Please describe the problem.

I am trying to import (and then reimport) a directory which I sync to from box.com shared with me folder. I have used --duplicate option to not delete original files upon import. But then upon-rerunning import command git-annex would error out if file already exists. --reinject-duplicates seems to be the option to use, but all those modes are "exclusive" so I cannot use --duplicate --reinject-duplicates, and using --reinject-duplicates alone would result in removing original files (as without --duplicates)

What version of git-annex are you using? On what operating system?

7.20190819+git2-g908476a9b-1~ndall+1

Please provide any additional information below.

my little demo snippet for import with using --duplicate and then both options at the same time:

$> mkdir /tmp/d-in /tmp/d-repo && touch /tmp/d-in/file && ( cd /tmp/d-repo && git init && git annex init && for r in 1 2; do echo "Run $r";  ls -l ../d-in && git annex import --duplicate ../d-in/.; done )  
Initialized empty Git repository in /tmp/d-repo/.git/
init  ok
(recording state in git...)
Run 1
total 0
-rw------- 1 yoh yoh 0 Oct 14 10:51 file
import ./file ok
(recording state in git...)
Run 2
total 0
-rw------- 1 yoh yoh 0 Oct 14 10:51 file
import ./file 
  not overwriting existing ./file (is a symlink)
failed
git-annex: import: 1 failed


$> cd d-repo 
$> git annex import ../d-in/. --reinject-duplicates --duplicate 2>&1 | head -n 3                                                                                                                           
Invalid option `--duplicate'                                                    

Usage: git-annex COMMAND

Or may be there is a better way to establish re-runnable import from a directory workflow?