PL/Ruby is a loadable procedural language for the Postgres database system that enable the Ruby language to create functions and trigger procedures
Functions and triggers are singleton methods of the module PLtemp.
All arguments (to the function or the triggers) are passed as string values, except for NULL values represented by Qnil. You must explicitely call a conversion function (like to_i) if you want to use an argument as an integer
To create a function in the PL/Ruby language use the syntax
CREATE FUNCTION funcname(arguments_type) RETURNS type AS ' # PL/Ruby function body ' LANGUAGE 'plruby';
when calling the function in a query, the arguments are given as string values in the array args. To create a little max function returning the higher of two int4 values write :
CREATE FUNCTION ruby_max(int4, int4) RETURNS int4 AS '
if args[0].to_i > args[1].to_i
return args[0]
else
return args[1]
end
' LANGUAGE 'plruby';
Tuple arguments are given as hash. Here is an example that defines the overpaid_2 function (as found in the older Postgres documentation) in PL/Ruby.
CREATE FUNCTION overpaid_2 (EMP) RETURNS bool AS '
args[0]["salary"].to_f > 200000 ||
(args[0]["salary"].to_f > 100000 && args[0]["age"].to_i < 30)
' LANGUAGE 'plruby';
For example to define a function (int4[], int4) and return int4[], in version < 7.4 you write
CREATE FUNCTION ruby_int4_accum(_int4, int4) RETURNS _int4 AS '
if /\\{(\\d+),(\\d+)\\}/ =~ args[0]
a, b = $1, $2
newsum = a.to_i + args[1].to_i
newcnt = b.to_i + 1
else
raise "unexpected value #{args[0]}"
end
"{#{newsum},#{newcnt}}"
' LANGUAGE 'plruby';
This must now (>= 7.4) be written
CREATE FUNCTION ruby_int4_accum(_int4, int4) RETURNS _int4 AS ' a = args[0] [a[0].to_i + args[1].to_i, a[1].to_i + 1] ' LANGUAGE 'plruby';
The return type must be declared as SETOF
The function must call yield to return rows or return a String which must be a valid SELECT statement
For example to concatenate 2 rows create the function
plruby_test=# CREATE FUNCTION tu(varchar) RETURNS setof record
plruby_test-# AS '
plruby_test'# size = PL.column_name(args[0]).size
plruby_test'# res = nil
plruby_test'# PL::Plan.new("select * from #{args[0]}",
plruby_test'# "block" => 50).each do |row|
plruby_test'# if res.nil?
plruby_test'# res = row.values
plruby_test'# else
plruby_test'# res.concat row.values
plruby_test'# yield res
plruby_test'# res = nil
plruby_test'# end
plruby_test'# end
plruby_test'# if res
plruby_test'# res.concat Array.new(size)
plruby_test'# yield res
plruby_test'# end
plruby_test'# ' language 'plruby';
CREATE FUNCTION
plruby_test=#
plruby_test=# select * from tt;
a | b
---+----
1 | 2
3 | 4
5 | 6
7 | 8
9 | 10
(5 rows)
plruby_test=# select * from tu('tt') as tbl(a int, b int, c int, d int);
a | b | c | d
---+----+---+---
1 | 2 | 3 | 4
5 | 6 | 7 | 8
9 | 10 | |
(3 rows)
plruby_test=#
Trigger procedures are defined in Postgres as functions without arguments and a return type of trigger. In PL/Ruby the procedure is called with 4 arguments :
The return value from a trigger procedure is one of the constant PL::OK or PL::SKIP, or an hash. If the return value is PL::OK, the normal operation (INSERT/UPDATE/DELETE) that fired this trigger will take place. Obviously, PL::SKIP tells the trigger manager to silently suppress the operation. The hash tells PL/Ruby to return a modified row to the trigger manager that will be inserted instead of the one given in new (INSERT/UPDATE only). Needless to say that all this is only meaningful when the trigger is BEFORE and FOR EACH ROW.
