def write_to_h5(infile, h5file, h5path='/', mode="a",
overwrite_data=False, link_type="soft",
create_dataset_args=None, min_size=500):
"""Write content of a h5py-like object into a HDF5 file.
:param infile: Path of input file, or :class:`commonh5.File` object
or :class:`commonh5.Group` object.
:param h5file: Path of output HDF5 file or HDF5 file handle
(`h5py.File` object)
:param str h5path: Target path in HDF5 file in which scan groups are created.
Default is root (``"/"``)
:param str mode: Can be ``"r+"`` (read/write, file must exist),
``"w"`` (write, existing file is lost), ``"w-"`` (write, fail
if exists) or ``"a"`` (read/write if exists, create otherwise).
This parameter is ignored if ``h5file`` is a file handle.
:param bool overwrite_data: If ``True``, existing groups and datasets can be
overwritten, if ``False`` they are skipped. This parameter is only
relevant if ``file_mode`` is ``"r+"`` or ``"a"``.
:param str link_type: *"soft"* (default) or *"hard"*
:param dict create_dataset_args: Dictionary of args you want to pass to
``h5py.File.create_dataset``. This allows you to specify filters and
compression parameters. Don't specify ``name`` and ``data``.
These arguments are only applied to datasets larger than 1MB.
:param int min_size: Minimum number of elements in a dataset to apply
chunking and compression. Default is 500.
The structure of the spec data in an HDF5 file is described in the
documentation of :mod:`silx.io.spech5`.
"""
writer = Hdf5Writer(h5path=h5path,
overwrite_data=overwrite_data,
link_type=link_type,
create_dataset_args=create_dataset_args,
min_size=min_size)
# both infile and h5file can be either file handle or a file name: 4 cases
if not isinstance(h5file, h5py.File) and not is_group(infile):
with silx.io.open(infile) as h5pylike:
if not _is_commonh5_group(h5pylike):
raise IOError("Cannot convert HDF5 file %s to HDF5" % infile)
with h5py.File(h5file, mode) as h5f:
writer.write(h5pylike, h5f)
elif isinstance(h5file, h5py.File) and not is_group(infile):
with silx.io.open(infile) as h5pylike:
if not _is_commonh5_group(h5pylike):
raise IOError("Cannot convert HDF5 file %s to HDF5" % infile)
writer.write(h5pylike, h5file)
elif is_group(infile) and not isinstance(h5file, h5py.File):
if not _is_commonh5_group(infile):
raise IOError("Cannot convert HDF5 file %s to HDF5" % infile.file.name)
with h5py.File(h5file, mode) as h5f:
writer.write(infile, h5f)
else:
if not _is_commonh5_group(infile):
raise IOError("Cannot convert HDF5 file %s to HDF5" % infile.file.name)
writer.write(infile, h5file)
def write_to_h5(infile, h5file, h5path='/', mode="a",
overwrite_data=False, link_type="soft",
create_dataset_args=None, min_size=500):
"""Write content of a h5py-like object into a HDF5 file.
:param infile: Path of input file, or :class:`commonh5.File` object
or :class:`commonh5.Group` object.
:param h5file: Path of output HDF5 file or HDF5 file handle
(`h5py.File` object)
:param str h5path: Target path in HDF5 file in which scan groups are created.
Default is root (``"/"``)
:param str mode: Can be ``"r+"`` (read/write, file must exist),
``"w"`` (write, existing file is lost), ``"w-"`` (write, fail
if exists) or ``"a"`` (read/write if exists, create otherwise).
This parameter is ignored if ``h5file`` is a file handle.
:param bool overwrite_data: If ``True``, existing groups and datasets can be
overwritten, if ``False`` they are skipped. This parameter is only
relevant if ``file_mode`` is ``"r+"`` or ``"a"``.
:param str link_type: *"soft"* (default) or *"hard"*
:param dict create_dataset_args: Dictionary of args you want to pass to
``h5py.File.create_dataset``. This allows you to specify filters and
compression parameters. Don't specify ``name`` and ``data``.
