Archive for March, 2022
MySQL 5-Table Procedure
A student wanted a better example of writing a MySQL Persistent Stored Module (PSM) that maintains transactional scope across a couple tables. Here’s the one I wrote about ten years ago to build the MySQL Video Store model. It looks I neglected to put it out there before, so here it is for reference.
-- Conditionally drop procedure if it exists. DROP PROCEDURE IF EXISTS contact_insert; -- Reset the delimiter so that a semicolon can be used as a statement and block terminator. DELIMITER $$ SELECT 'CREATE PROCEDURE contact_insert' AS "Statement"; CREATE PROCEDURE contact_insert ( pv_member_type CHAR(12) , pv_account_number CHAR(19) , pv_credit_card_number CHAR(19) , pv_credit_card_type CHAR(12) , pv_first_name CHAR(20) , pv_middle_name CHAR(20) , pv_last_name CHAR(20) , pv_contact_type CHAR(12) , pv_address_type CHAR(12) , pv_city CHAR(30) , pv_state_province CHAR(30) , pv_postal_code CHAR(20) , pv_street_address CHAR(30) , pv_telephone_type CHAR(12) , pv_country_code CHAR(3) , pv_area_code CHAR(6) , pv_telephone_number CHAR(10)) MODIFIES SQL DATA BEGIN /* Declare variables to manipulate auto generated sequence values. */ DECLARE member_id int unsigned; DECLARE contact_id int unsigned; DECLARE address_id int unsigned; DECLARE street_address_id int unsigned; DECLARE telephone_id int unsigned; /* Declare local constants for who-audit columns. */ DECLARE lv_created_by int unsigned DEFAULT 1001; DECLARE lv_creation_date DATE DEFAULT UTC_DATE(); DECLARE lv_last_updated_by int unsigned DEFAULT 1001; DECLARE lv_last_update_date DATE DEFAULT UTC_DATE(); /* Declare a locally scoped variable. */ DECLARE duplicate_key INT DEFAULT 0; /* Declare a duplicate key handler */ DECLARE CONTINUE HANDLER FOR 1062 SET duplicate_key = 1; /* Start the transaction context. */ START TRANSACTION; /* Create a SAVEPOINT as a recovery point. */ SAVEPOINT all_or_none; /* Insert into the first table in sequence based on inheritance of primary keys by foreign keys. */ INSERT INTO member ( member_type , account_number , credit_card_number , credit_card_type , created_by , creation_date , last_updated_by , last_update_date ) VALUES ((SELECT common_lookup_id FROM common_lookup WHERE common_lookup_context = 'MEMBER' AND common_lookup_type = pv_member_type) , pv_account_number , pv_credit_card_number ,(SELECT common_lookup_id FROM common_lookup WHERE common_lookup_context = 'MEMBER' AND common_lookup_type = pv_credit_card_type) , lv_created_by , lv_creation_date , lv_last_updated_by , lv_last_update_date ); /* Preserve the sequence by a table related variable name. */ SET member_id = last_insert_id(); /* Insert into the first table in sequence based on inheritance of primary keys by foreign keys. */ INSERT INTO contact VALUES ( null , member_id ,(SELECT common_lookup_id FROM common_lookup WHERE common_lookup_context = 'CONTACT' AND common_lookup_type = pv_contact_type) , pv_first_name , pv_middle_name , pv_last_name , lv_created_by , lv_creation_date , lv_last_updated_by , lv_last_update_date ); /* Preserve the sequence by a table related variable name. */ SET contact_id = last_insert_id(); /* Insert into the first table in sequence based on inheritance of primary keys by foreign keys. */ INSERT INTO address VALUES ( null , last_insert_id() ,(SELECT common_lookup_id FROM common_lookup WHERE common_lookup_context = 'MULTIPLE' AND common_lookup_type = pv_address_type) , pv_city , pv_state_province , pv_postal_code , lv_created_by , lv_creation_date , lv_last_updated_by , lv_last_update_date ); /* Preserve the sequence by a table related variable name. */ SET address_id = last_insert_id(); /* Insert into the first table in sequence based on inheritance of primary keys by foreign keys. */ INSERT INTO street_address VALUES ( null , last_insert_id() , pv_street_address , lv_created_by , lv_creation_date , lv_last_updated_by , lv_last_update_date ); /* Insert into the first table in sequence based on inheritance of primary keys by foreign keys. */ INSERT INTO telephone VALUES ( null , contact_id , address_id ,(SELECT common_lookup_id FROM common_lookup WHERE common_lookup_context = 'MULTIPLE' AND common_lookup_type = pv_telephone_type) , pv_country_code , pv_area_code , pv_telephone_number , lv_created_by , lv_creation_date , lv_last_updated_by , lv_last_update_date); /* This acts as an exception handling block. */ IF duplicate_key = 1 THEN /* This undoes all DML statements to this point in the procedure. */ ROLLBACK TO SAVEPOINT all_or_none; END IF; /* This commits the write when successful and is harmless otherwise. */ COMMIT; END; $$ -- Reset the standard delimiter to let the semicolon work as an execution command. DELIMITER ; |
