after migrating from PostgreSQL server version 9 to 8.4 I have encountered very strange error.
Short description:
If there is a trigger on a given table for each row before insert or update and one uses in conditional statement (if-else) TG_OP value check and OLD object, following error raises when doinng INSERT:
ERROR: record "old" is not assigned yet
DETAIL: The tuple structure of a not-yet-assigned record is indeterminate.
Detailed description:
There is following DB structure:
CREATE TABLE table1
(
id serial NOT NULL,
name character varying(256),
CONSTRAINT table1_pkey PRIMARY KEY (id)
)
WITH (OIDS=FALSE);
CREATE OR REPLACE FUNCTION exemplary_function()
RETURNS trigger AS
$BODY$ BEGIN
IF TG_OP = 'INSERT' OR OLD.name <> NEW.name THEN
NEW.name = 'someName';
END IF;
RETURN NEW;
END;
$BODY$
LANGUAGE plpgsql VOLATILE COST 100;
CREATE TRIGGER trigger1
BEFORE INSERT OR UPDATE
ON table1
FOR EACH ROW EXECUTE PROCEDURE exemplary_function();
and following SQL query that triggers error:
INSERT INTO table1 (name) VALUES ('other name')
It seems like parser is not stopping on TG_OP = 'INSERT' condition (and it should, because it is true) but checks another one and that triggers an error.
What's interesting, I was only able to reproduce it on version 8.4.
Postgres doesn't officially do short cuts on boolean statements (Unlike C for example)
It does say it that sometimes it can decide to short cut (see docs) but it might just easily decide to short cut on the second expression rather than the first.
It basically looks at how complicated the expressions on each side are before deciding the evaluation order. Then if that is TRUE it can decide not to bother with the other side.
In this case, it looks like its trying to interpret OLD while its still trying to decide the best order in which to evaluate the expression.
You should be able get around this by using a CASE to split the expressions eg.
IF (CASE WHEN TG_OP = 'INSERT' THEN TRUE ELSE OLD.name <> NEW.name END) THEN
Related
I'm new to PostgreSQL and really loving how constraints work with row level security, but I'm confused how to make them do what I want them to.
I have a column and I want add a constraint that creates a minimum length for a text column, this check works for that:
(length((column_name):: text) > 6)
BUT, it also then prevents users updating any rows where column_name is already under 6 characters.
I want to make it so they can't change that value TO that, but can still update a row where that is already happening, so they can change it as needed according to my new policy.
Is this possible?
BUT, it also then prevents users updating any rows where column_name is already under 6 characters.
Well, no. When you try to add that CHECK constraint, all existing rows are checked, and an exception is raised if any violation is found.
You would have to make it NOT VALID. Then yes.
You really need a trigger on INSERT or UPDATE that checks new values. Not as cheap and not as bullet-rpoof, but still pretty solid. Like:
CREATE OR REPLACE FUNCTION trg_col_min_len6()
RETURNS trigger
LANGUAGE plpgsql AS
$func$
BEGIN
IF TG_OP = 'UPDATE'
AND OLD.column_name IS NOT DISTINCT FROM NEW.column_name THEN
-- do nothing
ELSE
RAISE EXCEPTION 'New value for column "note" must have at least 6 characters.';
END IF;
RETURN NEW;
END
$func$;
-- trigger
CREATE TRIGGER tbl1_column_name_min_len6
BEFORE INSERT OR UPDATE ON tbl
FOR EACH ROW
WHEN (length(NEW.column_name) < 7)
EXECUTE FUNCTION trg_col_min_len6();
db<>fiddle here
It should be most efficient to check in a WHEN condition to the trigger directly. Then the trigger function is only ever called for short values and can be super simple.
See:
Trigger with multiple WHEN conditions
Fire trigger on update of columnA or ColumnB or ColumnC
You can create separate triggers for Insert and Update letting each completely define when it should fired. If completely different logic is required for the DML action this technique allows writing dedicated trigger functions. In this case that is not required the trigger function reduces to raise exception .... See Demo
-- Single trigger function for both Insert and Delete
create or replace function trg_col_min_len6()
returns trigger
language plpgsql
as $$
begin
raise exception 'Cannot % val = ''%''. Must have at least 6 characters.'
