在 SQL 中获取无循环的连续行值
Get successive row values without loop in SQL
我有以下 table:
AppId
Id
Direction
Text
Date
aaa
11
in
hello
11/2/2021 3:03:00 PM
aaa
22
out
yes?
11/2/2021 3:04:00 PM
aaa
33
in
need help!
11/3/2021 3:06:00 PM
aaa
44
in
you there?
11/4/2021 3:10:00 PM
aaa
55
out
yes!
11/5/2021 4:00:00 PM
bb
111
out
welcome!
11/6/2021 6:09:00 PM
bb
222
in
can i call?
11/6/2021 6:39:00 PM
bb
333
out
sure.
11/6/2021 8:22:00 PM
cc
1111
out
hello?
11/8/2021 2:22:00 PM
cc
2222
in
Whatsup!
11/8/2021 3:22:00 PM
Id 是一个主键，方向基本上告诉我们它是传入消息还是传出消息。 App id 用于识别属于单个对话的各种 id。我想将传入消息后的第一个响应时间标识为：
AppId
Id
Direction
Text
Date
ReplyDate
aaa
11
in
hello
11/2/2021 3:03:00 PM
11/2/2021 3:04:00 PM
aaa
22
out
yes?
11/2/2021 3:04:00 PM
null
aaa
33
in
need help!
11/3/2021 3:06:00 PM
11/5/2021 4:00:00 PM
aaa
44
in
you there?
11/4/2021 3:10:00 PM
11/5/2021 4:00:00 PM
aaa
55
out
yes!
11/5/2021 4:00:00 PM
null
bb
111
out
welcome!
11/6/2021 6:09:00 PM
null
bb
222
in
can i call?
11/6/2021 6:39:00 PM
11/6/2021 8:22:00 PM
bb
333
out
sure.
11/6/2021 8:22:00 PM
null
cc
1111
out
hello?
11/8/2021 2:22:00 PM
null
cc
2222
in
Whatsup!
11/8/2021 3:22:00 PM
null
对于任何 'out' 文本，回复列为空，但对于每个 'in' 文本，它会带来 'out' 的下一个最新时间戳。如果传入文本后没有传出文本，则传入的 'ReplyDate' 也为空，如 'cc'.
的情况
这可以在 SQL 中完成吗？我正在使用 vertica，它不允许递归或循环语句，所以我必须在没有它们的情况下实现这一点。
我已经能够使用 lead () 来获取下一个输出文本的时间，但无法为所有以前的输入文本填充它。
这是我迄今为止尝试过的方法，但这并没有让我得到所需的结果：
with cte as (
select
row_number() over(partition by AppId order by date asc) as rn,
Id,
AppId
Direction,
Text,
Date,
lead(Direction, 1) over(order by Date asc) as lead_direction,
lead(Date, 1) over (order by Date asc) as lead_date,
from table
order by Date desc)
select
Id,
AppId
Direction,
Text
Date,
case when Direction = 'Out' then null
when lead_direction = null then null
when rn <> 1 and Direction = 'In' and Direction = lead_direction then null
when rn <> 1 and Direction = 'In' and Direction <> lead_direction then lead_date
end as ReplyDate
from cte
任何帮助将不胜感激。
虽然这个解决方案不是最干净的解决方案，但它完成了工作。
WITH
main_data AS (
SELECT 'aaa' as AppId,11 as Id,'in' as Direction ,'hello' AS Text,'11/2/2021 3:03:00 PM' AS DATE_TEXT
UNION ALL SELECT 'aaa' as AppId,22 as Id,'out'as Direction ,'yes?' AS Text,'11/2/2021 3:04:00 PM' AS DATE_TEXT
UNION ALL SELECT 'aaa' as AppId,33 as Id,'in' as Direction ,'need help!' AS Text,'11/3/2021 3:06:00 PM' AS DATE_TEXT
UNION ALL SELECT 'aaa' as AppId,44 as Id,'in' as Direction ,'you there?' AS Text,'11/4/2021 3:10:00 PM' AS DATE_TEXT
