How are the deduplicated filenames normalised?

[tilusnet]

I notice deduplication during Import will generate filenames like YYYYMMDD_HHMMSS_<some-hash>.EXT.

How is <some-hash> computed?
Documentation says it's based on "file size and SHA1 of content" but no more detail.

I need this in order to identify the files left behind as duplicates. I need the left behind duplicate files for enriching metadata.
Thanks.

[tilusnet]

I'm answering my own question for the record, since I looked in the code.

A file's Canonical Name is computed using the file's creation date and Checksum.
The Creation Date is the timestamp of the image being taken, not necessarily the filesystem creation date -- check the fallback heuristic in the code.
The Checksum is the file's Castagnoli CRC32 checksum.

The final concatenation is here: it uses 'EEEEEEEE' if checksum is not the expected length of 8. The canonical name is prefixed with the YYYYMMDD_HHMMSS_ creation timestamp.