Coverage for /pythoncovmergedfiles/medio/medio/usr/local/lib/python3.8/site-packages/h5py/_hl/attrs.py: 20%
123 statements
« prev ^ index » next coverage.py v7.4.0, created at 2024-01-03 07:57 +0000
« prev ^ index » next coverage.py v7.4.0, created at 2024-01-03 07:57 +0000
1# This file is part of h5py, a Python interface to the HDF5 library.
2#
3# http://www.h5py.org
4#
5# Copyright 2008-2013 Andrew Collette and contributors
6#
7# License: Standard 3-clause BSD; see "license.txt" for full license terms
8# and contributor agreement.
10"""
11 Implements high-level operations for attributes.
13 Provides the AttributeManager class, available on high-level objects
14 as <obj>.attrs.
15"""
17import numpy
18import uuid
20from .. import h5, h5s, h5t, h5a, h5p
21from . import base
22from .base import phil, with_phil, Empty, is_empty_dataspace, product
23from .datatype import Datatype
26class AttributeManager(base.MutableMappingHDF5, base.CommonStateObject):
28 """
29 Allows dictionary-style access to an HDF5 object's attributes.
31 These are created exclusively by the library and are available as
32 a Python attribute at <object>.attrs
34 Like Group objects, attributes provide a minimal dictionary-
35 style interface. Anything which can be reasonably converted to a
36 Numpy array or Numpy scalar can be stored.
38 Attributes are automatically created on assignment with the
39 syntax <obj>.attrs[name] = value, with the HDF5 type automatically
40 deduced from the value. Existing attributes are overwritten.
42 To modify an existing attribute while preserving its type, use the
43 method modify(). To specify an attribute of a particular type and
44 shape, use create().
45 """
47 def __init__(self, parent):
48 """ Private constructor.
49 """
50 self._id = parent.id
52 @with_phil
53 def __getitem__(self, name):
54 """ Read the value of an attribute.
55 """
56 attr = h5a.open(self._id, self._e(name))
57 shape = attr.shape
59 # shape is None for empty dataspaces
60 if shape is None:
61 return Empty(attr.dtype)
63 dtype = attr.dtype
65 # Do this first, as we'll be fiddling with the dtype for top-level
66 # array types
67 htype = h5t.py_create(dtype)
69 # NumPy doesn't support top-level array types, so we have to "fake"
70 # the correct type and shape for the array. For example, consider
71 # attr.shape == (5,) and attr.dtype == '(3,)f'. Then:
72 if dtype.subdtype is not None:
73 subdtype, subshape = dtype.subdtype
74 shape = attr.shape + subshape # (5, 3)
75 dtype = subdtype # 'f'
77 arr = numpy.zeros(shape, dtype=dtype, order='C')
78 attr.read(arr, mtype=htype)
80 string_info = h5t.check_string_dtype(dtype)
81 if string_info and (string_info.length is None):
82 # Vlen strings: convert bytes to Python str
83 arr = numpy.array([
84 b.decode('utf-8', 'surrogateescape') for b in arr.flat
85 ], dtype=dtype).reshape(arr.shape)
87 if arr.ndim == 0:
88 return arr[()]
89 return arr
91 def get_id(self, name):
92 """Get a low-level AttrID object for the named attribute.
93 """
94 return h5a.open(self._id, self._e(name))
96 @with_phil
97 def __setitem__(self, name, value):
98 """ Set a new attribute, overwriting any existing attribute.
100 The type and shape of the attribute are determined from the data. To
101 use a specific type or shape, or to preserve the type of an attribute,
102 use the methods create() and modify().
103 """
104 self.create(name, data=value)
106 @with_phil
107 def __delitem__(self, name):
108 """ Delete an attribute (which must already exist). """
109 h5a.delete(self._id, self._e(name))
111 def create(self, name, data, shape=None, dtype=None):
112 """ Create a new attribute, overwriting any existing attribute.
114 name
115 Name of the new attribute (required)
116 data
117 An array to initialize the attribute (required)
118 shape
119 Shape of the attribute. Overrides data.shape if both are
120 given, in which case the total number of points must be unchanged.
121 dtype
122 Data type of the attribute. Overrides data.dtype if both
123 are given.
124 """
125 name = self._e(name)
127 with phil:
128 # First, make sure we have a NumPy array. We leave the data type
129 # conversion for HDF5 to perform.
130 if not isinstance(data, Empty):
131 data = base.array_for_new_object(data, specified_dtype=dtype)
133 if shape is None:
134 shape = data.shape
135 elif isinstance(shape, int):
136 shape = (shape,)
138 use_htype = None # If a committed type is given, we must use it
139 # in the call to h5a.create.
