From
https://cloud.google.com/bigquery/docs/partitioned-tables:
you can shard tables using a time-based naming approach such as [PREFIX]_YYYYMMDD
This enables me to do:
SELECT count(*) FROM `xxx.xxx.xxx_*`
and query across all the shards. Is there a special notation that queries only the latest shard? For example say I had:
xxx_20180726
xxx_20180801
could I do something along the lines of
SELECT count(*) FROM `xxx.xxx.xxx_{{ latest }}`
to query xxx_20180801?
SINGLE QUERY INSPIRED BY Mikhail Berlyant:
SELECT count(*) as c FROM `XXX.PREFIX_*` WHERE _TABLE_SUFFIX IN ( SELECT
SUBSTR(MAX(table_id), LENGTH('PREFIX_') + 2)
FROM
`XXX.__TABLES_SUMMARY__`
WHERE
table_id LIKE 'PREFIX_%')
If you do care about cost (meaning how many tables will be scaned by your query) - the only way to do so is to do in two steps like below
First query
#standardSQL
SELECT SUBSTR(MAX(table_id), LENGTH('PREFIX') + 1)
FROM `xxx.xxx.__TABLES_SUMMARY__`
WHERE table_id LIKE 'PREFIX%'
Second Query
#standardSQL
SELECT COUNT(*)
FROM `xxx.xxx.PREFIX_*`
WHERE _TABLE_SUFFIX = '<result of first query>'
so, if result of first query is 20180801 so, second query will obviously look like below
#standardSQL
SELECT COUNT(*)
FROM `xxx.xxx.PREFIX_*`
WHERE _TABLE_SUFFIX = '20180801'
If you don't care about cost but rather need just result - you can easily combine above two queries into one - but - again - remember - even though result will be out of last table - cost will be as you query all table that match xxx.xxx.PREFIX_*
Forgot to mention (even though it should be obvious): of course when you have only COUNT(1) in your SELECT - the cost will be 0(zero) for both options - but in reality - most likely you will have something more valuable than just count(1)
I know this is a kind of an old thread but I was surprised why no one offers an answer using Variables.
"Héctor Neri" already mentioned this in the comments but I thought might be better to have an actual answer with a sample code posted.
#standardSQL
DECLARE SHARD_DATE STRING;
SET SHARD_DATE=(
SELECT MAX(REPLACE(table_name,'{TABLE}_',''))
FROM `{PRJ}.{DATASET}.INFORMATION_SCHEMA.TABLES`
WHERE table_name LIKE '{TABLE}_20%'
);
SELECT * FROM `{PRJ}.{DATASET}.{TABLE}_*`
WHERE _TABLE_SUFFIX = SHARD_DATE
Make sure to replace {PRJ}, {DATASET}, and {TABLE} values with your table location.
If you run this on BigQuery Web UI, you will see this message:
WARNING: Could not compute bytes processed estimate for script.
But you can see that variable properly reduce the table scan to the latest partition and does not cause any extra cost after running the script.
Related
So I'd like make a query that shows all the datasets from a project, and the number of tables in each one. My problem is with the number of tables.
Here is what I'm stuck with :
SELECT
smt.catalog_name as `Project`,
smt.schema_name as `DataSet`,
( SELECT
COUNT(*)
FROM ***DataSet***.INFORMATION_SCHEMA.TABLES
) as `nbTable`,
smt.creation_time,
smt.location
FROM
INFORMATION_SCHEMA.SCHEMATA smt
ORDER BY DataSet
The view INFORMATION_SCHEMA.SCHEMATA lists all the datasets from the project the query is executed, and the view INFORMATION_SCHEMA.TABLES lists all the tables from a given dataset.
The thing is that the view INFORMATION_SCHEMA.TABLES needs to have the dataset specified like this give the tables informations : dataset.INFORMATION_SCHEMA.TABLES
So what I need is to replace the *** DataSet*** by the one I got from the query itself (smt.schema_name).
I am not sure if I can do it with a sub query, but I don't really know how to manage to do it.
I hope I'm clear enough, thanks in advance if you can help.
You can do this using some procedural language as follows:
CREATE TEMP TABLE table_counts (dataset_id STRING, table_count INT64);
FOR record IN
(
SELECT
catalog_name as project_id,
schema_name as dataset_id
FROM `elzagales.INFORMATION_SCHEMA.SCHEMATA`
)
DO
EXECUTE IMMEDIATE
CONCAT("INSERT table_counts (dataset_id, table_count) SELECT table_schema as dataset_id, count(table_name) from ", record.dataset_id,".INFORMATION_SCHEMA.TABLES GROUP BY dataset_id");
END FOR;
SELECT * FROM table_counts;
This will return something like:
I have a very large (3.5B records) table that I want to update/insert (upsert) using the MERGE statement in BigQuery. The source table is a staging table that contains only the new data, and I need to check if the record with a corresponding ID is in the target table, updating the row if so or inserting if not.
