The EBA COREP/FINREP taxonomies are about as complicated as XBRL gets. You can download the taxonomies and supporting documentation from their website. These taxonomies are updated twice a year (though the next release is delayed).
The taxonomies make use of (and therefore require you to have understanding of) XBRL 2.1, Dimensions 1.0, Table Linkbase (they still use a PWD version) and Formula 1.0. The level of understanding you would require depends on whether you use an existing XBRL processor or attempt this yourself. I'm not aware of any open source XBRL solution that's up to the task (most don't support Table or Formula and nearly all are poorly documented).
In terms of transformation, it's not straightforward. The EBA link above will lead you to some Excel templates. You then need to find some way of matching cells in this to the concepts and dimensions in the taxonomy. This already complex and unreliable process would be complicated by tables with open Y and Z axes. There may be some 'short cut' by using the Access DPM database (from which the taxonomy is generated).
There are then various filing rules for submission to EBA. Submission is via a country's NCA which then may apply its own specifc rules or may even extend and modify the taxonomies.
EBA also has a list of around 2,000 validations (implemented in Formula 1.0 and summarised in an Excel file) that the filing is required to pass else it will be rejected. As some of these have turned out to be erroneous, EBA tends to publish a list every few months activating or de-activating certain rules on this list. Validations are also added and removed in each release of the taxonomies.
This is not something you'd take on lightly. If you are working for a financial institution that's required to make COREP or FINREP submissions I'd strongly suggest you buy some proven reporting solution.