141 if isinstance(dtype, Datatype):
142 use_htype = dtype.id
143 dtype = dtype.dtype
144 elif dtype is None:
145 dtype = data.dtype
146 else:
147 dtype = numpy.dtype(dtype) # In case a string, e.g. 'i8' is passed
149 original_dtype = dtype # We'll need this for top-level array types
151 # Where a top-level array type is requested, we have to do some
152 # fiddling around to present the data as a smaller array of
153 # subarrays.
154 if dtype.subdtype is not None:
156 subdtype, subshape = dtype.subdtype
158 # Make sure the subshape matches the last N axes' sizes.
159 if shape[-len(subshape):] != subshape:
160 raise ValueError("Array dtype shape %s is incompatible with data shape %s" % (subshape, shape))
162 # New "advertised" shape and dtype
163 shape = shape[0:len(shape)-len(subshape)]
164 dtype = subdtype
166 # Not an array type; make sure to check the number of elements
167 # is compatible, and reshape if needed.
168 else:
170 if shape is not None and product(shape) != product(data.shape):
171 raise ValueError("Shape of new attribute conflicts with shape of data")
173 if shape != data.shape:
174 data = data.reshape(shape)
176 # We need this to handle special string types.
177 if not isinstance(data, Empty):
178 data = numpy.asarray(data, dtype=dtype)
180 # Make HDF5 datatype and dataspace for the H5A calls
181 if use_htype is None:
182 htype = h5t.py_create(original_dtype, logical=True)
183 htype2 = h5t.py_create(original_dtype) # Must be bit-for-bit representation rather than logical
184 else:
185 htype = use_htype
186 htype2 = None
188 if isinstance(data, Empty):
189 space = h5s.create(h5s.NULL)
190 else:
191 space = h5s.create_simple(shape)
193 # For a long time, h5py would create attributes with a random name
194 # and then rename them, imitating how you can atomically replace
195 # a file in a filesystem. But HDF5 does not offer atomic replacement
196 # (you have to delete the existing attribute first), and renaming
197 # exposes some bugs - see https://github.com/h5py/h5py/issues/1385
198 # So we've gone back to the simpler delete & recreate model.
199 if h5a.exists(self._id, name):
200 h5a.delete(self._id, name)
202 attr = h5a.create(self._id, name, htype, space)
203 try:
204 if not isinstance(data, Empty):
205 attr.write(data, mtype=htype2)
206 except:
207 attr.close()
208 h5a.delete(self._id, name)
209 raise
210 attr.close()
212 def modify(self, name, value):
213 """ Change the value of an attribute while preserving its type.
215 Differs from __setitem__ in that if the attribute already exists, its
216 type is preserved. This can be very useful for interacting with
217 externally generated files.
219 If the attribute doesn't exist, it will be automatically created.
220 """
221 with phil:
222 if not name in self:
223 self[name] = value
224 else:
225 attr = h5a.open(self._id, self._e(name))
227 if is_empty_dataspace(attr):
228 raise OSError("Empty attributes can't be modified")
230 # If the input data is already an array, let HDF5 do the conversion.
231 # If it's a list or similar, don't make numpy guess a dtype for it.
232 dt = None if isinstance(value, numpy.ndarray) else attr.dtype
233 value = numpy.asarray(value, order='C', dtype=dt)
235 # Allow the case of () <-> (1,)
236 if (value.shape != attr.shape) and not \
237 (value.size == 1 and product(attr.shape) == 1):
238 raise TypeError("Shape of data is incompatible with existing attribute")
239 attr.write(value)
241 @with_phil
242 def __len__(self):
243 """ Number of attributes attached to the object. """
244 # I expect we will not have more than 2**32 attributes
245 return h5a.get_num_attrs(self._id)
247 def __iter__(self):
248 """ Iterate over the names of attributes. """
249 with phil:
251 attrlist = []
252 def iter_cb(name, *args):
253 """ Callback to gather attribute names """
254 attrlist.append(self._d(name))
256 cpl = self._id.get_create_plist()
257 crt_order = cpl.get_attr_creation_order()
258 cpl.close()
259 if crt_order & h5p.CRT_ORDER_TRACKED:
260 idx_type = h5.INDEX_CRT_ORDER
261 else:
262 idx_type = h5.INDEX_NAME
264 h5a.iterate(self._id, iter_cb, index_type=idx_type)
266 for name in attrlist:
267 yield name
269 @with_phil
270 def __contains__(self, name):
271 """ Determine if an attribute exists, by name. """
272 return h5a.exists(self._id, self._e(name))
274 @with_phil
275 def __repr__(self):
276 if not self._id:
277 return "<Attributes of closed HDF5 object>"
278 return "<Attributes of HDF5 object at %s>" % id(self._id)