SQL hierarchy - getting newest grandchild - sql

Question

I have a H2 database with a LEFT JOIN query on Bird which returns all birds along with the newest Health_Check.Catch_Date. I want to extend this query and include in the results newest PIT.ID, PIT.CODE and TRANSMITTER.IDc, TRANSMITTER.CHANNEL which are related to Health_Check.
Note: not every Health_Check has a relationship to a Transmitter or PIT.
A few notes about how the data is structured.
Most birds have several health checks;
Not every health check has a associated PIT or TRANSMITTER;
The newest Health_Check for a bird may not contain a PIT or TRANSMITTER;
Most birds have one PIT;
Most birds have a few Transmitters;
The date the Transmitter or PIT was attached is based on the related Health_Check.Catch_Date;
Most of my queries will require returning a Bird with the newest Health Check, PIT and Transmitter.
The SQL to get the newest HEALTH_CHECK is:
SELECT b.NAME, b.ID as birdId, hc1.CATCH_DATE, hc1.id as healthCheckId
FROM BIRD b
LEFT OUTER JOIN
( HEALTH_CHECK hc1
INNER JOIN
(
SELECT BIRD_ID, MAX(CATCH_DATE) AS MAX_DATE
FROM HEALTH_CHECK
GROUP BY BIRD_ID
) hc2
ON hc2.BIRD_ID = hc1.BIRD_ID AND
hc2.MAX_DATE = hc1.CATCH_DATE
)
ON hc1.BIRD_ID = b.ID;
Question: how can I also return in the results the most recent PIT and TRANSMITTER?
NB: The newest PIT and TRANS is often not on the newest HealthCheck. I am open to a schema change/adding a second relationship but I must be able to determine which HealthCheck a PIT and TRANS was attached in.
As an example results would look like:
BIRD.ID | BIRD.NAME | NEWEST HEALTH CHCECK | PIT.ID | PIT.CODE | TRANS.ID | TRANS.CHNL |
---------|-------------|----------------------|--------|----------|----------|------------|
1 | Bob | 2022-03-01 | AB001 | 3 | 2 | 40 |
2 | Jim | NULL | NULL | NULL | NULL | NULL |
3 | Jane | 2022-01-02 | DC123 | 2 | 3 | 50 |

