Module H5A

Provides access to the low-level HDF5 “H5A” attribute interface.

Functional API

h5py.h5a.create(ObjectID loc, STRING name, TypeID tid, SpaceID space, **kwds) → AttrID

Create a new attribute, attached to an existing object.

STRING obj_name (“.”)

Attach attribute to this group member instead

PropID lapl

Link access property list for obj_name

h5py.h5a.open(ObjectID loc, STRING name=, INT index=, **kwds) → AttrID

Open an attribute attached to an existing object. You must specify exactly one of either name or idx. Keywords are:

STRING obj_name (“.”)

Attribute is attached to this group member

PropID lapl (None)

Link access property list for obj_name

INT index_type (h5.INDEX_NAME)

INT order (h5.ITER_INC)

h5py.h5a.exists(ObjectID loc, STRING name, **kwds) → BOOL

Determine if an attribute is attached to this object. Keywords:

STRING obj_name (“.”)

Look for attributes attached to this group member

PropID lapl (None):

Link access property list for obj_name

h5py.h5a.rename(ObjectID loc, STRING name, STRING new_name, **kwds)

Rename an attribute. Keywords:

STRING obj_name (“.”)

Attribute is attached to this group member

PropID lapl (None)

Link access property list for obj_name

h5py.h5a.delete(ObjectID loc, STRING name=, INT index=, **kwds)

Remove an attribute from an object. Specify exactly one of “name” or “index”. Keyword-only arguments:

STRING obj_name (“.”)

Attribute is attached to this group member

PropID lapl (None)

Link access property list for obj_name

INT index_type (h5.INDEX_NAME)

INT order (h5.ITER_INC)

h5py.h5a.get_num_attrs(ObjectID loc) → INT

Determine the number of attributes attached to an HDF5 object.

h5py.h5a.get_info(ObjectID loc, STRING name=, INT index=, **kwds) → AttrInfo

Get information about an attribute, in one of two ways:

1. If you have the attribute identifier, just pass it in

2. If you have the parent object, supply it and exactly one of either name or index.

STRING obj_name (“.”)

Use this group member instead

PropID lapl (None)

Link access property list for obj_name

INT index_type (h5.INDEX_NAME)

Which index to use

INT order (h5.ITER_INC)

What order the index is in

h5py.h5a.iterate(ObjectID loc, CALLABLE func, INT index=0, **kwds) → <Return value from func>

Iterate a callable (function, method or callable object) over the attributes attached to this object. You callable should have the signature:

func(STRING name) => Result

or if the keyword argument “info” is True:

func(STRING name, AttrInfo info) => Result

Returning None continues iteration; returning anything else aborts iteration and returns that value. Keywords:

BOOL info (False)

Callback is func(STRING name, AttrInfo info), not func(STRING name)

INT index_type (h5.INDEX_NAME)

Which index to use

INT order (h5.ITER_INC)

Index order to use

Info objects

class h5py.h5a.AttrInfo

corder

Creation order

corder_valid

Indicates if the creation order is valid

cset

Character set of attribute name (integer typecode from h5t)

data_size

Size of raw data

Attribute objects

class h5py.h5a.AttrID

Logical representation of an HDF5 attribute identifier.

Objects of this class can be used in any HDF5 function call which expects an attribute identifier. Additionally, all H5A* functions which always take an attribute instance as the first argument are presented as methods of this class.

• Hashable: No

• Equality: Identifier comparison

dtype

A Numpy-stype dtype object representing the attribute’s datatype

get_name() → STRING name

Determine the name of this attribute.

get_space() → SpaceID

Create and return a copy of the attribute’s dataspace.

get_storage_size() → INT

Get the amount of storage required for this attribute.

get_type() → TypeID

Create and return a copy of the attribute’s datatype.

name

The attribute’s name

read(NDARRAY arr, TypeID mtype=None)

Read the attribute data into the given Numpy array. Note that the Numpy array must have the same shape as the HDF5 attribute, and a conversion-compatible datatype.

The Numpy array must be writable and C-contiguous. If this is not the case, the read will fail with an exception.

If provided, the HDF5 TypeID mtype will override the array’s dtype.

shape

A Numpy-style shape tuple representing the attribute’s dataspace

write(NDARRAY arr)

Write the contents of a Numpy array to the attribute. Note that the Numpy array must have the same shape as the HDF5 attribute, and a conversion-compatible datatype.

The Numpy array must be C-contiguous. If this is not the case, the write will fail with an exception.