The target table is partitioned by an integer field called IdParent, and the matching is done on IdParent and another integer field called IdChild. My merge statement/script looks like this:
declare parentList array<int64>;
set parentList = array(select distinct IdParent from dataset.Staging);
merge into dataset.Target t
using dataset.Staging s
on
-- target is partitioned by IdParent, do this for partition pruning
t.IdParent in unnest(parentList)
and t.IdParent = s.IdParent
and t.IdChild = s.IdChild
when matched and t.IdParent in unnest(parentList) then
update
set t.Column1 = s.Column1,
t.Column2 = s.Column2,
...<more columns>
when not matched and IdParent in unnest(parentList) then
insert (<all the fields>)
values (<all the fields)
;
So I:
Pull the IdParent list from the staging table to know which partitions to prune
limit the partitions of the target table in the join predicate
also limit the partitions of the target table in the match/not matched conditions
The total size of dataset.Target is ~250GB. If I put this script in my BQ editor and remove all the IdParent in unnest(parentList) then it shows ~250GB to bill in the editor (as expected since there's no partition pruning). If I add the IdParent in unnest(parentList) back in so the script is exactly like you see it above i.e. attempting to partition prune, the editor shows ~97MB to bill. However, when I look at the query results, I see that it actually billed ~180GB:
The target table is also clustered on the two fields being matched, and I'm aware that the benefits of clustering are typically not shown in the editor's estimate. However, my understanding is that that should only make the bytes billed smaller... I can't think of any reason why this would happen.
Is this a BQ bug, or am I just missing something? BigQuery doesn't even say "the script is estimated to process XX MB", it says "This will process XX MB" and then it processes way more.
That's very interesting. What you did seems totally correct.
It seems BQ query planner could interpret your SQL correctly and know the partition pruning is provided, but when it executes. it failed to do so.
try removing t.IdParent in unnest(parentList) from both when matched clauses to see if the issue still happens, that is,
declare parentList array<int64>;
set parentList = array(select distinct IdParent from dataset.Staging);
merge into dataset.Target t
using dataset.Staging s
on
-- target is partitioned by IdParent, do this for partition pruning
t.IdParent in unnest(parentList)
and t.IdParent = s.IdParent
and t.IdChild = s.IdChild
when matched then
update
set t.Column1 = s.Column1,
t.Column2 = s.Column2,
...<more columns>
when not matched then
insert (<all the fields>)
values (<all the fields)
;
It would be a good idea to submit a bug to BigQuery if it couldn't be resolved.
which of these is more efficient query to run:
one where the INCLUDE / DON'T INCLUDE filter condition in WHERE clause and tested for each row
SELECT distinct fullvisitorid
FROM `google.com:analytics-bigquery.LondonCycleHelmet.ga_sessions_20130910` t, unnest(hits) as ht
WHERE (select max(if(cd.index = 1,cd.value,null))from unnest(ht.customDimensions) cd)
= 'high_worth'
one returning all rows and then outer SELECT clause doing all filtering test to INCLUDE / DON'T INCLUDE
SELECT distinct fullvisitorid
FROM
(
SELECT
fullvisitorid
, (select max(if(cd.index = 1,cd.value,null)) FROM unnest(ht.customDimensions) cd) hit_cd_1
FROM `google.com:analytics-bigquery.LondonCycleHelmet.ga_sessions_20130910` t
, unnest(hits) as ht
)
WHERE
hit_cd_1 = 'high_worth'
Both produce exactly same results!
the goal is: list of fullvisitorId, who ever sent hit Level Custom Dimension (index =1) with value = 'high_worth' users ()
Thanks for your inputs!
Cheers!
/Vibhor
I tried the two queries and compared their explanations, they are identical. I am assuming some sort of optimization magic occurs prior to the query being ran.
