h5config/storage/hdf5.py

   1 # Copyright (C) 2011 W. Trevor King <wking@drexel.edu>
   2 #
   3 # This file is part of h5config.
   4 #
   5 # h5config is free software; you can redistribute it and/or modify it
   6 # under the terms of the GNU General Public License as published by the
   7 # Free Software Foundation, either version 3 of the License, or (at your
   8 # option) any later version.
   9 #
  10 # h5config is distributed in the hope that it will be useful, but
  11 # WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13 # General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License
  16 # along with h5config.  If not, see <http://www.gnu.org/licenses/>.
  17
  18 """HDF5 backend implementation
  19 """
  20
  21 import types as _types
  22
  23 import h5py as _h5py
  24 import numpy as _numpy
  25
  26 from .. import LOG as _LOG
  27 from .. import config as _config
  28 from . import FileStorage as _FileStorage
  29
  30
  31 def pprint_HDF5(*args, **kwargs):
  32     print pformat_HDF5(*args, **kwargs)
  33
  34 def pformat_HDF5(filename, group='/'):
  35     with _h5py.File(filename, 'r') as f:
  36         cwg = f[group]
  37         ret = '\n'.join(_pformat_hdf5(cwg))
  38     return ret
  39
  40 def _pformat_hdf5(cwg, depth=0):
  41     lines = []
  42     lines.append('  '*depth + cwg.name)
  43     depth += 1
  44     for key,value in cwg.iteritems():
  45         if isinstance(value, _h5py.Group):
  46             lines.extend(_pformat_hdf5(value, depth))
  47         elif isinstance(value, _h5py.Dataset):
  48             lines.append('  '*depth + str(value))
  49             lines.append('  '*(depth+1) + str(value[...]))
  50         else:
  51             lines.append('  '*depth + str(value))
  52     return lines
  53
  54 def h5_create_group(cwg, path, force=False):
  55     "Create the group where the settings are stored (if necessary)."
  56     if path == '/':
  57         return cwg
  58     gpath = ['']
  59     for group in path.strip('/').split('/'):
  60         gpath.append(group)
  61         if group not in cwg.keys():
  62             _LOG.debug('creating group {} in {}'.format(
  63                     '/'.join(gpath), cwg.file))
  64             cwg.create_group(group)
  65         _cwg = cwg[group]
  66         if isinstance(_cwg, _h5py.Dataset):
  67             if force:
  68                 _LOG.info('overwrite {} in {} ({}) with a group'.format(
  69                         '/'.join(gpath), _cwg.file, _cwg))
  70                 del cwg[group]
  71                 _cwg = cwg.create_group(group)
  72             else:
  73                 raise ValueError(_cwg)
  74         cwg = _cwg
  75     return cwg
  76
  77
  78 class HDF5_Storage (_FileStorage):
  79     """Back a `Config` class with an HDF5 file.
  80
  81     The `.save` and `.load` methods have an optional `group` argument
  82     that allows you to save and load settings from an externally
  83     opened HDF5 file.  This can make it easier to stash several
  84     related `Config` classes in a single file.  For example
  85
  86     >>> import os
  87     >>> import tempfile
  88     >>> from ..test import TestConfig
  89     >>> fd,filename = tempfile.mkstemp(
  90     ...     suffix='.'+HDF5_Storage.extension, prefix='pypiezo-')
  91     >>> os.close(fd)
  92
  93     >>> f = _h5py.File(filename, 'a')
  94     >>> c = TestConfig(storage=HDF5_Storage(
  95     ...     filename='untouched_file.h5', group='/untouched/group'))
  96     >>> c['alive'] = True
  97     >>> group = f.create_group('base')
  98     >>> c.save(group=group)
  99     >>> pprint_HDF5(filename)  # doctest: +REPORT_UDIFF
 100     /
 101       /base
 102         <HDF5 dataset "age": shape (), type "<f8">
 103           1.3
 104         <HDF5 dataset "alive": shape (), type "|b1">
 105           True
 106         <HDF5 dataset "bids": shape (3,), type "<f8">
 107           [ 5.4  3.2  1. ]
 108         <HDF5 dataset "children": shape (), type "|S1">
 109     <BLANKLINE>
 110         <HDF5 dataset "daisies": shape (), type "<i4">
 111           13
 112         <HDF5 dataset "name": shape (), type "|S1">
 113     <BLANKLINE>
 114         <HDF5 dataset "species": shape (), type "|S14">
 115           Norwegian Blue
 116         <HDF5 dataset "spouse": shape (), type "|S1">
 117     <BLANKLINE>
 118     >>> c.clear()
 119     >>> c['alive']
 120     False
 121     >>> c.load(group=group)
 122     >>> c['alive']
 123     True
 124
 125     >>> f.close()
 126     >>> os.remove(filename)
 127     """
 128     extension = 'h5'
 129
 130     def __init__(self, group='/', **kwargs):
 131         super(HDF5_Storage, self).__init__(**kwargs)
 132         assert group.startswith('/'), group
 133         if not group.endswith('/'):
 134             group += '/'
 135         self.group = group
 136         self._file_checked = False
 137
 138     def _check_file(self):
 139         if self._file_checked:
 140             return
 141         self._setup_file()
 142         self._file_checked = True
 143
 144     def _setup_file(self):
 145         with _h5py.File(self._filename, 'a') as f:
 146             cwg = f  # current working group
