window view docs improvement

docs/en/sql-reference/functions/window-view-functions.md

@@ -5,11 +5,11 @@ toc_title: Window View
 
 # Window View Functions {#window-view-functions}
 
-Window functions indicate the lower and upper window bound of records in WindowView. The functions for working with WindowView are listed below.
+Window view functions return the inclusive lower and exclusive upper bound of the corresponding window. The functions for working with WindowView are listed below:
 
 ## tumble {#window-view-functions-tumble}
 
-A tumbling time window assigns records to non-overlapping, continuous windows with a fixed duration (interval). 
+A tumbling time window assigns records to non-overlapping, continuous windows with a fixed duration (`interval`). 
 
 ``` sql
 tumble(time_attr, interval [, timezone])
@@ -22,7 +22,7 @@ tumble(time_attr, interval [, timezone])
 
 **Returned values**
 
--  The lower and upper bound of the tumble window.
+- The inclusive lower and exclusive upper bound of the corresponding tumbling window.
 
 Type: `Tuple(DateTime, DateTime)`
 
@@ -59,9 +59,7 @@ hop(time_attr, hop_interval, window_interval [, timezone])
 
 **Returned values**
 
--  The lower and upper bound of the hop window. Since hop windows are
+- The inclusive lower and exclusive upper bound of the corresponding hopping window. Since one record can be assigned to multiple hop windows, the function only returns the bound of the **first** window when hop function is used **without** `WINDOW VIEW`.
-   overlapped, the function only returns the bound of the **first** window when
-   hop function is used **without** `WINDOW VIEW`.
 
 Type: `Tuple(DateTime, DateTime)`
 
@@ -83,7 +81,7 @@ Result:
 
 ## tumbleStart {#window-view-functions-tumblestart}
 
-Indicate the lower bound of a tumble function.
+Returns the inclusive lower bound of the corresponding tumbling window.
 
 ``` sql
 tumbleStart(time_attr, interval [, timezone]);
@@ -91,7 +89,7 @@ tumbleStart(time_attr, interval [, timezone]);
 
 ## tumbleEnd {#window-view-functions-tumbleend}
 
-Indicate the upper bound of a tumble function.
+Returns the exclusive upper bound of the corresponding tumbling window.
 
 ``` sql
 tumbleEnd(time_attr, interval [, timezone]);
@@ -99,7 +97,7 @@ tumbleEnd(time_attr, interval [, timezone]);
 
 ## hopStart {#window-view-functions-hopstart}
 
-Indicate the lower bound of a hop function.
+Returns the inclusive lower bound of the corresponding hopping window.
 
 ``` sql
 hopStart(time_attr, hop_interval, window_interval [, timezone]);
@@ -107,7 +105,7 @@ hopStart(time_attr, hop_interval, window_interval [, timezone]);
 
 ## hopEnd {#window-view-functions-hopend}
 
-Indicate the upper bound of a hop function.
+Returns the exclusive upper bound of the corresponding hopping window.
 
 ``` sql
 hopEnd(time_attr, hop_interval, window_interval [, timezone]);

docs/en/sql-reference/statements/create/view.md

@@ -254,13 +254,13 @@ Most common uses of live view tables include:
 CREATE WINDOW VIEW [IF NOT EXISTS] [db.]table_name [TO [db.]table_name] [ENGINE = engine] [WATERMARK = strategy] [ALLOWED_LATENESS = interval_function] AS SELECT ... GROUP BY window_view_function
 ```
 
-Window view can aggregate data by time window and output the results when the window is ready to fire. It stores the partial aggregation results in an inner(or specified) table and can push the processing result to a specified table or push notifications using the WATCH query.
+Window view can aggregate data by time window and output the results when the window is ready to fire. It stores the partial aggregation results in an inner(or specified) table to reduce latency and can push the processing result to a specified table or push notifications using the WATCH query.
 
 Creating a window view is similar to creating `MATERIALIZED VIEW`. Window view needs an inner storage engine to store intermediate data. The inner storage will use `AggregatingMergeTree` as the default engine.
 
 ### Window View Functions {#window-view-windowviewfunctions}
 
-[Window view functions](../../functions/window-view-functions.md) are used to indicate the lower and upper window bound of records. The window view needs to be used with a window view function.
+[Window view functions](../../functions/window-view-functions.md) are used to get the lower and upper window bound of records. The window view needs to be used with a window view function.
 