As of your original two queries: obviously - they are identical even though you slightly rearranged appearance. so from those two you should choose whatever easier for you to read/maintain. I would pick first query - but it is really matter of personal preferences
Meantime, try below (BigQuery Standard SQL) - it looks slightly optimized to me - but I didn't have chance to test on real data
SELECT DISTINCT fullvisitorid
FROM `google.com:analytics-bigquery.LondonCycleHelmet.ga_sessions_20130910` t,
UNNEST(hits) AS ht, UNNEST(ht.customDimensions) cd
WHERE cd.index = 1 AND cd.value = 'high_worth'
Obviously - it should produce same result as your two queries
Execution plan looks better to me and it (query) is faster is much easier to read / manage
How to get all the table names from the below Sql? My sql returns only the last table name.
with t as
(select 'select col1,
(select max(col3) from dd3) max_timestamp
from dd1,
dd2
where dd1.col1 = dd2.col1
and dd1.col1 in(select col1 from dd4)' sql_text from dual)
select regexp_substr(regexp_substr(upper(sql_text), '\sFROM\s*(\w|\.|_)*'), '(\w|_|\.)+', 1,2)
from t
Thanks,
DD.
This is a more of a regex question than an Oracle question.
If you can run the sql through REPLACE(REPLACE(sql,CHR(13),' '),CHR(10),NULL) to replace all newlines with a space, so that the query fits on a single line, here is regex that will return all the tables in group 1 (for the ones after FROM) and group 3 for subsequent items in a list:
/FROM ([A-Z0-9$#_]+)(,[\s]*([A-Z0-9$#_]+))*/gi
Having multiple groups is not ideal, so I would look at the full match instead, see https://regex101.com/r/OZUalH/1/ for an example (see full match on the right, where every match has from followed by one or more tables).
But let me warn you this is not going to be robust, as these valid FROM clause expressions are not handled:
"my_table"
MY_TABLE AS A
MY_TABLE AS "a"
etc...
If it were me, I would write a function to run the query through explain plan (execute immediate 'explain plan for ...') and extract the tables from the plan tables (or possibly using SYS.DBMS_XPLAN)
class Log:
project = ForeignKey(Project)
msg = CharField(...)
date = DateField(...)
I want to select the four most recent Log entries where each Log entry must have a unique project foreign key. I've tries the solutions on google search but none of them works and the django documentation isn't that very good for lookup..
I tried stuff like:
Log.objects.all().distinct('project')[:4]
Log.objects.values('project').distinct()[:4]
Log.objects.values_list('project').distinct('project')[:4]
But this either return nothing or Log entries of the same project..
Any help would be appreciated!
Queries don't work like that - either in Django's ORM or in the underlying SQL. If you want to get unique IDs, you can only query for the ID. So you'll need to do two queries to get the actual Log entries. Something like:
id_list = Log.objects.order_by('-date').values_list('project_id').distinct()[:4]
entries = Log.objects.filter(id__in=id_list)
Actually, you can get the project_ids in SQL. Assuming that you want the unique project ids for the four projects with the latest log entries, the SQL would look like this:
SELECT project_id, max(log.date) as max_date
FROM logs
GROUP BY project_id
ORDER BY max_date DESC LIMIT 4;
Now, you actually want all of the log information. In PostgreSQL 8.4 and later you can use windowing functions, but that doesn't work on other versions/databases, so I'll do it the more complex way:
SELECT logs.*
FROM logs JOIN (
SELECT project_id, max(log.date) as max_date
FROM logs
GROUP BY project_id
ORDER BY max_date DESC LIMIT 4 ) as latest
ON logs.project_id = latest.project_id
AND logs.date = latest.max_date;
Now, if you have access to windowing functions, it's a bit neater (I think anyway), and certainly faster to execute:
SELECT * FROM (
SELECT logs.field1, logs.field2, logs.field3, logs.date
rank() over ( partition by project_id
order by "date" DESC ) as dateorder
FROM logs ) as logsort
WHERE dateorder = 1
ORDER BY logs.date DESC LIMIT 1;
OK, maybe it's not easier to understand, but take my word for it, it runs worlds faster on a large database.
I'm not entirely sure how that translates to object syntax, though, or even if it does. Also, if you wanted to get other project data, you'd need to join against the projects table.
I know this is an old post, but in Django 2.0, I think you could just use:
Log.objects.values('project').distinct().order_by('project')[:4]
You need two querysets. The good thing is it still results in a single trip to the database (though there is a subquery involved).
latest_ids_per_project = Log.objects.values_list(
'project').annotate(latest=Max('date')).order_by(
'-latest').values_list('project')
log_objects = Log.objects.filter(
id__in=latest_ids_per_project[:4]).order_by('-date')
This looks a bit convoluted, but it actually results in a surprisingly compact query:
SELECT "log"."id",
"log"."project_id",
"log"."msg"
"log"."date"
FROM "log"
WHERE "log"."id" IN
(SELECT U0."id"
FROM "log" U0
GROUP BY U0."project_id"
ORDER BY MAX(U0."date") DESC
LIMIT 4)
ORDER BY "log"."date" DESC