@jifeon/pgcodebase
A tool for easy management of postgresql functions, views and triggers
Last updated 2 years ago by jifeon .
ISC · Repository · Bugs · Original npm · Tarball · package.json
$ cnpm install @jifeon/pgcodebase 
SYNC missed versions from official npm registry.

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 and views. Other variables (except for PGCODEBASE_CONNECTION_STRING) are standard for postgresql.

Problem

Have you ever worked with postgresql functions, triggers 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 requires (you can specify multiple) can be in any order at the start of a file.

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 requires.

Current Tags

  • 1.5.4                                ...           latest (2 years ago)

6 Versions

  • 1.5.4                                ...           2 years ago
  • 1.5.3                                ...           2 years ago
  • 1.5.2                                ...           2 years ago
  • 1.5.1                                ...           2 years ago
  • 1.5.0                                ...           2 years ago
  • 1.4.0                                ...           2 years ago
Maintainers (1)
Downloads
Today 0
This Week 0
This Month 2
Last Day 0
Last Week 1
Last Month 3
Dependencies (5)
Dev Dependencies (0)
None
Dependents (0)
None

Copyright 2014 - 2016 © taobao.org |