BigQuery，连续行相关问题之间的区别

Question

我很难找出正确的问题，但我有以下查询，我已经能够得到连续行之间的差异。

我的问题是（如图所示）当 icao_address 和 flight_number 更改（我想要）然后 我希望发生更改的行（更改 icao_address 和 flight_number）重置为零强>。

我的意图是按 icao_address 和 flight_number 对查询进行分组。

WITH
  temptable AS (
  SELECT
  date(timestamp) as flight_date,
    timestamp,
    UNIX_SECONDS(timestamp) AS timestamp_unix,
    icao_address,
    flight_number,
    LAG(UNIX_SECONDS(timestamp)) OVER (ORDER BY timestamp) AS value2
  FROM
    `table`
  WHERE
    (icao_address = '70C0C7' and flight_number = 'WY144') or (icao_address = '750467' and flight_number = 'AK1311') and 
    flight_number is not null and 
    timestamp BETWEEN '2020-01-01'
    AND '2020-01-02'),
  relation2 AS (
  SELECT
  flight_date,
    timestamp,
    timestamp_unix,
    timestamp_unix - value2 AS difference,
    icao_address,
    flight_number
  FROM
    temptable
  ) 
  select * 
  from relation2

在图像中，在第 11 行，在差异列下，考虑到 icao_address 和 flight_number 的变化，我希望那里有一个零。这是可能的，或任何其他方式来做到这一点。我正在尝试执行一个案例，但没有取得任何进展。

enter image description here

Answer 1

我想你想要partition by:

LAG(UNIX_SECONDS(timestamp)) OVER (PARTITION BY icao_address, flight_number ORDER BY timestamp) AS value2

如果您实际上想要0 而不是NULL，请使用LAG() 的三参数形式：

LAG(UNIX_SECONDS(timestamp), 1, 0) OVER (PARTITION BY icao_address, flight_number ORDER BY timestamp) AS value2