCurrentDataStream().header))
|
||||
{
|
||||
auto actions_dag = ActionsDAG::makeConvertingActions(
|
||||
query_plan.getCurrentDataStream().header.getColumnsWithTypeAndName(),
|
||||
sample_block_with_unique_names.getColumnsWithTypeAndName(),
|
||||
ActionsDAG::MatchColumnsMode::Position);
|
||||
auto converting_step = std::make_unique<ExpressionStep>(query_plan.getCurrentDataStream(), std::move(actions_dag));
|
||||
query_plan.addStep(std::move(converting_step));
|
||||
}
|
||||
|
||||
Block sample = interpreter.getSampleBlock();
|
||||
NamesAndTypesList columns = sample.getNamesAndTypesList();
|
||||
|
||||
auto external_storage_holder = TemporaryTableHolder(
|
||||
mutable_context,
|
||||
ColumnsDescription{columns},
|
||||
ConstraintsDescription{},
|
||||
nullptr /*query*/,
|
||||
true /*create_for_global_subquery*/);
|
||||
|
||||
StoragePtr external_storage = external_storage_holder.getTable();
|
||||
auto temporary_table_expression_node = std::make_shared<TableNode>(external_storage, mutable_context);
|
||||
temporary_table_expression_node->setTemporaryTableName(temporary_table_name);
|
||||
|
||||
auto table_out = external_storage->write({}, external_storage->getInMemoryMetadataPtr(), mutable_context);
|
||||
auto io = interpreter.execute();
|
||||
io.pipeline.complete(std::move(table_out));
|
||||
CompletedPipelineExecutor executor(io.pipeline);
|
||||
executor.execute();
|
||||
|
||||
mutable_context->addExternalTable(temporary_table_name, std::move(external_storage_holder));
|
||||
|
||||
return temporary_table_expression_node;
|
||||
}
|
||||
|
||||
QueryTreeNodePtr buildQueryTreeDistributed(SelectQueryInfo & query_info,
|
||||
const StorageSnapshotPtr & distributed_storage_snapshot,
|
||||
const StorageID & remote_storage_id,
|
||||
const ASTPtr & remote_table_function)
|
||||
{
|
||||
const auto & query_context = query_info.planner_context->getQueryContext();
|
||||
auto & planner_context = query_info.planner_context;
|
||||
const auto & query_context = planner_context->getQueryContext();
|
||||
|
||||
std::optional<TableExpressionModifiers> table_expression_modifiers;
|
||||
|
||||
if (auto * query_info_table_node = query_info.table_expression->as<TableNode>())
|
||||
table_expression_modifiers = query_info_table_node->getTableExpressionModifiers();
|
||||
else if (auto * query_info_table_function_node = query_info.table_expression->as<TableFunctionNode>())
|
||||
table_expression_modifiers = query_info_table_function_node->getTableExpressionModifiers();
|
||||
|
||||
QueryTreeNodePtr replacement_table_expression;
|
||||
|
||||
@ -651,6 +926,9 @@ QueryTreeNodePtr buildQueryTreeDistributedTableReplacedWithLocalTable(const Sele
|
||||
auto table_function_node = std::make_shared<TableFunctionNode>(remote_table_function_node.getFunctionName());
|
||||
table_function_node->getArgumentsNode() = remote_table_function_node.getArgumentsNode();
|
||||
|
||||
if (table_expression_modifiers)
|
||||
table_function_node->setTableExpressionModifiers(*table_expression_modifiers);
|
||||
|
||||
QueryAnalysisPass query_analysis_pass;
|
||||
query_analysis_pass.run(table_function_node, query_context);
|
||||
|
||||
@ -660,13 +938,89 @@ QueryTreeNodePtr buildQueryTreeDistributedTableReplacedWithLocalTable(const Sele
|
||||
{
|
||||
auto resolved_remote_storage_id = query_context->resolveStorageID(remote_storage_id);
|
||||
auto storage = std::make_shared<StorageDummy>(resolved_remote_storage_id, distributed_storage_snapshot->metadata->getColumns());
|
||||
auto table_node = std::make_shared<TableNode>(std::move(storage), query_context);
|
||||
|
||||
replacement_table_expression = std::make_shared<TableNode>(std::move(storage), query_context);
|
||||
if (table_expression_modifiers)
|
||||
table_node->setTableExpressionModifiers(*table_expression_modifiers);
|
||||
|
||||
replacement_table_expression = std::move(table_node);
|
||||
}
|
||||
|
||||
replacement_table_expression->setAlias(query_info.table_expression->getAlias());
|
||||
|
||||
return query_info.query_tree->cloneAndReplace(query_info.table_expression, std::move(replacement_table_expression));
|
||||
auto query_tree_to_modify = query_info.query_tree->cloneAndReplace(query_info.table_expression, std::move(replacement_table_expression));
|
||||
|
||||
CollectColumnSourceToColumnsVisitor collect_column_source_to_columns_visitor;
|
||||
collect_column_source_to_columns_visitor.visit(query_tree_to_modify);
|
||||
|
||||
const auto & column_source_to_columns = collect_column_source_to_columns_visitor.getColumnSourceToColumns();
|
||||
|
||||
DistributedProductModeRewriteInJoinVisitor visitor(query_info.planner_context->getQueryContext());
|
||||
visitor.visit(query_tree_to_modify);
|
||||
|
||||
auto replacement_map = visitor.getReplacementMap();
|
||||
const auto & global_in_or_join_nodes = visitor.getGlobalInOrJoinNodes();
|
||||
|
||||
for (const auto & global_in_or_join_node : global_in_or_join_nodes)
|
||||
{
|
||||
if (auto * join_node = global_in_or_join_node.query_node->as<JoinNode>())
|
||||
{
|
||||
auto join_right_table_expression = join_node->getRightTableExpression();
|
||||
auto join_right_table_expression_node_type = join_right_table_expression->getNodeType();
|
||||
|
||||
QueryTreeNodePtr subquery_node;
|
||||
|
||||
if (join_right_table_expression_node_type == QueryTreeNodeType::QUERY ||