 ### TIME ATTRIBUTES {#window-view-timeattributes}
 
@@ -274,13 +274,13 @@ CREATE WINDOW VIEW wv AS SELECT count(number), tumbleStart(w_id) as w_start from
 
 **Event time** is the time that each individual event occurred on its producing device. This time is typically embedded within the records when it is generated. Event time processing allows for consistent results even in case of out-of-order events or late events. Window view supports event time processing by using `WATERMARK` syntax.
 
-Window view provides three watermark strategies.
+Window view provides three watermark strategies:
 
 * `STRICTLY_ASCENDING`: Emits a watermark of the maximum observed timestamp so far. Rows that have a timestamp smaller to the max timestamp are not late.
 * `ASCENDING`: Emits a watermark of the maximum observed timestamp so far minus 1. Rows that have a timestamp equal and smaller to the max timestamp are not late.
 * `BOUNDED`: WATERMARK=INTERVAL. Emits watermarks, which are the maximum observed timestamp minus the specified delay.
 
-The following queries are examples of creating a window view with `WATERMARK`.
+The following queries are examples of creating a window view with `WATERMARK`:
 
 ``` sql
 CREATE WINDOW VIEW wv WATERMARK=STRICTLY_ASCENDING AS SELECT count(number) FROM date GROUP BY tumble(timestamp, INTERVAL '5' SECOND);

docs/zh/sql-reference/functions/window-view-functions.md

@@ -0,0 +1,112 @@
+---
+toc_priority: 68
+toc_title: Window View
+---
+
+# Window View 函数{#window-view-han-shu}
+
+Window view函数用于获取窗口的起始(包含边界)和结束时间(不包含边界)。系统支持的window view函数如下：
+
+## tumble {#window-view-functions-tumble}
+
+tumble窗口是连续的、不重叠的固定大小(`interval`)时间窗口。
+
+``` sql
+tumble(time_attr, interval [, timezone])
+```
+
+**参数**
+- `time_attr` - [DateTime](../../sql-reference/data-types/datetime.md)类型的时间数据。
+- `interval` - [Interval](../../sql-reference/data-types/special-data-types/interval.md)类型的窗口大小。
+- `timezone` — [Timezone name](../../operations/server-configuration-parameters/settings.md#server_configuration_parameters-timezone) 类型的时区(可选参数). 
+
+**返回值**
+
+-  tumble窗口的开始(包含边界)和结束时间(不包含边界)
+
+类型: `Tuple(DateTime, DateTime)`
+
+**示例**
+
+查询：
+
+``` sql
+SELECT tumble(now(), toIntervalDay('1'))
+```
+
+结果：
+
+``` text
+┌─tumble(now(), toIntervalDay('1'))─────────────┐
+│ ['2020-01-01 00:00:00','2020-01-02 00:00:00'] │
+└───────────────────────────────────────────────┘
+```
+
+## hop {#window-view-functions-hop}
+
+hop窗口是一个固定大小(`window_interval`)的时间窗口，并按照一个固定的滑动间隔(`hop_interval`)滑动。当滑动间隔小于窗口大小时，滑动窗口间存在重叠，此时一个数据可能存在于多个窗口。
+
+``` sql
+hop(time_attr, hop_interval, window_interval [, timezone])
+```
+
+**参数**
+
+- `time_attr` - [DateTime](../../sql-reference/data-types/datetime.md)类型的时间数据。
+- `hop_interval` - Hop interval in [Interval](../../sql-reference/data-types/special-data-types/interval.md)类型的滑动间隔，需要大于0。