UNION ALL SELECT 'aaa' as AppId,55 as Id,'out' as Direction ,'yes!' AS Text,'11/5/2021 4:00:00 PM' AS DATE_TEXT
UNION ALL SELECT 'bb' as AppId,111 AS ID,'out' as Direction ,'welcome!' AS Text,'11/6/2021 6:09:00 PM' AS DATE_TEXT
UNION ALL SELECT 'bb' as AppId,222 AS ID,'in' as Direction ,'can i call?' AS Text,'11/6/2021 6:39:00 PM' AS DATE_TEXT
UNION ALL SELECT 'bb' as AppId,333 AS ID,'out' as Direction ,'sure.' AS Text,'11/6/2021 8:22:00 PM' AS DATE_TEXT
UNION ALL SELECT 'cc' as AppId,1111AS ID,'out' as Direction ,'hello?' AS Text,'11/8/2021 2:22:00 PM' AS DATE_TEXT
UNION ALL SELECT 'cc' as AppId,2222AS ID,'in' as Direction ,'Whatsup!' AS Text,'11/8/2021 3:22:00 PM' AS DATE_TEXT
)
, temp_data AS (SELECT AppId
, Id
, Direction
, TEXT
, DATE_TEXT
, CONDITIONAL_TRUE_EVENT(Direction = 'out') OVER (PARTITION BY AppId ORDER BY Id) as rank_logic
FROM main_data
)
SELECT t1.AppId
, t1.Id
, t1.Direction
, t1.TEXT
, t1.DATE_TEXT
, t2.DATE_TEXT
FROM temp_data t1
LEFT JOIN temp_data t2
ON t1.AppId = t2.AppId
AND t1.rank_logic + 1 = t2.rank_logic
AND t2.Direction = 'out'
AND t1.Direction <> 'out';
输出
AppId | Id | Direction | TEXT | DATE_TEXT | DATE_TEXT
-------+------+-----------+-------------+----------------------+----------------------
aaa | 11 | in | hello | 11/2/2021 3:03:00 PM | 11/2/2021 3:04:00 PM
aaa | 22 | out | yes? | 11/2/2021 3:04:00 PM |
aaa | 33 | in | need help! | 11/3/2021 3:06:00 PM | 11/5/2021 4:00:00 PM
aaa | 44 | in | you there? | 11/4/2021 3:10:00 PM | 11/5/2021 4:00:00 PM
aaa | 55 | out | yes! | 11/5/2021 4:00:00 PM |
bb | 111 | out | welcome! | 11/6/2021 6:09:00 PM |
bb | 222 | in | can i call? | 11/6/2021 6:39:00 PM | 11/6/2021 8:22:00 PM
bb | 333 | out | sure. | 11/6/2021 8:22:00 PM |
cc | 1111 | out | hello? | 11/8/2021 2:22:00 PM |
cc | 2222 | in | Whatsup! | 11/8/2021 3:22:00 PM |
(10 rows)
尝试了几次，但现在我想我明白了。
WITH 子句中的第一个 CTE 不是最终查询的一部分 - 它只是将您的原始输入放入一个独立的演示查询中。
真正的查询，真正的WITH子句从那之后开始。
由于每个 'in' 行的结束 'out' 行可以在一、两或几行之后出现，这可以通过行为模式来解决：一个或多个 'in'行，后跟一 'out' 行。这就是 MATCH() 子句的用途。
在包含子句的查询中只返回满足模式的行。
依赖函数 PATTERN_ID() returns 在 PARTITION BY... ORDER BY 表达式中找到的模式的序号。
按 PARTITION BY 列和 PATTERN_ID 分组将帮助我获得模式的最后时间戳，我需要 replyts （我更改了列名以避免保留字，例如 DATE 或 TEXT).
最后，我只需要在 id 列和 [=24= 相等的情况下，将 indata CTE 与包含 MATCH() 子句的查询左连接] 等于 'in'，第二个查询与分组查询。
我将中间结果作为 CTE 表达式中机制的说明...
WITH
-- input from you ...
