Module H5T

HDF5 “H5T” data-type API

This module contains the datatype identifier class TypeID, and its subclasses which represent things like integer/float/compound identifiers. The majority of the H5T API is presented as methods on these identifiers.

Functions specific to h5py

h5py.h5t.py_create(OBJECT dtype_in, BOOL logical=False) TypeID

Given a Numpy dtype object, generate a byte-for-byte memory-compatible HDF5 datatype object. The result is guaranteed to be transient and unlocked.

Parameters:
  • dtype_in – may be a dtype object, or anything which can be converted to a dtype, including strings like ‘<i4’ or an “int”.

  • logical – when this flag is set, instead of returning a byte-for-byte identical representation of the type, the function returns the closest logically appropriate HDF5 type. For example, in the case of a “hinted” dtype of kind “O” representing a string, it would return an HDF5 variable- length string type.

h5py.h5t.string_dtype(encoding='utf-8', length=None)

Make a numpy dtype for HDF5 strings

encoding may be ‘utf-8’ or ‘ascii’.

length may be an integer for a fixed length string dtype, or None for variable length strings. String lengths for HDF5 are counted in bytes, not unicode code points.

For variable length strings, the data should be passed as Python str objects (unicode in Python 2) if the encoding is ‘utf-8’, and bytes if it is ‘ascii’. For fixed length strings, the data should be numpy fixed length bytes arrays, regardless of the encoding. Fixed length unicode data is not supported.

h5py.h5t.check_string_dtype(dt)

If the dtype represents an HDF5 string, returns a string_info object.

The returned string_info object holds the encoding and the length. The encoding can only be ‘utf-8’ or ‘ascii’. The length may be None for a variable-length string, or a fixed length in bytes.

Returns None if the dtype does not represent an HDF5 string.

h5py.h5t.vlen_dtype(basetype)

Make a numpy dtype for an HDF5 variable-length datatype

For variable-length string dtypes, use string_dtype() instead.

h5py.h5t.check_vlen_dtype(dt)

If the dtype represents an HDF5 vlen, returns the Python base class.

Returns None if the dtype does not represent an HDF5 vlen.

h5py.h5t.enum_dtype(values_dict, basetype=<class 'numpy.uint8'>)

Create a NumPy representation of an HDF5 enumerated type

values_dict maps string names to integer values. basetype is an appropriate integer base dtype large enough to hold the possible options.

h5py.h5t.check_enum_dtype(dt)

If the dtype represents an HDF5 enumerated type, returns the dictionary mapping string names to integer values.

Returns None if the dtype does not represent an HDF5 enumerated type.

h5py.h5t.special_dtype(**kwds)

Create a new h5py “special” type. Only one keyword may be given.

Legal keywords are:

vlen = basetype

Base type for HDF5 variable-length datatype. This can be Python str type or instance of np.dtype. Example: special_dtype( vlen=str )

enum = (basetype, values_dict)

Create a NumPy representation of an HDF5 enumerated type. Provide a 2-tuple containing an (integer) base dtype and a dict mapping string names to integer values.

ref = Reference | RegionReference

Create a NumPy representation of an HDF5 object or region reference type.

h5py.h5t.check_dtype(**kwds)

Check a dtype for h5py special type “hint” information. Only one keyword may be given.

vlen = dtype

If the dtype represents an HDF5 vlen, returns the Python base class. Currently only built-in string vlens (str) are supported. Returns None if the dtype does not represent an HDF5 vlen.

enum = dtype

If the dtype represents an HDF5 enumerated type, returns the dictionary mapping string names to integer values. Returns None if the dtype does not represent an HDF5 enumerated type.

ref = dtype

If the dtype represents an HDF5 reference type, returns the reference class (either Reference or RegionReference). Returns None if the dtype does not represent an HDF5 reference type.