These arguments are only applied to datasets larger than 1MB.
:param int min_size: Minimum number of elements in a dataset to apply
chunking and compression. Default is 500.
The structure of the spec data in an HDF5 file is described in the
documentation of :mod:`silx.io.spech5`.
"""
writer = Hdf5Writer(h5path=h5path,
overwrite_data=overwrite_data,
link_type=link_type,
create_dataset_args=create_dataset_args,
min_size=min_size)
# both infile and h5file can be either file handle or a file name: 4 cases
if not isinstance(h5file, h5py.File) and not is_group(infile):
with silx.io.open(infile) as h5pylike:
if not _is_commonh5_group(h5pylike):
raise IOError("Cannot convert HDF5 file %s to HDF5" % infile)
with h5py.File(h5file, mode) as h5f:
writer.write(h5pylike, h5f)
elif isinstance(h5file, h5py.File) and not is_group(infile):
with silx.io.open(infile) as h5pylike:
if not _is_commonh5_group(h5pylike):
raise IOError("Cannot convert HDF5 file %s to HDF5" % infile)
writer.write(h5pylike, h5file)
elif is_group(infile) and not isinstance(h5file, h5py.File):
if not _is_commonh5_group(infile):
raise IOError("Cannot convert HDF5 file %s to HDF5" % infile.file.name)
with h5py.File(h5file, mode) as h5f:
writer.write(infile, h5f)
else:
if not _is_commonh5_group(infile):
raise IOError("Cannot convert HDF5 file %s to HDF5" % infile.file.name)
writer.write(infile, h5file)
def append_member_to_h5(self, h5like_name, obj):
"""Add one group or one dataset to :attr:`h5f`"""
h5_name = self.h5path + h5like_name.lstrip("/")
if is_softlink(obj):
# links to be created after all groups and datasets
h5_target = self.h5path + obj.path.lstrip("/")
self._links.append((h5_name, h5_target))
elif is_dataset(obj):
_logger.debug("Saving dataset: " + h5_name)
member_initially_exists = h5_name in self._h5f
if self.overwrite_data and member_initially_exists:
_logger.warning("Overwriting dataset: " + h5_name)
del self._h5f[h5_name]
if self.overwrite_data or not member_initially_exists:
if fabioh5 is not None and \
isinstance(obj, fabioh5.FrameData) and \
len(obj.shape) > 2:
# special case of multiframe data
# write frame by frame to save memory usage low
ds = self._h5f.create_dataset(h5_name,
shape=obj.shape,
dtype=obj.dtype,
**self.create_dataset_args)
for i, frame in enumerate(obj):
ds[i] = frame
else:
# fancy arguments don't apply to small dataset
if obj.size < self.min_size:
ds = self._h5f.create_dataset(h5_name, data=obj.value)
else:
ds = self._h5f.create_dataset(h5_name, data=obj.value,
**self.create_dataset_args)
else:
ds = self._h5f[h5_name]
# add HDF5 attributes
for key in obj.attrs:
if self.overwrite_data or key not in ds.attrs:
ds.attrs.create(key,
_attr_utf8(obj.attrs[key]))
if not self.overwrite_data and member_initially_exists:
_logger.warning("Not overwriting existing dataset: " + h5_name)
elif is_group(obj):
if h5_name not in self._h5f:
_logger.debug("Creating group: " + h5_name)
grp = self._h5f.create_group(h5_name)
else:
grp = self._h5f[h5_name]
# add HDF5 attributes
for key in obj.attrs:
if self.overwrite_data or key not in grp.attrs:
grp.attrs.create(key,
_attr_utf8(obj.attrs[key]))
def append_member_to_h5(self, h5like_name, obj):
"""Add one group or one dataset to :attr:`h5f`"""
h5_name = self.h5path + h5like_name.lstrip("/")
if is_softlink(obj):
# links to be created after all groups and datasets
h5_target = self.h5path + obj.path.lstrip("/")
