@plyo/pgcodebase
v1.6.0
Published
A tool for easy management of postgresql functions, views and triggers
Downloads
807
Readme
Installation
npm install pgcodebase
Usage
Pgcodebase is a tool for easy management of postgresql functions, triggers and views inspired by PgRebase.
const pgcodebase = require('pgcodebase')
pgcodebase({
user: process.env.PGUSER,
host: process.env.PGHOST,
password: process.env.PGPASSWORD,
database: process.env.PGDATABASE,
port: process.env.PGPORT,
dir: path.join(__dirname, 'pgcodebase'),
schema: 'public',
createOnly: false,
dropOnly: false
}).catch(console.error)
pgcodebase
call returns Promise.
Console
Usage: npx pgcodebase [options]
Options:
-u, --user <user> Postgresql user
-H, --host <host> Postgresql host
-p, --password <password> Postgresql password
-d, --database <database> Postgresql database name
-P, --port <port> Postgresql port
-C, --connection-string <connectionString> Postgresql connection string
-s, --schema <schema> Postgresql schema to use
-D, --dir <dir> Sql files directory
-c, --create-only Only create entities in database, drop nothing
-r, --drop-only Only drop entities in database, create nothing
-h, --help output usage information
If createOnly
or dropOnly
is true, then the script will only create or only drop entities respectively. It is useful to specify dropOnly
before table migrations and createOnly
after.
You can also configure pgcodebase using .env
file. You will have to create .env
file in the directory where you execute the command. In .env
there may be variables PGUSER
, PGHOST
, PGPASSWORD
, PGDATABASE
, PGPORT
, PGCODEBASE_CONNECTION_STRING
and PGCODEBASE_DIR
. Last one should contain an absolute or relative path to the directory with functions, triggers, grants and views. Other variables (except for PGCODEBASE_CONNECTION_STRING
) are standard for postgresql.
Problem
Have you ever worked with postgresql functions, triggers, grants and views using migrations? It becomes increasingly complex when your codebase grows. If you want to change the signature of some function, postgresql requires you to drop and recreate all the functions that depend on your function.
Let me demonstrate you on an example. Usually you would create migrations to create functions. Migration 1 and Migration 2. Notice that function foo
calls function bar
therefore a dependency exists. Then if you need to modify body of the bar
you will need to drop and recreate both functions: Migration 3. Imagine now you have multiple functions that depend on bar()
. You will have to drop and recreate them all! This process becomes really painful as your codebase grows.
A solution to the problem is a tool that drops and recreates all the functions, triggers and views in the order that respects dependencies. In our example you just have to create a folder and 2 files in it: function_bar and function_foo. Then you can run pgcodebase
whenever any change is made in those files.
Internally pgcodebase after resolving dependencies drops all entities and starts filling a table named pgcodebase_dropcodes
with sql specified using drop-code
comment. Simultaneously it is creating entities in database. To drop entities it executes sql code from pgcodebase_dropcodes
table in reverse order.
Code base
All the sql that creates functions, triggers and views has to be in one directory at any depth. dir
in config should be the absolute path to this folder. At the start of every sql file there must be a line like
-- drop-code drop function foo(bar integer)
After drop-code
there should be a valid sql query that drops the entity created by this sql file.
If, for example, function baz()
depends on function foo(bar integer)
at the start of the file functions/baz.sql
there must be a line
-- require functions/foo.sql
Note that the path is relative to your dir
and also drop-code
and require
s (you can specify multiple) can be in any order at the start of a file.
Also folder require is supported
--- required functions
In this case all folder files would be required as dependencies.
Inspiration
This tool was inspired by PgRebase. PgRebase, however, cannot process dependencies between different types of entities. If you have a view that depends on some function, you are out of luck. Pgcodebase doesn't care about the types of your entities. It can create and drop them in any order specified with require
s.