You can then call the procedure, like:
SELECT 'CALL contact_insert() PROCEDURE 5 times' AS "Statement"; CALL contact_insert('INDIVIDUAL','R11-514-34','1111-1111-1111-1111','VISA_CARD','Goeffrey','Ward','Clinton','CUSTOMER','HOME','Provo','Utah','84606','118 South 9th East','HOME','011','801','423\-1234'); CALL contact_insert('INDIVIDUAL','R11-514-35','1111-2222-1111-1111','VISA_CARD','Wendy',null,'Moss','CUSTOMER','HOME','Provo','Utah','84606','1218 South 10th East','HOME','011','801','423-1234'); CALL contact_insert('INDIVIDUAL','R11-514-36','1111-1111-2222-1111','VISA_CARD','Simon','Jonah','Gretelz','CUSTOMER','HOME','Provo','Utah','84606','2118 South 7th East','HOME','011','801','423-1234'); CALL contact_insert('INDIVIDUAL','R11-514-37','1111-1111-1111-2222','MASTER_CARD','Elizabeth','Jane','Royal','CUSTOMER','HOME','Provo','Utah','84606','2228 South 14th East','HOME','011','801','423-1234'); CALL contact_insert('INDIVIDUAL','R11-514-38','1111-1111-3333-1111','VISA_CARD','Brian','Nathan','Smith','CUSTOMER','HOME','Spanish Fork','Utah','84606','333 North 2nd East','HOME','011','801','423-1234'); |
I hope this code complete approach helps those looking to learn how to write MySQL PSMs.
PostgreSQL+PowerShell
This post explains and demonstrates how to install, configure, and use the psqlODBC (32-bit) and psqlODBC (64-bit) libraries to connect your Microsoft PowerShell programs to a locally installed PostgreSQL 14 database. It relies on you previously installing and configuring a PostgreSQL 14 database. This post is a step-by-step guide to installing PostgreSQL 14 on Windows 10, and this post shows you how to configure the PostgreSQL 14 database.
If you didn’t follow the instructions to get the psqlODBC libraries in the installation blog post, you will need to get those libraries, as qualified by Microsoft with the PostgreSQL Stack Builder.
You can launch PostgreSQL Stack Builder after the install by clicking on Start -> PostgreSQL -> Stack Builder. Choose to enable Stack Builder to change your system and install the psqlODBC libraries. After you’ve installed the psqlODBC library, use Windows search field to find the ODBC Data Sources dialog and run it as administrator.
There are six steps to setup, test, and save your ODBC Data Source Name (DSN). You can click on the images on the right to launch them in a more readable format or simply read the instructions.
PostgreSQL ODBC Setup Steps
- The Microsoft DSN (Data Source Name) dialog automatically elects the User DSN tab. Click on the System DSN tab.
- The view under the System DSN is exactly like the User DSN tab. Click the Add button to start the workflow.
- The Create New Data Source dialog requires you select the PostgreSQL ODBC Driver(UNICODE) option from the list and click the Finish button to proceed.
- The PostgreSQL Unicode ODBC Driver Setup dialog should complete the prompts as follows below and consistent with the PostgreSQL 14 Configuration blog. If you opt for localhost as the server value because you have a DCHP IP address, make sure you’ve configured your hosts file in the C:\Windows\System32\drivers\etc directory. You should enter the following two lines in the hosts file:
127.0.0.1 localhost ::1 localhost
These are the string values you should enter in the PostgreSQL Unicode ODBC Driver Setup dialog:
Data Source: PostgreSQL35W Database: videodb Server: localhost User Name: student Description: PostgreSQL SSL Mode: disable Port: 5432 Password: student
After you complete the entry, click the Test button.
- The Connection Test dialog should return a “Connection successful” message. Click the OK button to continue.
- The ODBC Data Source Administrator dialog should show the PostgreSQL35W System Data Source. Click the OK button to continue.
