magicreplace is a search replace tool for migrating databases.
When moving databases, usually the url environment also changes. If the URL is hardcoded in the database (like WordPress does), those URLs have to be changed. If you now do a search and replace on your entire database to change the URLs, you will corrupt data that has been serialized. Just try out
unserialize(str_replace('www.foo.tld', 'www.barrr.tld', serialize('url=www.foo.tld')));
and you will get an ugly error.
- Velvet Blues Update URLs
- Better Search Replace
- Suchen & Ersetzen
- WP Migrate DB
- WP-CLI's search-replace
- Search-Replace-DB
- SerPlace
- Fast (~5sec runtime on 100mb database file with 300.000 rows)
- Lightweight: Only <10kb in size
- Works also on big files with small memory limit settings
- File based: Does not need a database or a wp installation - works on plain (sql) files
- Local usage: Does not need a remote server or a webservice
- Multi replace: Does multiple replaces
- Considers edge cases: Can handle objects and even references
- Ignores classes that are not available at runtime
- Can be used either with the command line or as a class
- Acts carefully: If serialization fails, nothing is changed
- Never changes data (out of bound ints are preserved, auto generated dates are not updated)
- Does its work in junks to overcome php limits
- Supports replacements in special base64 strings (e.g. in BeTheme)
This does not release you from taking backups. Use this script at your own risk!
brew install coreutils
Runs out of the box with WSL/WSL2/Cygwin.
–
wget https://raw.githubusercontent.com/vielhuber/magicreplace/master/src/magicreplace.php
php magicreplace.php input.sql output.sql search-1 replace-1 search-2 replace-2
composer require vielhuber/magicreplace
<?php
require __DIR__ . '/vendor/autoload.php';
use vielhuber\magicreplace\magicreplace;
magicreplace::run('input.sql', 'output.sql', ['search-1' => 'replace-2', 'search-2' => 'replace-2']);
If you want for example to replace http://www.foo.tld with https://www.bar.tld, the safest method to do so is with the following replacements (in the given order):
http://www.foo.tld
https://www.bar.tld
https://www.foo.tld
https://www.bar.tld
http://foo.tld
https://www.bar.tld
https://foo.tld
https://www.bar.tld
www.foo.tld
www.bar.tld
foo.tld
bar.tld
Just place these 3 files in a (optionally nested) subfolder of tests/data
:
input.sql
: The desired input fileoutput.sql
: The desired output filesettings.sql
: Define your replacements
Example settings.sql
file:
{
"replace": {
"http://www.foo.tld": "https://www.bar.tld",
"https://www.foo.tld": "https://www.bar.tld"
}
}
If a test fails, the expected output is stored in expected.sql
.
You can even auto generate test cases (that compares magicreplace to Search-Replace-DB and only gives you the diff) if you omit input.sql
and output.sql
and define a mysql database to dump from locally. Example settings.sql
file:
{
"source": {
"host": "localhost",
"port": "3306",
"database": "xxx",
"username": "xxx",
"password": "xxx",
},
"replace": {
"http://www.foo.tld": "https://www.bar.tld",
"https://www.foo.tld": "https://www.bar.tld"
}
}
input.sql
and output.sql
then get generated automatically. After you rerun the tests, these generated files are used. If you want to generate them again, just delete them before running the test. You also can provide a whitelist.sql
file that includes all lines from input.sql
that should be ignored (e.g. where magicreplace acts differently from Search-Replace-DB).