indata(AppId,Id,Direction,txt,ts) AS (
SELECT 'aaa',11,'in','hello',TIMESTAMP '11/2/2021 3:03:00 PM'
UNION ALL SELECT 'aaa',22,'out','yes?',TIMESTAMP '11/2/2021 3:04:00 PM'
UNION ALL SELECT 'aaa',33,'in','need help!',TIMESTAMP '11/3/2021 3:06:00 PM'
UNION ALL SELECT 'aaa',44,'in','you there?',TIMESTAMP '11/4/2021 3:10:00 PM'
UNION ALL SELECT 'aaa',55,'out','yes!',TIMESTAMP '11/5/2021 4:00:00 PM'
UNION ALL SELECT 'bb',111,'out','welcome!',TIMESTAMP '11/6/2021 6:09:00 PM'
UNION ALL SELECT 'bb',222,'in','can i call?',TIMESTAMP '11/6/2021 6:39:00 PM'
UNION ALL SELECT 'bb',333,'out','sure.',TIMESTAMP '11/6/2021 8:22:00 PM'
UNION ALL SELECT 'cc',1111,'out','hello?',TIMESTAMP '11/8/2021 2:22:00 PM'
UNION ALL SELECT 'cc',2222,'in','Whatsup!',TIMESTAMP '11/8/2021 3:22:00 PM'
)
-- real query starts here, replace following comma with "WITH" ...
,
-- the MATCH() clause in action - note the depending functions
-- PATTERN_ID(), MATCH_ID() and EVENT_NAME()
pattern_q AS (
SELECT
appid
, id
, direction
, txt
, ts
, PATTERN_ID()
, MATCH_ID()
, EVENT_NAME()
FROM indata
MATCH(
PARTITION BY appid ORDER BY id
DEFINE
inbound AS direction='in'
, outbound AS direction='out'
PATTERN
p AS (inbound+ outbound)
)
-- out appid | id | direction | txt | ts | PATTERN_ID | MATCH_ID | EVENT_NAME
-- out ---------+-----+-----------+-------------+---------------------+------------+----------+------------
-- out aaa | 11 | in | hello | 2021-11-02 15:03:00 | 1 | 1 | inbound
-- out aaa | 22 | out | yes? | 2021-11-02 15:04:00 | 1 | 2 | outbound
-- out aaa | 33 | in | need help! | 2021-11-03 15:06:00 | 2 | 1 | inbound
-- out aaa | 44 | in | you there? | 2021-11-04 15:10:00 | 2 | 2 | inbound
-- out aaa | 55 | out | yes! | 2021-11-05 16:00:00 | 2 | 3 | outbound
-- out bb | 222 | in | can i call? | 2021-11-06 18:39:00 | 1 | 1 | inbound
-- out bb | 333 | out | sure. | 2021-11-06 20:22:00 | 1 | 2 | outbound
)
,
-- need the last timestamp per PATTERN_ID() ... so grouping
pattern_grp AS (
SELECT
appid
, pattern_id
, MIN(ts) AS g_ts
, MAX(ts) AS replyts
FROM pattern_q
GROUP BY
appid
, pattern_id
-- out appid | pattern_id | ts | replyts
-- out -------+------------+---------------------+---------------------
-- out aaa | 1 | 2021-11-02 15:03:00 | 2021-11-02 15:04:00
-- out aaa | 2 | 2021-11-03 15:06:00 | 2021-11-05 16:00:00
-- out bb | 1 | 2021-11-06 18:39:00 | 2021-11-06 20:22:00
)
SELECT
i.*
, g.replyts
FROM indata i
LEFT JOIN pattern_q p ON i.id = p.id AND i.direction='in'
LEFT JOIN pattern_grp g ON p.appid = g.appid AND p.pattern_id = g.pattern_id
-- out Null display is "(null)".
