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 slow_compare() by CollectData 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 slow_compare(a, b):
 161     return (
 162         # Sort first by project:
 163         cmp(a.project, b.project)
 164         # Then by directory components:
 165         or cmp(a._get_dir_components(), b._get_dir_components())
 166         )
 167
 168   def __cmp__(a, b):
 169     """This method must only be called after ordinal has been set."""
 170
 171     return cmp(a.ordinal, b.ordinal)
 172
 173
 174 class CVSDirectory(CVSPath):
 175   """Represent a CVS directory.
 176
 177   Members:
 178
 179     id -- (int or None) unique id for this file.  If None, a new id is
 180         generated.
 181
 182     project -- (Project) the project containing this file.
 183
 184     parent_directory -- (CVSDirectory or None) the CVSDirectory
 185         containing this CVSDirectory.
 186
 187     basename -- (string) the base name of this CVSDirectory (no ',v').
 188
 189     ordinal -- (int) the order that this instance should be sorted
 190         relative to other CVSPath instances.  See CVSPath.ordinal.
 191
 192     empty_subdirectory_ids -- (list of int) a list of the ids of any
 193         direct subdirectories that are empty.  (An empty directory is
 194         defined to be a directory that doesn't contain any RCS files
 195         or non-empty subdirectories.
 196
 197   """
 198
 199   __slots__ = ['empty_subdirectory_ids']
 200
 201   def __init__(self, id, project, parent_directory, basename):
 202     """Initialize a new CVSDirectory object."""
 203
 204     CVSPath.__init__(self, id, project, parent_directory, basename)
 205
 206     # This member is filled in by CollectData.close():
 207     self.empty_subdirectory_ids = []
 208
 209   def _calculate_filename(self):
 210     """Return the filesystem path to this CVSPath in the CVS repository."""
 211
 212     if self.parent_directory is None:
 213       return self.project.project_cvs_repos_path
 214     else:
 215       return os.path.join(
 216           self.parent_directory.filename, self.basename
 217           )
 218
 219   def __getstate__(self):
 220     return (
 221         CVSPath.__getstate__(self),
 222         self.empty_subdirectory_ids,
 223         )
 224
 225   def __setstate__(self, state):
 226     (
 227         cvs_path_state,
 228         self.empty_subdirectory_ids,
 229         ) = state
 230     CVSPath.__setstate__(self, cvs_path_state)
 231
 232   def __str__(self):
 233     """For convenience only.  The format is subject to change at any time."""
 234
 235     return self.cvs_path + '/'
 236
 237   def __repr__(self):
 238     return 'CVSDirectory<%x>(%r)' % (self.id, str(self),)
 239
 240
 241 class CVSFile(CVSPath):
 242   """Represent a CVS file.
 243
 244   Members:
 245
 246     id -- (int) unique id for this file.
 247
 248     project -- (Project) the project containing this file.
 249
 250     parent_directory -- (CVSDirectory) the CVSDirectory containing
 251         this CVSFile.
 252
 253     basename -- (string) the base name of this CVSFile (no ',v').
 254
 255     ordinal -- (int) the order that this instance should be sorted
 256         relative to other CVSPath instances.  See CVSPath.ordinal.
 257
 258     _in_attic -- (bool) True if RCS file is in an Attic subdirectory
 259         that is not considered the parent directory.  (If a file is
 260         in-and-out-of-attic and one copy is to be left in Attic after
 261         the conversion, then the Attic directory is that file's
 262         PARENT_DIRECTORY and _IN_ATTIC is False.)
 263
 264     executable -- (bool) True iff RCS file has executable bit set.
 265
 266     file_size -- (long) size of the RCS file in bytes.
 267
 268     mode -- (string or None) 'kkv', 'kb', etc., as read from the CVS
 269         file.
 270
 271     description -- (string or None) the file description as read from
 272         the RCS file.
 273
 274     properties -- (dict) file properties that are preserved across
 275         this history of this file.  Keys are strings; values are
 276         strings (indicating the property value) or None (indicating
 277         that the property should be left unset).  These properties can
 278         be overridden by CVSRevision.properties.  Different backends
 279         can use these properties for different purposes; for cvs2svn
 280         they become SVN versioned properties.  Properties whose names
 281         start with underscore are reserved for internal cvs2svn
 282         purposes.
 283
 284   PARENT_DIRECTORY might contain an 'Attic' component if it should be
 285   retained in the SVN repository; i.e., if the same filename exists out
 286   of Attic and the --retain-conflicting-attic-files option was specified.
 287
 288   """
 289
 290   __slots__ = [
 291       '_in_attic',
 292       'executable',
 293       'file_size',
 294       'mode',
 295       'description',
 296       'properties',
 297       ]
 298
 299   def __init__(
 300         self, id, project, parent_directory, basename, in_attic,
 301         executable, file_size, mode, description
 302         ):
 303     """Initialize a new CVSFile object."""
 304
 305     assert parent_directory is not None
 306
 307     # This member is needed by _calculate_filename(), which is called
 308     # by CVSPath.__init__().  So initialize it before calling
 309     # CVSPath.__init__().
 310     self._in_attic = in_attic
 311     CVSPath.__init__(self, id, project, parent_directory, basename)
 312
 313     self.executable = executable
 314     self.file_size = file_size
 315     self.mode = mode
 316     self.description = description
 317     self.properties = None
 318
 319   def determine_file_properties(self, file_property_setters):
 320     """Determine the properties for this file from FILE_PROPERTY_SETTERS.
 321
 322     This must only be called after SELF.mode and SELF.description have
 323     been set by CollectData."""
 324
 325     self.properties = {}
 326
 327     for file_property_setter in file_property_setters:
 328       file_property_setter.set_properties(self)
 329
 330   def _calculate_filename(self):
 331     """Return the filesystem path to this CVSPath in the CVS repository."""
 332
 333     if self._in_attic:
 334       return os.path.join(
 335           self.parent_directory.filename, 'Attic', self.basename + ',v'
 336           )
 337     else:
 338       return os.path.join(
 339           self.parent_directory.filename, self.basename + ',v'
 340           )
 341
 342   def __getstate__(self):
 343     return (
 344         CVSPath.__getstate__(self),
 345         self._in_attic, self.executable, self.file_size, self.mode,
 346         self.description, self.properties,
 347         )
 348
 349   def __setstate__(self, state):
 350     (
 351         cvs_path_state,
 352         self._in_attic, self.executable, self.file_size, self.mode,
 353         self.description, self.properties,
 354         ) = state
 355     CVSPath.__setstate__(self, cvs_path_state)
 356
 357   def __str__(self):
 358     """For convenience only.  The format is subject to change at any time."""
 359
 360     return self.cvs_path
 361
 362   def __repr__(self):
 363     return 'CVSFile<%x>(%r)' % (self.id, str(self),)
 364
 365