Here's a little example trigger procedure that forces an integer value in a table to keep track of the # of updates that are performed on the row. For new row's inserted, the value is initialized to 0 and then incremented on every update operation :
CREATE FUNCTION trigfunc_modcount() RETURNS TRIGGER AS '
case tg["op"]
when PL::INSERT
new[args[0]] = 0
when PL::UPDATE
new[args[0]] = old[args[0]].to_i + 1
else
return PL::OK
end
new
' LANGUAGE 'plruby';
CREATE TABLE mytab (num int4, modcnt int4, descr text);
CREATE TRIGGER trig_mytab_modcount BEFORE INSERT OR UPDATE ON mytab
FOR EACH ROW EXECUTE PROCEDURE trigfunc_modcount('modcnt');
A more complex example (extract from test_setup.sql in the distribution) which use the global variable $Plans to store a prepared plan
create function trig_pkey2_after() returns trigger as '
if ! $Plans.key?("plan_dta2_upd")
$Plans["plan_dta2_upd"] =
PL::Plan.new("update T_dta2
set ref1 = $3, ref2 = $4
where ref1 = $1 and ref2 = $2",
["int4", "varchar", "int4", "varchar" ]).save
$Plans["plan_dta2_del"] =
PL::Plan.new("delete from T_dta2
where ref1 = $1 and ref2 = $2",
["int4", "varchar"]).save
end
old_ref_follow = false
old_ref_delete = false
case tg["op"]
when PL::UPDATE
new["key2"] = new["key2"].upcase
old_ref_follow = (new["key1"] != old["key1"]) ||
(new["key2"] != old["key2"])
when PL::DELETE
old_ref_delete = true
end
if old_ref_follow
n = $Plans["plan_dta2_upd"].exec([old["key1"], old["key2"], new["key1"],
new["key2"]])
warn "updated #{n} entries in T_dta2 for new key in T_pkey2" if n != 0
end
if old_ref_delete
n = $Plans["plan_dta2_del"].exec([old["key1"], old["key2"]])
warn "deleted #{n} entries from T_dta2" if n != 0
end
PL::OK
' language 'plruby';
create trigger pkey2_after after update or delete on T_pkey2
for each row execute procedure
trig_pkey2_after();
Sometime it can be usefull to define methods (in pure Ruby) which can be called from a PL/Ruby function or a PL/Ruby trigger.
In this case, you have 2 possibilities
:-) :-) :-)just close the current definition of the function (or trigger) with a
end and define your singleton method without the final end
Here a small and useless example
toto=> CREATE FUNCTION tutu() RETURNS int4 AS ' toto'> toto(1, 3) + toto(4, 4) toto'> end toto'> toto'> def PLtemp.toto(a, b) toto'> a + b toto'> ' LANGUAGE 'plruby'; CREATE toto=> select tutu(); tutu ---- 12 (1 row) toto=>
At load time, PL/Ruby look if it exist a table plruby_singleton_methods and if found try, for each row, to define singleton methods with the template :
def PLtemp.#{name} (#{args})
#{body}
end
The previous example can be written (you have a more complete example in test/plp/test_setup.sql)
toto=> SELECT * FROM plruby_singleton_methods; name|args|body ----+----+----- toto|a, b|a + b (1 row) toto=> CREATE FUNCTION tutu() RETURNS int4 AS ' toto'> toto(1, 3) + toto(4, 4) toto'> ' LANGUAGE 'plruby'; CREATE toto=> select tutu(); tutu ---- 12 (1 row) toto=>
warn [level], messageRuby interface to PostgreSQL elog()
Possible value for level are NOTICE, DEBUG and NOIND
Use raise() if you want to simulate elog(ERROR, "...")
$Plans (hash, tainted)args_typecolumn_name(table)column_type(table)quote(string)result_nameresult_typeresult_sizeresult_descriptionexec(string [, count [, type]])spi_exec(string [, count [, type]])If the query is a SELECT statement, an array is return (if count is not specified or with a value > 1). Each element of this array is an hash where the key is the column name. For example this procedure display all rows in the table pg_table.
CREATE FUNCTION pg_table_dis() RETURNS int4 AS '
res = PLruby.exec("select * from pg_class")
res.each do |x|
warn "======================"
x.each do |y, z|
warn "name = #{y} -- value = #{z}"
end
warn "======================"
end
return res.size
' LANGUAGE 'plruby';
if type is specified it can take the value
Example :
create table T_pkey1 (
skey1 int4,
skey2 varchar(20),
stxt varchar(40)
);
create function toto() returns bool as '
warn("=======")
PL.exec("select * from T_pkey1", 1, "hash") do |a|
warn(a.inspect)
end
warn("=======")
PL.exec("select * from T_pkey1", 1, "array") do |a|
warn(a.inspect)
end
warn("=======")
PL.exec("select * from T_pkey1", 1) do |a|
warn(a.inspect)
end
warn("=======")
return true
' language 'plruby';
plruby_test=# select toto();
NOTICE: =======
NOTICE: {"name"=>"skey1", "typeid"=>23, "type"=>"int4", "value"=>"12", "len"=>4}
NOTICE: {"name"=>"skey2", "typeid"=>1043, "type"=>"varchar", "value"=>"a", "len"=>20}
NOTICE: {"name"=>"stxt", "typeid"=>1043, "type"=>"varchar", "value"=>"b", "len"=>40}
NOTICE: =======
NOTICE: ["skey1", "12", "int4", 4, 23]
NOTICE: ["skey2", "a", "varchar", 20, 1043]
NOTICE: ["stxt", "b", "varchar", 40, 1043]
NOTICE: =======
NOTICE: ["skey1", "12"]
NOTICE: ["skey2", "a"]
NOTICE: ["stxt", "b"]
NOTICE: =======
toto
------
t
(1 row)
plruby_test=#
A block can be specified, in this case a call to yield() will be made.