+- `window_interval` - [Interval](../../sql-reference/data-types/special-data-types/interval.md)类型的窗口大小，需要大于0。
+- `timezone` — [Timezone name](../../operations/server-configuration-parameters/settings.md#server_configuration_parameters-timezone) 类型的时区(可选参数)。
+
+**返回值**
+
+- hop窗口的开始(包含边界)和结束时间(不包含边界)。由于一个数据可能存在于多个窗口，脱离window view单独调用该函数时只返回第一个窗口数据。
+
+类型: `Tuple(DateTime, DateTime)`
+
+**示例**
+
+查询：
+
+``` sql
+SELECT hop(now(), INTERVAL '1' SECOND, INTERVAL '2' SECOND)
+```
+
+结果：
+
+``` text
+┌─hop(now(), toIntervalSecond('1'), toIntervalSecond('2'))──┐
+│ ('2020-01-14 16:58:22','2020-01-14 16:58:24')             │
+└───────────────────────────────────────────────────────────┘
+```
+
+## tumbleStart {#window-view-functions-tumblestart}
+
+返回tumble窗口的开始时间(包含边界)。
+
+``` sql
+tumbleStart(time_attr, interval [, timezone]);
+```
+
+## tumbleEnd {#window-view-functions-tumbleend}
+
+返回tumble窗口的结束时间(不包含边界)。
+
+``` sql
+tumbleEnd(time_attr, interval [, timezone]);
+```
+
+## hopStart {#window-view-functions-hopstart}
+
+返回hop窗口的开始时间(包含边界)。
+
+``` sql
+hopStart(time_attr, hop_interval, window_interval [, timezone]);
+```
+
+## hopEnd {#window-view-functions-hopend}
+
+返回hop窗口的结束时间(不包含边界)。
+
+``` sql
+hopEnd(time_attr, hop_interval, window_interval [, timezone]);
+```

docs/zh/sql-reference/statements/create/view.md

@@ -5,7 +5,7 @@ toc_title: VIEW
 
 # CREATE VIEW {#create-view}
 
-创建一个新视图。 有两种类型的视图：普通视图和物化视图。
+创建一个新视图。 有两种类型的视图：普通视图，物化视图，Live视图和Window视图。
 
 ## Normal {#normal}
 
@@ -241,3 +241,120 @@ Code: 60. DB::Exception: Received from localhost:9000. DB::Exception: Table defa
 - 使用定期刷新从系统表中查看指标。
 
 [原始文章](https://clickhouse.com/docs/en/sql-reference/statements/create/view/) <!--hide-->
+
+## Window View [Experimental] {#window-view}
+
+!!! important "重要"
+    这是一项试验性功能，可能会在未来版本中以向后不兼容的方式进行更改。
+    通过[allow_experimental_window_view](../../../operations/settings/settings.md#allow-experimental-window-view)启用window view以及`WATCH`语句。输入命令
+    `set allow_experimental_window_view = 1`。
+
+``` sql
+CREATE WINDOW VIEW [IF NOT EXISTS] [db.]table_name [TO [db.]table_name] [ENGINE = engine] [WATERMARK = strategy] [ALLOWED_LATENESS = interval_function] AS SELECT ... GROUP BY window_view_function
+```
+
+Window view可以通过时间窗口聚合数据，并在满足窗口触发条件时自动触发对应窗口计算。其通过将计算状态保存降低处理延迟，支持将处理结果输出至目标表或通过`WATCH`语句输出至终端。
+
+创建window view的方式和创建物化视图类似。Window view使用默认为`AggregatingMergeTree`的内部存储引擎存储计算中间状态。
+
+### Window View 函数{#window-view-han-shu}
+
+[Window view函数](../../functions/window-view-functions.md)用于获取窗口的起始和结束时间。Window view需要和window view函数配合使用。
+
+### 时间属性{#window-view-shi-jian-shu-xing}
+
+Window view 支持**处理时间**和**事件时间**两种时间类型。
+
+**处理时间**为默认时间类型，该模式下window view使用本地机器时间计算窗口数据。“处理时间”时间类型计算简单，但具有不确定性。该模式下时间可以为window view函数的第一个参数`time_attr`，或通过函数`now()`使用当前机器时间。下面的例子展示了使用“处理时间”创建的window view的例子。
+
+``` sql