|
||||
join_right_table_expression_node_type == QueryTreeNodeType::UNION)
|
||||
{
|
||||
subquery_node = join_right_table_expression;
|
||||
}
|
||||
else if (join_right_table_expression_node_type == QueryTreeNodeType::TABLE ||
|
||||
join_right_table_expression_node_type == QueryTreeNodeType::TABLE_FUNCTION)
|
||||
{
|
||||
const auto & columns = column_source_to_columns.at(join_right_table_expression).columns;
|
||||
subquery_node = buildSubqueryToReadColumnsFromTableExpression(columns,
|
||||
join_right_table_expression,
|
||||
planner_context->getQueryContext());
|
||||
}
|
||||
else
|
||||
{
|
||||
throw Exception(ErrorCodes::LOGICAL_ERROR,
|
||||
"Expected JOIN right table expression to be table, table function, query or union node. Actual {}",
|
||||
join_right_table_expression->formatASTForErrorMessage());
|
||||
}
|
||||
|
||||
auto temporary_table_expression_node = executeSubqueryNode(subquery_node,
|
||||
planner_context->getMutableQueryContext(),
|
||||
global_in_or_join_node.subquery_depth);
|
||||
temporary_table_expression_node->setAlias(join_right_table_expression->getAlias());
|
||||
replacement_map.emplace(join_right_table_expression.get(), std::move(temporary_table_expression_node));
|
||||
continue;
|
||||
}
|
||||
else if (auto * in_function_node = global_in_or_join_node.query_node->as<FunctionNode>())
|
||||
{
|
||||
auto & in_function_subquery_node = in_function_node->getArguments().getNodes().at(1);
|
||||
auto in_function_node_type = in_function_subquery_node->getNodeType();
|
||||
if (in_function_node_type != QueryTreeNodeType::QUERY && in_function_node_type != QueryTreeNodeType::UNION)
|
||||
continue;
|
||||
|
||||
auto temporary_table_expression_node = executeSubqueryNode(in_function_subquery_node,
|
||||
planner_context->getMutableQueryContext(),
|
||||
global_in_or_join_node.subquery_depth);
|
||||
in_function_subquery_node = std::move(temporary_table_expression_node);
|
||||
}
|
||||
else
|
||||
{
|
||||
throw Exception(ErrorCodes::LOGICAL_ERROR,
|
||||
"Expected global IN or JOIN query node. Actual {}",
|
||||
global_in_or_join_node.query_node->formatASTForErrorMessage());
|
||||
}
|
||||
}
|
||||
|
||||
if (!replacement_map.empty())
|
||||
query_tree_to_modify = query_tree_to_modify->cloneAndReplace(replacement_map);
|
||||
|
||||
return query_tree_to_modify;
|
||||
}
|
||||
|
||||
}
|
||||
@ -694,17 +1048,13 @@ void StorageDistributed::read(
|
||||
if (!remote_table_function_ptr)
|
||||
remote_storage_id = StorageID{remote_database, remote_table};
|
||||
|
||||
auto query_tree_with_replaced_distributed_table = buildQueryTreeDistributedTableReplacedWithLocalTable(query_info,
|
||||
auto query_tree_distributed = buildQueryTreeDistributed(query_info,
|
||||
storage_snapshot,
|
||||
remote_storage_id,
|
||||
remote_table_function_ptr);
|
||||
|
||||
query_ast = queryNodeToSelectQuery(query_tree_with_replaced_distributed_table);
|
||||
|
||||
Planner planner(query_tree_with_replaced_distributed_table, SelectQueryOptions(processed_stage).analyze());
|
||||
planner.buildQueryPlanIfNeeded();
|
||||
|
||||
header = planner.getQueryPlan().getCurrentDataStream().header;
|
||||
query_ast = queryNodeToSelectQuery(query_tree_distributed);
|
||||
header = InterpreterSelectQueryAnalyzer::getSampleBlock(query_ast, local_context, SelectQueryOptions(processed_stage).analyze());
|
||||
}
|
||||
else
|
||||
{
|
||||
|
||||
@ -18,6 +18,7 @@
|
||||
#include <Interpreters/TransactionLog.h>
|
||||
#include <Interpreters/ClusterProxy/executeQuery.h>
|
||||
#include <Interpreters/ClusterProxy/SelectStreamFactory.h>
|
||||
#include <Interpreters/InterpreterSelectQueryAnalyzer.h>
|
||||
#include <IO/copyData.h>
|
||||
#include <Parsers/ASTCheckQuery.h>
|
||||
#include <Parsers/ASTFunction.h>
|
||||
@ -223,8 +224,12 @@ void StorageMergeTree::read(
|
||||
|
||||
auto cluster = local_context->getCluster(local_context->getSettingsRef().cluster_for_parallel_replicas);
|
||||
|
||||
Block header =
|
||||
InterpreterSelectQuery(modified_query_ast, local_context, SelectQueryOptions(processed_stage).analyze()).getSampleBlock();
|
||||
Block header;
|
||||
|
||||
if (local_context->getSettingsRef().allow_experimental_analyzer)
|
||||
header = InterpreterSelectQueryAnalyzer::getSampleBlock(modified_query_ast, local_context, SelectQueryOptions(processed_stage).analyze());
|
||||
else
|
||||
header = InterpreterSelectQuery(modified_query_ast, local_context, SelectQueryOptions(processed_stage).analyze()).getSampleBlock();
|
||||
|
||||
ClusterProxy::SelectStreamFactory select_stream_factory =
|
||||
ClusterProxy::SelectStreamFactory(
|
||||
|
||||
@ -17,6 +17,7 @@
|
||||
#include <Interpreters/InterpreterSelectQuery.h>
|
||||
#include <Interpreters/AddDefaultDatabaseVisitor.h>
|
||||
#include <Interpreters/TranslateQualifiedNamesVisitor.h>
|
||||
#include <Interpreters/InterpreterSelectQueryAnalyzer.h>
|
||||
#include <Processors/Transforms/AddingDefaultsTransform.h>
|
||||
#include <QueryPipeline/narrowPipe.h>
|
||||
#include <QueryPipeline/Pipe.h>
|
||||
@ -102,7 +103,20 @@ Pipe StorageS3Cluster::read(
|
||||
auto extension = getTaskIteratorExtension(query_info.query, context);
|
||||
|
||||
/// Calculate the header. This is significant, because some columns could be thrown away in some cases like query with count(*)