If count is specified with the value 1, only the first row (or FALSE if it fail) is returned as a hash. Here a little example :
CREATE FUNCTION pg_table_dis() RETURNS int4 AS '
PL.exec("select * from pg_class", 1) { |y, z|
warn "name = #{y} -- value = #{z}"
}
return 1
' LANGUAGE 'plruby';
prepare(string[, types])spi_prepare(string[, types])prepare(string, "types" => types, "count" => count, "output" => type, "tmp" => true)Deprecated : See PL::Plan::new and PL::Plan#save
Prepares AND SAVES a query plan for later execution. It is a bit different from the C level SPI_prepare in that the plan is automatically copied to the toplevel memory context.
If the query references arguments, the type names must be given as a Ruby array of strings. The return value from prepare is a PL::Plan object to be used in subsequent calls to PL::Plan#exec.
If the hash given has the keys count, output these values will be given to the subsequent calls to each
initialize(string, "types" => types, "count" => count, "output" => type, "save" => false)Prepares a query plan for later execution.
If the query references arguments, the type names must be given as a Ruby array of strings.
If the hash given has the keys output, count these values will be given to the subsequent calls to each
If "save" as a true value, the plan will be saved
exec(values, [count [, type]])execp(values, [count [, type]])exec("values" => values, "count" => count, "output" => type)execp("values" => values, "count" => count, "output" => type)Execute a prepared plan from PL::PLan::new with variable substitution. The optional count value tells PL::Plan#exec the maximum number of rows to be processed by the query.
If there was a typelist given to PL::Plan::new, an array of values of exactly the same length must be given to PL::Plan#exec as first argument. If the type list on PL::Plan::new was empty, this argument must be omitted.
If the query is a SELECT statement, the same as described for PL#exec happens for the loop-body and the variables for the fields selected.
If type is specified it can take the values
Here's an example for a PL/Ruby function using a prepared plan :
CREATE FUNCTION t1_count(int4, int4) RETURNS int4 AS '
if ! $Plans.key?("plan")
# prepare the saved plan on the first call
$Plans["plan"] = PL::Plan.new("SELECT count(*) AS cnt FROM t1
WHERE num >= $1 AND num <= $2",
["int4", "int4"]).save
end
n = $Plans["plan"].exec([args[0], args[1]], 1)
n["cnt"]
' LANGUAGE 'plruby';cursor(name = nil, "values" => values, "output" => type)Create a new object PL::Cursor
If output is specified it can take the values
If there was a typelist given to PL::Plan::new, an array of values of exactly the same length must be given to PL::Plan#cursor
each(values, [count [, type ]]) { ... }fetch(values, [count [, type ]]) { ... }each("values" => values, "count" => count, "output" => type) { ... }fetch("values" => values, "count" => count, "output" => type) { ... }Same then #exec but a call to SPI_cursor_open(), SPI_cursor_fetch() is made.
Can be used only with a block and a SELECT statement
create function toto() returns bool as '
plan = PL::Plan.new("select * from T_pkey1")
warn "=====> ALL"
plan.each do |x|
warn(x.inspect)
end
warn "=====> FIRST 2"
plan.each("count" => 2) do |x|
warn(x.inspect)
end
return true
' language 'plruby';
plruby_test=# select * from T_pkey1;
skey1 | skey2 | stxt
-------+-------+------
12 | a | b
24 | c | d
36 | e | f
(3 rows)
plruby_test=#
plruby_test=# select toto();
NOTICE: =====> ALL
NOTICE: {"skey1"=>"12", "skey2"=>"a", "stxt"=>"b"}
NOTICE: {"skey1"=>"24", "skey2"=>"c", "stxt"=>"d"}
NOTICE: {"skey1"=>"36", "skey2"=>"e", "stxt"=>"f"}
NOTICE: =====> FIRST 2
NOTICE: {"skey1"=>"12", "skey2"=>"a", "stxt"=>"b"}
NOTICE: {"skey1"=>"24", "skey2"=>"c", "stxt"=>"d"}
toto
------
t
(1 row)
plruby_test=# releasesavecloseeach {|row| ... }fetch(count = 1)row(count = 1)Fetches some rows from a cursor
if count > 0 fetch forward else backward
move(count)reverse_each {|row| ... }rewind