v2.3.0

What's Changed

• ✨ Support reading metadata of committed datatypes #96
• 🐛 Support Numpy 2 by allowing copy when converting scalar and nested sequences to numpy arrays #94 #95
• 🐛 Fix fetching boolean data as binary #99

Full Changelog: v2.2.0...v2.3.0

v2.2.0

What's Changed

• ✨ Add strpad to string dtype metadata #93

Full Changelog: v2.1.0...v2.2.0

v2.1.0

What's Changed

• ✨ Allow requesting value of non-numeric dataset as binary with format=bin, notably opaque datasets #89
• 🐛 Fix serialization of complex256 values to JSON #91
• ⚙️ Add / endpoint to FastAPI, Tornado and Flask routers, with support for both GET and HEAD requests, to check server status #92

Full Changelog: v2.0.0...v2.1.0

v2.0.0

⚠️ Breaking changes

• Now requires Python 3.8, h5py v3 and Pydantic v2 (for FastAPI) #82 #83
• For all entities metadata, the type field is renamed to kind #85
• For dataset and attribute metadata, the dtype field is replaced with a richer type field. #85
  • This type field is now an object that contains much more information about the type of the data.
    Ex: { dtype: '<f8' } becomes { type: {"class": 1, "dtype": "<f8", "size": 8, "order": 0}}
  • The old dtype field is still available in the type object (dtype field)
• The new type field brings support for non-obvious data types: opaque, variable-length, enum, etc.

Thanks to @axelboc for this big work on their first contribution 🎉

Other changes

• 🐛 Fix broken import of BaseSettings for pydantic v2 #82
• ✨ Datasets containing (u)int64 are no longer cast as (u)int32 when requesting data with dtype=safe #87
• ✨ The /paths endpoint now also includes links #88

Full Changelog: v1.3.0...v2.0.0

v1.3.0

What's Changed

• Add a /paths endpoint that returns the list of all the entities contained in the group. By default, the group is root so that it returns the list of all entities paths in the file #76
  • This relies on a new util get_list_of_paths that can be used in custom backends.
• Improve the error messages when query arguments are malformed/invalid (also return 422 instead of 500) #77
• Update the API doc to reflect these changes.

Full Changelog: v1.2.1...v1.3.0

v1.2.1

What's Changed

H5Grove now explicitly supports h5py >= 2.9.

• 🐛 Fix attribute and error handling for h5py 2.9 #75
• 🐛 Fix typing of path in get_content_from_file

Full Changelog: v1.2.0...v1.2.1

v1.2.0

What's Changed

⚠️ Dropped support for Python 3.6. Python 3.7 is now the minimal supported version.

Improved the handling of errors for backend utils and example implementations:

• Raise a 403 if a PermissionError is thrown when reading the HDF5 file #71
• Response on 404/403 errors is now consistent across backends (JSON payload: {message: <error message>}) #73

Full Changelog: v1.1.0...v1.2.0

v1.1.0

What's Changed

• The dtype field is now a dictionary of the dtypes of the compound fields for compound datasets and attributes #69

Full Changelog: v1.0.0...v1.1.0

v1.0.0 🎉

What's changed

• ✨ Added a FastAPI implementation in fastapi_utils #67 #68 ( thanks @shabihsherjeel ! )
  • ⚠️ Breaking change: flaskutils and tornadoutils were renamed to flask_utils and tornado_utils
• ✨ Added dtype argument to encode and /data/ endpoint #65. dtype can have two values:
  • origin (default): do not change the dtype of the returned data
  • safe: convert the dtype of the returned data to a dtype supported by JS TypedArray (if possible)
• ⚠️ Breaking change: Previously, the default behaviour of /data was to convert the dtype (safe). Now, by default, the dtype is unchanged (origin).

Full Changelog: v0.0.16...v1.0.0

v0.0.16

What's Changed

• Fix handling of np.float128 #62 ( thanks @cherbon1! )
• Include chunking and compression info in dataset metadata #64

Full Changelog: v0.0.15...v0.0.16