After you have created the System PostgreSQL ODBC Setup, it’s time to build a PowerShell Cmdlet (or, Commandlet). Some documentation and blog notes incorrectly suggest you need to write a connection string with a UID and password, like:
$ConnectionString = 'DSN=PostgreSQL35W;Uid=student;Pwd=student' |
The UID and password is unnecessary in the connection string. As a rule, the UID and password are only necessary in the ODBC DSN, like:
$ConnectionString = 'DSN=PostgreSQL35W' |
You can create a readcursor.ps1 Cmdlet like the following:
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 | # Define a ODBC DSN connection string. $ConnectionString = 'DSN=PostgreSQL35W' # Define a MySQL Command Object for a non-query. $Connection = New-Object System.Data.Odbc.OdbcConnection; $Connection.ConnectionString = $ConnectionString # Attempt connection. try { $Connection.Open() # Create a SQL command. $Command = $Connection.CreateCommand(); $Command.CommandText = "SELECT current_database();"; # Attempt to read SQL command. try { $Reader = $Command.ExecuteReader(); # Read while records are found. while ($Reader.Read()) { Write-Host "Current Database [" $Reader[0] "]"} } catch { Write-Error "Message: $($_.Exception.Message)" Write-Error "StackTrace: $($_.Exception.StackTrace)" Write-Error "LoaderExceptions: $($_.Exception.LoaderExceptions)" } finally { # Close the reader. $Reader.Close() } } catch { Write-Error "Message: $($_.Exception.Message)" Write-Error "StackTrace: $($_.Exception.StackTrace)" Write-Error "LoaderExceptions: $($_.Exception.LoaderExceptions)" } finally { $Connection.Close() } |
Line 14 assigns a SQL query that returns a single row with one column as the CommandText of a Command object. Line 22 reads the zero position of a row or record set with only one column.
You call the readcursor.ps1 Cmdlet with the following syntax:
powershell .\readcursor.ps1 |
It returns:
Current Database [ videodb ] |
A more realistic way to write a query would return multiple rows with a set of two or more columns. The following program queries a table with multiple rows of two columns, but the program logic can manage any number of columns.
# Define a ODBC DSN connection string. $ConnectionString = 'DSN=PostgreSQL35W' # Define a MySQL Command Object for a non-query. $Connection = New-Object System.Data.Odbc.OdbcConnection; $Connection.ConnectionString = $ConnectionString # Attempt connection. try { $Connection.Open() # Create a SQL command. $Command = $Connection.CreateCommand(); $Command.CommandText = "SELECT last_name, first_name FROM contact ORDER BY 1, 2"; # Attempt to read SQL command. try { $row = $Command.ExecuteReader(); # Read while records are found. while ($row.Read()) { # Initialize output for each row. $output = "" # Navigate across all columns (only two in this example). for ($column = 0; $column -lt $row.FieldCount; $column += 1) { # Mechanic for comma-delimit between last and first name. if ($output.length -eq 0) { $output += $row[$column] } else { $output += ", " + $row[$column] } } # Write the output from the database. Write-Host $output } } catch { Write-Error "Message: $($_.Exception.Message)" Write-Error "StackTrace: $($_.Exception.StackTrace)" Write-Error "LoaderExceptions: $($_.Exception.LoaderExceptions)" } finally { # Close the reader. $row.Close() } } catch { Write-Error "Message: $($_.Exception.Message)" Write-Error "StackTrace: $($_.Exception.StackTrace)" Write-Error "LoaderExceptions: $($_.Exception.LoaderExceptions)" } finally { $Connection.Close() } |
You call the readcontact.ps1 Cmdlet with the following syntax:
powershell .\readcontact.ps1 |
It returns an ordered set of comma-separated values, like
Clinton, Goeffrey Gretelz, Simon Moss, Wendy Royal, Elizabeth Smith, Brian Sweeney, Ian Sweeney, Matthew Sweeney, Meaghan Vizquel, Doreen Vizquel, Oscar Winn, Brian Winn, Randi |
As always, I hope this helps those looking for a complete concrete example of how to make Microsoft Powershell connect and query results from a PostgreSQL database.
PostgreSQL Arrays
If you’re wondering about this post, it shows the basic array of a set of integers and strings before showing you how to create nested tables of data in PostgreSQL. By the way, they’re not called nested tables in PostgreSQL, like they are in Oracle but perform like their Oracle cousins.