 147             h5_create_group(cwg, self.group)
 148
 149     def _load(self, config, group=None):
 150         f = None
 151         try:
 152             if group is None:
 153                 self._check_file()
 154                 f = _h5py.File(self._filename, 'r')
 155                 group = f[self.group]
 156             for s in config.settings:
 157                 if s.name not in group.keys():
 158                     continue
 159                 if isinstance(s, _config.ConfigListSetting):
 160                     try:
 161                         cwg = h5_create_group(group, s.name)
 162                     except ValueError:
 163                         pass
 164                     else:
 165                         value = []
 166                         for i in sorted(int(x) for x in cwg.keys()):
 167                             instance = s.config_class()
 168                             try:
 169                                 _cwg = h5_create_group(cwg, str(i))
 170                             except ValueError:
 171                                 pass
 172                             else:
 173                                 self._load(config=instance, group=_cwg)
 174                                 value.append(instance)
 175                         config[s.name] = value
 176                 elif isinstance(s, _config.ConfigSetting):
 177                     try:
 178                         cwg = h5_create_group(group, s.name)
 179                     except ValueError:
 180                         pass
 181                     else:
 182                         if not config[s.name]:
 183                             config[s.name] = s.config_class()
 184                         self._load(config=config[s.name], group=cwg)
 185                 else:
 186                     try:
 187                         v = group[s.name][...]
 188                     except Exception, e:
 189                         _LOG.error('Could not access {}/{}: {}'.format(
 190                                 group.name, s.name, e))
 191                         raise
 192                     if isinstance(v, _numpy.ndarray):
 193                         if isinstance(s, _config.BooleanSetting):
 194                             v = bool(v)  # array(True, dtype=bool) -> True
 195                         elif v.dtype.type == _numpy.string_:
 196                             v = str(v)  # array('abc', dtype='|S3') -> 'abc'
 197                         elif isinstance(s, _config.IntegerSetting):
 198                             v = int(v)  # array(3, dtpe='int32') -> 3
 199                         elif isinstance(s, _config.FloatSetting):
 200                             v = float(v)  # array(1.2, dtype='float64') -> 1.2
 201                         elif isinstance(s, _config.NumericSetting):
 202                             raise NotImplementedError(type(s))
 203                         elif isinstance(s, _config.FloatListSetting):
 204                             v = list(v)  # convert from numpy array
 205                     if isinstance(v, _types.StringTypes):
 206                         # convert back from None, etc.
 207                         v = s.convert_from_text(v)
 208                     config[s.name] = v
 209         finally:
 210             if f:
 211                 f.close()
 212
 213     def _save(self, config, group=None):
 214         f = None
 215         try:
 216             if group is None:
 217                 self._check_file()
 218                 f = _h5py.File(self._filename, 'a')
 219                 group = f[self.group]
 220             for s in config.settings:
 221                 value = None
 222                 if isinstance(s, (_config.BooleanSetting,
 223                                   _config.NumericSetting,
 224                                   _config.FloatListSetting)):
 225                     value = config[s.name]
 226                     if value in [None, []]:
 227                         value = s.convert_to_text(value)
 228                 elif isinstance(s, _config.ConfigListSetting):
 229                     configs = config[s.name]
 230                     if configs:
 231                         cwg = h5_create_group(group, s.name, force=True)
 232                         for i,cfg in enumerate(configs):
 233                             _cwg = h5_create_group(cwg, str(i), force=True)
 234                             self._save(config=cfg, group=_cwg)
 235                         continue
 236                 elif isinstance(s, _config.ConfigSetting):
 237                     cfg = config[s.name]
 238                     if cfg:
 239                         cwg = h5_create_group(group, s.name, force=True)
 240                         self._save(config=cfg, group=cwg)
 241                         continue
 242                 if value is None:  # not set yet, or invalid
 243                     value = s.convert_to_text(config[s.name])
 244                 try:
 245                     del group[s.name]
 246                 except KeyError:
 247                     pass
 248                 group[s.name] = value
 249         finally:
 250             if f:
 251                 f.close()