cvs2svn_lib/cvs_path.py

   1 # (Be in -*- python -*- mode.)
   2 #
   3 # ====================================================================
   4 # Copyright (c) 2000-2008 CollabNet.  All rights reserved.
   5 #
   6 # This software is licensed as described in the file COPYING, which
   7 # you should have received as part of this distribution.  The terms
   8 # are also available at http://subversion.tigris.org/license-1.html.
   9 # If newer versions of this license are posted there, you may use a
  10 # newer version instead, at your option.
  11 #
  12 # This software consists of voluntary contributions made by many
  13 # individuals.  For exact contribution history, see the revision
  14 # history and logs, available at http://cvs2svn.tigris.org/.
  15 # ====================================================================
  16
  17 """Classes that represent files and directories within CVS repositories."""
  18
  19 import os
  20
  21 from cvs2svn_lib.common import path_join
  22 from cvs2svn_lib.context import Ctx
  23
  24
  25 class CVSPath(object):
  26   """Represent a CVS file or directory.
  27
  28   Members:
  29
  30     id -- (int) unique ID for this CVSPath.  At any moment, there is
  31         at most one CVSPath instance with a particular ID.  (This
  32         means that object identity is the same as object equality, and
  33         objects can be used as map keys even though they don't have a
  34         __hash__() method).
  35
  36     project -- (Project) the project containing this CVSPath.
  37
  38     parent_directory -- (CVSDirectory or None) the CVSDirectory
  39         containing this CVSPath.
  40
  41     basename -- (string) the base name of this CVSPath (no ',v').  The
  42         basename of the root directory of a project is ''.
  43
  44     ordinal -- (int) the order that this instance should be sorted
  45         relative to other CVSPath instances.  This member is set based
  46         on the ordering imposed by sort_key() by CVSPathDatabase after
  47         all CVSFiles have been processed.  Comparisons of CVSPath
  48         using __cmp__() simply compare the ordinals.
  49
  50   """
  51
  52   __slots__ = [
  53       'id',
  54       'project',
  55       'parent_directory',
  56       'basename',
  57       'ordinal',
  58       'filename',
  59       ]
  60
  61   def __init__(self, id, project, parent_directory, basename):
  62     self.id = id
  63     self.project = project
  64     self.parent_directory = parent_directory
  65     self.basename = basename
  66
  67     self.filename = os.path.normpath(self._calculate_filename())
  68
  69   def __getstate__(self):
  70     """This method must only be called after ordinal has been set."""
  71
  72     return (
  73         self.id, self.project.id,
  74         self.parent_directory, self.basename,
  75         self.ordinal,
  76         )
  77
  78   def __setstate__(self, state):
  79     (
  80         self.id, project_id,
  81         self.parent_directory, self.basename,
  82         self.ordinal,
  83         ) = state
  84     self.project = Ctx()._projects[project_id]
  85     self.filename = os.path.normpath(self._calculate_filename())
  86
  87   def get_filename(self):
  88     """Return the filesystem path to this CVSPath in the CVS repository.
  89
  90     This is in native format, and already normalised the way
  91     os.path.normpath() normalises paths.
  92
  93     It starts with the repository path passed to run_options.add_project()
  94     in the options.py file."""
  95
  96     # This turns out to be a hot path through the code.
  97     # It's used by SubtreeSymbolTransform and similar transforms, so it's
  98     # called at least:
  99     #   (num_files * num_symbols_per_file * num_subtree_symbol_transforms)
 100     # times.  On a large repository with several subtree symbol transforms,
 101     # that can exceed 100,000,000 calls.  And _calculate_filename() is quite
 102     # complex, so doing that every time could add about 10 minutes to the
 103     # cvs2svn runtime.
 104     #
 105     # So now we precalculate this and just return it.
 106
 107     return self.filename
 108
 109   def get_ancestry(self):
 110     """Return a list of the CVSPaths leading from the root path to SELF.
 111
 112     Return the CVSPaths in a list, starting with
 113     self.project.get_root_cvs_directory() and ending with self."""
