✨ Features
- **io.plugins:** add loader for beamline ID21 ESM at NSLS-II ([c07e490](https://github.com/kmnhan/erlabpy/commit/c07e490c37b52706aa9407d6adb5aa7787e2c1b0))
This commit adds a new data loader for beamline ID21 ESM at NSLS-II, Brookhaven National Laboratory.
- **io.dataloader:** add formatters ([2ee9a4a](https://github.com/kmnhan/erlabpy/commit/2ee9a4a2361b91727c4e964adc00a26398863f2f))
A new attribute named `formatters` and a new method `get_formatted_attr_or_coord` has been added to loaders. This allows custom per-attribute pretty-printing behavior.
- **io:** add parallel argument to `load` ([88cd924](https://github.com/kmnhan/erlabpy/commit/88cd924efd9375ebe89417b504c80e55f7071404))
- **io:** add xarray backend for igor files ([1fe5ca5](https://github.com/kmnhan/erlabpy/commit/1fe5ca514777f4356f1b67251dc1a6f21b320d48))
`.pxt`, `.pxp`, and `.ibw` files can now be opened with xarray methods such as `xr.open_dataset` and `xr.open_dataarray`. See the updated user guide for more information.
🐞 Bug Fixes
- **io.dataloader:** properly reorder coordinates ([3ebfb0f](https://github.com/kmnhan/erlabpy/commit/3ebfb0fdbc21e690e51c185da92ed9c88921d3b1))
Coordinate order was broken for loaders which assign coordinates in inherited `post_process`. This is now fixed, and returned data will be consistently ordered with respect to the mapping, with the dimension coordinates coming first.
- **erlab.io.plugins.maestro:** temporary fix for xarray issue ([c2d04a3](https://github.com/kmnhan/erlabpy/commit/c2d04a31ee98cbbcc403dafce6cf8dd6e37a09e5))
- **io.plugins.da30:** properly handle output types ([6297aba](https://github.com/kmnhan/erlabpy/commit/6297aba2474ce4a28e865f48320d37e47c4ba1f7))
The DA30 loader now tries to return a dataset from `.zip` files only when there are no coordinate conflicts. In the case of conflicts, the loader will return a `DataTree`.
Also, single region DA30 `.pxt` files will now return a `DataArray` consistent with the equivalent `.ibw` file.
- **interactive.fermiedge:** allow transposed input to fermi edge fitting tool ([dcae75e](https://github.com/kmnhan/erlabpy/commit/dcae75e718dd6b0e2cc773e43407e3a6a2cd97f0))
- **interactive.imagetool:** retain attrs when exporting slice ([1bed572](https://github.com/kmnhan/erlabpy/commit/1bed572dc7309d723152237902b021d27d0dbb40))
When accessing the data of a single slice from the right-click menu of `ImageTool`, the attributes of the original data are now kept. This allows saved slices or data opened in other tools to retain their attributes.
- resolve gui script not working on windows with conda ([62253d0](https://github.com/kmnhan/erlabpy/commit/62253d0d64d9cf458ee3e58e20fc890914c13c18))
- **io.dataloader:** clear plot before loading in interactive summaries ([21d6dea](https://github.com/kmnhan/erlabpy/commit/21d6dea9876066c64b9a9d2dc62dd9d41ba3ec9b))
- **plotting.general:** fix `plot_slices` compatibility with slice object as argument ([5948a7b](https://github.com/kmnhan/erlabpy/commit/5948a7b64fb415cfe86668bb17f3c656df1466a6))
- **io:** disable memmapping when loading data ([c39da1b](https://github.com/kmnhan/erlabpy/commit/c39da1b64ad28447260b9131f38206131bd6c0cb))
Memmapping seemed to interfere loading multiple files when called through ipywidgets.
⚡️ Performance
- **interactive.imagetool:** improve manager speed ([891c4ee](https://github.com/kmnhan/erlabpy/commit/891c4eed0921fdb5b4ebd2cb17e6c932afa79ccc))
♻️ Code Refactor
- **io.utils:** use pathlib in `get_files` ([b7a0f5b](https://github.com/kmnhan/erlabpy/commit/b7a0f5b33e9a1f7a6840413f431fd60a10bb754e))
- **io.plugins.merlin:** combine ImageTool file menu into single entry ([6e28ac2](https://github.com/kmnhan/erlabpy/commit/6e28ac21d7b534f5243f25a40710910eefb3a2b8))
- **io:** add postprocessing and validation for `DataTree` objects ([3fb3ff5](https://github.com/kmnhan/erlabpy/commit/3fb3ff5ec3755a40d6f2a66e9f242e6be9529816))
- **io:** add warning when file is ambiguous ([8daabb8](https://github.com/kmnhan/erlabpy/commit/8daabb85eb4b4313fc05199fa0eb8ed4de6c80f6))
- **io:** remove renaming steps from `load_single` in multi-file loaders ([542f4f2](https://github.com/kmnhan/erlabpy/commit/542f4f2874f5255e71a53ecc58c0b87c68c84aaa))
Combining before renaming coords should be more straightforward
- **io:** allow missing alpha coord ([c9deed4](https://github.com/kmnhan/erlabpy/commit/c9deed43a0ecce20a70145d9fe6a21b2a1ea1693))
Validation checks will not warn about missing detector angle, allowing XPS measurements.
- **io.dataloader:** only allow real file or folder names as input ([b9a59cc](https://github.com/kmnhan/erlabpy/commit/b9a59cc4c4cd16c44c93237b24b519ab270aff79))
The previous behavior allowed passing `f_001.pxt` to load `f_001_S001.pxt`, `f_001_S002.pxt`... but this was confusing since there is no file named `f_001.pxt`. This commit disallows such input.
- cleanup erplot namespace ([007eedb](https://github.com/kmnhan/erlabpy/commit/007eedb8862fcdcd8a54bb170c5ec0272db3f194))
- **io:** implement metaclass ([e787b1c](https://github.com/kmnhan/erlabpy/commit/e787b1c63b1cf786797ddcd2c360438619574f80))
Whenever the `identify()` method failed to find any files, subclasses had to explicitly raise `FileNotFoundError`. This resulted in a lot of boilerplate code and ambiguous error messages. Now, all subclasses can just return `None` in `identify()` when no files are found. The appropriate error is automatically raised.
- **io.igor:** raise OSError on load fail ([6c7a4c4](https://github.com/kmnhan/erlabpy/commit/6c7a4c471874c8fb92c267f6c551e332e35d0bc6))
- move dataloader cell formatting implementation to utils module ([0f2cb1c](https://github.com/kmnhan/erlabpy/commit/0f2cb1c70b2c7dbadd5ab936770290486b50200a))
- **io:** deprecate calling igor functions from top level namespace ([bb8af7c](https://github.com/kmnhan/erlabpy/commit/bb8af7c76b1bf31f481de2a43f104f8c8a638225))
Calling `erlab.io.load_wave` and `erlab.io.load_experiment` is deprecated. When writing new code, use `xarray.load_dataarray` and `xarray.load_dataset` instead.