Skip to content


Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
Turn postgresql dump in a set of small sorted files
Ruby C
branch: master
Failed to load latest commit information.
bin move classes to library
ext/split_pgdump c: simpler capacity calculation
lib faster not empty file tracking
.gitignore no rules, no tables yet
README prepare for gem
split_pgdump.gemspec version 0.4.0


= 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  {split:<Split expr>} {sort:<Sort expr>}

<Split expr> examples:

<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 (c) 2011 by Sokolov Yura (
Released under the same terms of license as Ruby

== Homepage
Something went wrong with that request. Please try again.