 114
 115     ancestry = []
 116     p = self
 117     while p is not None:
 118       ancestry.append(p)
 119       p = p.parent_directory
 120
 121     ancestry.reverse()
 122     return ancestry
 123
 124   def get_cvs_path(self):
 125     """Return the canonical path within the Project.
 126
 127     The canonical path:
 128
 129     - Uses forward slashes
 130
 131     - Doesn't include ',v' for files
 132
 133     - This doesn't include the 'Attic' segment of the path unless the
 134       file is to be left in an Attic directory in the SVN repository;
 135       i.e., if a filename exists in and out of Attic and the
 136       --retain-conflicting-attic-files option was specified.
 137
 138     """
 139
 140     return path_join(*[p.basename for p in self.get_ancestry()[1:]])
 141
 142   cvs_path = property(get_cvs_path)
 143
 144   def _get_dir_components(self):
 145     """Return a list containing the components of the path leading to SELF.
 146
 147     The return value contains the base names of all of the parent
 148     directories (except for the root directory) and SELF."""
 149
 150     return [p.basename for p in self.get_ancestry()[1:]]
 151
 152   def __eq__(a, b):
 153     """Compare two CVSPath instances for equality.
 154
 155     This method is supplied to avoid using __cmp__() for comparing for
 156     equality."""
 157
 158     return a is b
 159
 160   def sort_key(self):
 161     """Return the key that should be used for sorting CVSPath instances.
 162
 163     This is a relatively expensive computation, so it is only used
 164     once, the the results are used to set the ordinal member."""
 165
 166     return (
 167         # Sort first by project:
 168         self.project,
 169         # Then by directory components:
 170         self._get_dir_components(),
 171         )
 172
 173   def __cmp__(a, b):
 174     """This method must only be called after ordinal has been set."""
 175
 176     return cmp(a.ordinal, b.ordinal)
 177
 178
 179 class CVSDirectory(CVSPath):
 180   """Represent a CVS directory.
 181
 182   Members:
 183
 184     id -- (int or None) unique id for this file.  If None, a new id is
 185         generated.
 186
 187     project -- (Project) the project containing this file.
 188
 189     parent_directory -- (CVSDirectory or None) the CVSDirectory
 190         containing this CVSDirectory.
 191
 192     basename -- (string) the base name of this CVSDirectory (no ',v').
 193
 194     ordinal -- (int) the order that this instance should be sorted
 195         relative to other CVSPath instances.  See CVSPath.ordinal.
 196
 197     empty_subdirectory_ids -- (list of int) a list of the ids of any
 198         direct subdirectories that are empty.  (An empty directory is
 199         defined to be a directory that doesn't contain any RCS files
 200         or non-empty subdirectories.
 201
 202   """
 203
 204   __slots__ = ['empty_subdirectory_ids']
 205
 206   def __init__(self, id, project, parent_directory, basename):
 207     """Initialize a new CVSDirectory object."""
 208
 209     CVSPath.__init__(self, id, project, parent_directory, basename)
 210
 211     # This member is filled in by CollectData.close():
 212     self.empty_subdirectory_ids = []
 213
 214   def _calculate_filename(self):
 215     """Return the filesystem path to this CVSPath in the CVS repository."""
 216
 217     if self.parent_directory is None:
 218       return self.project.project_cvs_repos_path
 219     else:
 220       return os.path.join(
 221           self.parent_directory.filename, self.basename
 222           )
 223
 224   def __getstate__(self):
 225     return (
 226         CVSPath.__getstate__(self),
 227         self.empty_subdirectory_ids,
 228         )
 229
 230   def __setstate__(self, state):
 231     (
 232         cvs_path_state,
 233         self.empty_subdirectory_ids,
 234         ) = state
 235     CVSPath.__setstate__(self, cvs_path_state)
 236
 237   def __str__(self):
 238     """For convenience only.  The format is subject to change at any time."""
 239
 240     return self.cvs_path + '/'
 241
 242   def __repr__(self):
 243     return 'CVSDirectory<%x>(%r)' % (self.id, str(self),)
 244
 245
 246 class CVSFile(CVSPath):
 247   """Represent a CVS file.
 248
 249   Members:
 250
 251     id -- (int) unique id for this file.