, tg_op, new.val;
return null;
end;
$$;
-- trigger before insert
create trigger tbl_val_min_len6_bir
before insert
on tbl
for each row
when (length(new.val) < 6)
execute function trg_col_min_len6();
-- trugger before update
create trigger tbl_val_min_len6_bur
before update
on tbl
for each row
when ( length(new.val) < 6
and new.val is distinct from old.val
)
execute function trg_col_min_len6();
Postgresql 10/11.
I need to delete row instead of update in case if target cell value is null.
So I created this trigger function:
CREATE OR REPLACE FUNCTION delete_on_update_related_table() RETURNS trigger
AS $$
DECLARE
refColumnName text = TG_ARGV[0];
BEGIN
IF TG_NARGS <> 1 THEN
RAISE EXCEPTION 'Trigger function expects 1 parameters, but got %', TG_NARGS;
END IF;
EXECUTE 'DELETE FROM ' || TG_TABLE_NAME || ' WHERE $1 = ''$2'''
USING refColumnName, OLD.id;
RETURN NULL;
END;
$$ LANGUAGE plpgsql;
And a BEFORE UPDATE trigger:
CREATE TRIGGER proper_delete
BEFORE UPDATE OF def_id
ON public.definition_products
FOR EACH ROW
WHEN (NEW.def_id IS NULL)
EXECUTE PROCEDURE delete_on_update_related_table('def_id');
Table is simple:
id uuid primary key
def_id uuid not null
Test:
UPDATE definition_products SET
def_id = NULL
WHERE id = 'f47415e8-6b00-4c65-aeb8-cadc15ca5890';
-- rows affected 0
Documentation says:
Row-level triggers fired BEFORE can return null to signal the trigger
manager to skip the rest of the operation for this row (i.e.,
subsequent triggers are not fired, and the INSERT/UPDATE/DELETE does
not occur for this row).
Previously, I used a RULE instead of the trigger. But there is no way to use WHERE & RETURNING clause in same rule.
You need an unconditional ON UPDATE DO INSTEAD rule with a RETURNING clause
So, is there a way?
While Jeremy's answer is good, there is still room for improvement.
Problems
You need to be very accurate in the definition of the objective. Your statement:
I need to delete row instead of update in case if target cell value is null.
... does not imply that the column was changed to NULL in the UPDATE at hand. Might have been NULL before, like, before you implemented the trigger. So not:
BEFORE UPDATE OF def_id ON public.definition_products
But just:
BEFORE UPDATE ON public.definition_products
Of course, if the column is defined NOT NULL (as it probably should be), there is no effective difference - except for the noise and an additional point of failure. The manual:
A column-specific trigger (one defined using the UPDATE OFcolumn_name syntax) will fire when any of its columns are listed as targets in the UPDATE command's SET list. It is possible for a column's value to change even when the trigger is not fired, because changes made to the row's contents by BEFORE UPDATE triggers are not considered.
Also, nothing in your question indicates the need for dynamic SQL. (That would be the case if you wanted to reuse the same trigger function for multiple triggers on different tables. And even then it's often better to just create several distinct trigger functions for multiple reason: simpler, faster, less error-prone, easier to read & maintain, ...)
As for "error-prone": your original dynamic statement was just invalid:
EXECUTE 'DELETE FROM ' || TG_TABLE_NAME || ' WHERE $1 = ''$2'''
USING refColumnName, OLD.id;
Can't pass a column name as value (refColumnName).
Can't put single quotes around $2, which is passed as value and hence needs no quoting.
An unqualified, unquoted TG_TABLE_NAME can go terribly wrong, which is especially critical for a heavy-weight function that deletes rows.
Jeremy's version fixes most, but still features the unqualified TG_TABLE_NAME.
This would be good:
EXECUTE format('DELETE FROM %s WHERE %I = $1', TG_RELID::regclass, refColumnName) -- refColumnName still unquoted
USING OLD.id;
Or:
EXECUTE format('DELETE FROM %I.%I WHERE %I = $1', TG_TABLE_SCHEMA, TG_TABLE_NAME, refColumnName)
USING OLD.id;
Related:
Why does a PostgreSQL SELECT query return different results when a schema name is specified?