-- out AppId | Id | Direction | txt | ts | replyts
-- out -------+------+-----------+-------------+---------------------+---------------------
-- out aaa | 11 | in | hello | 2021-11-02 15:03:00 | 2021-11-02 15:04:00
-- out aaa | 22 | out | yes? | 2021-11-02 15:04:00 | (null)
-- out aaa | 33 | in | need help! | 2021-11-03 15:06:00 | 2021-11-05 16:00:00
-- out aaa | 44 | in | you there? | 2021-11-04 15:10:00 | 2021-11-05 16:00:00
-- out aaa | 55 | out | yes! | 2021-11-05 16:00:00 | (null)
-- out bb | 111 | out | welcome! | 2021-11-06 18:09:00 | (null)
-- out bb | 222 | in | can i call? | 2021-11-06 18:39:00 | 2021-11-06 20:22:00
-- out bb | 333 | out | sure. | 2021-11-06 20:22:00 | (null)
-- out cc | 1111 | out | hello? | 2021-11-08 14:22:00 | (null)
-- out cc | 2222 | in | Whatsup! | 2021-11-08 15:22:00 | (null)
我有以下 table:
| AppId | Id | Direction | Text | Date |
|---|---|---|---|---|
| aaa | 11 | in | hello | 11/2/2021 3:03:00 PM |
| aaa | 22 | out | yes? | 11/2/2021 3:04:00 PM |
| aaa | 33 | in | need help! | 11/3/2021 3:06:00 PM |
| aaa | 44 | in | you there? | 11/4/2021 3:10:00 PM |
| aaa | 55 | out | yes! | 11/5/2021 4:00:00 PM |
| bb | 111 | out | welcome! | 11/6/2021 6:09:00 PM |
| bb | 222 | in | can i call? | 11/6/2021 6:39:00 PM |
| bb | 333 | out | sure. | 11/6/2021 8:22:00 PM |
| cc | 1111 | out | hello? | 11/8/2021 2:22:00 PM |
| cc | 2222 | in | Whatsup! | 11/8/2021 3:22:00 PM |
Id 是一个主键，方向基本上告诉我们它是传入消息还是传出消息。 App id 用于识别属于单个对话的各种 id。我想将传入消息后的第一个响应时间标识为：
| AppId | Id | Direction | Text | Date | ReplyDate |
|---|---|---|---|---|---|
| aaa | 11 | in | hello | 11/2/2021 3:03:00 PM | 11/2/2021 3:04:00 PM |
| aaa | 22 | out | yes? | 11/2/2021 3:04:00 PM | null |
| aaa | 33 | in | need help! | 11/3/2021 3:06:00 PM | 11/5/2021 4:00:00 PM |
| aaa | 44 | in | you there? | 11/4/2021 3:10:00 PM | 11/5/2021 4:00:00 PM |
| aaa | 55 | out | yes! | 11/5/2021 4:00:00 PM | null |
| bb | 111 | out | welcome! | 11/6/2021 6:09:00 PM | null |
| bb | 222 | in | can i call? | 11/6/2021 6:39:00 PM | 11/6/2021 8:22:00 PM |
| bb | 333 | out | sure. | 11/6/2021 8:22:00 PM | null |
| cc | 1111 | out | hello? | 11/8/2021 2:22:00 PM | null |
| cc | 2222 | in | Whatsup! | 11/8/2021 3:22:00 PM | null |
对于任何 'out' 文本，回复列为空，但对于每个 'in' 文本，它会带来 'out' 的下一个最新时间戳。如果传入文本后没有传出文本，则传入的 'ReplyDate' 也为空，如 'cc'.
的情况这可以在 SQL 中完成吗？我正在使用 vertica，它不允许递归或循环语句，所以我必须在没有它们的情况下实现这一点。
我已经能够使用 lead () 来获取下一个输出文本的时间，但无法为所有以前的输入文本填充它。
这是我迄今为止尝试过的方法，但这并没有让我得到所需的结果：
with cte as (
select
row_number() over(partition by AppId order by date asc) as rn,
Id,
AppId
Direction,
Text,
Date,
lead(Direction, 1) over(order by Date asc) as lead_direction,
lead(Date, 1) over (order by Date asc) as lead_date,
from table
order by Date desc)
select
Id,
AppId
Direction,
Text
Date,
case when Direction = 'Out' then null
when lead_direction = null then null
when rn <> 1 and Direction = 'In' and Direction = lead_direction then null