 252
 253     project -- (Project) the project containing this file.
 254
 255     parent_directory -- (CVSDirectory) the CVSDirectory containing
 256         this CVSFile.
 257
 258     basename -- (string) the base name of this CVSFile (no ',v').
 259
 260     ordinal -- (int) the order that this instance should be sorted
 261         relative to other CVSPath instances.  See CVSPath.ordinal.
 262
 263     _in_attic -- (bool) True if RCS file is in an Attic subdirectory
 264         that is not considered the parent directory.  (If a file is
 265         in-and-out-of-attic and one copy is to be left in Attic after
 266         the conversion, then the Attic directory is that file's
 267         PARENT_DIRECTORY and _IN_ATTIC is False.)
 268
 269     executable -- (bool) True iff RCS file has executable bit set.
 270
 271     file_size -- (long) size of the RCS file in bytes.
 272
 273     mode -- (string or None) 'kkv', 'kb', etc., as read from the CVS
 274         file.
 275
 276     description -- (string or None) the file description as read from
 277         the RCS file.
 278
 279     properties -- (dict) file properties that are preserved across
 280         this history of this file.  Keys are strings; values are
 281         strings (indicating the property value) or None (indicating
 282         that the property should be left unset).  These properties can
 283         be overridden by CVSRevision.properties.  Different backends
 284         can use these properties for different purposes; for cvs2svn
 285         they become SVN versioned properties.  Properties whose names
 286         start with underscore are reserved for internal cvs2svn
 287         purposes.
 288
 289   PARENT_DIRECTORY might contain an 'Attic' component if it should be
 290   retained in the SVN repository; i.e., if the same filename exists out
 291   of Attic and the --retain-conflicting-attic-files option was specified.
 292
 293   """
 294
 295   __slots__ = [
 296       '_in_attic',
 297       'executable',
 298       'file_size',
 299       'mode',
 300       'description',
 301       'properties',
 302       ]
 303
 304   def __init__(
 305         self, id, project, parent_directory, basename, in_attic,
 306         executable, file_size, mode, description
 307         ):
 308     """Initialize a new CVSFile object."""
 309
 310     assert parent_directory is not None
 311
 312     # This member is needed by _calculate_filename(), which is called
 313     # by CVSPath.__init__().  So initialize it before calling
 314     # CVSPath.__init__().
 315     self._in_attic = in_attic
 316     CVSPath.__init__(self, id, project, parent_directory, basename)
 317
 318     self.executable = executable
 319     self.file_size = file_size
 320     self.mode = mode
 321     self.description = description
 322     self.properties = None
 323
 324   def determine_file_properties(self, file_property_setters):
 325     """Determine the properties for this file from FILE_PROPERTY_SETTERS.
 326
 327     This must only be called after SELF.mode and SELF.description have
 328     been set by CollectData."""
 329
 330     self.properties = {}
 331
 332     for file_property_setter in file_property_setters:
 333       file_property_setter.set_properties(self)
 334
 335   def _calculate_filename(self):
 336     """Return the filesystem path to this CVSPath in the CVS repository."""
 337
 338     if self._in_attic:
 339       return os.path.join(
 340           self.parent_directory.filename, 'Attic', self.basename + ',v'
 341           )
 342     else:
 343       return os.path.join(
 344           self.parent_directory.filename, self.basename + ',v'
 345           )
 346
 347   def __getstate__(self):
 348     return (
 349         CVSPath.__getstate__(self),
 350         self._in_attic, self.executable, self.file_size, self.mode,
 351         self.description, self.properties,
 352         )
 353
 354   def __setstate__(self, state):
 355     (
 356         cvs_path_state,
 357         self._in_attic, self.executable, self.file_size, self.mode,
 358         self.description, self.properties,
 359         ) = state
 360     CVSPath.__setstate__(self, cvs_path_state)
 361
 362   def __str__(self):
 363     """For convenience only.  The format is subject to change at any time."""
 364
 365     return self.cvs_path
 366
 367   def __repr__(self):
 368     return 'CVSFile<%x>(%r)' % (self.id, str(self),)
 369
 370