|
||||
|
||||
Block sample_block;
|
||||
ASTPtr query_to_send = query_info.query;
|
||||
|
||||
if (context->getSettingsRef().allow_experimental_analyzer)
|
||||
{
|
||||
sample_block = InterpreterSelectQueryAnalyzer::getSampleBlock(query_info.query, context, SelectQueryOptions(processed_stage));
|
||||
}
|
||||
else
|
||||
{
|
||||
auto interpreter = InterpreterSelectQuery(query_info.query, context, SelectQueryOptions(processed_stage).analyze());
|
||||
sample_block = interpreter.getSampleBlock();
|
||||
query_to_send = interpreter.getQueryInfo().query->clone();
|
||||
}
|
||||
|
||||
const Scalars & scalars = context->hasQueryContext() ? context->getQueryContext()->getScalars() : Scalars{};
|
||||
|
||||
@ -110,7 +124,6 @@ Pipe StorageS3Cluster::read(
|
||||
|
||||
const bool add_agg_info = processed_stage == QueryProcessingStage::WithMergeableState;
|
||||
|
||||
ASTPtr query_to_send = interpreter.getQueryInfo().query->clone();
|
||||
if (!structure_argument_was_provided)
|
||||
addColumnsStructureToQueryWithClusterEngine(
|
||||
query_to_send, StorageDictionary::generateNamesAndTypesDescription(storage_snapshot->metadata->getColumns().getAll()), 5, getName());
|
||||
@ -136,7 +149,7 @@ Pipe StorageS3Cluster::read(
|
||||
shard_info.pool,
|
||||
std::vector<IConnectionPool::Entry>{try_result},
|
||||
queryToString(query_to_send),
|
||||
interpreter.getSampleBlock(),
|
||||
sample_block,
|
||||
context,
|
||||
/*throttler=*/nullptr,
|
||||
scalars,
|
||||
|
||||
@ -1,6 +1,7 @@
|
||||
#include <Interpreters/InterpreterSelectQuery.h>
|
||||
#include <Interpreters/InterpreterSelectWithUnionQuery.h>
|
||||
#include <Interpreters/InterpreterSelectQueryAnalyzer.h>
|
||||
#include <Interpreters/NormalizeSelectWithUnionQueryVisitor.h>
|
||||
#include <Interpreters/Context.h>
|
||||
#include <DataTypes/DataTypeLowCardinality.h>
|
||||
|
||||
@ -117,6 +118,10 @@ StorageView::StorageView(
|
||||
SelectQueryDescription description;
|
||||
|
||||
description.inner_query = query.select->ptr();
|
||||
|
||||
NormalizeSelectWithUnionQueryVisitor::Data data{SetOperationMode::Unspecified};
|
||||
NormalizeSelectWithUnionQueryVisitor{data}.visit(description.inner_query);
|
||||
|
||||
is_parameterized_view = query.isParameterizedView();
|
||||
parameter_types = analyzeReceiveQueryParamsWithType(description.inner_query);
|
||||
storage_metadata.setSelectQuery(description);
|
||||
|
||||
@ -78,6 +78,7 @@ namespace ErrorCodes
|
||||
extern const int SUPPORT_IS_DISABLED;
|
||||
extern const int TABLE_WAS_NOT_DROPPED;
|
||||
extern const int NOT_IMPLEMENTED;
|
||||
extern const int UNSUPPORTED_METHOD;
|
||||
}
|
||||
|
||||
namespace
|
||||
@ -1158,6 +1159,10 @@ StorageWindowView::StorageWindowView(
|
||||
, fire_signal_timeout_s(context_->getSettingsRef().wait_for_window_view_fire_signal_timeout.totalSeconds())
|
||||
, clean_interval_usec(context_->getSettingsRef().window_view_clean_interval.totalMicroseconds())
|
||||
{
|
||||
if (context_->getSettingsRef().allow_experimental_analyzer)
|
||||
throw Exception(ErrorCodes::UNSUPPORTED_METHOD,
|
||||
"Experimental WINDOW VIEW feature is not supported with new infrastructure for query analysis (the setting 'allow_experimental_analyzer')");
|
||||
|
||||
if (!query.select)
|
||||
throw Exception(ErrorCodes::INCORRECT_QUERY, "SELECT query is not specified for {}", getName());
|
||||
|
||||
|
||||
@ -53,7 +53,7 @@ std::vector<size_t> TableFunctionMerge::skipAnalysisForArguments(const QueryTree
|
||||
result.push_back(i);
|
||||
}
|
||||
|
||||
return {0};
|
||||
return result;
|
||||
}
|
||||
|
||||
void TableFunctionMerge::parseArguments(const ASTPtr & ast_function, ContextPtr context)
|
||||
|
||||
@ -9,6 +9,8 @@
|
||||
#include <Storages/StorageExternalDistributed.h>
|
||||
#include <Storages/NamedCollectionsHelpers.h>
|
||||
#include <TableFunctions/TableFunctionFactory.h>
|
||||
#include <Analyzer/FunctionNode.h>
|
||||
#include <Analyzer/TableFunctionNode.h>
|
||||
#include <Interpreters/parseColumnsListForTableFunction.h>
|
||||
#include <Interpreters/Context.h>
|
||||
#include <Formats/FormatFactory.h>
|
||||
@ -26,6 +28,24 @@ namespace ErrorCodes
|
||||
extern const int BAD_ARGUMENTS;
|
||||
}
|
||||
|
||||
std::vector<size_t> TableFunctionURL::skipAnalysisForArguments(const QueryTreeNodePtr & query_node_table_function, ContextPtr) const
|
||||
{
|
||||
auto & table_function_node = query_node_table_function->as<TableFunctionNode &>();
|
||||
auto & table_function_arguments_nodes = table_function_node.getArguments().getNodes();
|
||||
size_t table_function_arguments_size = table_function_arguments_nodes.size();
|
||||
|
||||
std::vector<size_t> result;
|
||||
|
||||
for (size_t i = 0; i < table_function_arguments_size; ++i)
|
||||
{
|
||||
auto * function_node = table_function_arguments_nodes[i]->as<FunctionNode>();
|
||||
if (function_node && function_node->getFunctionName() == "headers")
|
||||
result.push_back(i);
|
||||
}
|
||||
|
||||
return result;
|
||||
}
|
||||
|
||||
void TableFunctionURL::parseArguments(const ASTPtr & ast, ContextPtr context)
|
||||
{
|
||||
const auto & ast_function = assert_cast<const ASTFunction *>(ast.get());
|
||||
|
||||
@ -12,7 +12,7 @@ class Context;
|
||||
|
||||
/* url(source, format[, structure, compression]) - creates a temporary storage from url.