self._links.append((h5_name, h5_target))
elif is_dataset(obj):
_logger.debug("Saving dataset: " + h5_name)
member_initially_exists = h5_name in self._h5f
if self.overwrite_data and member_initially_exists:
_logger.warning("Overwriting dataset: " + h5_name)
del self._h5f[h5_name]
if self.overwrite_data or not member_initially_exists:
if fabioh5 is not None and \
isinstance(obj, fabioh5.FrameData) and \
len(obj.shape) > 2:
# special case of multiframe data
# write frame by frame to save memory usage low
ds = self._h5f.create_dataset(h5_name,
shape=obj.shape,
dtype=obj.dtype,
**self.create_dataset_args)
for i, frame in enumerate(obj):
ds[i] = frame
else:
# fancy arguments don't apply to small dataset
if obj.size < self.min_size:
ds = self._h5f.create_dataset(h5_name, data=obj.value)
else:
ds = self._h5f.create_dataset(
h5_name,
data=obj.value,
**self.create_dataset_args)
else:
ds = self._h5f[h5_name]
# add HDF5 attributes
for key in obj.attrs:
if self.overwrite_data or key not in ds.attrs:
ds.attrs.create(key, _attr_utf8(obj.attrs[key]))
if not self.overwrite_data and member_initially_exists:
_logger.warning("Not overwriting existing dataset: " + h5_name)
elif is_group(obj):
if h5_name not in self._h5f:
_logger.debug("Creating group: " + h5_name)
grp = self._h5f.create_group(h5_name)
else:
grp = self._h5f[h5_name]
# add HDF5 attributes
for key in obj.attrs:
if self.overwrite_data or key not in grp.attrs:
grp.attrs.create(key, _attr_utf8(obj.attrs[key]))
def isGroup(item):
if isinstance(item, Group):
return True
elif hasattr(item, "keys"):
return True
elif is_group(item):
return True
else:
return False
def __init__(self, ffile, node, parent=None, path=None):
self.__sorting = False
self.__sorting_list = None
self.__sorting_order = qt.Qt.AscendingOrder
if 1: #with ffile.plock:
self._file = ffile
self._parent = parent
if hasattr(node, '_posixPath'):
self._name = node._posixPath
else:
self._name = node.name
"""
if hasattr(node, "_sourceName"):
self._name = node._sourceName
else:
self._name = posixpath.basename(node.name)
"""
self._type = type(node).__name__
self._hasChildren = is_group(node)
#self._attrs = []
self._color = qt.QColor(qt.Qt.black)
if hasattr(node, 'attrs'):
attrs = list(node.attrs)
for cname in ['class', 'NX_class']:
if cname in attrs:
nodeattr = node.attrs[cname]
if sys.version < '3.0':
_type = "%s" % nodeattr
elif hasattr(nodeattr, "decode"):
_type = nodeattr.decode('utf=8')
else:
_type = "%s" % nodeattr
self._type = _type
if _type in ["NXdata"]:
self._color = qt.QColor(qt.Qt.blue)
elif ("default" in attrs):
self._color = qt.QColor(qt.Qt.blue)
#self._attrs = attrs
break
#self._type = _type[2].upper() + _type[3:]
self._children = []
if hasattr(node, 'dtype'):
self._dtype = safe_str(node.dtype)
else:
self._dtype = ""
if hasattr(node, 'shape'):
if 0:
self._shape = safe_str(node.shape)
else:
self._shape = node.shape
else:
self._shape = ""
def __init__(self, ffile, node, parent=None, path=None):
if 1:#with ffile.plock:
self._file = ffile
self._parent = parent
if hasattr(node, '_posixPath'):
self._name = node._posixPath
else:
self._name = node.name
"""
if hasattr(node, "_sourceName"):
self._name = node._sourceName
else:
self._name = posixpath.basename(node.name)
"""
self._type = type(node).__name__
self._hasChildren = is_group(node)
#self._attrs = []
self._color = qt.QColor(qt.Qt.black)
if hasattr(node, 'attrs'):
attrs = list(node.attrs)
