Sort keys in metadata .json files (#211)

* wip: draft mmstack ome-tiff fov

* MM FOV base class

* tests

* bump tifffile

* comment

* fix indent after rebase

* use get default

* test pixel indexing

* set MM metadata

* style

* update dependencies

* add xarray

* move old readers to the `_deprecated` namespace

* uapi for ndtiff

* refactor test setup to parametrize by dataset
use globals instead of fixtures since parametrization happens before fixture evaluation

* convert mmstack

* fix and test chunking

* fix metadata conversion and test ndtiff

* update cli

* fix scaling

* test 1.4 and incomplete ome-tiffs

* move reader tests

* deprecate reader tests

* update deprecated tests

* update ngff tests

* isort

* update black target to 3.10

* lint

* fix download paths

* update docs references and theme

* untrack autogenerated file

* ignore execution time file

* add github icon

* update docstring

* update docstring

* show channel names and chunk size in info

* print plate chunk size if verbose

* fallback for pixel size

* remove log level setting

* do not filter logs and warnings in reader

* avoid root logger

* isort

* set default logging level to INFO

* format docstring

* improve conversion messages

* black

* fix ome-tiff channel name indexing

* fix ndtiff channel name indexing

* update converter test

* remove use of os.path in `reader`

* expand _check_ndtiff checks

* fix iteration

* fix python 3.10

using `Path.glob(*/)` to get subdirs was added in 3.11

* bump zarr version to include resizing fix
zarr-developers/zarr-python#1540

* fix cli default

* set log level with an environment variable

* fix unset

* catch non-existent page

* implement fallback for incomplete channel names
workaround for the Dragonfly microscope where the multi-camera setup only has one channel name written

* add debug logs

* handle virtual frames

* try reading pages from TiffFile directly

* filter error logs about ImageJ metadata being broken
this is a known MM limitation when writing OME-TIFFs

* fix regex

* remove use of os.path in `convert.py`

* better channel indexing in `_get_summary_metadata`

* style

* safer NoneType check

* private default axis names for NDTiff

* sort metadata keys

* Squashed commit of the following:

commit fac2c13
Author: Ivan Ivanov <[email protected]>
Date:   Tue Apr 9 11:25:36 2024 -0700

    Fix bug reading dragonfly acquisitions (#215)

    * fix bug reading dragonfly acquisitions

    * fix typo

    * style

    * bugfix

commit 0c6984e
Author: Ivan Ivanov <[email protected]>
Date:   Mon Mar 11 12:35:51 2024 -0700

    Fix bug determining number of rows and cols (#214)

    * fix bug determining number of rows and cols

    * add another XY Stage variation

    * add docs and fix style

commit 3ab89ba
Author: Ziwen Liu <[email protected]>
Date:   Mon Mar 4 11:02:49 2024 -0800

    Universal API implementations for Micro-Manager OME-TIFF and NDTiff (#185)

    * wip: draft mmstack ome-tiff fov

    * MM FOV base class

    * tests

    * bump tifffile

    * comment

    * fix indent after rebase

    * use get default

    * test pixel indexing

    * set MM metadata

    * style

    * update dependencies

    * add xarray

    * move old readers to the `_deprecated` namespace

    * uapi for ndtiff

    * refactor test setup to parametrize by dataset
    use globals instead of fixtures since parametrization happens before fixture evaluation

    * convert mmstack

    * fix and test chunking

    * fix metadata conversion and test ndtiff

    * update cli

    * fix scaling

    * test 1.4 and incomplete ome-tiffs

    * move reader tests

    * deprecate reader tests

    * update deprecated tests

    * update ngff tests

    * isort

    * update black target to 3.10

    * lint

    * fix download paths

    * update docs references and theme

    * untrack autogenerated file

    * ignore execution time file

    * add github icon

    * update docstring

    * update docstring

    * show channel names and chunk size in info

    * print plate chunk size if verbose

    * fallback for pixel size

    * remove log level setting

    * do not filter logs and warnings in reader

    * avoid root logger

    * isort

    * set default logging level to INFO

    * format docstring

    * improve conversion messages

    * black

    * fix ome-tiff channel name indexing

    * fix ndtiff channel name indexing

    * update converter test

    * remove use of os.path in `reader`

    * expand _check_ndtiff checks

    * fix iteration

    * fix python 3.10

    using `Path.glob(*/)` to get subdirs was added in 3.11

    * bump zarr version to include resizing fix
    zarr-developers/zarr-python#1540

    * fix cli default

    * set log level with an environment variable

    * fix unset

    * catch non-existent page

    * implement fallback for incomplete channel names
    workaround for the Dragonfly microscope where the multi-camera setup only has one channel name written

    * add debug logs

    * handle virtual frames

    * try reading pages from TiffFile directly

    * filter error logs about ImageJ metadata being broken
    this is a known MM limitation when writing OME-TIFFs

    * fix regex

    * remove use of os.path in `convert.py`

    * better channel indexing in `_get_summary_metadata`

    * style

    * safer NoneType check

    * private default axis names for NDTiff

    * update documentation to reflect new entry point

    * add repr to MM FOV and dataset types

    * rename mm_meta and expose summary metadata

    * add MicroManagerFOVMapping.root

    * add MicroManagerFOVMapping.zyx_scale

    * add warning log for failed position grid

    * fix grid layout

    * suppress hypothesis flakiness

    * different health check suppression

    ---------

    Co-authored-by: Ivan Ivanov <[email protected]>

* black

* bugfix

---------

Co-authored-by: Ziwen Liu <[email protected]>

iohub/convert.py

@@ -320,6 +320,7 @@ def _create_zeros_array(
 
     def _convert_image_plane_metadata(self, fov, zarr_name: str):
         position_image_plane_metadata = {}
+        sorted_keys = []
         for t_idx, c_idx in product(
             range(self.t),
             range(self.c),
@@ -344,9 +345,16 @@ def _convert_image_plane_metadata(self, fov, zarr_name: str):
                         "incomplete."
                     )
                     continue
+                if not sorted_keys:
+                    # Sort keys, ordering keys without dashes first
+                    sorted_keys = sorted(
+                        metadata.keys(), key=lambda x: ('-' in x, x)
+                    )
+
+                sorted_metadata = {key: metadata[key] for key in sorted_keys}
                 # T/C/Z
                 frame_key = "/".join([str(i) for i in (t_idx, c_idx, z_idx)])
-                position_image_plane_metadata[frame_key] = metadata
+                position_image_plane_metadata[frame_key] = sorted_metadata
         with open(
             self.output_dir / zarr_name / "image_plane_metadata.json",
             mode="x",