|
||||
*/
|
||||
class TableFunctionURL : public ITableFunctionFileLike
|
||||
class TableFunctionURL final: public ITableFunctionFileLike
|
||||
{
|
||||
public:
|
||||
static constexpr auto name = "url";
|
||||
@ -23,10 +23,11 @@ public:
|
||||
|
||||
ColumnsDescription getActualTableStructure(ContextPtr context) const override;
|
||||
|
||||
protected:
|
||||
private:
|
||||
std::vector<size_t> skipAnalysisForArguments(const QueryTreeNodePtr & query_node_table_function, ContextPtr context) const override;
|
||||
|
||||
void parseArguments(const ASTPtr & ast, ContextPtr context) override;
|
||||
|
||||
private:
|
||||
StoragePtr getStorage(
|
||||
const String & source, const String & format_, const ColumnsDescription & columns, ContextPtr global_context,
|
||||
const std::string & table_name, const String & compression_method_) const override;
|
||||
|
||||
@ -36,7 +36,8 @@ def get_options(i, upgrade_check):
|
||||
client_options.append("join_algorithm='partial_merge'")
|
||||
if join_alg_num % 5 == 2:
|
||||
client_options.append("join_algorithm='full_sorting_merge'")
|
||||
if join_alg_num % 5 == 3:
|
||||
if join_alg_num % 5 == 3 and not upgrade_check:
|
||||
# Some crashes are not fixed in 23.2 yet, so ignore the setting in Upgrade check
|
||||
client_options.append("join_algorithm='grace_hash'")
|
||||
if join_alg_num % 5 == 4:
|
||||
client_options.append("join_algorithm='auto'")
|
||||
@ -224,6 +225,20 @@ def prepare_for_hung_check(drop_databases):
|
||||
return True
|
||||
|
||||
|
||||
def is_ubsan_build():
|
||||
try:
|
||||
query = """clickhouse client -q "SELECT value FROM system.build_options WHERE name = 'CXX_FLAGS'" """
|
||||
output = (
|
||||
check_output(query, shell=True, stderr=STDOUT, timeout=30)
|
||||
.decode("utf-8")
|
||||
.strip()
|
||||
)
|
||||
return "-fsanitize=undefined" in output
|
||||
except Exception as e:
|
||||
logging.info("Failed to get build flags: %s", str(e))
|
||||
return False
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
logging.basicConfig(level=logging.INFO, format="%(asctime)s %(message)s")
|
||||
parser = argparse.ArgumentParser(
|
||||
@ -243,6 +258,10 @@ if __name__ == "__main__":
|
||||
args = parser.parse_args()
|
||||
if args.drop_databases and not args.hung_check:
|
||||
raise Exception("--drop-databases only used in hung check (--hung-check)")
|
||||
|
||||
# FIXME Hung check with ubsan is temporarily disabled due to https://github.com/ClickHouse/ClickHouse/issues/45372
|
||||
suppress_hung_check = is_ubsan_build()
|
||||
|
||||
func_pipes = []
|
||||
func_pipes = run_func_test(
|
||||
args.test_cmd,
|
||||
@ -307,7 +326,7 @@ if __name__ == "__main__":
|
||||
res = call(cmd, shell=True, stdout=tee.stdin, stderr=STDOUT)
|
||||
if tee.stdin is not None:
|
||||
tee.stdin.close()
|
||||
if res != 0 and have_long_running_queries:
|
||||
if res != 0 and have_long_running_queries and not suppress_hung_check:
|
||||
logging.info("Hung check failed with exit code %d", res)
|
||||
else:
|
||||
hung_check_status = "No queries hung\tOK\t\\N\t\n"
|
||||
|
||||
@ -1,40 +0,0 @@
|
||||
import pytest
|
||||
|
||||
from helpers.cluster import ClickHouseCluster
|
||||
|
||||
cluster = ClickHouseCluster(__file__)
|
||||
|
||||
node = cluster.add_instance("node")
|
||||
|
||||
|
||||
@pytest.fixture(scope="module", autouse=True)
|
||||
def start_cluster():
|
||||
try:
|
||||
cluster.start()
|
||||
yield cluster
|
||||
finally:
|
||||
cluster.shutdown()
|
||||
|
||||
|
||||
def test_memory_usage():
|
||||
node.query(
|
||||
"CREATE TABLE async_table(data Array(UInt64)) ENGINE=MergeTree() ORDER BY data"
|
||||
)
|
||||
|
||||
node.get_query_request("SELECT count() FROM system.numbers")
|
||||
|
||||
INSERT_QUERY = "INSERT INTO async_table SETTINGS async_insert=1, wait_for_async_insert=1 VALUES ({})"
|
||||
for iter in range(10):
|
||||
values = list(range(iter * 5000000, (iter + 1) * 5000000))
|
||||
node.query(INSERT_QUERY.format(values))