Functional API

h5py.h5t.create(INT classtype, UINT size) TypeID

Create a new HDF5 type object. Legal class values are COMPOUND and OPAQUE. Use enum_create for enums.

h5py.h5t.open(ObjectID group, STRING name) TypeID

Open a named datatype from a file. If present, tapl must be a datatype access property list.

h5py.h5t.array_create(TypeID base, TUPLE dimensions) TypeArrayID

Create a new array datatype, using and HDF5 parent type and dimensions given via a tuple of positive integers. “Unlimited” dimensions are not allowed.

h5py.h5t.enum_create(TypeID base) TypeID

Create a new enumerated type based on an (integer) parent type.

h5py.h5t.vlen_create(TypeID base) TypeID

Create a new variable-length datatype, using any HDF5 type as a base.

Although the Python interface can manipulate these types, there is no provision for reading/writing vlen data.

h5py.h5t.decode(STRING buf) TypeID

Deserialize an HDF5 type. You can also do this with the native Python pickling machinery.

h5py.h5t.convert(TypeID src, TypeID dst, UINT n, NDARRAY buf, NDARRAY bkg=None, PropID dxpl=None)

Convert n contiguous elements of a buffer in-place. The array dtype is ignored. The backing buffer is optional; for conversion of compound types, a temporary copy of conversion buffer will used for backing if one is not supplied.

h5py.h5t.find(TypeID src, TypeID dst) TUPLE or None

Determine if a conversion path exists from src to dst. Result is None or a tuple describing the conversion path. Currently tuple entries are:

  1. INT need_bkg: Whether this routine requires a backing buffer.

    Values are BKG_NO, BKG_TEMP and BKG_YES.

Type classes

class h5py.h5t.TypeID

Base class for type identifiers (implements common operations)

  • Hashable: If committed or locked

  • Equality: Logical H5T comparison

commit(ObjectID group, STRING name, PropID lcpl=None)

Commit this (transient) datatype to a named datatype in a file. If present, lcpl may be a link creation property list.

committed() BOOL is_comitted

Determine if a given type object is named (T) or transient (F).

copy() TypeID

Create a copy of this type object.

detect_class(INT classtype) BOOL class_is_present

Determine if a member of the given class exists in a compound datatype. The search is recursive.

dtype

A Numpy-style dtype object representing this object.

encode() STRING

Serialize an HDF5 type. Bear in mind you can also use the native Python pickle/unpickle machinery to do this. The returned string may contain binary values, including NULLs.

equal(TypeID typeid) BOOL

Logical comparison between datatypes. Also called by Python’s “==” operator.

get_class() INT classcode

Determine the datatype’s class code.

get_create_plist() PropTCID

Create and return a new copy of the datatype creation property list used when this datatype was created.

get_size() INT size

Determine the total size of a datatype, in bytes.

get_super() TypeID

Determine the parent type of an array, enumeration or vlen datatype.

lock()

Lock this datatype, which makes it immutable and indestructible. Once locked, it can’t be unlocked.

set_size(UINT size)

Set the total size of the datatype, in bytes.

Atomic classes

Atomic types are integers and floats. Much of the functionality for each is inherited from the base class TypeAtomicID.

class h5py.h5t.TypeAtomicID

Bases: TypeID

Base class for atomic datatypes (float or integer)

get_offset() INT offset

Get the offset of the first significant bit.

get_order() INT order

Obtain the byte order of the datatype; one of:

get_pad() -> (INT lsb_pad_code, INT msb_pad_code)

Determine the padding type. Possible values are:

get_precision() UINT precision

Get the number of significant bits (excludes padding).

set_offset(UINT offset)

Set the offset of the first significant bit.

set_order(INT order)

Set the byte order of the datatype; one of:

set_pad(INT lsb_pad_code, INT msb_pad_code)

Set the padding type. Possible values are:

set_precision(UINT precision)

Set the number of significant bits (excludes padding).

class h5py.h5t.TypeIntegerID

Bases: TypeAtomicID

Integer atomic datatypes

get_sign() INT sign

Get the “signedness” of the datatype; one of:

SGN_NONE

Unsigned

SGN_2

Signed 2’s complement

set_sign(INT sign)