Table name as a PostgreSQL function parameter
Solution
Simpler trigger function:
CREATE OR REPLACE FUNCTION delete_on_update_related_table()
RETURNS trigger AS
$func$
BEGIN
DELETE FROM public.definition_products WHERE id = OLD.id; -- def_id?
RETURN NULL;
END
$func$ LANGUAGE plpgsql;
Simpler trigger:
CREATE TRIGGER proper_delete
BEFORE UPDATE ON public.definition_products
FOR EACH ROW
WHEN (NEW.def_id IS NULL) -- that's the defining condition!
EXECUTE PROCEDURE delete_on_update_related_table(); -- no parameter
You probably want to use OLD.id, not OLD.def_id. (The row to delete is best defined by it's PK, not by the column changed to NULL.) But that's not entirely clear.
This works for me, with a few small changes:
CREATE OR REPLACE FUNCTION delete_on_update_related_table() RETURNS trigger
AS $$
DECLARE
refColumnName text = quote_ident(TG_ARGV[0]);
BEGIN
IF TG_NARGS <> 1 THEN RAISE EXCEPTION 'Trigger function expects 1 parameters, but got %', TG_NARGS; END IF;
EXECUTE format('DELETE FROM %s WHERE %s = %s', quote_ident(TG_TABLE_NAME), refColumnName, quote_literal(OLD.id));
RETURN NULL;
END;
$$ LANGUAGE plpgsql;
-- create trigger
CREATE TRIGGER proper_delete
BEFORE UPDATE OF def_id
ON public.definition_products
FOR EACH ROW
WHEN (NEW.def_id IS NULL)
EXECUTE PROCEDURE delete_on_update_related_table('id'); --Note id, not def_id
I want to create a trigger which counts rows and updates a field in an other table. My current solution works for INSERT statements but failes when I DELETE a row.
My current function:
CREATE OR REPLACE FUNCTION update_table_count()
RETURNS trigger AS
$$
DECLARE updatecount INT;
BEGIN
Select count(*) into updatecount
From source_table
Where id = new.id;
Update dest_table set count=updatecount
Where id = new.id;
RETURN NEW;
END;
$$
LANGUAGE 'plpgsql';
The trigger is a pretty basic one, looking like.
CREATE TRIGGER count_trigger
AFTER INSERT OR DELETE
ON source_table
FOR EACH ROW
EXECUTE PROCEDURE update_table_count();
When I excute a DELETE statement the following error occurs:
ERROR: record "new" is not assigned yet
DETAIL: The tuple structure of a not-yet-assigned record is indeterminate.
I know one solution could be to create just one set of trigger and function for the DELETE and one for the INSERT statement. But I want to do it a bit more elegant and want to know, if there is a solution to check if NEW or OLD is present in the current context and just implement an IF ELSE block. But I dont know how to check for this context sensitive items.
Thanks for your help
The usual approach to make a trigger function do different things depending on how the trigger was fired is to check the trigger operation through TG_OP
CREATE OR REPLACE FUNCTION update_table_count()
RETURNS trigger AS
$$
DECLARE
updatecount INT;
BEGIN
if tg_op = 'UPDATE' then
select count(*) into updatecount from source_table where id = new.id;
update dest_table set count=updatecount where id = new.id;
elsif tg_op = 'DELETE' then
... do something else
end if;
RETURN NEW;
END;
$$
LANGUAGE plpgsql;
Unrelated, but: the language name is an identifier. Do not quote it using single quotes.
From PostgreSQL's documentation:
NEW
Data type RECORD; variable holding the new database row for INSERT/UPDATE operations in row-level triggers. This variable is null in statement-level triggers and for DELETE operations.
OLD
Data type RECORD; variable holding the old database row for UPDATE/DELETE operations in row-level triggers. This variable is null in statement-level triggers and for INSERT operations.
So, for example, if NEW is NULL, then the trigger was invoked on DELETE.
I need to check if a value inside a jsonb is already present in a array. I'm trying to achieve this with a trigger but i'm new to this language and I don't know how to write the query.
CREATE TABLE merchants (
key uuid PRIMARY KEY,
data jsonb NOT NULL
)
Here is the trigger. I think the NEW.data.ids part is wrong.