for cname in ['class', 'NX_class']:
if cname in attrs:
nodeattr = node.attrs[cname]
if sys.version <'3.0':
_type = "%s" % nodeattr
elif hasattr(nodeattr, "decode"):
_type = nodeattr.decode('utf=8')
else:
_type = "%s" % nodeattr
self._type = _type
if _type in ["NXdata"]:
self._color = qt.QColor(qt.Qt.blue)
elif ("default" in attrs):
self._color = qt.QColor(qt.Qt.blue)
#self._attrs = attrs
break
#self._type = _type[2].upper() + _type[3:]
self._children = []
if hasattr(node, 'dtype'):
self._dtype = safe_str(node.dtype)
else:
self._dtype = ""
if hasattr(node, 'shape'):
if 0:
self._shape = safe_str(node.shape)
else:
self._shape = node.shape
else:
self._shape = ""
def append_member_to_h5(self, h5like_name, obj):
"""Add one group or one dataset to :attr:`h5f`"""
h5_name = self.h5path + h5like_name.lstrip("/")
if is_softlink(obj):
# links to be created after all groups and datasets
h5_target = self.h5path + obj.path.lstrip("/")
self._links.append((h5_name, h5_target))
elif is_dataset(obj):
_logger.debug("Saving dataset: " + h5_name)
member_initially_exists = h5_name in self._h5f
if self.overwrite_data and member_initially_exists:
_logger.warn("Overwriting dataset: " + h5_name)
del self._h5f[h5_name]
if self.overwrite_data or not member_initially_exists:
# fancy arguments don't apply to small dataset
if obj.size < self.min_size:
ds = self._h5f.create_dataset(h5_name, data=obj.value)
else:
ds = self._h5f.create_dataset(h5_name, data=obj.value,
**self.create_dataset_args)
else:
ds = self._h5f[h5_name]
# add HDF5 attributes
for key in obj.attrs:
if self.overwrite_data or key not in ds.attrs:
ds.attrs.create(key,
_attr_utf8(obj.attrs[key]))
if not self.overwrite_data and member_initially_exists:
_logger.warn("Not overwriting existing dataset: " + h5_name)
elif is_group(obj):
if h5_name not in self._h5f:
_logger.debug("Creating group: " + h5_name)
grp = self._h5f.create_group(h5_name)
else:
grp = self._h5f[h5_name]
# add HDF5 attributes
for key in obj.attrs:
if self.overwrite_data or key not in grp.attrs:
grp.attrs.create(key,
_attr_utf8(obj.attrs[key]))
def __init__(self, ffile, node, parent=None, path=None):
if 1: #with ffile.plock:
self._file = ffile
self._parent = parent
if hasattr(node, '_posixPath'):
self._name = node._posixPath
else:
self._name = node.name
"""
if hasattr(node, "_sourceName"):
self._name = node._sourceName
else:
self._name = posixpath.basename(node.name)
"""
self._type = type(node).__name__
self._hasChildren = is_group(node)
if hasattr(node, 'attrs'):
attrs = list(node.attrs)
for cname in ['class', 'NX_class']:
if cname in attrs:
nodeattr = node.attrs[cname]
if sys.version < '3.0':
_type = "%s" % nodeattr
elif hasattr(nodeattr, "decode"):
_type = nodeattr.decode('utf=8')
else:
_type = "%s" % nodeattr
self._type = _type
break
#self._type = _type[2].upper() + _type[3:]
self._children = []
if hasattr(node, 'dtype'):
self._dtype = safe_str(node.dtype)
else:
self._dtype = ""
if hasattr(node, 'shape'):
if 0:
self._shape = safe_str(node.shape)
else:
self._shape = node.shape
else:
self._shape = ""
def _is_commonh5_group(grp):
"""Return True if grp is a commonh5 group.
(h5py.Group objects are not commonh5 groups)"""
return is_group(grp) and not isinstance(grp, h5py.Group)
def _is_commonh5_group(grp):
"""Return True if grp is a commonh5 group.
(h5py.Group objects are not commonh5 groups)"""
return is_group(grp) and not isinstance(grp, h5py.Group)