+CREATE WINDOW VIEW wv AS SELECT count(number), tumbleStart(w_id) as w_start from date GROUP BY tumble(now(), INTERVAL '5' SECOND) as w_id
+```
+
+**事件时间** 是事件真实发生的时间，该时间往往在事件发生时便嵌入数据记录。事件时间处理提供较高的确定性，可以处理乱序数据以及迟到数据。Window view 通过水位线(`WATERMARK`)启用事件时间处理。
+
+Window view提供如下三种水位线策略：
+
+* `STRICTLY_ASCENDING`: 提交观测到的最大时间作为水位线，小于最大观测时间的数据不算迟到。
+* `ASCENDING`: 提交观测到的最大时间减1作为水位线。小于或等于最大观测时间的数据不算迟到。
+* `BOUNDED`: WATERMARK=INTERVAL. 提交最大观测时间减去固定间隔(`INTERVAL`)做为水位线。
+
+以下为使用`WATERMARK`创建window view的示例：
+
+``` sql
+CREATE WINDOW VIEW wv WATERMARK=STRICTLY_ASCENDING AS SELECT count(number) FROM date GROUP BY tumble(timestamp, INTERVAL '5' SECOND);
+CREATE WINDOW VIEW wv WATERMARK=ASCENDING AS SELECT count(number) FROM date GROUP BY tumble(timestamp, INTERVAL '5' SECOND);
+CREATE WINDOW VIEW wv WATERMARK=INTERVAL '3' SECOND AS SELECT count(number) FROM date GROUP BY tumble(timestamp, INTERVAL '5' SECOND);
+```
+
+通常，窗口会在水位线到达时触发，水位线到达之后的数据会被丢弃。Window view可以通过设置`ALLOWED_LATENESS=INTERVAL`来开启迟到消息处理。示例如下：
+
+``` sql
+CREATE WINDOW VIEW test.wv TO test.dst WATERMARK=ASCENDING ALLOWED_LATENESS=INTERVAL '2' SECOND AS SELECT count(a) AS count, tumbleEnd(wid) AS w_end FROM test.mt GROUP BY tumble(timestamp, INTERVAL '5' SECOND) AS wid;
+```
+
+需要注意的是，迟到消息需要更新之前的处理结果。与在窗口结束时触发不同，迟到消息到达时window view会立即触发计算。因此，会导致同一个窗口输出多次计算结果。用户需要注意这种情况，并消除重复结果。
+
+### 新窗口监控{#window-view-xin-chuang-kou-jian-kong}
+
+Window view可以通过`WATCH`语句将处理结果推送至终端，或通过`TO`语句将结果推送至数据表。
+
+``` sql
+WATCH [db.]name [LIMIT n]
+```
+
+`WATCH`语句和`LIVE VIEW`中的类似。支持设置`LIMIT`参数，输出消息数目达到`LIMIT`限制时结束查询。
+
+### 设置{#window-view-she-zhi}
+
+- `window_view_clean_interval`: window view清除过期数据间隔(单位为秒)。系统会定期清除过期数据，尚未触发的窗口数据不会被清除。
+- `window_view_heartbeat_interval`: 用于判断watch查询活跃的心跳时间间隔。
+
+### 示例{#window-view-shi-li}
+
+假设我们需要每10秒统计一次`data`表中的点击日志，且`data`表的结构如下：
+
+``` sql
+CREATE TABLE data ( `id` UInt64, `timestamp` DateTime) ENGINE = Memory;
+```
+
+首先，使用10秒大小的tumble函数创建window view。
+
+``` sql
+CREATE WINDOW VIEW wv as select count(id), tumbleStart(w_id) as window_start from data group by tumble(timestamp, INTERVAL '10' SECOND) as w_id
+```
+
+随后，我们使用`WATCH`语句获取计算结果。
+
+``` sql
+WATCH wv
+```
+
+当日志插入表`data`时，
+
+``` sql
+INSERT INTO data VALUES(1,now())
+```
+
+`WATCH`语句会输出如下结果：
+
+``` text
+┌─count(id)─┬────────window_start─┐
+│         1 │ 2020-01-14 16:56:40 │
+└───────────┴─────────────────────┘
+```
+
+或者，我们可以通过`TO`关键字将处理结果输出至另一张表。
+
+``` sql
+CREATE WINDOW VIEW wv TO dst AS SELECT count(id), tumbleStart(w_id) as window_start FROM data GROUP BY tumble(timestamp, INTERVAL '10' SECOND) as w_id
+```
+
+ClickHouse测试中提供了更多的示例(以`*window_view*`命名)。
+
+### Window View 使用场景{#window-view-shi-yong-chang-jing}
+
+Window view 在以下场景有用：
+
+* **监控**: 以时间维度聚合及处理数据，并将处理结果输出至目标表。用户可通过目标表获取并操作计算结果。
+* **分析**: 以时间维度进行数据分析. 当数据源非常庞大时，window view可以减少重复全表查询的计算量。