|
||||
|
||||
response = node.get_query_request(
|
||||
"SELECT groupArray(number) FROM numbers(1000000) SETTINGS max_memory_usage_for_user={}".format(
|
||||
30 * (2**23)
|
||||
)
|
||||
)
|
||||
|
||||
_, err = response.get_answer_and_error()
|
||||
assert err == "", "Query failed with error {}".format(err)
|
||||
|
||||
node.query("DROP TABLE async_table")
|
||||
@ -48,10 +48,10 @@
|
||||
{
|
||||
"i0": "0",
|
||||
"u0": "0",
|
||||
"ip": "9223372036854775807",
|
||||
"in": "-9223372036854775808",
|
||||
"up": "18446744073709551615",
|
||||
"arr": ["0"],
|
||||
"ip": "0",
|
||||
"in": "0",
|
||||
"up": "0",
|
||||
"arr": [],
|
||||
"tuple": ["0","0"]
|
||||
},
|
||||
|
||||
@ -119,7 +119,7 @@
|
||||
["0", "0", "9223372036854775807", "-9223372036854775808", "18446744073709551615", ["0"], ["0","0"]]
|
||||
],
|
||||
|
||||
"totals": ["0", "0", "9223372036854775807", "-9223372036854775808", "18446744073709551615", ["0"], ["0","0"]],
|
||||
"totals": ["0", "0", "0", "0", "0", [], ["0","0"]],
|
||||
|
||||
"extremes":
|
||||
{
|
||||
@ -180,10 +180,10 @@
|
||||
{
|
||||
"i0": 0,
|
||||
"u0": 0,
|
||||
"ip": 9223372036854775807,
|
||||
"in": -9223372036854775808,
|
||||
"up": 18446744073709551615,
|
||||
"arr": [0],
|
||||
"ip": 0,
|
||||
"in": 0,
|
||||
"up": 0,
|
||||
"arr": [],
|
||||
"tuple": [0,0]
|
||||
},
|
||||
|
||||
@ -251,7 +251,7 @@
|
||||
[0, 0, 9223372036854775807, -9223372036854775808, 18446744073709551615, [0], [0,0]]
|
||||
],
|
||||
|
||||
"totals": [0, 0, 9223372036854775807, -9223372036854775808, 18446744073709551615, [0], [0,0]],
|
||||
"totals": [0, 0, 0, 0, 0, [], [0,0]],
|
||||
|
||||
"extremes":
|
||||
{
|
||||
|
||||
@ -2,6 +2,7 @@
|
||||
|
||||
SET output_format_write_statistics = 0;
|
||||
SET extremes = 1;
|
||||
SET allow_experimental_analyzer = 1;
|
||||
|
||||
SET output_format_json_quote_64bit_integers = 1;
|
||||
SELECT toInt64(0) as i0, toUInt64(0) as u0, toInt64(9223372036854775807) as ip, toInt64(-9223372036854775808) as in, toUInt64(18446744073709551615) as up, [toInt64(0)] as arr, (toUInt64(0), toUInt64(0)) as tuple GROUP BY i0, u0, ip, in, up, arr, tuple WITH TOTALS FORMAT JSON;
|
||||
|
||||
@ -22,13 +22,13 @@
|
||||
13 13
|
||||
14 14
|
||||
\N 8
|
||||
0 0
|
||||
0 2
|
||||
0 4
|
||||
0 6
|
||||
0 8
|
||||
1 1
|
||||
3 3
|
||||
5 5
|
||||
7 7
|
||||
9 9
|
||||
\N 0
|
||||
\N 2
|
||||
\N 4
|
||||
\N 6
|
||||
\N 8
|
||||
|
||||
@ -1,3 +1,4 @@
|
||||
SET allow_experimental_analyzer = 1;
|
||||
SET join_use_nulls = 0;
|
||||
SET any_join_distinct_right_table_keys = 1;
|
||||
|
||||
|
||||
@ -16,24 +16,24 @@
|
||||
┌─x──────┬─name─┐
|
||||
│ system │ one │
|
||||
└────────┴──────┘
|
||||
┌─database─┬─t.name─┐
|
||||
┌─database─┬─name─┐
|
||||
│ system │ one │
|
||||
└──────────┴────────┘
|
||||
└──────────┴──────┘
|
||||
┌─db.x───┬─name─┐
|
||||
│ system │ one │
|
||||
└────────┴──────┘
|
||||
┌─db.name─┬─t.name─┐
|
||||
┌─db.name─┬─name─┐
|
||||
│ system │ one │
|
||||
└─────────┴────────┘
|
||||
┌─db.name─┬─t.name─┐
|
||||
└─────────┴──────┘
|
||||
┌─db.name─┬─name─┐
|
||||
│ system │ one │
|
||||
└─────────┴────────┘
|
||||
┌─t.database─┬─t.name─┐
|
||||
└─────────┴──────┘
|
||||
┌─database─┬─name─┐
|
||||
│ system │ one │
|
||||
└────────────┴────────┘
|
||||
┌─database─┬─t.name─┐
|
||||
└──────────┴──────┘
|
||||
┌─database─┬─name─┐
|
||||
│ system │ one │
|
||||
└──────────┴────────┘
|
||||
└──────────┴──────┘
|
||||
2
|
||||
2
|
||||
2
|
||||
|
||||
@ -1,3 +1,5 @@
|
||||
-- Tags: no-parallel
|
||||
|
||||
SET allow_experimental_analyzer = 1;
|
||||
|
||||
DROP TABLE IF EXISTS one;
|
||||
|
||||
@ -10,13 +10,13 @@ l \N \N String Nullable(String)
|
||||
\N \N
|
||||
\N \N
|
||||
using
|
||||
l \N String Nullable(String)
|
||||
\N String Nullable(String)
|
||||
l \N String Nullable(String)
|
||||
l \N Nullable(String) Nullable(String)
|
||||
l \N Nullable(String) Nullable(String)
|
||||
\N \N Nullable(String) Nullable(String)
|
||||