Let’s create a table with an auto-incrementing column and two arrays, one array of integers and another of strings:
-- Conditionally drop the demo table. DROP TABLE IF EXISTS demo; -- Create the test table. CREATE TABLE demo ( demo_id serial , demo_number integer[5] , demo_string varchar(5)[7]); |
You can insert test values like this:
INSERT INTO demo (demo_number, demo_string) VALUES ( array[1,2,3,4,5] , array['One','Two','Three','Four','Five','Six','Seven']); |
Then, you can query them with this unnest function, like:
SELECT unnest(demo_number) AS numbers , unnest(demo_string) AS strings FROM demo; |
It returns:
numbers | strings
---------+---------
1 | One
2 | Two
3 | Three
4 | Four
5 | Five
| Six
| Seven
(7 rows) |
You may note that the two arrays are asymmetrical. It only becomes an issue when you navigate the result in a PL/pgSQL cursor or imperative programming language, like Python.
Now, let’s do something more interesting like work with a composite user-defined type, like the player structure. You would create the composite user-defined type with this syntax:
-- Conditionally drop the player type. DROP TYPE IF EXISTS player; -- Create the player type. CREATE TYPE player AS ( player_no integer , player_name varchar(24) , player_position varchar(14) , ab integer , r integer , h integer , bb integer , rbi integer ); |
You can create a world_series table that include a players column that uses an array of player type, like
-- Conditionally drop the world_series table. DROP TABLE IF EXISTS world_series; -- Create the player type. CREATE TABLE world_series ( world_series_id serial , team varchar(24) , players player[30] , game_no integer , year integer ); |
If you’re familiar with the Oracle Database, you’d have to specify a tested table in the syntax. Fortunately, PostgreSQL doesn’t require that.
Insert two rows with the following statement:
INSERT INTO world_series ( team , players , game_no , year ) VALUES ('San Francisco Giants' , array[(24,'Willie Mayes','Center Fielder',5,0,1,0,0)::player ,(5,'Tom Haller','Catcher',4,1,2,0,2)::player] , 4 , 1962 ); |
You can append to the array with the following syntax. A former student and I have a disagreement on whether this is shown in the PostgreSQL 8.15.4 Modifying Array documentation. I believe it can be inferred from the document and he doesn’t believe so. Anyway, here’s how you add an element to an existing array in a table with the UPDATE statement:
UPDATE world_series SET players = (SELECT array_append(players,(7,'Henry Kuenn','Right Fielder',3,0,0,1,0)::player) FROM world_series) WHERE team = 'San Francisco Giants' AND year = 1962 AND game_no = 4; |
Like Oracle’s nested tables, PostgreSQL’s arrays of composite user-defined types requires writing a PL/pgSQL function. I’ll try to add one of those shortly in another blog entry to show you how to edit and replace entries in stored arrays of composite user-defined types.
You can query the unnested rows and get a return set like a Python tuple with the following query:
SELECT unnest(players) AS player_list FROM world_series WHERE team = 'San Francisco Giants' AND year = 1962 AND game_no = 4; |
It returns the three rows from the players array:
player_list ---------------------------------------------- (24,"Willie Mayes","Center Field",5,0,1,0,0) (5,"Tom Haller",Catcher,4,1,2,0,2) (7,"Henry Kuenn","Right Fielde",3,0,0,1,0) (3 rows) |
It returns the data set in entry-order. If we step outside of the standard 8.15 Arrays PostgreSQL Documentation, you can do much more with arrays (or nested tables). The balance of this example demonstrates some new syntax that helps you achieve constructive outcomes in PostgreSQL.
You can use a Common Table Expression (CTE) to get the columnar display of the player composite user-defined type. This type of solution is beyond the standard , like:
WITH list AS (SELECT unnest(players) AS row_result FROM world_series WHERE team = 'San Francisco Giants' AND year = 1962 AND game_no = 4) SELECT (row_result).player_name , (row_result).player_no , (row_result).player_position FROM list; |
If you’re unfamiliar with accessing composite user-defined types, I wrote a post on that 7 years ago. You can find the older blog entry PostgreSQL Composites on my blog.
It returns only the three requested columns of the player composite user-defined type:
player_name | player_no | player_position --------------+-----------+----------------- Willie Mayes | 24 | Center Fielder Tom Haller | 5 | Catcher Henry Kuenn | 7 | Right Fielder (3 rows) |
You should note that the data is presented in an entry-ordered manner when unnested alone in the SELECT-list. That behavior changes when the SELECT-list includes non-array data.