when rn <> 1 and Direction = 'In' and Direction <> lead_direction then lead_date
end as ReplyDate
from cte
任何帮助将不胜感激。
虽然这个解决方案不是最干净的解决方案，但它完成了工作。
WITH
main_data AS (
SELECT 'aaa' as AppId,11 as Id,'in' as Direction ,'hello' AS Text,'11/2/2021 3:03:00 PM' AS DATE_TEXT
UNION ALL SELECT 'aaa' as AppId,22 as Id,'out'as Direction ,'yes?' AS Text,'11/2/2021 3:04:00 PM' AS DATE_TEXT
UNION ALL SELECT 'aaa' as AppId,33 as Id,'in' as Direction ,'need help!' AS Text,'11/3/2021 3:06:00 PM' AS DATE_TEXT
UNION ALL SELECT 'aaa' as AppId,44 as Id,'in' as Direction ,'you there?' AS Text,'11/4/2021 3:10:00 PM' AS DATE_TEXT
UNION ALL SELECT 'aaa' as AppId,55 as Id,'out' as Direction ,'yes!' AS Text,'11/5/2021 4:00:00 PM' AS DATE_TEXT
UNION ALL SELECT 'bb' as AppId,111 AS ID,'out' as Direction ,'welcome!' AS Text,'11/6/2021 6:09:00 PM' AS DATE_TEXT
UNION ALL SELECT 'bb' as AppId,222 AS ID,'in' as Direction ,'can i call?' AS Text,'11/6/2021 6:39:00 PM' AS DATE_TEXT
UNION ALL SELECT 'bb' as AppId,333 AS ID,'out' as Direction ,'sure.' AS Text,'11/6/2021 8:22:00 PM' AS DATE_TEXT
UNION ALL SELECT 'cc' as AppId,1111AS ID,'out' as Direction ,'hello?' AS Text,'11/8/2021 2:22:00 PM' AS DATE_TEXT
UNION ALL SELECT 'cc' as AppId,2222AS ID,'in' as Direction ,'Whatsup!' AS Text,'11/8/2021 3:22:00 PM' AS DATE_TEXT
)
, temp_data AS (SELECT AppId
, Id
, Direction
, TEXT
, DATE_TEXT
, CONDITIONAL_TRUE_EVENT(Direction = 'out') OVER (PARTITION BY AppId ORDER BY Id) as rank_logic
FROM main_data
)
SELECT t1.AppId
, t1.Id
, t1.Direction
, t1.TEXT
, t1.DATE_TEXT
, t2.DATE_TEXT
FROM temp_data t1
LEFT JOIN temp_data t2
ON t1.AppId = t2.AppId
AND t1.rank_logic + 1 = t2.rank_logic
AND t2.Direction = 'out'
AND t1.Direction <> 'out';
输出
AppId | Id | Direction | TEXT | DATE_TEXT | DATE_TEXT
-------+------+-----------+-------------+----------------------+----------------------
aaa | 11 | in | hello | 11/2/2021 3:03:00 PM | 11/2/2021 3:04:00 PM
aaa | 22 | out | yes? | 11/2/2021 3:04:00 PM |
aaa | 33 | in | need help! | 11/3/2021 3:06:00 PM | 11/5/2021 4:00:00 PM
aaa | 44 | in | you there? | 11/4/2021 3:10:00 PM | 11/5/2021 4:00:00 PM
aaa | 55 | out | yes! | 11/5/2021 4:00:00 PM |
bb | 111 | out | welcome! | 11/6/2021 6:09:00 PM |
bb | 222 | in | can i call? | 11/6/2021 6:39:00 PM | 11/6/2021 8:22:00 PM
bb | 333 | out | sure. | 11/6/2021 8:22:00 PM |
cc | 1111 | out | hello? | 11/8/2021 2:22:00 PM |
cc | 2222 | in | Whatsup! | 11/8/2021 3:22:00 PM |
(10 rows)
尝试了几次，但现在我想我明白了。
WITH 子句中的第一个 CTE 不是最终查询的一部分 - 它只是将您的原始输入放入一个独立的演示查询中。
真正的查询，真正的WITH子句从那之后开始。
由于每个 'in' 行的结束 'out' 行可以在一、两或几行之后出现，这可以通过行为模式来解决：一个或多个 'in'行，后跟一 'out' 行。这就是 MATCH() 子句的用途。
在包含子句的查询中只返回满足模式的行。
依赖函数 PATTERN_ID() returns 在 PARTITION BY... ORDER BY 表达式中找到的模式的序号。
按 PARTITION BY 列和 PATTERN_ID 分组将帮助我获得模式的最后时间戳，我需要 replyts （我更改了列名以避免保留字，例如 DATE 或 TEXT).
最后，我只需要在 id 列和 [=24= 相等的情况下，将 indata CTE 与包含 MATCH() 子句的查询左连接] 等于 'in'，第二个查询与分组查询。
我将中间结果作为 CTE 表达式中机制的说明...
