If you know regex and are good at scripting in bash, hledger is fucking based. Importing csvs doesn't require any scripting per se, but obviously, where a system is cli / text based, you can do a lot with stdout.

For example, I have one account for receivables. All invoices are dumped into it. Clients are "payees". They don't have subaccounts.

If I want to get reports on all clients balances / transactions, I pull the list of all payees, grep for client number patten, read the result and recursively run each "payee" back into hledger one at a time. So I can treat clients like they each have an account without having to have 500 subaccounts.

The only thing that throws most people for a loop is double entry accounting in cash, not accrual based business, e.g. receivables vs revenues. I still double entry account, because it's easier in most systems — cash accounting usually requires extra steps to avoid the default double entry systems — but my reports are effective cash, e.g. I never look at revenues.