CraftRoq Relocate DB

CraftRoq Relocate DB changes text across your database: moving a site from staging to production, switching to HTTPS, changing a domain, or renaming something everywhere it appears.

It is built for doing that on real sites without breaking them.

Preview first, always

Every replacement starts as a dry run. It searches the tables you choose, changes nothing, and shows exactly what would change: rows and replacements per table and column, values it would leave alone and why, and before/after examples. Only a completed dry run can be applied, and applying it repeats exactly what was previewed.

Serialized data and JSON handled properly

Plugins and themes store settings as serialized PHP. A plain text replacement breaks those values whenever the length changes. Relocate rewrites serialized data token by token and recalculates every length, without unserializing it, so no plugin code runs and objects, private properties, references and numbers stay exactly as they were. URLs inside JSON (as page builders store them, with escaped slashes) are found and replaced too.

A value is left untouched, and reported, when changing it could corrupt it: serialized data that is already broken, a custom serialized format only its own class understands, or JSON that would become invalid.

Made for large databases

Tables are processed in small batches by primary key, a few seconds per request, so there are no timeouts or memory limits to hit however big the tables are. Progress is saved after every batch. If the page is closed or the connection drops, the job can be continued from the last completed batch.

Safety measures

• Each batch of a replacement is one database transaction: it is saved completely or not at all, so an interruption never leaves a batch half applied or applies it twice. This needs InnoDB tables, which WordPress uses by default; the confirmation step warns about any that are not.
• The original value of everything a replacement changes is saved to a downloadable file of SQL statements. Importing it puts those values back.
• The site address (siteurl and home) is changed last, so you are not logged out part-way through.
• Tables without a primary key or suitable unique key are skipped, because their rows cannot be updated one at a time safely.
• Post GUIDs are left alone unless you ask for them to be changed.
• Only administrators who can post unfiltered HTML can use it.

Also included

• Up to 20 search and replace pairs in one job, applied together in a single pass.
• Choose the tables to search, and leave out individual columns.
• Case-insensitive and whole-word matching, and matching the http:// and protocol-relative versions of a URL.
• A live progress view with the current table, rows scanned, changes found and time remaining.
• A searchable, sortable history of every job, a log, and automatic clean-up. Old jobs can also be deleted by hand.
• Import / Export: download the database as a .sql.gz file and import .sql or .sql.gz files, in resumable steps, with the site address changed on the way out or on the way in. Imported tables are built next to the ones in use and put in place together at the end, so the site keeps working during an import, and one that fails or is cancelled leaves them as they were.
• Export and import the plugin’s settings as a JSON file.
• WP-CLI commands: wp crq search-replace and wp crq resume.