Answer 1

One solution can be, to get the highest ID from every TRANs ANd PIT, i thinl that is the only ORDER that ic could find
Row_number would give every row for a specific chekc id, and the nimber 1 should be last one(ORDER BY ID DESC), it counts for every channel a new
SELECT b.NAME, b.ID as birdId, hc1.CATCH_DATE, hc1.id as healthCheckId
,PIT.ID , PIT.CODE,TRANS.ID , TRANS.CHANNEL
FROM BIRD b
LEFT OUTER JOIN
( HEALTH_CHECK hc1
INNER JOIN
(
SELECT BIRD_ID, MAX(CATCH_DATE) AS MAX_DATE
FROM HEALTH_CHECK
GROUP BY BIRD_ID
) hc2
ON hc2.BIRD_ID = hc1.BIRD_ID AND
hc2.MAX_DATE = hc1.CATCH_DATE
)
ON hc1.BIRD_ID = b.ID
LEFT JOIN
(SELECT ID, CODE,HEALTH_CHECK_ID
FROM
(SELECT ID , CODE,HEALTH_CHECK_ID
, ROW_NUMBER() OVER( PARTITION BY HEALTH_CHECK_ID ORDER BY ID DESC) rn FROM PIT) Pt
WHERE rn = 1) PIT ON PIT.HEALTH_CHeCK_ID = hc1.ID
LEFT JOIN
(SELECT ID, CHANNEL,HEALTH_CHECK_ID
FROM
(SELECT ID , CHANNEL,HEALTH_CHECK_ID
, ROW_NUMBER() OVER( PARTITION BY HEALTH_CHECK_ID ORDER BY ID DESC) rn FROM PIT) tr
WHERE rn = 1) TRANS ON TRANS.HEALTH_CHECK_ID = hc1.ID;
If there is a 1:1 relaionship you need only to join them,
SELECT b.NAME, b.ID as birdId, hc1.CATCH_DATE, hc1.id as healthCheckId
,PIT.ID , PIT.CODE,TRANS.ID , TRANS.CHANNEL
FROM BIRD b
LEFT OUTER JOIN
( HEALTH_CHECK hc1
INNER JOIN
(
SELECT BIRD_ID, MAX(CATCH_DATE) AS MAX_DATE
FROM HEALTH_CHECK
GROUP BY BIRD_ID
) hc2
ON hc2.BIRD_ID = hc1.BIRD_ID AND
hc2.MAX_DATE = hc1.CATCH_DATE
)
ON hc1.BIRD_ID = b.ID
LEFT JOIN
PIT ON PIT.HEALTH_CHeCK_ID = hc1.ID
LEFT JOIN
TRANS ON TRANS.HEALTH_CHECK_ID = hc1.ID;
if you need another heahc_ckes tarns and Pit that are not in the latest healthcheck, you need to add to the sunselect of the first query all healch ckes of that bird
SELECT b.NAME, b.ID as birdId, hc1.CATCH_DATE, hc1.id as healthCheckId
,PIT.ID , PIT.CODE,TRANS.ID , TRANS.CHANNEL
FROM BIRD b
LEFT OUTER JOIN
( HEALTH_CHECK hc1
INNER JOIN
(
SELECT BIRD_ID, MAX(CATCH_DATE) AS MAX_DATE
FROM HEALTH_CHECK
GROUP BY BIRD_ID
) hc2
ON hc2.BIRD_ID = hc1.BIRD_ID AND
hc2.MAX_DATE = hc1.CATCH_DATE
)
ON hc1.BIRD_ID = b.ID
LEFT JOIN
(SELECT ID, CODE,HEALTH_CHECK_ID,BIRD_ID
FROM
(SELECT ID , CODE,hc3.BIRD_ID,HEALTH_CHECK_ID
, ROW_NUMBER() OVER( PARTITION BY HEALTH_CHECK_ID ORDER BY hc3.CATCH_DATE DESC) rn FROM HEALTH_CHECK hc3 JOIN PIT ON hc3.ID = PIT.HEALTH_CHeCK_ID
WHERE hc3.BIRD_ID = hc1.BIRD_ID) Pt
WHERE rn = 1) PIT ON PIT.BIRD_ID = hc1.BIRD_ID
LEFT JOIN
(SELECT ID, CHANNEL,HEALTH_CHECK_ID,BIRD_ID
FROM
(SELECT ID , CHANNEL,HEALTH_CHECK_ID,BIRD_ID
, ROW_NUMBER() OVER( PARTITION BY HEALTH_CHECK_ID ORDER BY hc4.CATCH_DATE DESC) rn FROM HEALTH_CHECK hc4 JOIN PIT ON hc4.ID = PIT.HEALTH_CHeCK_ID
WHERE hc4.BIRD_ID = hc1.BIRD_ID) tr
WHERE rn = 1) TRANS ON TRANS.BIRD_ID = hc1.BIRD_ID;

Answer 2

For each bird (left join) to return a result set containing:
Newest health check (by date)
Recursively getting newest PIT (determined by date from related health check)
Recursively getting newest Trans (determined by date from related health check)
Noting that the PIT and Trans are often not on the newest Health Check can be done by:
SELECT *
FROM BIRD AS b
LEFT JOIN (
SELECT *
FROM (
SELECT id AS hc_id, bird_id, catch_date
, ROW_NUMBER() OVER (PARTITION BY bird_id ORDER BY catch_date DESC) AS n
FROM HEALTH_CHECK
) AS cte1a
WHERE n = 1
) AS cte1
ON cte1.bird_id = b.id
LEFT JOIN (
SELECT *
FROM (
SELECT p.id AS pit_id
, hc.bird_id
, p.Health_Check_Id
, p.code
, hc.catch_date as catchdate
, RANK() OVER (PARTITION BY hc.bird_id ORDER BY catch_date DESC) AS n
FROM PIT AS p
JOIN HEALTH_CHECK AS hc
ON p.Health_Check_Id = hc.id
) AS cte2a
WHERE n = 1
) AS cte2
ON cte2.bird_id = b.id
LEFT JOIN (
SELECT *
FROM (
SELECT t.id AS transmitter_id
, hc.bird_id
, t.Health_Check_Id
, t.channel
, hc.catch_date as catchdate
, RANK() OVER (PARTITION BY hc.bird_id ORDER BY catch_date DESC) AS n
FROM TRANSMITTER AS t
JOIN HEALTH_CHECK AS hc
ON t.Health_Check_Id = hc.id
) AS cte3a
WHERE n = 1
A fiddle is here.