The easiest way to display data from the non-array and array columns is to list them inside the SELECT-list of the CTE, like:
WITH list AS (SELECT game_no AS game , year , unnest(players) AS row_result FROM world_series WHERE team = 'San Francisco Giants' AND year = 1962 AND game_no = 4) SELECT game , year , (row_result).player_name , (row_result).player_no , (row_result).player_position FROM list; |
It returns an ordered set of unnested rows when you include non-array columns, like:
game | year | player_name | player_no | player_position
------+------+--------------+-----------+-----------------
4 | 1962 | Henry Kuenn | 7 | Right Fielder
4 | 1962 | Tom Haller | 5 | Catcher
4 | 1962 | Willie Mayes | 24 | Center Fielder
(3 rows) |
While you can join the world_series table to the unnested array rows (returned as a derived table, its a bad idea. The mechanics to do it require you to return the primary key column in the same SELECT-list of the CTE. Then, you join the CTE list to the world_series table by using the world_series_id primary key.
However, there is no advantage to an inner join approach and it imposes unnecessary processing on the database server. The odd rationale that I heard when I noticed somebody was using a CTE to base-table join was: “That’s necessary so they could use column aliases for the non-array columns.” That’s not true because you can use the aliases inside the CTE, as shown above when game is an alias to the game_no column.
As always, I hope this helps those looking to solve a problem in PostgreSQL.
Setting SQL_MODE
In MySQL, the @@sql_mode parameter should generally use ONLY_FULL_GROUP_BY. If it doesn’t include it and you don’t have the ability to change the database parameters, you can use a MySQL PSM (Persistent Stored Module), like:
Create the set_full_group_by procedure:
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 | -- Drop procedure conditionally on whether it exists already. DROP PROCEDURE IF EXISTS set_full_group_by; -- Reset delimter to allow semicolons to terminate statements. DELIMITER $$ -- Create a procedure to verify and set connection parameter. CREATE PROCEDURE set_full_group_by() LANGUAGE SQL NOT DETERMINISTIC SQL SECURITY DEFINER COMMENT 'Set connection parameter when not set.' BEGIN /* Check whether full group by is set in the connection and if unset, set it in the scope of the connection. */ IF NOT EXISTS (SELECT NULL WHERE REGEXP_LIKE(@@SQL_MODE,'ONLY_FULL_GROUP_BY')) THEN SET SQL_MODE=(SELECT CONCAT(@@sql_mode,',ONLY_FULL_GROUP_BY')); END IF; END; $$ -- Reset the default delimiter. DELIMITER ; |
Run the following SQL command before you attempt the exercises in the same session scope:
CALL set_full_group_by(); |
As always, I hope this helps those looking for a solution. Naturally, you can simply use the SET command on line #21 above.
Drop Overloaded Routine
In October 2019, I wrote a post with anonymous block programs to drop tables, sequences, routines, and triggers. Two weeks later, I wrote another post to drop all overloaded routines. However, I recognized the other day that I should have written a function that let you target which function or procedure you want to drop.
The older code only let you drop all of your functions or procedures. That was overkill when you’re working on new functions or procedures.
This post provides a utility for those writing functions and procedures in a public schema of any database in a PostgreSQL installation. It is designed to drop functions or procedures from the public schema.
The code follows below:
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 | CREATE OR REPLACE FUNCTION drop_routine( IN pv_routine_name VARCHAR(64) , IN pv_routine_type VARCHAR(64)) RETURNS INTEGER AS $$ DECLARE /* Declare the current catalog. */ lv_local_catalog VARCHAR(64) := current_database(); /* Declare return type variable. */ lv_retval INTEGER := 1; /* Manage data dictionary case mechanics: ====================================== routine_name is always in lowercase. routine_type is always in uppercase. */ lv_routine_name VARCHAR(64) := LOWER(pv_routine_name); lv_routine_type VARCHAR(64) := UPPER(pv_routine_type); /* Declare an indefinite length string for SQL statement. */ sql VARCHAR; /* Declare variables to manage cursor return values. */ row RECORD; arg VARCHAR; /* Declare parameter list. */ list VARCHAR; /* Declare a routine cursor. */ routine_cursor CURSOR( cv_routine_name VARCHAR , cv_routine_type VARCHAR ) FOR SELECT r.routine_name , r.specific_name , r.routine_type FROM information_schema.routines r WHERE r.specific_catalog = current_database() AND r.routine_schema = 'public' AND r.routine_type = cv_routine_type AND r.routine_name = cv_routine_name; /* Declare