\N \N Nullable(String) Nullable(String)
|
||||
l \N Nullable(String) Nullable(String)
|
||||
l \N Nullable(String) Nullable(String)
|
||||
\N \N Nullable(String) Nullable(String)
|
||||
l \N String Nullable(String)
|
||||
\N String Nullable(String)
|
||||
l \N String Nullable(String)
|
||||
\N \N Nullable(String) Nullable(String)
|
||||
\N \N
|
||||
\N \N
|
||||
@ -32,13 +32,13 @@ l \N \N Nullable(String) Nullable(String)
|
||||
\N \N
|
||||
\N \N
|
||||
using + join_use_nulls
|
||||
l \N String Nullable(String)
|
||||
l \N Nullable(String) Nullable(String)
|
||||
\N \N Nullable(String) Nullable(String)
|
||||
\N \N Nullable(String) Nullable(String)
|
||||
l \N String Nullable(String)
|
||||
l \N Nullable(String) Nullable(String)
|
||||
\N \N Nullable(String) Nullable(String)
|
||||
\N \N Nullable(String) Nullable(String)
|
||||
r \N Nullable(String) Nullable(String)
|
||||
r \N Nullable(String) Nullable(String)
|
||||
l \N Nullable(String) Nullable(String)
|
||||
l \N Nullable(String) Nullable(String)
|
||||
r \N Nullable(String) Nullable(String)
|
||||
r \N Nullable(String) Nullable(String)
|
||||
\N \N
|
||||
\N \N
|
||||
|
||||
@ -1,4 +1,5 @@
|
||||
SET any_join_distinct_right_table_keys = 1;
|
||||
SET allow_experimental_analyzer = 1;
|
||||
|
||||
DROP TABLE IF EXISTS t1_00848;
|
||||
DROP TABLE IF EXISTS t2_00848;
|
||||
@ -53,16 +54,16 @@ SELECT t3.id = 'l', t3.not_id = 'l' FROM t1_00848 t1 LEFT JOIN t3_00848 t3 ON t1
|
||||
|
||||
SELECT 'using + join_use_nulls';
|
||||
|
||||
SELECT *, toTypeName(t1.id), toTypeName(t3.id) FROM t1_00848 t1 ANY LEFT JOIN t3_00848 t3 USING(id) ORDER BY t1.id, t3.id;
|
||||
SELECT *, toTypeName(t1.id), toTypeName(t3.id) FROM t1_00848 t1 ANY FULL JOIN t3_00848 t3 USING(id) ORDER BY t1.id, t3.id;
|
||||
SELECT *, toTypeName(t2.id), toTypeName(t3.id) FROM t2_00848 t2 ANY FULL JOIN t3_00848 t3 USING(id) ORDER BY t2.id, t3.id;
|
||||
SELECT *, toTypeName(t1.id), toTypeName(t3.id) FROM t1_00848 t1 ANY LEFT JOIN t3_00848 t3 USING(id) ORDER BY id;
|
||||
SELECT *, toTypeName(t1.id), toTypeName(t3.id) FROM t1_00848 t1 ANY FULL JOIN t3_00848 t3 USING(id) ORDER BY id;
|
||||
SELECT *, toTypeName(t2.id), toTypeName(t3.id) FROM t2_00848 t2 ANY FULL JOIN t3_00848 t3 USING(id) ORDER BY id;
|
||||
|
||||
SELECT *, toTypeName(t1.id), toTypeName(t3.id) FROM t1_00848 t1 LEFT JOIN t3_00848 t3 USING(id) ORDER BY t1.id, t3.id;
|
||||
SELECT *, toTypeName(t1.id), toTypeName(t3.id) FROM t1_00848 t1 FULL JOIN t3_00848 t3 USING(id) ORDER BY t1.id, t3.id;
|
||||
SELECT *, toTypeName(t2.id), toTypeName(t3.id) FROM t2_00848 t2 FULL JOIN t3_00848 t3 USING(id) ORDER BY t2.id, t3.id;
|
||||
SELECT *, toTypeName(t1.id), toTypeName(t3.id) FROM t1_00848 t1 LEFT JOIN t3_00848 t3 USING(id) ORDER BY id;
|
||||
SELECT *, toTypeName(t1.id), toTypeName(t3.id) FROM t1_00848 t1 FULL JOIN t3_00848 t3 USING(id) ORDER BY id;
|
||||
SELECT *, toTypeName(t2.id), toTypeName(t3.id) FROM t2_00848 t2 FULL JOIN t3_00848 t3 USING(id) ORDER BY id;
|
||||
|
||||
SELECT t3.id = 'l', t3.not_id = 'l' FROM t1_00848 t1 ANY LEFT JOIN t3_00848 t3 USING(id) ORDER BY t1.id, t3.id;
|
||||
SELECT t3.id = 'l', t3.not_id = 'l' FROM t1_00848 t1 LEFT JOIN t3_00848 t3 USING(id) ORDER BY t1.id, t3.id;
|
||||
SELECT t3.id = 'l', t3.not_id = 'l' FROM t1_00848 t1 ANY LEFT JOIN t3_00848 t3 USING(id) ORDER BY id;
|
||||
SELECT t3.id = 'l', t3.not_id = 'l' FROM t1_00848 t1 LEFT JOIN t3_00848 t3 USING(id) ORDER BY id;
|
||||
|
||||
DROP TABLE t1_00848;
|
||||
DROP TABLE t2_00848;
|
||||
|
||||
@ -15,5 +15,5 @@ bar bar 1 2 String Nullable(String)
|
||||
\N 0 1 String Nullable(String)
|
||||
foo 2 0 String
|
||||
bar 1 2 String
|
||||
test 0 1 String
|
||||
0 1 String
|
||||
0 1 String
|
||||
|
||||
@ -27,7 +27,7 @@ SELECT s1.other, s2.other, count_a, count_b, toTypeName(s1.other), toTypeName(s2
|
||||
( SELECT other, count() AS count_a FROM table_a GROUP BY other ) s1
|
||||
ALL FULL JOIN
|
||||
( SELECT other, count() AS count_b FROM table_b GROUP BY other ) s2
