Tool for splitting postgresql dump in a set of files
I wish to use git or mercurial for managing my database history. Unfortunately, every single data change force them to store whole dump again. Even if you data actually not changed, rows order is not promised to be stable.
split_pgdump splits dump in a set of small sorted files, so that git could track changes only of atcually changed data.
Also, it allows rsync to effectevely transmit backup changes over network.
Usage
Simplest example:
> pg_dump my_base | split_pgdump
It produces:
`dump.sql` - file with schema and psql copy instructions,
`dump.sql-tables/#{table}.dat` - 'copy data' for each table in a dump,
sorted numerically (I hope, it is `id`)
You can change file name by ‘-f` option.
Rules
Rules are read from ‘split.rules` file (could be changed by `-r` option). File could contain set of lines:
table_regexp expr> expr>
<Split expr> examples:
split:$field_name!
split:$field_name!_$other_field!
split:$client_id%00100!-$id%0025000!
split:$some_field[2..-1]!/$other_field[10..30]%0005!
<Sort expr> is space separated list of fields, optionally with options for gnu ‘sort` –key parameters (on my machine they are MbdfghinRrV):
sort:client_id uid
sort:client_id:n id:n
Example for redmines wiki_content_versions:
wiki_content_versions split:$page_id%0025!/$id%0000250! sort:page_id:n id:n
Either ‘split:` or `sort:` option could be skipped.
Author and Copyright
Copyright © 2011 by Sokolov Yura ([email protected]) Released under the same terms of license as Ruby