CREATE FUNCTION validate_id_constraint() returns trigger as $$
DECLARE merchants_count int;
BEGIN
merchants_count := (SELECT count(*) FROM merchants WHERE data->'ids' #> NEW.data.ids);
IF (merchants_count != 0) THEN
RAISE EXCEPTION 'Duplicate id';
END IF;
RETURN NEW;
END;
$$ language plpgsql;
CREATE TRIGGER validate_id_constraint_trigger BEFORE INSERT OR UPDATE ON merchants
FOR EACH ROW EXECUTE PROCEDURE validate_id_constraint();
When i insert into the table i get this error message
ERROR: missing FROM-clause entry for table "data"
LINE 1: ...LECT count(*) FROM merchants WHERE data->'ids' #> NEW.data.i...
^
I have done the query outside the trigger and it works fine
SELECT count(*) FROM merchants WHERE data->'ids' #> '["11176", "11363"]'
You get that error because you are using . instead of -> to extract the ids array in the expression NEW.data.ids.
But your trigger won't work anyway because you are not trying to avoid containment, but overlaps in the arrays.
One way you could write the trigger function is:
CREATE OR REPLACE FUNCTION validate_id_constraint() RETURNS trigger
LANGUAGE plpgsql AS
$$DECLARE
j jsonb;
BEGIN
FOR j IN
SELECT jsonb_array_elements(NEW.data->'ids')
LOOP
IF EXISTS
(SELECT 1 FROM merchants WHERE j <# (data->'ids'))
THEN
RAISE EXCEPTION 'Duplicate IDs';
END IF;
END LOOP;
RETURN NEW;
END;$$;
You have to loop because there is no “overlaps” operator on jsonb arrays.
This is all slow and cumbersome because of your table design.
Note 1: You would be much better off if you only store data in jsonb that you do not need to manipulate in the database. In particular, you should store the ids array as field in the table. Then you can use the “overlaps” operator && and speed that up with a gin index.
You would be even faster if you normalized the table structure and stored the individual array entries in a separate table, then a regular unique constraint would do.
Note 2: Any constraint enabled by a trigger suffers from a race condition: if two concurrent INSERTs conflict with each other, the trigger function will not see the values from the concurrent INSERT and you may end up with inconsistent data.
I have a table called assignments. I would like to be able to read/write to all the columns in this table using either assignments.column or homework.column, how can I do this?
I know this is not something you would normally do. I need to be able to do this to provide backwards compatibility for a short period of time.
We have an iOS app that currently does direct postgresql queries against the DB. We're updating all of our apps to use an API. In the process of building the API the developer decided to change the name of the tables because we (foolishly) thought we didn't need backwards compatibility.
Now, V1.0 and the API both need to be able to write to this table so I don't have to do some voodoo later to transfer/combine data later...
We're using Ruby on Rails for the API.
With Postgres 9.3 the following should be enough:
CREATE VIEW homework AS SELECT * FROM assignments;
It works because simple views are automatically updatable (see docs).
In Postgres 9.3 or later, a simple VIEW is "updatable" automatically. The manual:
Simple views are automatically updatable: the system will allow
INSERT, UPDATE and DELETE statements to be used on the view in
the same way as on a regular table. A view is automatically updatable
if it satisfies all of the following conditions:
The view must have exactly one entry in its FROM list, which must be a table or another updatable view.
The view definition must not contain WITH, DISTINCT, GROUP BY, HAVING, LIMIT, or OFFSET clauses at the top level.
The view definition must not contain set operations (UNION, INTERSECT or EXCEPT) at the top level.
The view's select list must not contain any aggregates, window functions or set-returning functions.
If one of these conditions is not met (or for the now outdated Postgres 9.2 or older), a manual setup may do the job.
Building on your work in progress:
Trigger function
CREATE OR REPLACE FUNCTION trg_ia_insupdel()
RETURNS trigger
LANGUAGE plpgsql AS
$func$
DECLARE
_tbl CONSTANT regclass := 'iassignments_assignments';
_cols text;
_vals text;
BEGIN
CASE TG_OP
WHEN 'INSERT' THEN
INSERT INTO iassignments_assignments
VALUES (NEW.*);
RETURN NEW;
WHEN 'UPDATE' THEN
SELECT INTO _cols, _vals
string_agg(quote_ident(attname), ', ') -- incl. pk col!