Set the “signedness” of the datatype; one of:

SGN_NONE

Unsigned

SGN_2

Signed 2’s complement

class h5py.h5t.TypeFloatID

Bases: TypeAtomicID

Floating-point atomic datatypes

get_ebias() UINT ebias

Get the exponent bias.

get_fields() TUPLE field_info

Get information about floating-point bit fields. See the HDF5 docs for a full description. Tuple has the following members:

  1. UINT spos

  2. UINT epos

  3. UINT esize

  4. UINT mpos

  5. UINT msize

get_inpad() INT pad_code

Determine the internal padding strategy. Legal values are:

get_norm() INT normalization_code

Get the normalization strategy. Legal values are:

set_ebias(UINT ebias)

Set the exponent bias.

set_fields(UINT spos, UINT epos, UINT esize, UINT mpos, UINT msize)

Set floating-point bit fields. Refer to the HDF5 docs for argument definitions.

set_inpad(INT pad_code)

Set the internal padding strategy. Legal values are:

set_norm(INT normalization_code)

Set the normalization strategy. Legal values are:

Strings

class h5py.h5t.TypeStringID

Bases: TypeID

String datatypes, both fixed and vlen.

get_cset() INT character_set

Retrieve the character set used for a string.

get_strpad() INT padding_type

Get the padding type. Legal values are:

STR_NULLTERM

NULL termination only (C style)

STR_NULLPAD

Pad buffer with NULLs

STR_SPACEPAD

Pad buffer with spaces (FORTRAN style)

is_variable_str() BOOL is_variable

Determine if the given string datatype is a variable-length string.

set_cset(INT character_set)

Set the character set used for a string.

set_strpad(INT pad)

Set the padding type. Legal values are:

STR_NULLTERM

NULL termination only (C style)

STR_NULLPAD

Pad buffer with NULLs

STR_SPACEPAD

Pad buffer with spaces (FORTRAN style)

Compound Types

Traditional compound type (like NumPy record type) and enumerated types share a base class, TypeCompositeID.

class h5py.h5t.TypeCompositeID

Bases: TypeID

Base class for enumerated and compound types.

get_member_index(STRING name) INT index

Determine the index of a member of a compound or enumerated datatype identified by a string name.

get_member_name(INT member) STRING name

Determine the name of a member of a compound or enumerated type, identified by its index (0 <= member < nmembers).

get_nmembers() INT number_of_members

Determine the number of members in a compound or enumerated type.

class h5py.h5t.TypeCompoundID

Bases: TypeCompositeID

Represents a compound datatype

get_member_class(INT member) INT class

Determine the datatype class of the member of a compound type, identified by its index (0 <= member < nmembers).

get_member_offset(INT member) INT offset

Determine the offset, in bytes, of the beginning of the specified member of a compound datatype.

get_member_type(INT member) TypeID

Create a copy of a member of a compound datatype, identified by its index.

insert(STRING name, UINT offset, TypeID field)

Add a named member datatype to a compound datatype. The parameter offset indicates the offset from the start of the compound datatype, in bytes.

pack()

Recursively removes padding (introduced on account of e.g. compiler alignment rules) from a compound datatype.

class h5py.h5t.TypeEnumID

Bases: TypeCompositeID

Represents an enumerated type

enum_insert(STRING name, INT/LONG value)

Define a new member of an enumerated type. The value will be automatically converted to the base type defined for this enum. If the conversion results in overflow, the value will be silently clipped.

enum_nameof(LONG value) STRING name

Determine the name associated with the given value. Due to a limitation of the HDF5 library, this can only retrieve names up to 1023 characters in length.

enum_valueof(STRING name) LONG value

Get the value associated with an enum name.

get_member_value(UINT index) LONG value

Determine the value for the member at the given zero-based index.

