Repeatable Output during Dev Testing

For testing I am creating a framework that crawls an entire EE site, stores the URLs and then creates an MD5 hash for each URL.

I intend to use this so I can make simple changes and have instant view of what the effects were across the whole site.

After doing two runs with no changes, it transpires that EE is producing different output for each run. For example, a banner may be produced in a different order, or an email address encrypted with a different salt, which changes the underlying HTML.

This means that the MD5 sums will never match and I won’t get a repeatable view of what changes between code updates.

Is there a way that I can force EE to always use the same “random” choices, so I can make the output consistent between test runs? Even non-randomness is fine in this context.

The test environment is all local within Docker containers which gives me a high level of control.