a parameter cursor. */ parameter_cursor CURSOR( cv_specific_name VARCHAR ) FOR SELECT args.data_type FROM information_schema.parameters args WHERE args.specific_catalog = current_database() AND args.specific_schema = 'public' AND args.specific_name = cv_specific_name; BEGIN /* Open the cursor. */ OPEN routine_cursor(lv_routine_name, lv_routine_type); <<row_loop>> LOOP /* Fetch table names. */ FETCH routine_cursor INTO row; /* Exit when no more records are found. */ EXIT row_loop WHEN NOT FOUND; /* Initialize parameter list. */ list := '('; /* Open the parameter cursor. */ OPEN parameter_cursor(row.specific_name::varchar); <<parameter_loop>> LOOP FETCH parameter_cursor INTO arg; /* Exit the parameter loop. */ EXIT parameter_loop WHEN NOT FOUND; /* Add parameter and delimit more than one parameter with a comma. */ IF LENGTH(list) > 1 THEN list := CONCAT(list,',',arg); ELSE list := CONCAT(list,arg); END IF; END LOOP; /* Close the parameter list. */ list := CONCAT(list,')'); /* Close the parameter cursor. */ CLOSE parameter_cursor; /* Concatenate together a DDL to drop the table with prejudice. */ sql := 'DROP '||row.routine_type||' IF EXISTS '||row.routine_name||list; /* Execute the DDL statement. */ EXECUTE sql; /* Assign success flag of 0. */ lv_retval := 0; END LOOP; /* Close the routine_cursor. */ CLOSE routine_cursor; /* Return the output text variable. */ RETURN lv_retval; END $$ LANGUAGE plpgsql; |
If you now create a series of hello overloaded functions, like:
CREATE OR REPLACE FUNCTION hello() RETURNS text AS $$ DECLARE output VARCHAR; BEGIN SELECT 'Hello World!' INTO output; RETURN output; END $$ LANGUAGE plpgsql; CREATE OR REPLACE FUNCTION hello(whom text) RETURNS text AS $$ DECLARE output VARCHAR; BEGIN SELECT CONCAT('Hello ',whom,'!') INTO output; RETURN output; END $$ LANGUAGE plpgsql; CREATE OR REPLACE FUNCTION hello(id int, whom text) RETURNS text AS $$ DECLARE output VARCHAR; BEGIN SELECT CONCAT('[',id,'] Hello ',whom,'!') INTO output; RETURN output; END $$ LANGUAGE plpgsql; |
After you create the overloaded functions, you can query their status from the information_schema.routines table in the data dictionary:
SELECT routine_name , specific_name , routine_type FROM information_schema.routines WHERE specific_catalog = current_setting('videodb.catalog_name') AND routine_schema = 'public' AND routine_name = 'hello'; |
Which shows you the three versions of the hello function:
routine_name | specific_name | routine_type --------------+---------------+-------------- hello | hello_18100 | FUNCTION hello | hello_18101 | FUNCTION hello | hello_18102 | FUNCTION (3 rows) |
You can drop all versions of the hello functions by calling the drop_routine function:
SELECT CASE WHEN drop_routine('hello','function') = 0 THEN 'Success' ELSE 'Failure' END AS drop_routine; |
It returns the following:
drop_routine -------------- Success (1 row) |
As always, I hope this helps those looking for how to routinely test new functions and procedures.
Install Python on Windows
A number of my students want to run the databases and development environments on Windows rather than Linux. Some of the students have various problems configuring a virtual machine or Docker environment on their laptops.
Installing Python on Windows is quite straightforward. You simply open a Windows Command Line Interface (CLI) with Administrator privileges and type python at the command line. It will launch a dialog that lets you download and install Python.
After the installation, you are returned to the Windows CLI where you can type python to launch the Python IDE. You’ll learn that it’s an older version when it opens.
Python 3.10.2 (tags/v3.10.2:a58ebcc, Jan 17 2022, 14:12:15) [MSC v.1929 64 bit (AMD64)] on win32 Type "help", "copyright", "credits" or "license" for more information. >>> |
After installing Python, I started to install drivers with the Python package-management utility – PIP. I loaded the psycopg2 driver first and discovered that Windows installed an older version of PIP, as qualified when I installed the Python library:
pip install psycopg2 Collecting psycopg2 Downloading psycopg2-2.9.3-cp310-cp310-win_amd64.whl (1.2 MB) |████████████████████████████████| 1.2 MB 1.6 MB/s Installing collected packages: psycopg2 Successfully installed psycopg2-2.9.3 WARNING: You are using pip version 21.2.4; however, version 22.0.3 is available. You should consider upgrading via the 'C:\Users\mclaughlinm\AppData\Local\Microsoft\WindowsApps\PythonSoftwareFoundation.Python.3.10_qbz5n2kfra8p0\python.exe -m pip install --upgrade pip' command. |
So, I immediately installed the new version of PIP and discovered that I have to manually reset the %PATH% environment variable.