|
||||
USING other
|
||||
ON s1.other = s2.other
|
||||
ORDER BY s2.other DESC, count_a, s1.other;
|
||||
|
||||
SELECT s1.something, s2.something, count_a, count_b, toTypeName(s1.something), toTypeName(s2.something) FROM
|
||||
@ -41,7 +41,7 @@ SELECT s1.something, s2.something, count_a, count_b, toTypeName(s1.something), t
|
||||
( SELECT something, count() AS count_a FROM table_a GROUP BY something ) s1
|
||||
ALL RIGHT JOIN
|
||||
( SELECT something, count() AS count_b FROM table_b GROUP BY something ) s2
|
||||
USING (something)
|
||||
ON s1.something = s2.something
|
||||
ORDER BY count_a DESC, s1.something, s2.something;
|
||||
|
||||
SET joined_subquery_requires_alias = 0;
|
||||
@ -50,7 +50,7 @@ SELECT something, count_a, count_b, toTypeName(something) FROM
|
||||
( SELECT something, count() AS count_a FROM table_a GROUP BY something ) as s1
|
||||
ALL FULL JOIN
|
||||
( SELECT something, count() AS count_b FROM table_b GROUP BY something ) as s2
|
||||
USING (something)
|
||||
ON s1.something = s2.something
|
||||
ORDER BY count_a DESC, something DESC;
|
||||
|
||||
DROP TABLE table_a;
|
||||
|
||||
@ -1 +1,3 @@
|
||||
1
|
||||
1
|
||||
1
|
||||
|
||||
@ -1,3 +1,4 @@
|
||||
set allow_experimental_analyzer = 1;
|
||||
set distributed_product_mode = 'local';
|
||||
|
||||
drop table if exists shard1;
|
||||
@ -21,7 +22,7 @@ where distr1.id in
|
||||
from distr1
|
||||
join distr2 on distr1.id = distr2.id
|
||||
where distr1.id > 0
|
||||
); -- { serverError 288 }
|
||||
);
|
||||
|
||||
select distinct(d0.id) from distr1 d0
|
||||
where d0.id in
|
||||
@ -32,15 +33,14 @@ where d0.id in
|
||||
where d1.id > 0
|
||||
);
|
||||
|
||||
-- TODO
|
||||
--select distinct(distr1.id) from distr1
|
||||
--where distr1.id in
|
||||
--(
|
||||
-- select distr1.id
|
||||
-- from distr1 as d1
|
||||
-- join distr2 as d2 on distr1.id = distr2.id
|
||||
-- where distr1.id > 0
|
||||
--);
|
||||
select distinct(distr1.id) from distr1
|
||||
where distr1.id in
|
||||
(
|
||||
select distr1.id
|
||||
from distr1 as d1
|
||||
join distr2 as d2 on distr1.id = distr2.id
|
||||
where distr1.id > 0
|
||||
);
|
||||
|
||||
drop table shard1;
|
||||
drop table shard2;
|
||||
|
||||
@ -1 +1,3 @@
|
||||
99999
|
||||
99999
|
||||
0 0 13
|
||||
|
||||
@ -13,15 +13,24 @@ $CLICKHOUSE_CLIENT --query="CREATE TABLE small_table (a UInt64 default 0, n UInt
|
||||
$CLICKHOUSE_CLIENT --query="INSERT INTO small_table (n) SELECT * from system.numbers limit 100000;"
|
||||
$CLICKHOUSE_CLIENT --query="OPTIMIZE TABLE small_table FINAL;"
|
||||
|
||||
cached_query="SELECT count() FROM small_table where n > 0;"
|
||||
cached_query="SELECT count() FROM small_table WHERE n > 0;"
|
||||
|
||||
$CLICKHOUSE_CLIENT --use_uncompressed_cache=1 --query="$cached_query" &> /dev/null
|
||||
|
||||
$CLICKHOUSE_CLIENT --use_uncompressed_cache=1 --allow_prefetched_read_pool_for_remote_filesystem=0 --allow_prefetched_read_pool_for_local_filesystem=0 --query_id="test-query-uncompressed-cache" --query="$cached_query" &> /dev/null
|
||||
$CLICKHOUSE_CLIENT --log_queries 1 --use_uncompressed_cache 1 --query="$cached_query"
|
||||
$CLICKHOUSE_CLIENT --log_queries 1 --use_uncompressed_cache 1 --allow_prefetched_read_pool_for_remote_filesystem 0 --allow_prefetched_read_pool_for_local_filesystem 0 --query_id="test-query-uncompressed-cache" --query="$cached_query"
|
||||
|
||||
$CLICKHOUSE_CLIENT --query="SYSTEM FLUSH LOGS"
|
||||
|
||||
|
||||
$CLICKHOUSE_CLIENT --query="SELECT ProfileEvents['Seek'], ProfileEvents['ReadCompressedBytes'], ProfileEvents['UncompressedCacheHits'] AS hit FROM system.query_log WHERE (query_id = 'test-query-uncompressed-cache') and current_database = currentDatabase() AND (type = 2) AND event_date >= yesterday() ORDER BY event_time DESC LIMIT 1"
|
||||
$CLICKHOUSE_CLIENT --query="
|
||||
SELECT
|
||||
ProfileEvents['Seek'],
|
||||
ProfileEvents['ReadCompressedBytes'],
|
||||
