WordPress¶

The cms server hosts the corporate (www.open-contracting.org) and coalition (www.open-spending.eu) WordPress sites. To work on a site locally, you need its files and its database.

The sites’ theme repositories contain a Makefile to load a local environment from these dumps:

Copy a site from the server¶

script/snapshot-site dumps the database. For example:

./script/snapshot-site corporate
./script/snapshot-site coalition path/to/theme

Logs and Relevanssi search indexes are skipped, as a local site regenerates or doesn’t need them. Pass --all-tables for a complete copy.

Pass --files to download the public_html directory as a .tar.gz file, or --rsync it as a directory, which is incremental and cheaper to repeat. wp-content/uploads is skipped; add it with --uploads.

./script/snapshot-site corporate --files
./script/snapshot-site corporate --rsync --uploads

It connects as the site’s own user, so anyone with deploy access can run it, and reads the database credentials from wp-config.php. The dump uses --single-transaction, to not lock tables on the live site. The archive is streamed, so the server needs no temporary space for a second copy of the site.

Load the site locally¶

The script prints the commands in this section, filled in for the site you copied. wp db export omits CREATE DATABASE, so the database must exist first:

mysql -e 'CREATE DATABASE IF NOT EXISTS corporate_wp'
gunzip -c corporate-snapshot/corporate_wp.sql.gz | mysql corporate_wp

The database stores the site’s URL and its absolute paths, so WordPress redirects to the production site until both are rewritten:

wp search-replace 'https://www.open-contracting.org' 'http://localhost:8090' --all-tables
wp search-replace '/home/corporate/public_html' "$(pwd)" --all-tables

Note

The snapshot includes wp-config.php, whose credentials are for the server’s database. Point it at your local database before loading the site.

Download a backup instead¶

Use a backup to recover a deleted file or to see a site as it was, rather than as it is. Backups run daily. (The bucket name is narrower than its contents.)

export AWS_DEFAULT_REGION=eu-west-2

aws s3 ls s3://ocp-coalition-backup/database/
aws s3 ls s3://ocp-coalition-backup/site/

Databases are named {timestamp}_{database}.sql.gz. Site files are named after the directory, with non-alphanumeric characters replaced by underscores. For example:

database/20260827T044506Z_corporate_wp.sql.gz
site/home_corporate_public_html_backup_20260827T041526Z.tar.gz

To download the most recent of each:

latest() { aws s3 ls "s3://ocp-coalition-backup/$1/" | awk -v p="$2" '$4 ~ p {print $4}' | sort | tail -1; }

aws s3 cp "s3://ocp-coalition-backup/database/$(latest database corporate_wp)" .
aws s3 cp "s3://ocp-coalition-backup/site/$(latest site home_corporate)" .

The corporate site tarball is about 5 GB. If you only need the database, download that alone.