├── README.markdown
└── split-mysql-dump.rb
/README.markdown:
--------------------------------------------------------------------------------
1 | What is it?
2 | ===========
3 |
4 | A simple script that splits a MySQL dump into lots of smaller files.
5 | It works both with data definitions and data only dumps.
6 |
7 | Usage:
8 | ------
9 |
10 | First you need a mysqldump file, put it into the directory you want
11 | all the split files in:
12 |
13 |
14 | $ ruby split-mysql-dump.rb db.sql
15 | Found a new db: app
16 | Found a new table: administrator_log
17 | writing line: 229 200.494MB in 4 seconds 50.124MB/sec
18 |
19 | Found a new table: auth_strings
20 | writing line: 239 205.482MB in 6 seconds 34.247MB/sec
21 |
22 |
23 | Alternatively, you can pipe in via STDIN in using '-s'. Great
24 | for working with large gzipped backups:
25 |
26 |
27 | $ gunzip -c db.sql.gz | ruby split-mysql-dump.rb -s
28 |
29 |
30 | You can also limit the dump to particular tables using '-t'
31 | or exclude tables using '-i'.
32 |
33 |
34 | $ ruby split-mysql-dump.rb -t auth_strings, administrator_log db.sql
35 |
36 |
37 | and
38 |
39 |
40 | $ ruby split-mysql-dump.rb -i auth_strings
41 |
42 |
43 | When you're done you should have lots of files like this:
44 |
45 |
46 | -rw-r--r-- 1 rip rip 210233252 May 17 18:06 administrator_log.sql
47 | -rw-r--r-- 1 rip rip 215463582 May 17 18:06 auth_strings.sql
48 |
49 |
50 | The first bit of the files will be the database that the tables are in
51 | based on the _USE_ statements in the dump.
52 |
53 | Contact:
54 | --------
55 | You can contact me on rip@devco.net or follow my blog at http://www.devco.net I am also on twitter as ripienaar
56 |
57 |
--------------------------------------------------------------------------------
/split-mysql-dump.rb:
--------------------------------------------------------------------------------
1 | #!/usr/bin/env ruby
2 |
3 | require 'optparse'
4 |
5 | tables = []
6 | ignore = []
7 | use_database = nil
8 | dumpfile = ""
9 |
10 | cmds = OptionParser.new do |opts|
11 | opts.banner = "Usage: split-mysql-dump.rb [options] [FILE]"
12 |
13 | opts.on("-s", "Read from stdin") do
14 | dumpfile = $stdin
15 | end
16 |
17 | opts.on("-t", '--tables TABLES', Array, "Extract only these tables") do |t|
18 | tables = t
19 | end
20 |
21 | opts.on("-i", '--ignore-tables TABLES', Array, "Ignore these tables") do |i|
22 | ignore = i
23 | end
24 |
25 | opts.on("-u", '--use-database NAME', String, "Assume NAME as database name") do |n|
26 | use_database = n
27 | end
28 |
29 | opts.on_tail("-h", "--help") do
30 | puts opts
31 | end
32 |
33 | end.parse!
34 |
35 | if dumpfile == ""
36 | dumpfile = ARGV.shift
37 | if not dumpfile
38 | puts "Nothing to do"
39 | exit
40 | end
41 | end
42 |
43 | STDOUT.sync = true
44 |
45 | class Numeric
46 | def bytes_to_human
47 | units = %w{B KB MB GB TB}
48 | e = self > 0 ? (Math.log(self)/Math.log(1024)).floor : 0
49 | s = "%.3f" % (to_f / 1024**e)
50 | s.sub(/\.?0*$/, units[e])
51 | end
52 | end
53 |
54 | if File.exist?(dumpfile)
55 | if dumpfile == $stdin
56 | d = $stdin
57 | else
58 | d = File.new(dumpfile, "r:binary")
59 | end
60 |
61 | outfile = nil
62 | table = nil
63 | db = use_database
64 | linecount = tablecount = starttime = 0
65 |
66 | while (line = d.gets)
67 | # Detect table changes
68 | if line =~ /^-- Table structure for table .(.+)./ or line =~ /^-- Dumping data for table .(.+)./ or line =~ /^# Dump of table.(.+)/
69 | is_new_table = table != $1
70 | table = $1
71 |
72 | # previous file should be closed
73 | if is_new_table
74 | outfile.close if outfile and !outfile.closed?
75 |
76 | puts("\n\nFound a new table: #{table}")
77 |
78 | if (tables != [] and not tables.include?(table))
79 | puts"`#{table}` not in list, ignoring"
80 | table = nil
81 | elsif (ignore != [] and ignore.include?(table))
82 | puts"`#{table}` will be ignored"
83 | table = nil
84 | else
85 | starttime = Time.now
86 | linecount = 0
87 | tablecount += 1
88 | path = (db.to_s == "" ? "" : "#{db}/") + "tables";
89 | Dir.mkdir(path) unless File.exists?(path)
90 | outfile = File.new("#{path}/#{table}.sql", "w")
91 | outfile.syswrite("USE `#{db}`;\n\n")
92 | end
93 | end
94 | elsif line =~ /^-- Current Database: .(.+)./
95 | db = $1
96 | table = nil
97 | outfile.close if outfile and !outfile.closed?
98 | Dir.mkdir(db)
99 | Dir.mkdir("#{db}/tables")
100 | outfile = File.new("#{db}/create.sql", "w")
101 | puts("\n\nFound a new db: #{db}")
102 | elsif line =~ /^-- Position to start replication or point-in-time recovery from/
103 | db = nil
104 | table = nil
105 | outfile.close if outfile and !outfile.closed?
106 | outfile = File.new("1replication.sql", "w")
107 | puts("\n\nFound replication data")
108 | end
109 |
110 | # Write line to outfile
111 | if outfile and !outfile.closed?
112 | outfile.syswrite(line)
113 | linecount += 1
114 | elapsed = Time.now.to_i - starttime.to_i + 1
115 | print(" writing line: #{linecount} #{outfile.stat.size.bytes_to_human} in #{elapsed} seconds #{(outfile.stat.size / elapsed).bytes_to_human}/sec \r")
116 | end
117 | end
118 | end
119 |
120 | puts
121 |
--------------------------------------------------------------------------------