python -m pip install --upgrade pip Requirement already satisfied: pip in c:\program files\windowsapps\pythonsoftwarefoundation.python.3.10_3.10.752.0_x64__qbz5n2kfra8p0\lib\site-packages (21.2.4) Collecting pip Downloading pip-22.0.3-py3-none-any.whl (2.1 MB) |████████████████████████████████| 2.1 MB 656 kB/s Installing collected packages: pip WARNING: The scripts pip.exe, pip3.10.exe and pip3.exe are installed in 'C:\Users\mclaughlinm\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.10_qbz5n2kfra8p0\LocalCache\local-packages\Python310\Scripts' which is not on PATH. Consider adding this directory to PATH or, if you prefer to suppress this warning, use --no-warn-script-location. Successfully installed pip-22.0.3 |
The default location for the PIP.exe and PIP3.exe are found in the:
C:\Users\mclaughlinm\AppData\Local\Microsoft\WindowsApps |
Rather than put the newer directory in the System Environment %PATH% variable, I created a batch file that lets me set it only when I need it.
SET PATH=C:\Users\mclaughlinm\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.10_qbz5n2kfra8p0\LocalCache\local-packages\Python310\Scripts;%PATH% |
After installing the psycopg2 driver, I tested the new_hire.sql and new_hire.py files from this earlier Linux post successfully. I hope this helps those looking to work with Python on Windows.
Python on PostgreSQL
The ODBC library you use when connecting Python to PostgreSQL is the psycopg2 Python library. This blog post will show use how to use it in Python and install it on your Fedora Linux installation. It leverages a videodb database that I show you how to build in this earlier post on configuring PostgreSQL 14.
You would import psycopg2 as follows in your Python code:
import psycopg2 |
Unfortunately, that only works on Linux servers when you’ve installed the library. That library isn’t installed with generic Python libraries. You get the following error when the psycopg2 library isn’t installed on your server.
Traceback (most recent call last):
File "python_new_hire.sql", line 1, in <module>
import psycopg2
ModuleNotFoundError: No module named 'psycopg2' |
You can install it on Fedora Linux with the following command:
yum install python3-psycopg2 |
It will install:
==================================================================================== Package Architecture Version Repository Size ==================================================================================== Installing: python3-psycopg2 x86_64 2.7.7-1.fc30 fedora 160 k Transaction Summary ==================================================================================== Install 1 Package Total download size: 160 k Installed size: 593 k Is this ok [y/N]: y Downloading Packages: python3-psycopg2-2.7.7-1.fc30.x86_64.rpm 364 kB/s | 160 kB 00:00 ------------------------------------------------------------------------------------ Total 167 kB/s | 160 kB 00:00 Running transaction check Transaction check succeeded. Running transaction test Transaction test succeeded. Running transaction Preparing : 1/1 Installing : python3-psycopg2-2.7.7-1.fc30.x86_64 1/1 Running scriptlet: python3-psycopg2-2.7.7-1.fc30.x86_64 1/1 Verifying : python3-psycopg2-2.7.7-1.fc30.x86_64 1/1 Installed: python3-psycopg2-2.7.7-1.fc30.x86_64 Complete! |
Here’s a quick test case that you can run in PostgreSQL and Python to test all the pieces. The first SQL script creates a new_hire table and inserts two rows, and the Python program queries data from the new_hire table.
