1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
|
--
-- First, create a slave database and define the functions and test data
-- therein.
--
-- This initial hackery is to allow successive runs without failures.
--
-- Adjust this setting to control where the objects get created.
SET search_path = public;
SET autocommit TO 'on';
CREATE OR REPLACE FUNCTION conditional_drop()
RETURNS text AS '
DECLARE
dbname text;
BEGIN
SELECT INTO dbname datname FROM pg_database WHERE datname = ''regression_slave'';
IF FOUND THEN
DROP DATABASE regression_slave;
END IF;
RETURN ''OK'';
END;
' LANGUAGE 'plpgsql';
SELECT conditional_drop();
CREATE DATABASE regression_slave;
\connect regression_slave
-- Turn off echoing so that expected file does not depend on
-- contents of dblink.sql.
\set ECHO none
SET autocommit TO 'on';
\i dblink.sql
\set ECHO all
CREATE TABLE foo(f1 int, f2 text, f3 text[], primary key (f1,f2));
INSERT INTO foo VALUES (0,'a','{"a0","b0","c0"}');
INSERT INTO foo VALUES (1,'b','{"a1","b1","c1"}');
INSERT INTO foo VALUES (2,'c','{"a2","b2","c2"}');
INSERT INTO foo VALUES (3,'d','{"a3","b3","c3"}');
INSERT INTO foo VALUES (4,'e','{"a4","b4","c4"}');
INSERT INTO foo VALUES (5,'f','{"a5","b5","c5"}');
INSERT INTO foo VALUES (6,'g','{"a6","b6","c6"}');
INSERT INTO foo VALUES (7,'h','{"a7","b7","c7"}');
INSERT INTO foo VALUES (8,'i','{"a8","b8","c8"}');
INSERT INTO foo VALUES (9,'j','{"a9","b9","c9"}');
-- misc utilities
-- show the currently executing query
SELECT 'hello' AS hello, dblink_current_query() AS query;
-- list the primary key fields
SELECT *
FROM dblink_get_pkey('foo');
-- build an insert statement based on a local tuple,
-- replacing the primary key values with new ones
SELECT dblink_build_sql_insert('foo','1 2',2,'{"0", "a"}','{"99", "xyz"}');
-- build an update statement based on a local tuple,
-- replacing the primary key values with new ones
SELECT dblink_build_sql_update('foo','1 2',2,'{"0", "a"}','{"99", "xyz"}');
-- build a delete statement based on a local tuple,
SELECT dblink_build_sql_delete('foo','1 2',2,'{"0", "a"}');
--
-- Connect back to the regression database and define the functions.
-- Turn off echoing so that expected file does not depend on
-- contents of dblink.sql.
--
\connect regression
\set ECHO none
SET autocommit TO 'on';
\i dblink.sql
\set ECHO all
-- regular old dblink
SELECT *
FROM dblink('dbname=regression_slave','SELECT * FROM foo') AS t(a int, b text, c text[])
WHERE t.a > 7;
-- should generate "no connection available" error
SELECT *
FROM dblink('SELECT * FROM foo') AS t(a int, b text, c text[])
WHERE t.a > 7;
-- create a persistent connection
SELECT dblink_connect('dbname=regression_slave');
-- use the persistent connection
SELECT *
FROM dblink('SELECT * FROM foo') AS t(a int, b text, c text[])
WHERE t.a > 7;
-- open a cursor
SELECT dblink_open('rmt_foo_cursor','SELECT * FROM foo');
-- fetch some data
SELECT *
FROM dblink_fetch('rmt_foo_cursor',4) AS t(a int, b text, c text[]);
SELECT *
FROM dblink_fetch('rmt_foo_cursor',4) AS t(a int, b text, c text[]);
-- this one only finds two rows left
SELECT *
FROM dblink_fetch('rmt_foo_cursor',4) AS t(a int, b text, c text[]);
-- close the cursor
SELECT dblink_close('rmt_foo_cursor');
-- should generate "cursor rmt_foo_cursor does not exist" error
SELECT *
FROM dblink_fetch('rmt_foo_cursor',4) AS t(a int, b text, c text[]);
-- close the persistent connection
SELECT dblink_disconnect();
-- should generate "no connection available" error
SELECT *
FROM dblink('SELECT * FROM foo') AS t(a int, b text, c text[])
WHERE t.a > 7;
-- put more data into our slave table, first using arbitrary connection syntax
-- but truncate the actual return value so we can use diff to check for success
SELECT substr(dblink_exec('dbname=regression_slave','SET autocommit TO ''on'';INSERT INTO foo VALUES(10,''k'',''{"a10","b10","c10"}'')'),1,6);
-- create a persistent connection
SELECT dblink_connect('dbname=regression_slave');
-- put more data into our slave table, using persistent connection syntax
-- but truncate the actual return value so we can use diff to check for success
SELECT substr(dblink_exec('INSERT INTO foo VALUES(11,''l'',''{"a11","b11","c11"}'')'),1,6);
-- let's see it
SELECT *
FROM dblink('SELECT * FROM foo') AS t(a int, b text, c text[]);
-- change some data
SELECT dblink_exec('UPDATE foo SET f3[2] = ''b99'' WHERE f1 = 11');
-- let's see it
SELECT *
FROM dblink('SELECT * FROM foo') AS t(a int, b text, c text[])
WHERE a = 11;
-- delete some data
SELECT dblink_exec('DELETE FROM foo WHERE f1 = 11');
-- let's see it
SELECT *
FROM dblink('SELECT * FROM foo') AS t(a int, b text, c text[])
WHERE a = 11;
-- close the persistent connection
SELECT dblink_disconnect();
|