ProfileEvents['UncompressedCacheHits'] AS hit
|
||||
FROM system.query_log
|
||||
WHERE query_id = 'test-query-uncompressed-cache'
|
||||
AND current_database = currentDatabase()
|
||||
AND type = 2
|
||||
AND event_date >= yesterday()
|
||||
ORDER BY event_time DESC
|
||||
LIMIT 1"
|
||||
|
||||
$CLICKHOUSE_CLIENT --query="DROP TABLE IF EXISTS small_table"
|
||||
|
||||
@ -1 +1,3 @@
|
||||
SELECT 1 AS a, a + a AS b, b + b AS c, c + c AS d, d + d AS e, e + e AS f, f + f AS g, g + g AS h, h + h AS i, i + i AS j, j + j AS k, k + k AS l, l + l AS m, m + m AS n, n + n AS o, o + o AS p, p + p AS q, q + q AS r, r + r AS s, s + s AS t, t + t AS u, u + u AS v, v + v AS w, w + w AS x, x + x AS y, y + y AS z; -- { serverError 168 }
|
||||
SET allow_experimental_analyzer = 1;
|
||||
|
||||
SELECT 1 AS a, a + a AS b, b + b AS c, c + c AS d, d + d AS e, e + e AS f, f + f AS g, g + g AS h, h + h AS i, i + i AS j, j + j AS k, k + k AS l, l + l AS m, m + m AS n, n + n AS o, o + o AS p, p + p AS q, q + q AS r, r + r AS s, s + s AS t, t + t AS u, u + u AS v, v + v AS w, w + w AS x, x + x AS y, y + y AS z; -- { serverError 36 }
|
||||
|
||||
@ -3,7 +3,10 @@
|
||||
SET max_memory_usage = 32000000;
|
||||
SET join_on_disk_max_files_to_merge = 4;
|
||||
|
||||
SELECT number * 200000 as n, j FROM numbers(5) nums
|
||||
SELECT n, j FROM
|
||||
(
|
||||
SELECT number * 200000 as n FROM numbers(5)
|
||||
) nums
|
||||
ANY LEFT JOIN (
|
||||
SELECT number * 2 AS n, number AS j
|
||||
FROM numbers(1000000)
|
||||
@ -13,14 +16,20 @@ USING n; -- { serverError 241 }
|
||||
SET join_algorithm = 'partial_merge';
|
||||
SET default_max_bytes_in_join = 0;
|
||||
|
||||
SELECT number * 200000 as n, j FROM numbers(5) nums
|
||||
SELECT n, j FROM
|
||||
(
|
||||
SELECT number * 200000 as n FROM numbers(5)
|
||||
) nums
|
||||
ANY LEFT JOIN (
|
||||
SELECT number * 2 AS n, number AS j
|
||||
FROM numbers(1000000)
|
||||
) js2
|
||||
USING n; -- { serverError 12 }
|
||||
|
||||
SELECT number * 200000 as n, j FROM numbers(5) nums
|
||||
SELECT n, j FROM
|
||||
(
|
||||
SELECT number * 200000 as n FROM numbers(5)
|
||||
) nums
|
||||
ANY LEFT JOIN (
|
||||
SELECT number * 2 AS n, number AS j
|
||||
FROM numbers(1000000)
|
||||
@ -28,7 +37,10 @@ ANY LEFT JOIN (
|
||||
USING n
|
||||
SETTINGS max_bytes_in_join = 30000000; -- { serverError 241 }
|
||||
|
||||
SELECT number * 200000 as n, j FROM numbers(5) nums
|
||||
SELECT n, j FROM
|
||||
(
|
||||
SELECT number * 200000 as n FROM numbers(5)
|
||||
) nums
|
||||
ANY LEFT JOIN (
|
||||
SELECT number * 2 AS n, number AS j
|
||||
FROM numbers(1000000)
|
||||
@ -39,7 +51,10 @@ SETTINGS max_bytes_in_join = 10000000;
|
||||
|
||||
SET partial_merge_join_optimizations = 1;
|
||||
|
||||
SELECT number * 200000 as n, j FROM numbers(5) nums
|
||||
SELECT n, j FROM
|
||||
(
|
||||
SELECT number * 200000 as n FROM numbers(5)
|
||||
) nums
|
||||
LEFT JOIN (
|
||||
SELECT number * 2 AS n, number AS j
|
||||
FROM numbers(1000000)
|
||||
@ -50,7 +65,10 @@ SETTINGS max_rows_in_join = 100000;
|
||||
|
||||
SET default_max_bytes_in_join = 10000000;
|
||||
|
||||
SELECT number * 200000 as n, j FROM numbers(5) nums
|
||||
SELECT n, j FROM
|
||||
(
|
||||
SELECT number * 200000 as n FROM numbers(5)
|
||||
) nums
|
||||
JOIN (
|
||||
SELECT number * 2 AS n, number AS j
|
||||
FROM numbers(1000000)
|
||||
|
||||
@ -1,12 +1,15 @@
|
||||
0 0
|
||||
0 0
|
||||
0 0
|
||||
0
|
||||
0
|
||||
0
|
||||
0
|
||||
┌─[1mone.dummy[0m─┬─[1mA.dummy[0m─┬─[1mB.dummy[0m─┐
|
||||
0
|
||||
0
|
||||
┌─[1msystem.one.dummy[0m─┬─[1mA.dummy[0m─┬─[1mB.dummy[0m─┐
|
||||
│ 0 │ 0 │ 0 │
|
||||
└───────────┴─────────┴─────────┘
|
||||
└──────────────────┴─────────┴─────────┘
|
||||
┌─[1mA.dummy[0m─┬─[1mone.dummy[0m─┬─[1mtwo.dummy[0m─┐
|
||||
│ 0 │ 0 │ 0 │
|
||||
└─────────┴───────────┴───────────┘
|
||||
|
||||
Some files were not shown because too many files have changed in this diff Show More
Loading…
Reference in New Issue
Block a user