, string_agg('n.' || quote_ident(attname), ', ')
FROM pg_attribute
WHERE attrelid = _tbl -- _tbl converted to oid automatically
AND attnum > 0 -- no system columns
AND NOT attisdropped; -- no dropped (dead) columns
EXECUTE format('
UPDATE %s t
SET (%s) = (%s)
FROM (SELECT ($1).*) n
WHERE t.published_assignment_id
= ($2).published_assignment_id' -- match to OLD value of pk
, _tbl, _cols, _vals) -- _tbl converted to text automatically
USING NEW, OLD;
RETURN NEW;
WHEN 'DELETE' THEN
DELETE FROM iassignments_assignments
WHERE published_assignment_id = OLD.published_assignment_id;
RETURN OLD;
END CASE;
RETURN NULL; -- control should never reach this
END
$func$;
Trigger
CREATE TRIGGER insupbef
INSTEAD OF INSERT OR UPDATE OR DELETE ON assignments_published
FOR EACH ROW EXECUTE PROCEDURE trg_ia_insupdel();
Notes
assignments_published must be a VIEW, an INSTEAD OF trigger is only allowed for views.
Dynamic SQL (in the UPDATE section) is not strictly necessary, only to cover future changes to the table layout automatically. The names of table and PK are still hard coded.
Simpler and probably cheaper without sub-block (like you had).
Using (SELECT ($1).*) instead of the shorter VALUES ($1.*) to preserve column names.
My naming convention: I prepend trg_ for trigger functions, followed by an abbreviation indicating the target table and finally one or more of the the tokens ins, up and del for INSERT, UPDATE and DELETE respectively. The name of the trigger is a copy of the function name, stripped of the first two parts. This is purely a matter of convention and taste but has proven useful for me since the names tell the purpose and are still short.
More explanation in the related answer that has already been mentioned:
Update multiple columns in a trigger function in plpgsql
This is where I am with the trigger functions so far, any feedback would be greatly appreciated. It's a combination of http://vibhorkumar.wordpress.com/2011/10/28/instead-of-trigger/ and Update multiple columns in a trigger function in plpgsql
Table: iassignments_assignments
Columns:
published_assignment_id
name
filepath
filename
link
teacher
due date
description
published
classrooms
View: assignments_published - SELECT * FROM iassignments_assignments
Trigger Function for assignments_published
CREATE OR REPLACE FUNCTION assignments_published_trigger_func()
RETURNS TRIGGER
LANGUAGE plpgsql
AS $function$
BEGIN
IF TG_OP = 'INSERT' THEN
EXECUTE format('INSERT INTO %s SELECT ($1).*', 'iassignments_assignments')
USING NEW;
RETURN NEW;
ELSIF TG_OP = 'UPDATE' THEN
DECLARE
tbl = 'iassignments_assignments';
cols text;
vals text;
BEGIN
SELECT INTO cols, vals
string_agg(quote_ident(attname), ', ')
,string_agg('x.' || quote_ident(attname), ', ')
FROM pg_attribute
WHERE attrelid = tbl
AND NOT attisdropped -- no dropped (dead) columns
AND attnum > 0; -- no system columns
EXECUTE format('
UPDATE %s t
SET (%s) = (%s)
FROM (SELECT ($1).*) x
WHERE t.published_assignment_id = ($2).published_assignment_id'
, tbl, cols, vals)
USING NEW, OLD;
RETURN NEW;
END
ELSIF TG_OP = 'DELETE' THEN
DELETE FROM iassignments_assignments WHERE published_assignment_id=OLD.published_assignment_id;
RETURN NULL;
END IF;
RETURN NEW;
END;
$function$;
Trigger
CREATE TRIGGER assignments_published_trigger
INSTEAD OF INSERT OR UPDATE OR DELETE ON
assignments_published FOR EACH ROW EXECUTE PROCEDURE assignments_published_trigger_func();
Table: iassignments_classes
Columns:
class_assignment_id
guid
assignment_published_id
View: assignments_class - SELECT * FROM assignments_classes
Trigger Function for assignments_class
**I'll create this function once I have received feedback on the other and know it's create, so I (hopefully) need very little changes to this function.
Trigger
CREATE TRIGGER assignments_class_trigger
INSTEAD OF INSERT OR UPDATE OR DELETE ON
assignments_class FOR EACH ROW EXECUTE PROCEDURE assignments_class_trigger_func();