Other types

class h5py.h5t.TypeArrayID

Bases: TypeID

Represents an array datatype

get_array_dims() TUPLE dimensions

Get the dimensions of the given array datatype as a tuple of integers.

get_array_ndims() INT rank

Get the rank of the given array datatype.

class h5py.h5t.TypeOpaqueID

Bases: TypeID

Represents an opaque type

get_tag() STRING tag

Get the tag associated with an opaque datatype.

set_tag(STRING tag)

Set a string describing the contents of an opaque datatype. Limited to 256 characters.

class h5py.h5t.TypeVlenID

Bases: TypeID

Non-string vlen datatypes.

class h5py.h5t.TypeBitfieldID

Bases: TypeID

HDF5 bitfield type

get_order() INT order

Obtain the byte order of the datatype; one of:

class h5py.h5t.TypeReferenceID

Bases: TypeID

HDF5 object or region reference

Predefined Datatypes

These locked types are pre-allocated by the library.

Floating-point

h5py.h5t.IEEE_F32LE
h5py.h5t.IEEE_F32BE
h5py.h5t.IEEE_F64LE
h5py.h5t.IEEE_F64BE

Integer types

h5py.h5t.STD_I8LE
h5py.h5t.STD_I16LE
h5py.h5t.STD_I32LE
h5py.h5t.STD_I64LE
h5py.h5t.STD_I8BE
h5py.h5t.STD_I16BE
h5py.h5t.STD_I32BE
h5py.h5t.STD_I64BE
h5py.h5t.STD_U8LE
h5py.h5t.STD_U16LE
h5py.h5t.STD_U32LE
h5py.h5t.STD_U64LE
h5py.h5t.STD_U8BE
h5py.h5t.STD_U16BE
h5py.h5t.STD_U32BE
h5py.h5t.STD_U64BE
h5py.h5t.NATIVE_INT8
h5py.h5t.NATIVE_UINT8
h5py.h5t.NATIVE_INT16
h5py.h5t.NATIVE_UINT16
h5py.h5t.NATIVE_INT32
h5py.h5t.NATIVE_UINT32
h5py.h5t.NATIVE_INT64
h5py.h5t.NATIVE_UINT64
h5py.h5t.NATIVE_FLOAT
h5py.h5t.NATIVE_DOUBLE

Reference types

h5py.h5t.STD_REF_OBJ
h5py.h5t.STD_REF_DSETREG

String types

h5py.h5t.C_S1

Null-terminated fixed-length string

h5py.h5t.FORTRAN_S1

Zero-padded fixed-length string

h5py.h5t.VARIABLE

Variable-length string

Python object type

h5py.h5t.PYTHON_OBJECT

Module constants

Datatype class codes

h5py.h5t.NO_CLASS
h5py.h5t.INTEGER
h5py.h5t.FLOAT
h5py.h5t.TIME
h5py.h5t.STRING
h5py.h5t.BITFIELD
h5py.h5t.OPAQUE
h5py.h5t.COMPOUND
h5py.h5t.REFERENCE
h5py.h5t.ENUM
h5py.h5t.VLEN
h5py.h5t.ARRAY

API Constants

h5py.h5t.SGN_NONE
h5py.h5t.SGN_2
h5py.h5t.ORDER_LE
h5py.h5t.ORDER_BE
h5py.h5t.ORDER_VAX
h5py.h5t.ORDER_NONE
h5py.h5t.ORDER_NATIVE
h5py.h5t.DIR_DEFAULT
h5py.h5t.DIR_ASCEND
h5py.h5t.DIR_DESCEND
h5py.h5t.STR_NULLTERM
h5py.h5t.STR_NULLPAD
h5py.h5t.STR_SPACEPAD
h5py.h5t.NORM_IMPLIED
h5py.h5t.NORM_MSBSET
h5py.h5t.NORM_NONE
h5py.h5t.CSET_ASCII
h5py.h5t.CSET_UTF8
h5py.h5t.PAD_ZERO
h5py.h5t.PAD_ONE
h5py.h5t.PAD_BACKGROUND
h5py.h5t.BKG_NO
h5py.h5t.BKG_TEMP
h5py.h5t.BKG_YES