In place import

[joshmoore]

In combination with https://github.com/joshmoore/omero-user-scripts/blob/in-place-import/InPlaceImport.py, this PR allows for in-place imports. There are several further changes that may should happen to that script:

• Permit mv rather than ln -s
• Add optional call to chmod a-w
• Allow specifying target dataset (unneeded in Inplace Import via SymlinkFileTransfer (See #11573) #1989)
• Properly fill in version info (unneded in Inplace Import via SymlinkFileTransfer (See #11573) #1989)
• Allow globing
• Return log file as download
• Permit earlier disconnect

Once these are complete, it may should be moved into the http://github.com/ome/scripts repository.

Dry runs

By default, execution simulates bin/omero import -f and lists which files would be imported. This allows the detection of directories with multiple filesets.

Testing

• git clone https://github.com/joshmoore/omero-user-scripts.git lib/scripts/josh
• bin/omero script list
• bin/omero script launch /josh/InPlaceImport.py Path=/home/data/sample/leica-lif/Beta Catenin.lif Dry_Run=False

Expected linkage

/OMERO/ManagedRepository/root_0/2013-12/19/20-42-18.901/data/sample/leica-lif$ ls -ltra
lrwxrwxrwx 1 jamoore jamoore   44 Dec 19 20:42 Beta Catenin.lif -> /home/data/sample/leica-lif/Beta\ Catenin.lif
...

Web example

[ximenesuk]

This all works as expected both from the cli and through the web. One question though, why is more of the path used here? ie a normal import will give me a repo path of:

root_0/2014-01/01/14-08-46.811/IAGFP-Noc01_R3D.dv
root_0/2014-01/01/14-08-46.811/IAGFP-Noc01_R3D.dv.log

while an in-place import results in:

root_0/2014-01/01/14-01-06.863/Work/Images/dv/IAGFP-Noc01_R3D.dv
root_0/2014-01/01/14-01-06.863/Work/Images/dv/IAGFP-Noc01_R3D.dv.log

[joshmoore]

@ximenesuk : the difference is almost certainly the Python vs. the Java implementation. In thinking over this, I'm wondering if it wouldn't make more sense to teach CommandLineImporter how to do the symlink-ing so that the two implementations would be identical (no need for all the "TBDs")

[mtbc]

The "normal" import has ManagedRepositoryI.trimPaths utilizing FormatReader.getRequiredDirectories, falling back to retaining three parents if there is no help from Bio-Formats.

[joshmoore]

Thanks for the initial review, all. I'm closing this PR since it was mostly for discussion and investigation. I'll open a PR with these changes on dev_5_0 though so a proper in-place import PR can be built on top of it.

[joshmoore]

Marked off some of the TODOs since gh-1989 supports them.