The new_hire.sql file creates the new_hire table and inserts two rows:
-- Environment settings for the script. SET SESSION "videodb.table_name" = 'new_hire'; SET CLIENT_MIN_MESSAGES TO ERROR; -- Verify table name. SELECT current_setting('videodb.table_name'); -- ------------------------------------------------------------------ -- Conditionally drop table. -- ------------------------------------------------------------------ DROP TABLE IF EXISTS new_hire CASCADE; -- ------------------------------------------------------------------ -- Create table. -- ------------------------------------------------------------------- CREATE TABLE new_hire ( new_hire_id SERIAL , first_name VARCHAR(20) NOT NULL , middle_name VARCHAR(20) , last_name VARCHAR(20) NOT NULL , hire_date TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT CURRENT_TIMESTAMP , PRIMARY KEY (new_hire_id)); -- Alter the sequence by restarting it at 1001. ALTER SEQUENCE new_hire_new_hire_id_seq RESTART WITH 1001; -- Display the table organization. SELECT tc.table_catalog || '.' || tc.constraint_name AS constraint_name , tc.table_catalog || '.' || tc.table_name AS table_name , kcu.column_name , ccu.table_catalog || '.' || ccu.table_name AS foreign_table_name , ccu.column_name AS foreign_column_name FROM information_schema.table_constraints AS tc JOIN information_schema.key_column_usage AS kcu ON tc.constraint_name = kcu.constraint_name AND tc.table_schema = kcu.table_schema JOIN information_schema.constraint_column_usage AS ccu ON ccu.constraint_name = tc.constraint_name AND ccu.table_schema = tc.table_schema WHERE tc.constraint_type = 'FOREIGN KEY' AND tc.table_name = current_setting('videodb.table_name') ORDER BY 1; SELECT c1.table_name , c1.ordinal_position , c1.column_name , CASE WHEN c1.is_nullable = 'NO' AND c2.column_name IS NOT NULL THEN 'PRIMARY KEY' WHEN c1.is_nullable = 'NO' AND c2.column_name IS NULL THEN 'NOT NULL' END AS is_nullable , CASE WHEN data_type = 'character varying' THEN data_type||'('||character_maximum_length||')' WHEN data_type = 'numeric' THEN CASE WHEN numeric_scale != 0 AND numeric_scale IS NOT NULL THEN data_type||'('||numeric_precision||','||numeric_scale||')' ELSE data_type||'('||numeric_precision||')' END ELSE data_type END AS data_type FROM information_schema.columns c1 LEFT JOIN (SELECT trim(regexp_matches(column_default,current_setting('videodb.table_name'))::text,'{}')||'_id' column_name FROM information_schema.columns) c2 ON c1.column_name = c2.column_name WHERE c1.table_name = current_setting('videodb.table_name') ORDER BY c1.ordinal_position; -- Display primary key and unique constraints. SELECT constraint_name , lower(constraint_type) AS constraint_type FROM information_schema.table_constraints WHERE table_name = current_setting('videodb.table_name') AND constraint_type IN ('PRIMARY KEY','UNIQUE'); -- Insert two test records. INSERT INTO new_hire ( first_name, middle_name, last_name, hire_date ) VALUES ('Malcolm','Jacob','Lewis','2018-2-14') ,('Henry',null,'Chabot','1990-07-31'); |
You can put it into a local directory, connect as the student user to a videodb database, and run the following command (or any database you’ve created).
\i new_hire.sql |
The new_hire.py file creates the new_hire table and inserts two rows:
# Import the PostgreSQL connector library. import psycopg2 try: # Open a connection to the database. connection = psycopg2.connect( user="student" , password="student" , port="5432" , dbname="videodb") # Open a cursor. cursor = connection.cursor() # Assign a static query. query = "SELECT new_hire_id, first_name, last_name " \ "FROM new_hire" # Parse and execute the query. cursor.execute(query) # Fetch all rows from a table. records = cursor.fetchall() # Read through and print the rows as tuples. for row in range(0, len(records)): print(records[row]) except (Exception, psycopg2.Error) as error : print("Error while fetching data from PostgreSQL", error) finally: # Close the database connection. if (connection): cursor.close() connection.close() |
You run it from the command line, like:
python3 ./new_hire.py |
It should print:
(1001, 'Malcolm', 'Lewis') (1002, 'Henry', 'Chabot') |
As always, I hope this helps those trying to sort out how to connect Python to PostgreSQL.
PostgreSQL CLI Error
Problems get reported to me all the time on installations for my students, this one was interesting. They got an error complaining about a missing libpq.so.5 library.
psql: /usr/pgsql-11/lib/libpq.so.5: no version information available (required by psql) psql: /usr/pgsql-11/lib/libpq.so.5: no version information available (required by psql) could not change directory to "/root": Permission denied psql (11.7, server 11.8) Type "help" for help. postgres=# |
It appeared as a mismatch of libraries but it’s not that. For reference, this was a Fedora instance. I ran the rpm utility:
rpm -qa | grep postgres |
It returned:
postgresql11-libs-11.8-1PGDG.f30.x86_64 postgresql-11.7-2.fc30.x86_64 postgresql-server-11.7-2.fc30.x86_64 |
Then, I had them run the rpm utility again looking for the Python driver for PostgreSQL:
rpm -qa | grep psycopg2 |
It returned:
python3-psycopg2-2.7.7-1.fc30.x86_64 |
Then, it was easy to explain. The Python psycopg2 library uses both PostgreSQL 11.7 and 11.8 dependent libraries and the libpq.so.5 library is missing version information. You must ignore the error, which is really only a warning message, when you want to work on Fedora, PostgreSQL 11, and Python 3.
Subscribe via RSS