WITH
-- input from you ...
indata(AppId,Id,Direction,txt,ts) AS (
SELECT 'aaa',11,'in','hello',TIMESTAMP '11/2/2021 3:03:00 PM'
UNION ALL SELECT 'aaa',22,'out','yes?',TIMESTAMP '11/2/2021 3:04:00 PM'
UNION ALL SELECT 'aaa',33,'in','need help!',TIMESTAMP '11/3/2021 3:06:00 PM'
UNION ALL SELECT 'aaa',44,'in','you there?',TIMESTAMP '11/4/2021 3:10:00 PM'
UNION ALL SELECT 'aaa',55,'out','yes!',TIMESTAMP '11/5/2021 4:00:00 PM'
UNION ALL SELECT 'bb',111,'out','welcome!',TIMESTAMP '11/6/2021 6:09:00 PM'
UNION ALL SELECT 'bb',222,'in','can i call?',TIMESTAMP '11/6/2021 6:39:00 PM'
UNION ALL SELECT 'bb',333,'out','sure.',TIMESTAMP '11/6/2021 8:22:00 PM'
UNION ALL SELECT 'cc',1111,'out','hello?',TIMESTAMP '11/8/2021 2:22:00 PM'
UNION ALL SELECT 'cc',2222,'in','Whatsup!',TIMESTAMP '11/8/2021 3:22:00 PM'
)
-- real query starts here, replace following comma with "WITH" ...
,
-- the MATCH() clause in action - note the depending functions
-- PATTERN_ID(), MATCH_ID() and EVENT_NAME()
pattern_q AS (
SELECT
appid
, id
, direction
, txt
, ts
, PATTERN_ID()
, MATCH_ID()
, EVENT_NAME()
FROM indata
MATCH(
PARTITION BY appid ORDER BY id
DEFINE
inbound AS direction='in'
, outbound AS direction='out'
PATTERN
p AS (inbound+ outbound)
)
-- out appid | id | direction | txt | ts | PATTERN_ID | MATCH_ID | EVENT_NAME
-- out ---------+-----+-----------+-------------+---------------------+------------+----------+------------
-- out aaa | 11 | in | hello | 2021-11-02 15:03:00 | 1 | 1 | inbound
-- out aaa | 22 | out | yes? | 2021-11-02 15:04:00 | 1 | 2 | outbound
-- out aaa | 33 | in | need help! | 2021-11-03 15:06:00 | 2 | 1 | inbound
-- out aaa | 44 | in | you there? | 2021-11-04 15:10:00 | 2 | 2 | inbound
-- out aaa | 55 | out | yes! | 2021-11-05 16:00:00 | 2 | 3 | outbound
-- out bb | 222 | in | can i call? | 2021-11-06 18:39:00 | 1 | 1 | inbound
-- out bb | 333 | out | sure. | 2021-11-06 20:22:00 | 1 | 2 | outbound
)
,
-- need the last timestamp per PATTERN_ID() ... so grouping
pattern_grp AS (
SELECT
appid
, pattern_id
, MIN(ts) AS g_ts
, MAX(ts) AS replyts
FROM pattern_q
GROUP BY
appid
, pattern_id
-- out appid | pattern_id | ts | replyts
-- out -------+------------+---------------------+---------------------
-- out aaa | 1 | 2021-11-02 15:03:00 | 2021-11-02 15:04:00
-- out aaa | 2 | 2021-11-03 15:06:00 | 2021-11-05 16:00:00
-- out bb | 1 | 2021-11-06 18:39:00 | 2021-11-06 20:22:00
)
SELECT
i.*
, g.replyts
FROM indata i
LEFT JOIN pattern_q p ON i.id = p.id AND i.direction='in'
LEFT JOIN pattern_grp g ON p.appid = g.appid AND p.pattern_id = g.pattern_id
-- out Null display is "(null)".
-- out AppId | Id | Direction | txt | ts | replyts
-- out -------+------+-----------+-------------+---------------------+---------------------
-- out aaa | 11 | in | hello | 2021-11-02 15:03:00 | 2021-11-02 15:04:00
-- out aaa | 22 | out | yes? | 2021-11-02 15:04:00 | (null)
-- out aaa | 33 | in | need help! | 2021-11-03 15:06:00 | 2021-11-05 16:00:00
-- out aaa | 44 | in | you there? | 2021-11-04 15:10:00 | 2021-11-05 16:00:00
-- out aaa | 55 | out | yes! | 2021-11-05 16:00:00 | (null)
-- out bb | 111 | out | welcome! | 2021-11-06 18:09:00 | (null)
-- out bb | 222 | in | can i call? | 2021-11-06 18:39:00 | 2021-11-06 20:22:00
-- out bb | 333 | out | sure. | 2021-11-06 20:22:00 | (null)
-- out cc | 1111 | out | hello? | 2021-11-08 14:22:00 | (null)
-- out cc | 2222 | in | Whatsup! | 2021-11-08 15:22:00 | (null)